diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,86666 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0237002109442859,
+  "eval_steps": 500,
+  "global_step": 123750,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 8.272324936923522e-05,
+      "grad_norm": 80462.0859375,
+      "learning_rate": 8.080808080808081e-09,
+      "loss": 6007.0633,
+      "step": 10
+    },
+    {
+      "epoch": 0.00016544649873847045,
+      "grad_norm": 360423.46875,
+      "learning_rate": 1.6161616161616162e-08,
+      "loss": 16793.1375,
+      "step": 20
+    },
+    {
+      "epoch": 0.0002481697481077057,
+      "grad_norm": 260210.0,
+      "learning_rate": 2.4242424242424243e-08,
+      "loss": 8807.757,
+      "step": 30
+    },
+    {
+      "epoch": 0.0003308929974769409,
+      "grad_norm": 333457.03125,
+      "learning_rate": 3.2323232323232324e-08,
+      "loss": 6780.607,
+      "step": 40
+    },
+    {
+      "epoch": 0.0004136162468461761,
+      "grad_norm": 240292.671875,
+      "learning_rate": 4.040404040404041e-08,
+      "loss": 10353.4664,
+      "step": 50
+    },
+    {
+      "epoch": 0.0004963394962154114,
+      "grad_norm": 205636.09375,
+      "learning_rate": 4.8484848484848486e-08,
+      "loss": 9239.3687,
+      "step": 60
+    },
+    {
+      "epoch": 0.0005790627455846465,
+      "grad_norm": 91692.9375,
+      "learning_rate": 5.656565656565657e-08,
+      "loss": 5793.2141,
+      "step": 70
+    },
+    {
+      "epoch": 0.0006617859949538818,
+      "grad_norm": 128958.5859375,
+      "learning_rate": 6.464646464646465e-08,
+      "loss": 7645.5141,
+      "step": 80
+    },
+    {
+      "epoch": 0.000744509244323117,
+      "grad_norm": 198792.46875,
+      "learning_rate": 7.272727272727274e-08,
+      "loss": 7540.1,
+      "step": 90
+    },
+    {
+      "epoch": 0.0008272324936923522,
+      "grad_norm": 143177.3125,
+      "learning_rate": 8.080808080808082e-08,
+      "loss": 5863.3023,
+      "step": 100
+    },
+    {
+      "epoch": 0.0009099557430615875,
+      "grad_norm": 69092.4453125,
+      "learning_rate": 8.88888888888889e-08,
+      "loss": 9028.2383,
+      "step": 110
+    },
+    {
+      "epoch": 0.0009926789924308227,
+      "grad_norm": 135931.546875,
+      "learning_rate": 9.696969696969697e-08,
+      "loss": 7231.0602,
+      "step": 120
+    },
+    {
+      "epoch": 0.0010754022418000579,
+      "grad_norm": 142479.5625,
+      "learning_rate": 1.0505050505050506e-07,
+      "loss": 4912.673,
+      "step": 130
+    },
+    {
+      "epoch": 0.001158125491169293,
+      "grad_norm": 310996.5625,
+      "learning_rate": 1.1313131313131314e-07,
+      "loss": 4637.1719,
+      "step": 140
+    },
+    {
+      "epoch": 0.0012408487405385284,
+      "grad_norm": 195854.375,
+      "learning_rate": 1.2121212121212122e-07,
+      "loss": 5191.2895,
+      "step": 150
+    },
+    {
+      "epoch": 0.0013235719899077636,
+      "grad_norm": 98090.03125,
+      "learning_rate": 1.292929292929293e-07,
+      "loss": 6609.8664,
+      "step": 160
+    },
+    {
+      "epoch": 0.0014062952392769987,
+      "grad_norm": 69154.9765625,
+      "learning_rate": 1.3737373737373738e-07,
+      "loss": 6437.2246,
+      "step": 170
+    },
+    {
+      "epoch": 0.001489018488646234,
+      "grad_norm": 161934.78125,
+      "learning_rate": 1.4545454545454548e-07,
+      "loss": 7987.9906,
+      "step": 180
+    },
+    {
+      "epoch": 0.0015717417380154692,
+      "grad_norm": 72492.5390625,
+      "learning_rate": 1.5353535353535356e-07,
+      "loss": 5339.8434,
+      "step": 190
+    },
+    {
+      "epoch": 0.0016544649873847044,
+      "grad_norm": 28318.041015625,
+      "learning_rate": 1.6161616161616163e-07,
+      "loss": 3778.6113,
+      "step": 200
+    },
+    {
+      "epoch": 0.0017371882367539398,
+      "grad_norm": 78588.4375,
+      "learning_rate": 1.6969696969696974e-07,
+      "loss": 5780.0262,
+      "step": 210
+    },
+    {
+      "epoch": 0.001819911486123175,
+      "grad_norm": 69777.796875,
+      "learning_rate": 1.777777777777778e-07,
+      "loss": 3939.0852,
+      "step": 220
+    },
+    {
+      "epoch": 0.00190263473549241,
+      "grad_norm": 26142.01953125,
+      "learning_rate": 1.858585858585859e-07,
+      "loss": 3253.984,
+      "step": 230
+    },
+    {
+      "epoch": 0.0019853579848616455,
+      "grad_norm": 77199.8828125,
+      "learning_rate": 1.9393939393939395e-07,
+      "loss": 3297.4723,
+      "step": 240
+    },
+    {
+      "epoch": 0.0020680812342308806,
+      "grad_norm": 79269.9921875,
+      "learning_rate": 2.0202020202020205e-07,
+      "loss": 3044.4252,
+      "step": 250
+    },
+    {
+      "epoch": 0.0021508044836001158,
+      "grad_norm": 39357.45703125,
+      "learning_rate": 2.1010101010101013e-07,
+      "loss": 3016.1178,
+      "step": 260
+    },
+    {
+      "epoch": 0.002233527732969351,
+      "grad_norm": 37892.91015625,
+      "learning_rate": 2.181818181818182e-07,
+      "loss": 2370.4621,
+      "step": 270
+    },
+    {
+      "epoch": 0.002316250982338586,
+      "grad_norm": 40799.3984375,
+      "learning_rate": 2.2626262626262628e-07,
+      "loss": 2173.1178,
+      "step": 280
+    },
+    {
+      "epoch": 0.0023989742317078217,
+      "grad_norm": 16129.89453125,
+      "learning_rate": 2.343434343434344e-07,
+      "loss": 2872.4018,
+      "step": 290
+    },
+    {
+      "epoch": 0.002481697481077057,
+      "grad_norm": 12510.9287109375,
+      "learning_rate": 2.4242424242424244e-07,
+      "loss": 1385.131,
+      "step": 300
+    },
+    {
+      "epoch": 0.002564420730446292,
+      "grad_norm": 21687.78515625,
+      "learning_rate": 2.505050505050505e-07,
+      "loss": 1238.8357,
+      "step": 310
+    },
+    {
+      "epoch": 0.002647143979815527,
+      "grad_norm": 26216.89453125,
+      "learning_rate": 2.585858585858586e-07,
+      "loss": 1459.9193,
+      "step": 320
+    },
+    {
+      "epoch": 0.0027298672291847623,
+      "grad_norm": 54759.859375,
+      "learning_rate": 2.666666666666667e-07,
+      "loss": 1641.5232,
+      "step": 330
+    },
+    {
+      "epoch": 0.0028125904785539974,
+      "grad_norm": 11365.4833984375,
+      "learning_rate": 2.7474747474747475e-07,
+      "loss": 1190.3224,
+      "step": 340
+    },
+    {
+      "epoch": 0.002895313727923233,
+      "grad_norm": 16354.1533203125,
+      "learning_rate": 2.828282828282829e-07,
+      "loss": 885.4668,
+      "step": 350
+    },
+    {
+      "epoch": 0.002978036977292468,
+      "grad_norm": 11645.587890625,
+      "learning_rate": 2.9090909090909096e-07,
+      "loss": 1400.4568,
+      "step": 360
+    },
+    {
+      "epoch": 0.0030607602266617033,
+      "grad_norm": 8614.1328125,
+      "learning_rate": 2.9898989898989904e-07,
+      "loss": 814.0722,
+      "step": 370
+    },
+    {
+      "epoch": 0.0031434834760309385,
+      "grad_norm": 11007.8818359375,
+      "learning_rate": 3.070707070707071e-07,
+      "loss": 970.882,
+      "step": 380
+    },
+    {
+      "epoch": 0.0032262067254001736,
+      "grad_norm": 6238.4267578125,
+      "learning_rate": 3.151515151515152e-07,
+      "loss": 946.8882,
+      "step": 390
+    },
+    {
+      "epoch": 0.003308929974769409,
+      "grad_norm": 5815.009765625,
+      "learning_rate": 3.2323232323232327e-07,
+      "loss": 792.6276,
+      "step": 400
+    },
+    {
+      "epoch": 0.003391653224138644,
+      "grad_norm": 34222.5390625,
+      "learning_rate": 3.3131313131313135e-07,
+      "loss": 787.8205,
+      "step": 410
+    },
+    {
+      "epoch": 0.0034743764735078795,
+      "grad_norm": 4778.3984375,
+      "learning_rate": 3.393939393939395e-07,
+      "loss": 653.9669,
+      "step": 420
+    },
+    {
+      "epoch": 0.0035570997228771147,
+      "grad_norm": 2678.25244140625,
+      "learning_rate": 3.474747474747475e-07,
+      "loss": 793.3054,
+      "step": 430
+    },
+    {
+      "epoch": 0.00363982297224635,
+      "grad_norm": 1457.694580078125,
+      "learning_rate": 3.555555555555556e-07,
+      "loss": 630.5825,
+      "step": 440
+    },
+    {
+      "epoch": 0.003722546221615585,
+      "grad_norm": 5033.21923828125,
+      "learning_rate": 3.6363636363636366e-07,
+      "loss": 759.3085,
+      "step": 450
+    },
+    {
+      "epoch": 0.00380526947098482,
+      "grad_norm": 3704.6376953125,
+      "learning_rate": 3.717171717171718e-07,
+      "loss": 644.2665,
+      "step": 460
+    },
+    {
+      "epoch": 0.0038879927203540553,
+      "grad_norm": 5654.28662109375,
+      "learning_rate": 3.7979797979797987e-07,
+      "loss": 710.7244,
+      "step": 470
+    },
+    {
+      "epoch": 0.003970715969723291,
+      "grad_norm": 1483.4691162109375,
+      "learning_rate": 3.878787878787879e-07,
+      "loss": 694.1961,
+      "step": 480
+    },
+    {
+      "epoch": 0.004053439219092526,
+      "grad_norm": 2019.6363525390625,
+      "learning_rate": 3.9595959595959597e-07,
+      "loss": 573.5805,
+      "step": 490
+    },
+    {
+      "epoch": 0.004136162468461761,
+      "grad_norm": 1998.4979248046875,
+      "learning_rate": 4.040404040404041e-07,
+      "loss": 624.7649,
+      "step": 500
+    },
+    {
+      "epoch": 0.004218885717830996,
+      "grad_norm": 2387.564453125,
+      "learning_rate": 4.121212121212122e-07,
+      "loss": 626.6972,
+      "step": 510
+    },
+    {
+      "epoch": 0.0043016089672002315,
+      "grad_norm": 3881.398681640625,
+      "learning_rate": 4.2020202020202026e-07,
+      "loss": 759.5687,
+      "step": 520
+    },
+    {
+      "epoch": 0.004384332216569467,
+      "grad_norm": 1924.41845703125,
+      "learning_rate": 4.282828282828283e-07,
+      "loss": 700.7449,
+      "step": 530
+    },
+    {
+      "epoch": 0.004467055465938702,
+      "grad_norm": 2266.10693359375,
+      "learning_rate": 4.363636363636364e-07,
+      "loss": 640.943,
+      "step": 540
+    },
+    {
+      "epoch": 0.004549778715307937,
+      "grad_norm": 7472.12548828125,
+      "learning_rate": 4.444444444444445e-07,
+      "loss": 806.1628,
+      "step": 550
+    },
+    {
+      "epoch": 0.004632501964677172,
+      "grad_norm": 1539.118408203125,
+      "learning_rate": 4.5252525252525257e-07,
+      "loss": 556.9574,
+      "step": 560
+    },
+    {
+      "epoch": 0.004715225214046408,
+      "grad_norm": 1867.2452392578125,
+      "learning_rate": 4.6060606060606064e-07,
+      "loss": 453.8387,
+      "step": 570
+    },
+    {
+      "epoch": 0.004797948463415643,
+      "grad_norm": 6622.47119140625,
+      "learning_rate": 4.686868686868688e-07,
+      "loss": 585.511,
+      "step": 580
+    },
+    {
+      "epoch": 0.0048806717127848785,
+      "grad_norm": 1117.4276123046875,
+      "learning_rate": 4.767676767676768e-07,
+      "loss": 591.8928,
+      "step": 590
+    },
+    {
+      "epoch": 0.004963394962154114,
+      "grad_norm": 1792.2279052734375,
+      "learning_rate": 4.848484848484849e-07,
+      "loss": 659.099,
+      "step": 600
+    },
+    {
+      "epoch": 0.005046118211523349,
+      "grad_norm": 952.1702880859375,
+      "learning_rate": 4.929292929292929e-07,
+      "loss": 558.1813,
+      "step": 610
+    },
+    {
+      "epoch": 0.005128841460892584,
+      "grad_norm": 1103.0098876953125,
+      "learning_rate": 5.01010101010101e-07,
+      "loss": 488.6359,
+      "step": 620
+    },
+    {
+      "epoch": 0.005211564710261819,
+      "grad_norm": 2055.734130859375,
+      "learning_rate": 5.090909090909092e-07,
+      "loss": 611.8516,
+      "step": 630
+    },
+    {
+      "epoch": 0.005294287959631054,
+      "grad_norm": 1028.85986328125,
+      "learning_rate": 5.171717171717172e-07,
+      "loss": 568.1,
+      "step": 640
+    },
+    {
+      "epoch": 0.005377011209000289,
+      "grad_norm": 1175.14892578125,
+      "learning_rate": 5.252525252525253e-07,
+      "loss": 407.9692,
+      "step": 650
+    },
+    {
+      "epoch": 0.005459734458369525,
+      "grad_norm": 1459.8382568359375,
+      "learning_rate": 5.333333333333335e-07,
+      "loss": 496.3047,
+      "step": 660
+    },
+    {
+      "epoch": 0.00554245770773876,
+      "grad_norm": 3791.671630859375,
+      "learning_rate": 5.414141414141415e-07,
+      "loss": 492.2279,
+      "step": 670
+    },
+    {
+      "epoch": 0.005625180957107995,
+      "grad_norm": 1465.2242431640625,
+      "learning_rate": 5.494949494949495e-07,
+      "loss": 518.7601,
+      "step": 680
+    },
+    {
+      "epoch": 0.00570790420647723,
+      "grad_norm": 1057.6175537109375,
+      "learning_rate": 5.575757575757576e-07,
+      "loss": 496.9481,
+      "step": 690
+    },
+    {
+      "epoch": 0.005790627455846466,
+      "grad_norm": 1338.5599365234375,
+      "learning_rate": 5.656565656565658e-07,
+      "loss": 493.3083,
+      "step": 700
+    },
+    {
+      "epoch": 0.005873350705215701,
+      "grad_norm": 1229.9342041015625,
+      "learning_rate": 5.737373737373738e-07,
+      "loss": 507.9572,
+      "step": 710
+    },
+    {
+      "epoch": 0.005956073954584936,
+      "grad_norm": 2031.885498046875,
+      "learning_rate": 5.818181818181819e-07,
+      "loss": 545.9699,
+      "step": 720
+    },
+    {
+      "epoch": 0.0060387972039541715,
+      "grad_norm": 1317.18408203125,
+      "learning_rate": 5.898989898989899e-07,
+      "loss": 515.3469,
+      "step": 730
+    },
+    {
+      "epoch": 0.006121520453323407,
+      "grad_norm": 1839.4345703125,
+      "learning_rate": 5.979797979797981e-07,
+      "loss": 377.9613,
+      "step": 740
+    },
+    {
+      "epoch": 0.006204243702692642,
+      "grad_norm": 1210.0321044921875,
+      "learning_rate": 6.060606060606061e-07,
+      "loss": 465.0116,
+      "step": 750
+    },
+    {
+      "epoch": 0.006286966952061877,
+      "grad_norm": 2228.900390625,
+      "learning_rate": 6.141414141414142e-07,
+      "loss": 336.6599,
+      "step": 760
+    },
+    {
+      "epoch": 0.006369690201431112,
+      "grad_norm": 1690.2906494140625,
+      "learning_rate": 6.222222222222223e-07,
+      "loss": 402.5558,
+      "step": 770
+    },
+    {
+      "epoch": 0.006452413450800347,
+      "grad_norm": 1048.852783203125,
+      "learning_rate": 6.303030303030304e-07,
+      "loss": 460.7069,
+      "step": 780
+    },
+    {
+      "epoch": 0.0065351367001695825,
+      "grad_norm": 1623.499755859375,
+      "learning_rate": 6.383838383838384e-07,
+      "loss": 322.7708,
+      "step": 790
+    },
+    {
+      "epoch": 0.006617859949538818,
+      "grad_norm": 4279.55908203125,
+      "learning_rate": 6.464646464646465e-07,
+      "loss": 430.4449,
+      "step": 800
+    },
+    {
+      "epoch": 0.006700583198908053,
+      "grad_norm": 1473.4871826171875,
+      "learning_rate": 6.545454545454547e-07,
+      "loss": 474.9226,
+      "step": 810
+    },
+    {
+      "epoch": 0.006783306448277288,
+      "grad_norm": 1544.477783203125,
+      "learning_rate": 6.626262626262627e-07,
+      "loss": 459.8634,
+      "step": 820
+    },
+    {
+      "epoch": 0.006866029697646524,
+      "grad_norm": 991.1500854492188,
+      "learning_rate": 6.707070707070708e-07,
+      "loss": 339.08,
+      "step": 830
+    },
+    {
+      "epoch": 0.006948752947015759,
+      "grad_norm": 1124.6473388671875,
+      "learning_rate": 6.78787878787879e-07,
+      "loss": 476.3061,
+      "step": 840
+    },
+    {
+      "epoch": 0.007031476196384994,
+      "grad_norm": 1139.6890869140625,
+      "learning_rate": 6.868686868686869e-07,
+      "loss": 504.3425,
+      "step": 850
+    },
+    {
+      "epoch": 0.007114199445754229,
+      "grad_norm": 1275.02294921875,
+      "learning_rate": 6.94949494949495e-07,
+      "loss": 431.0188,
+      "step": 860
+    },
+    {
+      "epoch": 0.0071969226951234646,
+      "grad_norm": 1159.5965576171875,
+      "learning_rate": 7.03030303030303e-07,
+      "loss": 455.7138,
+      "step": 870
+    },
+    {
+      "epoch": 0.0072796459444927,
+      "grad_norm": 1401.516357421875,
+      "learning_rate": 7.111111111111112e-07,
+      "loss": 529.1278,
+      "step": 880
+    },
+    {
+      "epoch": 0.007362369193861935,
+      "grad_norm": 1474.188720703125,
+      "learning_rate": 7.191919191919193e-07,
+      "loss": 326.5665,
+      "step": 890
+    },
+    {
+      "epoch": 0.00744509244323117,
+      "grad_norm": 917.8528442382812,
+      "learning_rate": 7.272727272727273e-07,
+      "loss": 436.4797,
+      "step": 900
+    },
+    {
+      "epoch": 0.007527815692600405,
+      "grad_norm": 941.6436157226562,
+      "learning_rate": 7.353535353535354e-07,
+      "loss": 357.6453,
+      "step": 910
+    },
+    {
+      "epoch": 0.00761053894196964,
+      "grad_norm": 934.293701171875,
+      "learning_rate": 7.434343434343436e-07,
+      "loss": 433.9422,
+      "step": 920
+    },
+    {
+      "epoch": 0.0076932621913388755,
+      "grad_norm": 1530.59814453125,
+      "learning_rate": 7.515151515151516e-07,
+      "loss": 506.7857,
+      "step": 930
+    },
+    {
+      "epoch": 0.007775985440708111,
+      "grad_norm": 3313.063720703125,
+      "learning_rate": 7.595959595959597e-07,
+      "loss": 386.7442,
+      "step": 940
+    },
+    {
+      "epoch": 0.007858708690077347,
+      "grad_norm": 9505.775390625,
+      "learning_rate": 7.676767676767677e-07,
+      "loss": 393.8395,
+      "step": 950
+    },
+    {
+      "epoch": 0.007941431939446582,
+      "grad_norm": 9777.1015625,
+      "learning_rate": 7.757575757575758e-07,
+      "loss": 444.2824,
+      "step": 960
+    },
+    {
+      "epoch": 0.008024155188815817,
+      "grad_norm": 952.7528686523438,
+      "learning_rate": 7.838383838383839e-07,
+      "loss": 463.8062,
+      "step": 970
+    },
+    {
+      "epoch": 0.008106878438185052,
+      "grad_norm": 1653.4957275390625,
+      "learning_rate": 7.919191919191919e-07,
+      "loss": 422.011,
+      "step": 980
+    },
+    {
+      "epoch": 0.008189601687554287,
+      "grad_norm": 1222.734619140625,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 370.3142,
+      "step": 990
+    },
+    {
+      "epoch": 0.008272324936923522,
+      "grad_norm": 1304.0595703125,
+      "learning_rate": 8.080808080808082e-07,
+      "loss": 423.2224,
+      "step": 1000
+    },
+    {
+      "epoch": 0.008355048186292758,
+      "grad_norm": 1451.695556640625,
+      "learning_rate": 8.161616161616162e-07,
+      "loss": 333.3123,
+      "step": 1010
+    },
+    {
+      "epoch": 0.008437771435661993,
+      "grad_norm": 1295.0885009765625,
+      "learning_rate": 8.242424242424244e-07,
+      "loss": 395.7991,
+      "step": 1020
+    },
+    {
+      "epoch": 0.008520494685031228,
+      "grad_norm": 24490.919921875,
+      "learning_rate": 8.323232323232324e-07,
+      "loss": 484.7744,
+      "step": 1030
+    },
+    {
+      "epoch": 0.008603217934400463,
+      "grad_norm": 1630.39306640625,
+      "learning_rate": 8.404040404040405e-07,
+      "loss": 507.0378,
+      "step": 1040
+    },
+    {
+      "epoch": 0.008685941183769698,
+      "grad_norm": 1444.76708984375,
+      "learning_rate": 8.484848484848486e-07,
+      "loss": 424.0987,
+      "step": 1050
+    },
+    {
+      "epoch": 0.008768664433138933,
+      "grad_norm": 9745.6103515625,
+      "learning_rate": 8.565656565656566e-07,
+      "loss": 356.5501,
+      "step": 1060
+    },
+    {
+      "epoch": 0.008851387682508169,
+      "grad_norm": 1603.8682861328125,
+      "learning_rate": 8.646464646464647e-07,
+      "loss": 363.1298,
+      "step": 1070
+    },
+    {
+      "epoch": 0.008934110931877404,
+      "grad_norm": 1848.8529052734375,
+      "learning_rate": 8.727272727272728e-07,
+      "loss": 453.0991,
+      "step": 1080
+    },
+    {
+      "epoch": 0.009016834181246639,
+      "grad_norm": 1721.5032958984375,
+      "learning_rate": 8.808080808080808e-07,
+      "loss": 352.8759,
+      "step": 1090
+    },
+    {
+      "epoch": 0.009099557430615874,
+      "grad_norm": 1215.8017578125,
+      "learning_rate": 8.88888888888889e-07,
+      "loss": 358.2751,
+      "step": 1100
+    },
+    {
+      "epoch": 0.00918228067998511,
+      "grad_norm": 3627.594482421875,
+      "learning_rate": 8.96969696969697e-07,
+      "loss": 373.2328,
+      "step": 1110
+    },
+    {
+      "epoch": 0.009265003929354344,
+      "grad_norm": 1093.44677734375,
+      "learning_rate": 9.050505050505051e-07,
+      "loss": 305.7944,
+      "step": 1120
+    },
+    {
+      "epoch": 0.00934772717872358,
+      "grad_norm": 5584.5224609375,
+      "learning_rate": 9.131313131313133e-07,
+      "loss": 371.0138,
+      "step": 1130
+    },
+    {
+      "epoch": 0.009430450428092816,
+      "grad_norm": 1449.30419921875,
+      "learning_rate": 9.212121212121213e-07,
+      "loss": 374.5949,
+      "step": 1140
+    },
+    {
+      "epoch": 0.009513173677462052,
+      "grad_norm": 1450.5250244140625,
+      "learning_rate": 9.292929292929294e-07,
+      "loss": 370.9519,
+      "step": 1150
+    },
+    {
+      "epoch": 0.009595896926831287,
+      "grad_norm": 1153.1729736328125,
+      "learning_rate": 9.373737373737376e-07,
+      "loss": 311.7418,
+      "step": 1160
+    },
+    {
+      "epoch": 0.009678620176200522,
+      "grad_norm": 1182.4305419921875,
+      "learning_rate": 9.454545454545455e-07,
+      "loss": 340.1112,
+      "step": 1170
+    },
+    {
+      "epoch": 0.009761343425569757,
+      "grad_norm": 2788.3564453125,
+      "learning_rate": 9.535353535353536e-07,
+      "loss": 402.735,
+      "step": 1180
+    },
+    {
+      "epoch": 0.009844066674938992,
+      "grad_norm": 1240.61669921875,
+      "learning_rate": 9.616161616161617e-07,
+      "loss": 430.115,
+      "step": 1190
+    },
+    {
+      "epoch": 0.009926789924308227,
+      "grad_norm": 7811.59228515625,
+      "learning_rate": 9.696969696969698e-07,
+      "loss": 294.6839,
+      "step": 1200
+    },
+    {
+      "epoch": 0.010009513173677462,
+      "grad_norm": 1490.346923828125,
+      "learning_rate": 9.77777777777778e-07,
+      "loss": 314.4158,
+      "step": 1210
+    },
+    {
+      "epoch": 0.010092236423046698,
+      "grad_norm": 1153.1387939453125,
+      "learning_rate": 9.858585858585858e-07,
+      "loss": 422.2901,
+      "step": 1220
+    },
+    {
+      "epoch": 0.010174959672415933,
+      "grad_norm": 1359.1773681640625,
+      "learning_rate": 9.93939393939394e-07,
+      "loss": 396.2289,
+      "step": 1230
+    },
+    {
+      "epoch": 0.010257682921785168,
+      "grad_norm": 1238.34814453125,
+      "learning_rate": 1.002020202020202e-06,
+      "loss": 432.7112,
+      "step": 1240
+    },
+    {
+      "epoch": 0.010340406171154403,
+      "grad_norm": 1150.58154296875,
+      "learning_rate": 1.01010101010101e-06,
+      "loss": 533.3193,
+      "step": 1250
+    },
+    {
+      "epoch": 0.010423129420523638,
+      "grad_norm": 2107.6708984375,
+      "learning_rate": 1.0181818181818183e-06,
+      "loss": 386.3309,
+      "step": 1260
+    },
+    {
+      "epoch": 0.010505852669892873,
+      "grad_norm": 1610.2342529296875,
+      "learning_rate": 1.0262626262626264e-06,
+      "loss": 378.974,
+      "step": 1270
+    },
+    {
+      "epoch": 0.010588575919262109,
+      "grad_norm": 10111.8642578125,
+      "learning_rate": 1.0343434343434344e-06,
+      "loss": 440.8467,
+      "step": 1280
+    },
+    {
+      "epoch": 0.010671299168631344,
+      "grad_norm": 3038.983642578125,
+      "learning_rate": 1.0424242424242426e-06,
+      "loss": 531.3744,
+      "step": 1290
+    },
+    {
+      "epoch": 0.010754022418000579,
+      "grad_norm": 2078.825927734375,
+      "learning_rate": 1.0505050505050506e-06,
+      "loss": 358.3208,
+      "step": 1300
+    },
+    {
+      "epoch": 0.010836745667369814,
+      "grad_norm": 1494.41943359375,
+      "learning_rate": 1.0585858585858587e-06,
+      "loss": 368.8838,
+      "step": 1310
+    },
+    {
+      "epoch": 0.01091946891673905,
+      "grad_norm": 913.7642211914062,
+      "learning_rate": 1.066666666666667e-06,
+      "loss": 343.7515,
+      "step": 1320
+    },
+    {
+      "epoch": 0.011002192166108284,
+      "grad_norm": 1475.224853515625,
+      "learning_rate": 1.0747474747474747e-06,
+      "loss": 311.1521,
+      "step": 1330
+    },
+    {
+      "epoch": 0.01108491541547752,
+      "grad_norm": 6028.0986328125,
+      "learning_rate": 1.082828282828283e-06,
+      "loss": 346.5072,
+      "step": 1340
+    },
+    {
+      "epoch": 0.011167638664846755,
+      "grad_norm": 1047.4739990234375,
+      "learning_rate": 1.090909090909091e-06,
+      "loss": 371.3439,
+      "step": 1350
+    },
+    {
+      "epoch": 0.01125036191421599,
+      "grad_norm": 2157.152099609375,
+      "learning_rate": 1.098989898989899e-06,
+      "loss": 396.0025,
+      "step": 1360
+    },
+    {
+      "epoch": 0.011333085163585225,
+      "grad_norm": 1551.3135986328125,
+      "learning_rate": 1.1070707070707072e-06,
+      "loss": 423.25,
+      "step": 1370
+    },
+    {
+      "epoch": 0.01141580841295446,
+      "grad_norm": 1080.8287353515625,
+      "learning_rate": 1.1151515151515153e-06,
+      "loss": 255.2609,
+      "step": 1380
+    },
+    {
+      "epoch": 0.011498531662323695,
+      "grad_norm": 835.9887084960938,
+      "learning_rate": 1.1232323232323233e-06,
+      "loss": 284.983,
+      "step": 1390
+    },
+    {
+      "epoch": 0.011581254911692932,
+      "grad_norm": 774.7077026367188,
+      "learning_rate": 1.1313131313131315e-06,
+      "loss": 274.2369,
+      "step": 1400
+    },
+    {
+      "epoch": 0.011663978161062167,
+      "grad_norm": 1325.56982421875,
+      "learning_rate": 1.1393939393939395e-06,
+      "loss": 797.6222,
+      "step": 1410
+    },
+    {
+      "epoch": 0.011746701410431402,
+      "grad_norm": 1981.0302734375,
+      "learning_rate": 1.1474747474747476e-06,
+      "loss": 366.0973,
+      "step": 1420
+    },
+    {
+      "epoch": 0.011829424659800638,
+      "grad_norm": 2644.739501953125,
+      "learning_rate": 1.1555555555555556e-06,
+      "loss": 327.9804,
+      "step": 1430
+    },
+    {
+      "epoch": 0.011912147909169873,
+      "grad_norm": 1665.543212890625,
+      "learning_rate": 1.1636363636363638e-06,
+      "loss": 425.4298,
+      "step": 1440
+    },
+    {
+      "epoch": 0.011994871158539108,
+      "grad_norm": 1862.1712646484375,
+      "learning_rate": 1.1717171717171719e-06,
+      "loss": 324.925,
+      "step": 1450
+    },
+    {
+      "epoch": 0.012077594407908343,
+      "grad_norm": 1164.2347412109375,
+      "learning_rate": 1.1797979797979799e-06,
+      "loss": 298.3789,
+      "step": 1460
+    },
+    {
+      "epoch": 0.012160317657277578,
+      "grad_norm": 1282.484375,
+      "learning_rate": 1.187878787878788e-06,
+      "loss": 399.5682,
+      "step": 1470
+    },
+    {
+      "epoch": 0.012243040906646813,
+      "grad_norm": 1316.5758056640625,
+      "learning_rate": 1.1959595959595961e-06,
+      "loss": 338.0538,
+      "step": 1480
+    },
+    {
+      "epoch": 0.012325764156016049,
+      "grad_norm": 875.5802001953125,
+      "learning_rate": 1.2040404040404042e-06,
+      "loss": 308.0409,
+      "step": 1490
+    },
+    {
+      "epoch": 0.012408487405385284,
+      "grad_norm": 1360.3880615234375,
+      "learning_rate": 1.2121212121212122e-06,
+      "loss": 292.9588,
+      "step": 1500
+    },
+    {
+      "epoch": 0.012491210654754519,
+      "grad_norm": 1129.11083984375,
+      "learning_rate": 1.2202020202020202e-06,
+      "loss": 322.4768,
+      "step": 1510
+    },
+    {
+      "epoch": 0.012573933904123754,
+      "grad_norm": 2066.356689453125,
+      "learning_rate": 1.2282828282828285e-06,
+      "loss": 271.5606,
+      "step": 1520
+    },
+    {
+      "epoch": 0.012656657153492989,
+      "grad_norm": 1669.0103759765625,
+      "learning_rate": 1.2363636363636365e-06,
+      "loss": 283.8805,
+      "step": 1530
+    },
+    {
+      "epoch": 0.012739380402862224,
+      "grad_norm": 2613.5205078125,
+      "learning_rate": 1.2444444444444445e-06,
+      "loss": 376.5511,
+      "step": 1540
+    },
+    {
+      "epoch": 0.01282210365223146,
+      "grad_norm": 1188.38427734375,
+      "learning_rate": 1.2525252525252527e-06,
+      "loss": 429.8225,
+      "step": 1550
+    },
+    {
+      "epoch": 0.012904826901600695,
+      "grad_norm": 1084.84814453125,
+      "learning_rate": 1.2606060606060608e-06,
+      "loss": 289.73,
+      "step": 1560
+    },
+    {
+      "epoch": 0.01298755015096993,
+      "grad_norm": 2206.3232421875,
+      "learning_rate": 1.268686868686869e-06,
+      "loss": 316.7855,
+      "step": 1570
+    },
+    {
+      "epoch": 0.013070273400339165,
+      "grad_norm": 2453.128662109375,
+      "learning_rate": 1.2767676767676768e-06,
+      "loss": 304.9484,
+      "step": 1580
+    },
+    {
+      "epoch": 0.0131529966497084,
+      "grad_norm": 976.6162719726562,
+      "learning_rate": 1.2848484848484848e-06,
+      "loss": 303.9162,
+      "step": 1590
+    },
+    {
+      "epoch": 0.013235719899077635,
+      "grad_norm": 1493.1630859375,
+      "learning_rate": 1.292929292929293e-06,
+      "loss": 356.7861,
+      "step": 1600
+    },
+    {
+      "epoch": 0.01331844314844687,
+      "grad_norm": 1568.7215576171875,
+      "learning_rate": 1.301010101010101e-06,
+      "loss": 556.4589,
+      "step": 1610
+    },
+    {
+      "epoch": 0.013401166397816106,
+      "grad_norm": 2514.0361328125,
+      "learning_rate": 1.3090909090909093e-06,
+      "loss": 414.0774,
+      "step": 1620
+    },
+    {
+      "epoch": 0.01348388964718534,
+      "grad_norm": 1089.7327880859375,
+      "learning_rate": 1.3171717171717172e-06,
+      "loss": 330.1746,
+      "step": 1630
+    },
+    {
+      "epoch": 0.013566612896554576,
+      "grad_norm": 1226.1904296875,
+      "learning_rate": 1.3252525252525254e-06,
+      "loss": 408.1962,
+      "step": 1640
+    },
+    {
+      "epoch": 0.013649336145923813,
+      "grad_norm": 2240.143310546875,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 344.259,
+      "step": 1650
+    },
+    {
+      "epoch": 0.013732059395293048,
+      "grad_norm": 1064.2305908203125,
+      "learning_rate": 1.3414141414141417e-06,
+      "loss": 348.617,
+      "step": 1660
+    },
+    {
+      "epoch": 0.013814782644662283,
+      "grad_norm": 1232.1429443359375,
+      "learning_rate": 1.3494949494949497e-06,
+      "loss": 406.3739,
+      "step": 1670
+    },
+    {
+      "epoch": 0.013897505894031518,
+      "grad_norm": 1123.156982421875,
+      "learning_rate": 1.357575757575758e-06,
+      "loss": 396.9652,
+      "step": 1680
+    },
+    {
+      "epoch": 0.013980229143400753,
+      "grad_norm": 1280.202880859375,
+      "learning_rate": 1.3656565656565657e-06,
+      "loss": 391.5166,
+      "step": 1690
+    },
+    {
+      "epoch": 0.014062952392769988,
+      "grad_norm": 1420.6954345703125,
+      "learning_rate": 1.3737373737373738e-06,
+      "loss": 437.0295,
+      "step": 1700
+    },
+    {
+      "epoch": 0.014145675642139224,
+      "grad_norm": 2150.417236328125,
+      "learning_rate": 1.381818181818182e-06,
+      "loss": 346.5058,
+      "step": 1710
+    },
+    {
+      "epoch": 0.014228398891508459,
+      "grad_norm": 1029.7105712890625,
+      "learning_rate": 1.38989898989899e-06,
+      "loss": 326.7657,
+      "step": 1720
+    },
+    {
+      "epoch": 0.014311122140877694,
+      "grad_norm": 1207.562744140625,
+      "learning_rate": 1.3979797979797982e-06,
+      "loss": 279.2056,
+      "step": 1730
+    },
+    {
+      "epoch": 0.014393845390246929,
+      "grad_norm": 919.50244140625,
+      "learning_rate": 1.406060606060606e-06,
+      "loss": 264.9025,
+      "step": 1740
+    },
+    {
+      "epoch": 0.014476568639616164,
+      "grad_norm": 1641.6435546875,
+      "learning_rate": 1.4141414141414143e-06,
+      "loss": 239.4691,
+      "step": 1750
+    },
+    {
+      "epoch": 0.0145592918889854,
+      "grad_norm": 1205.822998046875,
+      "learning_rate": 1.4222222222222223e-06,
+      "loss": 333.4085,
+      "step": 1760
+    },
+    {
+      "epoch": 0.014642015138354635,
+      "grad_norm": 1508.9305419921875,
+      "learning_rate": 1.4303030303030306e-06,
+      "loss": 300.8089,
+      "step": 1770
+    },
+    {
+      "epoch": 0.01472473838772387,
+      "grad_norm": 1087.9306640625,
+      "learning_rate": 1.4383838383838386e-06,
+      "loss": 292.6564,
+      "step": 1780
+    },
+    {
+      "epoch": 0.014807461637093105,
+      "grad_norm": 2015.249755859375,
+      "learning_rate": 1.4464646464646464e-06,
+      "loss": 277.2897,
+      "step": 1790
+    },
+    {
+      "epoch": 0.01489018488646234,
+      "grad_norm": 4033.002685546875,
+      "learning_rate": 1.4545454545454546e-06,
+      "loss": 336.2396,
+      "step": 1800
+    },
+    {
+      "epoch": 0.014972908135831575,
+      "grad_norm": 1528.888916015625,
+      "learning_rate": 1.4626262626262627e-06,
+      "loss": 310.2283,
+      "step": 1810
+    },
+    {
+      "epoch": 0.01505563138520081,
+      "grad_norm": 1040.3310546875,
+      "learning_rate": 1.470707070707071e-06,
+      "loss": 296.3051,
+      "step": 1820
+    },
+    {
+      "epoch": 0.015138354634570046,
+      "grad_norm": 3019.808837890625,
+      "learning_rate": 1.478787878787879e-06,
+      "loss": 278.5342,
+      "step": 1830
+    },
+    {
+      "epoch": 0.01522107788393928,
+      "grad_norm": 1161.161376953125,
+      "learning_rate": 1.4868686868686872e-06,
+      "loss": 357.7166,
+      "step": 1840
+    },
+    {
+      "epoch": 0.015303801133308516,
+      "grad_norm": 2254.221435546875,
+      "learning_rate": 1.494949494949495e-06,
+      "loss": 325.2632,
+      "step": 1850
+    },
+    {
+      "epoch": 0.015386524382677751,
+      "grad_norm": 3298.8662109375,
+      "learning_rate": 1.5030303030303032e-06,
+      "loss": 340.6655,
+      "step": 1860
+    },
+    {
+      "epoch": 0.015469247632046986,
+      "grad_norm": 1297.3563232421875,
+      "learning_rate": 1.5111111111111112e-06,
+      "loss": 327.4712,
+      "step": 1870
+    },
+    {
+      "epoch": 0.015551970881416221,
+      "grad_norm": 1163.013671875,
+      "learning_rate": 1.5191919191919195e-06,
+      "loss": 302.4008,
+      "step": 1880
+    },
+    {
+      "epoch": 0.015634694130785456,
+      "grad_norm": 3574.491943359375,
+      "learning_rate": 1.5272727272727275e-06,
+      "loss": 304.6065,
+      "step": 1890
+    },
+    {
+      "epoch": 0.015717417380154693,
+      "grad_norm": 953.6083984375,
+      "learning_rate": 1.5353535353535353e-06,
+      "loss": 283.16,
+      "step": 1900
+    },
+    {
+      "epoch": 0.015800140629523927,
+      "grad_norm": 579.375732421875,
+      "learning_rate": 1.5434343434343435e-06,
+      "loss": 294.0983,
+      "step": 1910
+    },
+    {
+      "epoch": 0.015882863878893164,
+      "grad_norm": 904.5838012695312,
+      "learning_rate": 1.5515151515151516e-06,
+      "loss": 277.4686,
+      "step": 1920
+    },
+    {
+      "epoch": 0.015965587128262397,
+      "grad_norm": 878.8187255859375,
+      "learning_rate": 1.5595959595959598e-06,
+      "loss": 348.8653,
+      "step": 1930
+    },
+    {
+      "epoch": 0.016048310377631634,
+      "grad_norm": 1161.1363525390625,
+      "learning_rate": 1.5676767676767678e-06,
+      "loss": 374.7675,
+      "step": 1940
+    },
+    {
+      "epoch": 0.016131033627000867,
+      "grad_norm": 1667.3831787109375,
+      "learning_rate": 1.5757575757575759e-06,
+      "loss": 359.4971,
+      "step": 1950
+    },
+    {
+      "epoch": 0.016213756876370104,
+      "grad_norm": 2197.3330078125,
+      "learning_rate": 1.5838383838383839e-06,
+      "loss": 367.9104,
+      "step": 1960
+    },
+    {
+      "epoch": 0.016296480125739338,
+      "grad_norm": 1425.8123779296875,
+      "learning_rate": 1.5919191919191921e-06,
+      "loss": 367.0379,
+      "step": 1970
+    },
+    {
+      "epoch": 0.016379203375108575,
+      "grad_norm": 4604.79052734375,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 462.412,
+      "step": 1980
+    },
+    {
+      "epoch": 0.016461926624477808,
+      "grad_norm": 1806.8238525390625,
+      "learning_rate": 1.6080808080808084e-06,
+      "loss": 350.7875,
+      "step": 1990
+    },
+    {
+      "epoch": 0.016544649873847045,
+      "grad_norm": 1519.6781005859375,
+      "learning_rate": 1.6161616161616164e-06,
+      "loss": 272.1075,
+      "step": 2000
+    },
+    {
+      "epoch": 0.01662737312321628,
+      "grad_norm": 2001.111572265625,
+      "learning_rate": 1.6242424242424242e-06,
+      "loss": 264.7772,
+      "step": 2010
+    },
+    {
+      "epoch": 0.016710096372585515,
+      "grad_norm": 2383.3994140625,
+      "learning_rate": 1.6323232323232325e-06,
+      "loss": 293.2167,
+      "step": 2020
+    },
+    {
+      "epoch": 0.016792819621954752,
+      "grad_norm": 670.5611572265625,
+      "learning_rate": 1.6404040404040405e-06,
+      "loss": 379.9431,
+      "step": 2030
+    },
+    {
+      "epoch": 0.016875542871323986,
+      "grad_norm": 1432.4300537109375,
+      "learning_rate": 1.6484848484848487e-06,
+      "loss": 362.749,
+      "step": 2040
+    },
+    {
+      "epoch": 0.016958266120693222,
+      "grad_norm": 6461.77880859375,
+      "learning_rate": 1.6565656565656567e-06,
+      "loss": 407.5534,
+      "step": 2050
+    },
+    {
+      "epoch": 0.017040989370062456,
+      "grad_norm": 1288.83544921875,
+      "learning_rate": 1.6646464646464648e-06,
+      "loss": 315.7526,
+      "step": 2060
+    },
+    {
+      "epoch": 0.017123712619431693,
+      "grad_norm": 907.4447021484375,
+      "learning_rate": 1.6727272727272728e-06,
+      "loss": 292.4799,
+      "step": 2070
+    },
+    {
+      "epoch": 0.017206435868800926,
+      "grad_norm": 1921.0576171875,
+      "learning_rate": 1.680808080808081e-06,
+      "loss": 297.3491,
+      "step": 2080
+    },
+    {
+      "epoch": 0.017289159118170163,
+      "grad_norm": 1620.2481689453125,
+      "learning_rate": 1.688888888888889e-06,
+      "loss": 291.1318,
+      "step": 2090
+    },
+    {
+      "epoch": 0.017371882367539396,
+      "grad_norm": 892.2626342773438,
+      "learning_rate": 1.6969696969696973e-06,
+      "loss": 337.7454,
+      "step": 2100
+    },
+    {
+      "epoch": 0.017454605616908633,
+      "grad_norm": 1724.4583740234375,
+      "learning_rate": 1.705050505050505e-06,
+      "loss": 386.2352,
+      "step": 2110
+    },
+    {
+      "epoch": 0.017537328866277867,
+      "grad_norm": 1373.0406494140625,
+      "learning_rate": 1.7131313131313131e-06,
+      "loss": 312.9512,
+      "step": 2120
+    },
+    {
+      "epoch": 0.017620052115647104,
+      "grad_norm": 1802.06884765625,
+      "learning_rate": 1.7212121212121214e-06,
+      "loss": 392.6038,
+      "step": 2130
+    },
+    {
+      "epoch": 0.017702775365016337,
+      "grad_norm": 1823.2747802734375,
+      "learning_rate": 1.7292929292929294e-06,
+      "loss": 276.9635,
+      "step": 2140
+    },
+    {
+      "epoch": 0.017785498614385574,
+      "grad_norm": 2303.46484375,
+      "learning_rate": 1.7373737373737376e-06,
+      "loss": 363.1688,
+      "step": 2150
+    },
+    {
+      "epoch": 0.017868221863754807,
+      "grad_norm": 939.9644165039062,
+      "learning_rate": 1.7454545454545456e-06,
+      "loss": 371.3494,
+      "step": 2160
+    },
+    {
+      "epoch": 0.017950945113124044,
+      "grad_norm": 803.9099731445312,
+      "learning_rate": 1.7535353535353537e-06,
+      "loss": 325.9621,
+      "step": 2170
+    },
+    {
+      "epoch": 0.018033668362493278,
+      "grad_norm": 1147.0989990234375,
+      "learning_rate": 1.7616161616161617e-06,
+      "loss": 339.9603,
+      "step": 2180
+    },
+    {
+      "epoch": 0.018116391611862515,
+      "grad_norm": 1110.0518798828125,
+      "learning_rate": 1.76969696969697e-06,
+      "loss": 311.3122,
+      "step": 2190
+    },
+    {
+      "epoch": 0.018199114861231748,
+      "grad_norm": 1060.744140625,
+      "learning_rate": 1.777777777777778e-06,
+      "loss": 350.8291,
+      "step": 2200
+    },
+    {
+      "epoch": 0.018281838110600985,
+      "grad_norm": 2863.481201171875,
+      "learning_rate": 1.7858585858585862e-06,
+      "loss": 347.886,
+      "step": 2210
+    },
+    {
+      "epoch": 0.01836456135997022,
+      "grad_norm": 1049.102294921875,
+      "learning_rate": 1.793939393939394e-06,
+      "loss": 255.8531,
+      "step": 2220
+    },
+    {
+      "epoch": 0.018447284609339455,
+      "grad_norm": 1447.9163818359375,
+      "learning_rate": 1.802020202020202e-06,
+      "loss": 270.9402,
+      "step": 2230
+    },
+    {
+      "epoch": 0.01853000785870869,
+      "grad_norm": 5666.75634765625,
+      "learning_rate": 1.8101010101010103e-06,
+      "loss": 265.4302,
+      "step": 2240
+    },
+    {
+      "epoch": 0.018612731108077925,
+      "grad_norm": 901.1162109375,
+      "learning_rate": 1.8181818181818183e-06,
+      "loss": 327.2734,
+      "step": 2250
+    },
+    {
+      "epoch": 0.01869545435744716,
+      "grad_norm": 1582.9307861328125,
+      "learning_rate": 1.8262626262626265e-06,
+      "loss": 295.4947,
+      "step": 2260
+    },
+    {
+      "epoch": 0.018778177606816396,
+      "grad_norm": 910.0087280273438,
+      "learning_rate": 1.8343434343434343e-06,
+      "loss": 254.5832,
+      "step": 2270
+    },
+    {
+      "epoch": 0.018860900856185633,
+      "grad_norm": 1553.7003173828125,
+      "learning_rate": 1.8424242424242426e-06,
+      "loss": 312.1798,
+      "step": 2280
+    },
+    {
+      "epoch": 0.018943624105554866,
+      "grad_norm": 1657.673828125,
+      "learning_rate": 1.8505050505050506e-06,
+      "loss": 333.7471,
+      "step": 2290
+    },
+    {
+      "epoch": 0.019026347354924103,
+      "grad_norm": 2194.053466796875,
+      "learning_rate": 1.8585858585858588e-06,
+      "loss": 252.629,
+      "step": 2300
+    },
+    {
+      "epoch": 0.019109070604293336,
+      "grad_norm": 968.1193237304688,
+      "learning_rate": 1.8666666666666669e-06,
+      "loss": 317.2808,
+      "step": 2310
+    },
+    {
+      "epoch": 0.019191793853662573,
+      "grad_norm": 2099.283203125,
+      "learning_rate": 1.874747474747475e-06,
+      "loss": 318.3145,
+      "step": 2320
+    },
+    {
+      "epoch": 0.019274517103031807,
+      "grad_norm": 1018.1463012695312,
+      "learning_rate": 1.882828282828283e-06,
+      "loss": 310.5461,
+      "step": 2330
+    },
+    {
+      "epoch": 0.019357240352401044,
+      "grad_norm": 1319.893310546875,
+      "learning_rate": 1.890909090909091e-06,
+      "loss": 403.0323,
+      "step": 2340
+    },
+    {
+      "epoch": 0.019439963601770277,
+      "grad_norm": 3296.14404296875,
+      "learning_rate": 1.8989898989898992e-06,
+      "loss": 277.1991,
+      "step": 2350
+    },
+    {
+      "epoch": 0.019522686851139514,
+      "grad_norm": 2120.728515625,
+      "learning_rate": 1.9070707070707072e-06,
+      "loss": 266.7476,
+      "step": 2360
+    },
+    {
+      "epoch": 0.019605410100508747,
+      "grad_norm": 1736.4859619140625,
+      "learning_rate": 1.9151515151515154e-06,
+      "loss": 385.987,
+      "step": 2370
+    },
+    {
+      "epoch": 0.019688133349877984,
+      "grad_norm": 1218.7796630859375,
+      "learning_rate": 1.9232323232323235e-06,
+      "loss": 337.6978,
+      "step": 2380
+    },
+    {
+      "epoch": 0.019770856599247218,
+      "grad_norm": 4797.71337890625,
+      "learning_rate": 1.9313131313131315e-06,
+      "loss": 255.6176,
+      "step": 2390
+    },
+    {
+      "epoch": 0.019853579848616455,
+      "grad_norm": 1956.2137451171875,
+      "learning_rate": 1.9393939393939395e-06,
+      "loss": 283.0651,
+      "step": 2400
+    },
+    {
+      "epoch": 0.019936303097985688,
+      "grad_norm": 918.0914916992188,
+      "learning_rate": 1.9474747474747475e-06,
+      "loss": 312.682,
+      "step": 2410
+    },
+    {
+      "epoch": 0.020019026347354925,
+      "grad_norm": 2107.04736328125,
+      "learning_rate": 1.955555555555556e-06,
+      "loss": 384.748,
+      "step": 2420
+    },
+    {
+      "epoch": 0.02010174959672416,
+      "grad_norm": 861.828125,
+      "learning_rate": 1.9636363636363636e-06,
+      "loss": 340.8865,
+      "step": 2430
+    },
+    {
+      "epoch": 0.020184472846093395,
+      "grad_norm": 6480.6171875,
+      "learning_rate": 1.9717171717171716e-06,
+      "loss": 416.4972,
+      "step": 2440
+    },
+    {
+      "epoch": 0.02026719609546263,
+      "grad_norm": 1926.79296875,
+      "learning_rate": 1.97979797979798e-06,
+      "loss": 260.5365,
+      "step": 2450
+    },
+    {
+      "epoch": 0.020349919344831865,
+      "grad_norm": 1225.735107421875,
+      "learning_rate": 1.987878787878788e-06,
+      "loss": 261.068,
+      "step": 2460
+    },
+    {
+      "epoch": 0.0204326425942011,
+      "grad_norm": 857.5597534179688,
+      "learning_rate": 1.995959595959596e-06,
+      "loss": 241.7558,
+      "step": 2470
+    },
+    {
+      "epoch": 0.020515365843570336,
+      "grad_norm": 42987.5546875,
+      "learning_rate": 2.004040404040404e-06,
+      "loss": 380.0271,
+      "step": 2480
+    },
+    {
+      "epoch": 0.02059808909293957,
+      "grad_norm": 1689.1239013671875,
+      "learning_rate": 2.012121212121212e-06,
+      "loss": 390.1416,
+      "step": 2490
+    },
+    {
+      "epoch": 0.020680812342308806,
+      "grad_norm": 4134.138671875,
+      "learning_rate": 2.02020202020202e-06,
+      "loss": 265.3958,
+      "step": 2500
+    },
+    {
+      "epoch": 0.02076353559167804,
+      "grad_norm": 1039.45703125,
+      "learning_rate": 2.0282828282828286e-06,
+      "loss": 297.8323,
+      "step": 2510
+    },
+    {
+      "epoch": 0.020846258841047276,
+      "grad_norm": 2275.5380859375,
+      "learning_rate": 2.0363636363636367e-06,
+      "loss": 326.8338,
+      "step": 2520
+    },
+    {
+      "epoch": 0.02092898209041651,
+      "grad_norm": 1341.4542236328125,
+      "learning_rate": 2.0444444444444447e-06,
+      "loss": 255.4626,
+      "step": 2530
+    },
+    {
+      "epoch": 0.021011705339785747,
+      "grad_norm": 1704.446044921875,
+      "learning_rate": 2.0525252525252527e-06,
+      "loss": 311.6551,
+      "step": 2540
+    },
+    {
+      "epoch": 0.021094428589154984,
+      "grad_norm": 1272.9852294921875,
+      "learning_rate": 2.0606060606060607e-06,
+      "loss": 278.1134,
+      "step": 2550
+    },
+    {
+      "epoch": 0.021177151838524217,
+      "grad_norm": 788.3895263671875,
+      "learning_rate": 2.0686868686868688e-06,
+      "loss": 282.1122,
+      "step": 2560
+    },
+    {
+      "epoch": 0.021259875087893454,
+      "grad_norm": 1385.612548828125,
+      "learning_rate": 2.0767676767676768e-06,
+      "loss": 314.3244,
+      "step": 2570
+    },
+    {
+      "epoch": 0.021342598337262687,
+      "grad_norm": 1955.3095703125,
+      "learning_rate": 2.0848484848484852e-06,
+      "loss": 262.7597,
+      "step": 2580
+    },
+    {
+      "epoch": 0.021425321586631924,
+      "grad_norm": 1727.3680419921875,
+      "learning_rate": 2.092929292929293e-06,
+      "loss": 264.8611,
+      "step": 2590
+    },
+    {
+      "epoch": 0.021508044836001158,
+      "grad_norm": 1702.6575927734375,
+      "learning_rate": 2.1010101010101013e-06,
+      "loss": 286.7952,
+      "step": 2600
+    },
+    {
+      "epoch": 0.021590768085370395,
+      "grad_norm": 1000.8258666992188,
+      "learning_rate": 2.1090909090909093e-06,
+      "loss": 259.9386,
+      "step": 2610
+    },
+    {
+      "epoch": 0.021673491334739628,
+      "grad_norm": 1225.5982666015625,
+      "learning_rate": 2.1171717171717173e-06,
+      "loss": 248.9738,
+      "step": 2620
+    },
+    {
+      "epoch": 0.021756214584108865,
+      "grad_norm": 2182.8955078125,
+      "learning_rate": 2.1252525252525254e-06,
+      "loss": 290.0493,
+      "step": 2630
+    },
+    {
+      "epoch": 0.0218389378334781,
+      "grad_norm": 946.896484375,
+      "learning_rate": 2.133333333333334e-06,
+      "loss": 285.7407,
+      "step": 2640
+    },
+    {
+      "epoch": 0.021921661082847335,
+      "grad_norm": 965.7489624023438,
+      "learning_rate": 2.1414141414141414e-06,
+      "loss": 277.0386,
+      "step": 2650
+    },
+    {
+      "epoch": 0.02200438433221657,
+      "grad_norm": 1855.605712890625,
+      "learning_rate": 2.1494949494949494e-06,
+      "loss": 308.2425,
+      "step": 2660
+    },
+    {
+      "epoch": 0.022087107581585805,
+      "grad_norm": 945.2372436523438,
+      "learning_rate": 2.157575757575758e-06,
+      "loss": 314.9084,
+      "step": 2670
+    },
+    {
+      "epoch": 0.02216983083095504,
+      "grad_norm": 2101.0771484375,
+      "learning_rate": 2.165656565656566e-06,
+      "loss": 277.5356,
+      "step": 2680
+    },
+    {
+      "epoch": 0.022252554080324276,
+      "grad_norm": 931.0570678710938,
+      "learning_rate": 2.173737373737374e-06,
+      "loss": 380.9075,
+      "step": 2690
+    },
+    {
+      "epoch": 0.02233527732969351,
+      "grad_norm": 849.65380859375,
+      "learning_rate": 2.181818181818182e-06,
+      "loss": 266.7404,
+      "step": 2700
+    },
+    {
+      "epoch": 0.022418000579062746,
+      "grad_norm": 795.3492431640625,
+      "learning_rate": 2.18989898989899e-06,
+      "loss": 301.0031,
+      "step": 2710
+    },
+    {
+      "epoch": 0.02250072382843198,
+      "grad_norm": 998.2992553710938,
+      "learning_rate": 2.197979797979798e-06,
+      "loss": 236.0826,
+      "step": 2720
+    },
+    {
+      "epoch": 0.022583447077801216,
+      "grad_norm": 3029.24462890625,
+      "learning_rate": 2.2060606060606064e-06,
+      "loss": 284.6466,
+      "step": 2730
+    },
+    {
+      "epoch": 0.02266617032717045,
+      "grad_norm": 991.4857177734375,
+      "learning_rate": 2.2141414141414145e-06,
+      "loss": 480.9545,
+      "step": 2740
+    },
+    {
+      "epoch": 0.022748893576539687,
+      "grad_norm": 610.8831787109375,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 280.1348,
+      "step": 2750
+    },
+    {
+      "epoch": 0.02283161682590892,
+      "grad_norm": 1224.61083984375,
+      "learning_rate": 2.2303030303030305e-06,
+      "loss": 287.0002,
+      "step": 2760
+    },
+    {
+      "epoch": 0.022914340075278157,
+      "grad_norm": 967.6873779296875,
+      "learning_rate": 2.2383838383838385e-06,
+      "loss": 302.4269,
+      "step": 2770
+    },
+    {
+      "epoch": 0.02299706332464739,
+      "grad_norm": 1249.24169921875,
+      "learning_rate": 2.2464646464646466e-06,
+      "loss": 324.5018,
+      "step": 2780
+    },
+    {
+      "epoch": 0.023079786574016627,
+      "grad_norm": 2674.385498046875,
+      "learning_rate": 2.254545454545455e-06,
+      "loss": 253.8143,
+      "step": 2790
+    },
+    {
+      "epoch": 0.023162509823385864,
+      "grad_norm": 1356.0936279296875,
+      "learning_rate": 2.262626262626263e-06,
+      "loss": 287.3596,
+      "step": 2800
+    },
+    {
+      "epoch": 0.023245233072755098,
+      "grad_norm": 1003.8848266601562,
+      "learning_rate": 2.2707070707070706e-06,
+      "loss": 259.5809,
+      "step": 2810
+    },
+    {
+      "epoch": 0.023327956322124335,
+      "grad_norm": 1609.186767578125,
+      "learning_rate": 2.278787878787879e-06,
+      "loss": 243.4104,
+      "step": 2820
+    },
+    {
+      "epoch": 0.023410679571493568,
+      "grad_norm": 1286.5640869140625,
+      "learning_rate": 2.286868686868687e-06,
+      "loss": 303.2196,
+      "step": 2830
+    },
+    {
+      "epoch": 0.023493402820862805,
+      "grad_norm": 1285.9228515625,
+      "learning_rate": 2.294949494949495e-06,
+      "loss": 274.7912,
+      "step": 2840
+    },
+    {
+      "epoch": 0.02357612607023204,
+      "grad_norm": 1015.9469604492188,
+      "learning_rate": 2.303030303030303e-06,
+      "loss": 255.5114,
+      "step": 2850
+    },
+    {
+      "epoch": 0.023658849319601275,
+      "grad_norm": 1524.3387451171875,
+      "learning_rate": 2.311111111111111e-06,
+      "loss": 394.2463,
+      "step": 2860
+    },
+    {
+      "epoch": 0.02374157256897051,
+      "grad_norm": 4101.37548828125,
+      "learning_rate": 2.3191919191919192e-06,
+      "loss": 343.4752,
+      "step": 2870
+    },
+    {
+      "epoch": 0.023824295818339745,
+      "grad_norm": 1837.815673828125,
+      "learning_rate": 2.3272727272727277e-06,
+      "loss": 306.1534,
+      "step": 2880
+    },
+    {
+      "epoch": 0.02390701906770898,
+      "grad_norm": 1742.3822021484375,
+      "learning_rate": 2.3353535353535357e-06,
+      "loss": 382.2382,
+      "step": 2890
+    },
+    {
+      "epoch": 0.023989742317078216,
+      "grad_norm": 922.0031127929688,
+      "learning_rate": 2.3434343434343437e-06,
+      "loss": 230.2729,
+      "step": 2900
+    },
+    {
+      "epoch": 0.02407246556644745,
+      "grad_norm": 860.6764526367188,
+      "learning_rate": 2.3515151515151517e-06,
+      "loss": 257.8272,
+      "step": 2910
+    },
+    {
+      "epoch": 0.024155188815816686,
+      "grad_norm": 3094.787841796875,
+      "learning_rate": 2.3595959595959598e-06,
+      "loss": 247.0388,
+      "step": 2920
+    },
+    {
+      "epoch": 0.02423791206518592,
+      "grad_norm": 1400.10009765625,
+      "learning_rate": 2.367676767676768e-06,
+      "loss": 350.3922,
+      "step": 2930
+    },
+    {
+      "epoch": 0.024320635314555156,
+      "grad_norm": 1306.7431640625,
+      "learning_rate": 2.375757575757576e-06,
+      "loss": 318.7807,
+      "step": 2940
+    },
+    {
+      "epoch": 0.02440335856392439,
+      "grad_norm": 859.3858032226562,
+      "learning_rate": 2.3838383838383843e-06,
+      "loss": 292.0828,
+      "step": 2950
+    },
+    {
+      "epoch": 0.024486081813293627,
+      "grad_norm": 1207.4444580078125,
+      "learning_rate": 2.3919191919191923e-06,
+      "loss": 268.6456,
+      "step": 2960
+    },
+    {
+      "epoch": 0.02456880506266286,
+      "grad_norm": 1388.7242431640625,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 327.8167,
+      "step": 2970
+    },
+    {
+      "epoch": 0.024651528312032097,
+      "grad_norm": 1601.5413818359375,
+      "learning_rate": 2.4080808080808083e-06,
+      "loss": 217.4461,
+      "step": 2980
+    },
+    {
+      "epoch": 0.02473425156140133,
+      "grad_norm": 1094.072509765625,
+      "learning_rate": 2.4161616161616164e-06,
+      "loss": 237.2505,
+      "step": 2990
+    },
+    {
+      "epoch": 0.024816974810770567,
+      "grad_norm": 1064.4403076171875,
+      "learning_rate": 2.4242424242424244e-06,
+      "loss": 274.2577,
+      "step": 3000
+    },
+    {
+      "epoch": 0.0248996980601398,
+      "grad_norm": 2712.60986328125,
+      "learning_rate": 2.432323232323233e-06,
+      "loss": 336.2089,
+      "step": 3010
+    },
+    {
+      "epoch": 0.024982421309509038,
+      "grad_norm": 657.0144653320312,
+      "learning_rate": 2.4404040404040404e-06,
+      "loss": 285.5267,
+      "step": 3020
+    },
+    {
+      "epoch": 0.02506514455887827,
+      "grad_norm": 1818.3843994140625,
+      "learning_rate": 2.4484848484848485e-06,
+      "loss": 345.3333,
+      "step": 3030
+    },
+    {
+      "epoch": 0.025147867808247508,
+      "grad_norm": 1924.058349609375,
+      "learning_rate": 2.456565656565657e-06,
+      "loss": 254.806,
+      "step": 3040
+    },
+    {
+      "epoch": 0.025230591057616745,
+      "grad_norm": 926.4883422851562,
+      "learning_rate": 2.464646464646465e-06,
+      "loss": 248.8595,
+      "step": 3050
+    },
+    {
+      "epoch": 0.025313314306985978,
+      "grad_norm": 1217.055419921875,
+      "learning_rate": 2.472727272727273e-06,
+      "loss": 258.9547,
+      "step": 3060
+    },
+    {
+      "epoch": 0.025396037556355215,
+      "grad_norm": 837.9513549804688,
+      "learning_rate": 2.480808080808081e-06,
+      "loss": 239.7319,
+      "step": 3070
+    },
+    {
+      "epoch": 0.02547876080572445,
+      "grad_norm": 702.333740234375,
+      "learning_rate": 2.488888888888889e-06,
+      "loss": 290.1943,
+      "step": 3080
+    },
+    {
+      "epoch": 0.025561484055093685,
+      "grad_norm": 1217.6317138671875,
+      "learning_rate": 2.496969696969697e-06,
+      "loss": 282.1116,
+      "step": 3090
+    },
+    {
+      "epoch": 0.02564420730446292,
+      "grad_norm": 1642.226806640625,
+      "learning_rate": 2.5050505050505055e-06,
+      "loss": 242.3538,
+      "step": 3100
+    },
+    {
+      "epoch": 0.025726930553832156,
+      "grad_norm": 1036.6883544921875,
+      "learning_rate": 2.5131313131313135e-06,
+      "loss": 339.3384,
+      "step": 3110
+    },
+    {
+      "epoch": 0.02580965380320139,
+      "grad_norm": 2547.31396484375,
+      "learning_rate": 2.5212121212121215e-06,
+      "loss": 268.7113,
+      "step": 3120
+    },
+    {
+      "epoch": 0.025892377052570626,
+      "grad_norm": 2499.7900390625,
+      "learning_rate": 2.5292929292929296e-06,
+      "loss": 371.6728,
+      "step": 3130
+    },
+    {
+      "epoch": 0.02597510030193986,
+      "grad_norm": 835.4087524414062,
+      "learning_rate": 2.537373737373738e-06,
+      "loss": 331.6685,
+      "step": 3140
+    },
+    {
+      "epoch": 0.026057823551309096,
+      "grad_norm": 1723.9932861328125,
+      "learning_rate": 2.5454545454545456e-06,
+      "loss": 241.2475,
+      "step": 3150
+    },
+    {
+      "epoch": 0.02614054680067833,
+      "grad_norm": 892.8982543945312,
+      "learning_rate": 2.5535353535353536e-06,
+      "loss": 239.1675,
+      "step": 3160
+    },
+    {
+      "epoch": 0.026223270050047567,
+      "grad_norm": 1679.0596923828125,
+      "learning_rate": 2.5616161616161617e-06,
+      "loss": 251.3017,
+      "step": 3170
+    },
+    {
+      "epoch": 0.0263059932994168,
+      "grad_norm": 1314.4605712890625,
+      "learning_rate": 2.5696969696969697e-06,
+      "loss": 291.6142,
+      "step": 3180
+    },
+    {
+      "epoch": 0.026388716548786037,
+      "grad_norm": 1959.02490234375,
+      "learning_rate": 2.577777777777778e-06,
+      "loss": 335.8238,
+      "step": 3190
+    },
+    {
+      "epoch": 0.02647143979815527,
+      "grad_norm": 1912.5706787109375,
+      "learning_rate": 2.585858585858586e-06,
+      "loss": 374.3031,
+      "step": 3200
+    },
+    {
+      "epoch": 0.026554163047524507,
+      "grad_norm": 3296.021728515625,
+      "learning_rate": 2.593939393939394e-06,
+      "loss": 258.0628,
+      "step": 3210
+    },
+    {
+      "epoch": 0.02663688629689374,
+      "grad_norm": 1115.4268798828125,
+      "learning_rate": 2.602020202020202e-06,
+      "loss": 334.8846,
+      "step": 3220
+    },
+    {
+      "epoch": 0.026719609546262978,
+      "grad_norm": 973.0841064453125,
+      "learning_rate": 2.6101010101010107e-06,
+      "loss": 275.7638,
+      "step": 3230
+    },
+    {
+      "epoch": 0.02680233279563221,
+      "grad_norm": 1405.8780517578125,
+      "learning_rate": 2.6181818181818187e-06,
+      "loss": 290.556,
+      "step": 3240
+    },
+    {
+      "epoch": 0.026885056045001448,
+      "grad_norm": 2512.458740234375,
+      "learning_rate": 2.6262626262626267e-06,
+      "loss": 255.2991,
+      "step": 3250
+    },
+    {
+      "epoch": 0.02696777929437068,
+      "grad_norm": 1015.4577026367188,
+      "learning_rate": 2.6343434343434343e-06,
+      "loss": 293.1441,
+      "step": 3260
+    },
+    {
+      "epoch": 0.027050502543739918,
+      "grad_norm": 1590.4383544921875,
+      "learning_rate": 2.6424242424242423e-06,
+      "loss": 315.3852,
+      "step": 3270
+    },
+    {
+      "epoch": 0.02713322579310915,
+      "grad_norm": 1241.1497802734375,
+      "learning_rate": 2.6505050505050508e-06,
+      "loss": 279.314,
+      "step": 3280
+    },
+    {
+      "epoch": 0.02721594904247839,
+      "grad_norm": 803.2537841796875,
+      "learning_rate": 2.658585858585859e-06,
+      "loss": 237.7475,
+      "step": 3290
+    },
+    {
+      "epoch": 0.027298672291847625,
+      "grad_norm": 1308.884765625,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 289.3862,
+      "step": 3300
+    },
+    {
+      "epoch": 0.02738139554121686,
+      "grad_norm": 1037.373291015625,
+      "learning_rate": 2.674747474747475e-06,
+      "loss": 278.7418,
+      "step": 3310
+    },
+    {
+      "epoch": 0.027464118790586096,
+      "grad_norm": 1414.7493896484375,
+      "learning_rate": 2.6828282828282833e-06,
+      "loss": 388.9876,
+      "step": 3320
+    },
+    {
+      "epoch": 0.02754684203995533,
+      "grad_norm": 1686.8880615234375,
+      "learning_rate": 2.6909090909090913e-06,
+      "loss": 224.3872,
+      "step": 3330
+    },
+    {
+      "epoch": 0.027629565289324566,
+      "grad_norm": 1130.1907958984375,
+      "learning_rate": 2.6989898989898994e-06,
+      "loss": 219.274,
+      "step": 3340
+    },
+    {
+      "epoch": 0.0277122885386938,
+      "grad_norm": 1604.05419921875,
+      "learning_rate": 2.7070707070707074e-06,
+      "loss": 313.1014,
+      "step": 3350
+    },
+    {
+      "epoch": 0.027795011788063036,
+      "grad_norm": 1515.8199462890625,
+      "learning_rate": 2.715151515151516e-06,
+      "loss": 351.9699,
+      "step": 3360
+    },
+    {
+      "epoch": 0.02787773503743227,
+      "grad_norm": 1105.636474609375,
+      "learning_rate": 2.7232323232323234e-06,
+      "loss": 258.1985,
+      "step": 3370
+    },
+    {
+      "epoch": 0.027960458286801507,
+      "grad_norm": 2343.5537109375,
+      "learning_rate": 2.7313131313131315e-06,
+      "loss": 247.2701,
+      "step": 3380
+    },
+    {
+      "epoch": 0.02804318153617074,
+      "grad_norm": 3364.70068359375,
+      "learning_rate": 2.7393939393939395e-06,
+      "loss": 320.3825,
+      "step": 3390
+    },
+    {
+      "epoch": 0.028125904785539977,
+      "grad_norm": 1226.61279296875,
+      "learning_rate": 2.7474747474747475e-06,
+      "loss": 308.3539,
+      "step": 3400
+    },
+    {
+      "epoch": 0.02820862803490921,
+      "grad_norm": 1365.4581298828125,
+      "learning_rate": 2.755555555555556e-06,
+      "loss": 265.8813,
+      "step": 3410
+    },
+    {
+      "epoch": 0.028291351284278447,
+      "grad_norm": 1471.4437255859375,
+      "learning_rate": 2.763636363636364e-06,
+      "loss": 246.5465,
+      "step": 3420
+    },
+    {
+      "epoch": 0.02837407453364768,
+      "grad_norm": 1522.9873046875,
+      "learning_rate": 2.771717171717172e-06,
+      "loss": 294.6199,
+      "step": 3430
+    },
+    {
+      "epoch": 0.028456797783016918,
+      "grad_norm": 2295.73095703125,
+      "learning_rate": 2.77979797979798e-06,
+      "loss": 471.7754,
+      "step": 3440
+    },
+    {
+      "epoch": 0.02853952103238615,
+      "grad_norm": 1030.5484619140625,
+      "learning_rate": 2.7878787878787885e-06,
+      "loss": 315.6531,
+      "step": 3450
+    },
+    {
+      "epoch": 0.028622244281755388,
+      "grad_norm": 1534.0970458984375,
+      "learning_rate": 2.7959595959595965e-06,
+      "loss": 236.7922,
+      "step": 3460
+    },
+    {
+      "epoch": 0.02870496753112462,
+      "grad_norm": 1193.41748046875,
+      "learning_rate": 2.804040404040404e-06,
+      "loss": 301.1876,
+      "step": 3470
+    },
+    {
+      "epoch": 0.028787690780493858,
+      "grad_norm": 884.4568481445312,
+      "learning_rate": 2.812121212121212e-06,
+      "loss": 248.0099,
+      "step": 3480
+    },
+    {
+      "epoch": 0.02887041402986309,
+      "grad_norm": 1417.0870361328125,
+      "learning_rate": 2.82020202020202e-06,
+      "loss": 214.0175,
+      "step": 3490
+    },
+    {
+      "epoch": 0.02895313727923233,
+      "grad_norm": 906.0728149414062,
+      "learning_rate": 2.8282828282828286e-06,
+      "loss": 337.6909,
+      "step": 3500
+    },
+    {
+      "epoch": 0.029035860528601562,
+      "grad_norm": 720.5711059570312,
+      "learning_rate": 2.8363636363636366e-06,
+      "loss": 237.9402,
+      "step": 3510
+    },
+    {
+      "epoch": 0.0291185837779708,
+      "grad_norm": 2159.7626953125,
+      "learning_rate": 2.8444444444444446e-06,
+      "loss": 290.2204,
+      "step": 3520
+    },
+    {
+      "epoch": 0.029201307027340032,
+      "grad_norm": 1583.6937255859375,
+      "learning_rate": 2.8525252525252527e-06,
+      "loss": 339.3829,
+      "step": 3530
+    },
+    {
+      "epoch": 0.02928403027670927,
+      "grad_norm": 1711.7625732421875,
+      "learning_rate": 2.860606060606061e-06,
+      "loss": 338.8203,
+      "step": 3540
+    },
+    {
+      "epoch": 0.029366753526078506,
+      "grad_norm": 1867.3909912109375,
+      "learning_rate": 2.868686868686869e-06,
+      "loss": 295.4354,
+      "step": 3550
+    },
+    {
+      "epoch": 0.02944947677544774,
+      "grad_norm": 1350.51708984375,
+      "learning_rate": 2.876767676767677e-06,
+      "loss": 255.1478,
+      "step": 3560
+    },
+    {
+      "epoch": 0.029532200024816976,
+      "grad_norm": 1835.9559326171875,
+      "learning_rate": 2.884848484848485e-06,
+      "loss": 235.7582,
+      "step": 3570
+    },
+    {
+      "epoch": 0.02961492327418621,
+      "grad_norm": 1273.539794921875,
+      "learning_rate": 2.892929292929293e-06,
+      "loss": 305.7688,
+      "step": 3580
+    },
+    {
+      "epoch": 0.029697646523555447,
+      "grad_norm": 1092.1051025390625,
+      "learning_rate": 2.9010101010101012e-06,
+      "loss": 229.7812,
+      "step": 3590
+    },
+    {
+      "epoch": 0.02978036977292468,
+      "grad_norm": 1232.7423095703125,
+      "learning_rate": 2.9090909090909093e-06,
+      "loss": 308.136,
+      "step": 3600
+    },
+    {
+      "epoch": 0.029863093022293917,
+      "grad_norm": 860.4628295898438,
+      "learning_rate": 2.9171717171717173e-06,
+      "loss": 245.5094,
+      "step": 3610
+    },
+    {
+      "epoch": 0.02994581627166315,
+      "grad_norm": 1105.772705078125,
+      "learning_rate": 2.9252525252525253e-06,
+      "loss": 225.044,
+      "step": 3620
+    },
+    {
+      "epoch": 0.030028539521032387,
+      "grad_norm": 1092.992431640625,
+      "learning_rate": 2.9333333333333338e-06,
+      "loss": 240.9013,
+      "step": 3630
+    },
+    {
+      "epoch": 0.03011126277040162,
+      "grad_norm": 1877.3671875,
+      "learning_rate": 2.941414141414142e-06,
+      "loss": 304.4779,
+      "step": 3640
+    },
+    {
+      "epoch": 0.030193986019770858,
+      "grad_norm": 1553.412109375,
+      "learning_rate": 2.94949494949495e-06,
+      "loss": 253.1373,
+      "step": 3650
+    },
+    {
+      "epoch": 0.03027670926914009,
+      "grad_norm": 1163.033447265625,
+      "learning_rate": 2.957575757575758e-06,
+      "loss": 272.4007,
+      "step": 3660
+    },
+    {
+      "epoch": 0.030359432518509328,
+      "grad_norm": 1105.1661376953125,
+      "learning_rate": 2.9656565656565663e-06,
+      "loss": 243.5992,
+      "step": 3670
+    },
+    {
+      "epoch": 0.03044215576787856,
+      "grad_norm": 1370.2781982421875,
+      "learning_rate": 2.9737373737373743e-06,
+      "loss": 232.3612,
+      "step": 3680
+    },
+    {
+      "epoch": 0.030524879017247798,
+      "grad_norm": 925.8299560546875,
+      "learning_rate": 2.981818181818182e-06,
+      "loss": 194.9076,
+      "step": 3690
+    },
+    {
+      "epoch": 0.03060760226661703,
+      "grad_norm": 4565.89208984375,
+      "learning_rate": 2.98989898989899e-06,
+      "loss": 317.9731,
+      "step": 3700
+    },
+    {
+      "epoch": 0.03069032551598627,
+      "grad_norm": 1785.587890625,
+      "learning_rate": 2.997979797979798e-06,
+      "loss": 261.7598,
+      "step": 3710
+    },
+    {
+      "epoch": 0.030773048765355502,
+      "grad_norm": 1177.1761474609375,
+      "learning_rate": 3.0060606060606064e-06,
+      "loss": 211.625,
+      "step": 3720
+    },
+    {
+      "epoch": 0.03085577201472474,
+      "grad_norm": 980.7220458984375,
+      "learning_rate": 3.0141414141414144e-06,
+      "loss": 198.16,
+      "step": 3730
+    },
+    {
+      "epoch": 0.030938495264093972,
+      "grad_norm": 1450.34033203125,
+      "learning_rate": 3.0222222222222225e-06,
+      "loss": 281.4796,
+      "step": 3740
+    },
+    {
+      "epoch": 0.03102121851346321,
+      "grad_norm": 917.07275390625,
+      "learning_rate": 3.0303030303030305e-06,
+      "loss": 266.2501,
+      "step": 3750
+    },
+    {
+      "epoch": 0.031103941762832443,
+      "grad_norm": 1657.814697265625,
+      "learning_rate": 3.038383838383839e-06,
+      "loss": 279.4682,
+      "step": 3760
+    },
+    {
+      "epoch": 0.03118666501220168,
+      "grad_norm": 1294.77685546875,
+      "learning_rate": 3.046464646464647e-06,
+      "loss": 259.5207,
+      "step": 3770
+    },
+    {
+      "epoch": 0.03126938826157091,
+      "grad_norm": 1622.716796875,
+      "learning_rate": 3.054545454545455e-06,
+      "loss": 298.1313,
+      "step": 3780
+    },
+    {
+      "epoch": 0.031352111510940146,
+      "grad_norm": 1660.8116455078125,
+      "learning_rate": 3.0626262626262626e-06,
+      "loss": 250.9933,
+      "step": 3790
+    },
+    {
+      "epoch": 0.03143483476030939,
+      "grad_norm": 799.0031127929688,
+      "learning_rate": 3.0707070707070706e-06,
+      "loss": 310.6956,
+      "step": 3800
+    },
+    {
+      "epoch": 0.03151755800967862,
+      "grad_norm": 1199.9678955078125,
+      "learning_rate": 3.078787878787879e-06,
+      "loss": 283.1158,
+      "step": 3810
+    },
+    {
+      "epoch": 0.031600281259047854,
+      "grad_norm": 1449.6212158203125,
+      "learning_rate": 3.086868686868687e-06,
+      "loss": 249.93,
+      "step": 3820
+    },
+    {
+      "epoch": 0.031683004508417094,
+      "grad_norm": 1299.2960205078125,
+      "learning_rate": 3.094949494949495e-06,
+      "loss": 224.7813,
+      "step": 3830
+    },
+    {
+      "epoch": 0.03176572775778633,
+      "grad_norm": 2239.721435546875,
+      "learning_rate": 3.103030303030303e-06,
+      "loss": 223.2453,
+      "step": 3840
+    },
+    {
+      "epoch": 0.03184845100715556,
+      "grad_norm": 1760.5396728515625,
+      "learning_rate": 3.1111111111111116e-06,
+      "loss": 252.7971,
+      "step": 3850
+    },
+    {
+      "epoch": 0.031931174256524794,
+      "grad_norm": 682.6892700195312,
+      "learning_rate": 3.1191919191919196e-06,
+      "loss": 216.5466,
+      "step": 3860
+    },
+    {
+      "epoch": 0.032013897505894034,
+      "grad_norm": 2052.32861328125,
+      "learning_rate": 3.1272727272727276e-06,
+      "loss": 288.775,
+      "step": 3870
+    },
+    {
+      "epoch": 0.03209662075526327,
+      "grad_norm": 1602.2225341796875,
+      "learning_rate": 3.1353535353535357e-06,
+      "loss": 328.757,
+      "step": 3880
+    },
+    {
+      "epoch": 0.0321793440046325,
+      "grad_norm": 681.2459106445312,
+      "learning_rate": 3.143434343434344e-06,
+      "loss": 190.3252,
+      "step": 3890
+    },
+    {
+      "epoch": 0.032262067254001735,
+      "grad_norm": 1182.54052734375,
+      "learning_rate": 3.1515151515151517e-06,
+      "loss": 268.2503,
+      "step": 3900
+    },
+    {
+      "epoch": 0.032344790503370975,
+      "grad_norm": 1219.3660888671875,
+      "learning_rate": 3.1595959595959597e-06,
+      "loss": 239.1843,
+      "step": 3910
+    },
+    {
+      "epoch": 0.03242751375274021,
+      "grad_norm": 1041.707275390625,
+      "learning_rate": 3.1676767676767678e-06,
+      "loss": 273.016,
+      "step": 3920
+    },
+    {
+      "epoch": 0.03251023700210944,
+      "grad_norm": 1269.7681884765625,
+      "learning_rate": 3.1757575757575758e-06,
+      "loss": 249.6547,
+      "step": 3930
+    },
+    {
+      "epoch": 0.032592960251478675,
+      "grad_norm": 1372.903564453125,
+      "learning_rate": 3.1838383838383842e-06,
+      "loss": 311.8126,
+      "step": 3940
+    },
+    {
+      "epoch": 0.032675683500847916,
+      "grad_norm": 2617.830322265625,
+      "learning_rate": 3.1919191919191923e-06,
+      "loss": 299.153,
+      "step": 3950
+    },
+    {
+      "epoch": 0.03275840675021715,
+      "grad_norm": 1350.3302001953125,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 238.4413,
+      "step": 3960
+    },
+    {
+      "epoch": 0.03284112999958638,
+      "grad_norm": 2270.457763671875,
+      "learning_rate": 3.2080808080808083e-06,
+      "loss": 267.5316,
+      "step": 3970
+    },
+    {
+      "epoch": 0.032923853248955616,
+      "grad_norm": 1409.05712890625,
+      "learning_rate": 3.2161616161616168e-06,
+      "loss": 237.9799,
+      "step": 3980
+    },
+    {
+      "epoch": 0.033006576498324856,
+      "grad_norm": 1641.050537109375,
+      "learning_rate": 3.2242424242424248e-06,
+      "loss": 352.6279,
+      "step": 3990
+    },
+    {
+      "epoch": 0.03308929974769409,
+      "grad_norm": 1178.6591796875,
+      "learning_rate": 3.232323232323233e-06,
+      "loss": 200.0928,
+      "step": 4000
+    },
+    {
+      "epoch": 0.03317202299706332,
+      "grad_norm": 895.1983032226562,
+      "learning_rate": 3.2404040404040404e-06,
+      "loss": 274.1366,
+      "step": 4010
+    },
+    {
+      "epoch": 0.03325474624643256,
+      "grad_norm": 2193.698486328125,
+      "learning_rate": 3.2484848484848484e-06,
+      "loss": 205.6771,
+      "step": 4020
+    },
+    {
+      "epoch": 0.0333374694958018,
+      "grad_norm": 1705.0743408203125,
+      "learning_rate": 3.256565656565657e-06,
+      "loss": 253.3518,
+      "step": 4030
+    },
+    {
+      "epoch": 0.03342019274517103,
+      "grad_norm": 1552.232666015625,
+      "learning_rate": 3.264646464646465e-06,
+      "loss": 243.4265,
+      "step": 4040
+    },
+    {
+      "epoch": 0.033502915994540264,
+      "grad_norm": 4037.46875,
+      "learning_rate": 3.272727272727273e-06,
+      "loss": 229.4004,
+      "step": 4050
+    },
+    {
+      "epoch": 0.033585639243909504,
+      "grad_norm": 3386.515869140625,
+      "learning_rate": 3.280808080808081e-06,
+      "loss": 338.8573,
+      "step": 4060
+    },
+    {
+      "epoch": 0.03366836249327874,
+      "grad_norm": 1145.6063232421875,
+      "learning_rate": 3.2888888888888894e-06,
+      "loss": 254.9263,
+      "step": 4070
+    },
+    {
+      "epoch": 0.03375108574264797,
+      "grad_norm": 2452.519775390625,
+      "learning_rate": 3.2969696969696974e-06,
+      "loss": 365.3762,
+      "step": 4080
+    },
+    {
+      "epoch": 0.033833808992017204,
+      "grad_norm": 1752.8304443359375,
+      "learning_rate": 3.3050505050505054e-06,
+      "loss": 242.0698,
+      "step": 4090
+    },
+    {
+      "epoch": 0.033916532241386445,
+      "grad_norm": 1975.453857421875,
+      "learning_rate": 3.3131313131313135e-06,
+      "loss": 294.2446,
+      "step": 4100
+    },
+    {
+      "epoch": 0.03399925549075568,
+      "grad_norm": 1159.05615234375,
+      "learning_rate": 3.321212121212121e-06,
+      "loss": 254.9454,
+      "step": 4110
+    },
+    {
+      "epoch": 0.03408197874012491,
+      "grad_norm": 904.747802734375,
+      "learning_rate": 3.3292929292929295e-06,
+      "loss": 291.38,
+      "step": 4120
+    },
+    {
+      "epoch": 0.034164701989494145,
+      "grad_norm": 1563.5689697265625,
+      "learning_rate": 3.3373737373737375e-06,
+      "loss": 212.5943,
+      "step": 4130
+    },
+    {
+      "epoch": 0.034247425238863385,
+      "grad_norm": 1204.7821044921875,
+      "learning_rate": 3.3454545454545456e-06,
+      "loss": 265.8186,
+      "step": 4140
+    },
+    {
+      "epoch": 0.03433014848823262,
+      "grad_norm": 904.9209594726562,
+      "learning_rate": 3.3535353535353536e-06,
+      "loss": 260.1104,
+      "step": 4150
+    },
+    {
+      "epoch": 0.03441287173760185,
+      "grad_norm": 1248.727783203125,
+      "learning_rate": 3.361616161616162e-06,
+      "loss": 291.2336,
+      "step": 4160
+    },
+    {
+      "epoch": 0.034495594986971086,
+      "grad_norm": 893.3284912109375,
+      "learning_rate": 3.36969696969697e-06,
+      "loss": 233.5644,
+      "step": 4170
+    },
+    {
+      "epoch": 0.034578318236340326,
+      "grad_norm": 1158.58984375,
+      "learning_rate": 3.377777777777778e-06,
+      "loss": 246.782,
+      "step": 4180
+    },
+    {
+      "epoch": 0.03466104148570956,
+      "grad_norm": 1969.4088134765625,
+      "learning_rate": 3.385858585858586e-06,
+      "loss": 237.9205,
+      "step": 4190
+    },
+    {
+      "epoch": 0.03474376473507879,
+      "grad_norm": 1035.88427734375,
+      "learning_rate": 3.3939393939393946e-06,
+      "loss": 321.7581,
+      "step": 4200
+    },
+    {
+      "epoch": 0.034826487984448026,
+      "grad_norm": 918.1465454101562,
+      "learning_rate": 3.4020202020202026e-06,
+      "loss": 272.7782,
+      "step": 4210
+    },
+    {
+      "epoch": 0.03490921123381727,
+      "grad_norm": 1070.0615234375,
+      "learning_rate": 3.41010101010101e-06,
+      "loss": 224.1239,
+      "step": 4220
+    },
+    {
+      "epoch": 0.0349919344831865,
+      "grad_norm": 1230.9432373046875,
+      "learning_rate": 3.4181818181818182e-06,
+      "loss": 276.4893,
+      "step": 4230
+    },
+    {
+      "epoch": 0.035074657732555733,
+      "grad_norm": 1131.748046875,
+      "learning_rate": 3.4262626262626262e-06,
+      "loss": 217.2287,
+      "step": 4240
+    },
+    {
+      "epoch": 0.03515738098192497,
+      "grad_norm": 2427.749267578125,
+      "learning_rate": 3.4343434343434347e-06,
+      "loss": 251.6459,
+      "step": 4250
+    },
+    {
+      "epoch": 0.03524010423129421,
+      "grad_norm": 1668.80810546875,
+      "learning_rate": 3.4424242424242427e-06,
+      "loss": 317.6187,
+      "step": 4260
+    },
+    {
+      "epoch": 0.03532282748066344,
+      "grad_norm": 372.29510498046875,
+      "learning_rate": 3.4505050505050507e-06,
+      "loss": 179.444,
+      "step": 4270
+    },
+    {
+      "epoch": 0.035405550730032674,
+      "grad_norm": 1011.8233032226562,
+      "learning_rate": 3.4585858585858588e-06,
+      "loss": 219.8903,
+      "step": 4280
+    },
+    {
+      "epoch": 0.03548827397940191,
+      "grad_norm": 1418.88232421875,
+      "learning_rate": 3.4666666666666672e-06,
+      "loss": 286.1539,
+      "step": 4290
+    },
+    {
+      "epoch": 0.03557099722877115,
+      "grad_norm": 2228.23095703125,
+      "learning_rate": 3.4747474747474752e-06,
+      "loss": 293.0923,
+      "step": 4300
+    },
+    {
+      "epoch": 0.03565372047814038,
+      "grad_norm": 715.6218872070312,
+      "learning_rate": 3.4828282828282833e-06,
+      "loss": 282.1017,
+      "step": 4310
+    },
+    {
+      "epoch": 0.035736443727509615,
+      "grad_norm": 742.7416381835938,
+      "learning_rate": 3.4909090909090913e-06,
+      "loss": 204.9258,
+      "step": 4320
+    },
+    {
+      "epoch": 0.035819166976878855,
+      "grad_norm": 1228.9534912109375,
+      "learning_rate": 3.498989898989899e-06,
+      "loss": 259.2133,
+      "step": 4330
+    },
+    {
+      "epoch": 0.03590189022624809,
+      "grad_norm": 967.5274047851562,
+      "learning_rate": 3.5070707070707073e-06,
+      "loss": 288.3385,
+      "step": 4340
+    },
+    {
+      "epoch": 0.03598461347561732,
+      "grad_norm": 1587.67724609375,
+      "learning_rate": 3.5151515151515154e-06,
+      "loss": 277.6203,
+      "step": 4350
+    },
+    {
+      "epoch": 0.036067336724986555,
+      "grad_norm": 1431.6448974609375,
+      "learning_rate": 3.5232323232323234e-06,
+      "loss": 213.064,
+      "step": 4360
+    },
+    {
+      "epoch": 0.036150059974355796,
+      "grad_norm": 1135.779052734375,
+      "learning_rate": 3.5313131313131314e-06,
+      "loss": 221.4357,
+      "step": 4370
+    },
+    {
+      "epoch": 0.03623278322372503,
+      "grad_norm": 1519.3890380859375,
+      "learning_rate": 3.53939393939394e-06,
+      "loss": 277.6575,
+      "step": 4380
+    },
+    {
+      "epoch": 0.03631550647309426,
+      "grad_norm": 1360.25048828125,
+      "learning_rate": 3.547474747474748e-06,
+      "loss": 195.2055,
+      "step": 4390
+    },
+    {
+      "epoch": 0.036398229722463496,
+      "grad_norm": 2059.44287109375,
+      "learning_rate": 3.555555555555556e-06,
+      "loss": 324.8405,
+      "step": 4400
+    },
+    {
+      "epoch": 0.036480952971832736,
+      "grad_norm": 780.7919921875,
+      "learning_rate": 3.563636363636364e-06,
+      "loss": 212.4319,
+      "step": 4410
+    },
+    {
+      "epoch": 0.03656367622120197,
+      "grad_norm": 1413.5367431640625,
+      "learning_rate": 3.5717171717171724e-06,
+      "loss": 255.6041,
+      "step": 4420
+    },
+    {
+      "epoch": 0.0366463994705712,
+      "grad_norm": 637.59521484375,
+      "learning_rate": 3.57979797979798e-06,
+      "loss": 193.1592,
+      "step": 4430
+    },
+    {
+      "epoch": 0.03672912271994044,
+      "grad_norm": 1240.098876953125,
+      "learning_rate": 3.587878787878788e-06,
+      "loss": 331.8386,
+      "step": 4440
+    },
+    {
+      "epoch": 0.03681184596930968,
+      "grad_norm": 980.0120239257812,
+      "learning_rate": 3.595959595959596e-06,
+      "loss": 276.4505,
+      "step": 4450
+    },
+    {
+      "epoch": 0.03689456921867891,
+      "grad_norm": 2805.927001953125,
+      "learning_rate": 3.604040404040404e-06,
+      "loss": 269.0381,
+      "step": 4460
+    },
+    {
+      "epoch": 0.036977292468048144,
+      "grad_norm": 2359.113525390625,
+      "learning_rate": 3.6121212121212125e-06,
+      "loss": 258.4454,
+      "step": 4470
+    },
+    {
+      "epoch": 0.03706001571741738,
+      "grad_norm": 1293.63720703125,
+      "learning_rate": 3.6202020202020205e-06,
+      "loss": 225.2379,
+      "step": 4480
+    },
+    {
+      "epoch": 0.03714273896678662,
+      "grad_norm": 2243.848876953125,
+      "learning_rate": 3.6282828282828286e-06,
+      "loss": 213.6965,
+      "step": 4490
+    },
+    {
+      "epoch": 0.03722546221615585,
+      "grad_norm": 770.7854614257812,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 231.6815,
+      "step": 4500
+    },
+    {
+      "epoch": 0.037308185465525084,
+      "grad_norm": 1308.199462890625,
+      "learning_rate": 3.644444444444445e-06,
+      "loss": 232.9941,
+      "step": 4510
+    },
+    {
+      "epoch": 0.03739090871489432,
+      "grad_norm": 883.5875244140625,
+      "learning_rate": 3.652525252525253e-06,
+      "loss": 216.6295,
+      "step": 4520
+    },
+    {
+      "epoch": 0.03747363196426356,
+      "grad_norm": 672.0443115234375,
+      "learning_rate": 3.660606060606061e-06,
+      "loss": 201.8103,
+      "step": 4530
+    },
+    {
+      "epoch": 0.03755635521363279,
+      "grad_norm": 1314.2528076171875,
+      "learning_rate": 3.6686868686868687e-06,
+      "loss": 256.3278,
+      "step": 4540
+    },
+    {
+      "epoch": 0.037639078463002025,
+      "grad_norm": 1359.0018310546875,
+      "learning_rate": 3.6767676767676767e-06,
+      "loss": 216.419,
+      "step": 4550
+    },
+    {
+      "epoch": 0.037721801712371265,
+      "grad_norm": 1884.207763671875,
+      "learning_rate": 3.684848484848485e-06,
+      "loss": 211.731,
+      "step": 4560
+    },
+    {
+      "epoch": 0.0378045249617405,
+      "grad_norm": 1213.398193359375,
+      "learning_rate": 3.692929292929293e-06,
+      "loss": 272.2958,
+      "step": 4570
+    },
+    {
+      "epoch": 0.03788724821110973,
+      "grad_norm": 2207.1171875,
+      "learning_rate": 3.701010101010101e-06,
+      "loss": 265.4331,
+      "step": 4580
+    },
+    {
+      "epoch": 0.037969971460478966,
+      "grad_norm": 1944.73876953125,
+      "learning_rate": 3.7090909090909092e-06,
+      "loss": 272.1659,
+      "step": 4590
+    },
+    {
+      "epoch": 0.038052694709848206,
+      "grad_norm": 1226.4189453125,
+      "learning_rate": 3.7171717171717177e-06,
+      "loss": 226.3827,
+      "step": 4600
+    },
+    {
+      "epoch": 0.03813541795921744,
+      "grad_norm": 1100.8829345703125,
+      "learning_rate": 3.7252525252525257e-06,
+      "loss": 221.5738,
+      "step": 4610
+    },
+    {
+      "epoch": 0.03821814120858667,
+      "grad_norm": 1859.399169921875,
+      "learning_rate": 3.7333333333333337e-06,
+      "loss": 229.0128,
+      "step": 4620
+    },
+    {
+      "epoch": 0.038300864457955906,
+      "grad_norm": 1107.61572265625,
+      "learning_rate": 3.7414141414141418e-06,
+      "loss": 224.0653,
+      "step": 4630
+    },
+    {
+      "epoch": 0.03838358770732515,
+      "grad_norm": 1617.0789794921875,
+      "learning_rate": 3.74949494949495e-06,
+      "loss": 240.6666,
+      "step": 4640
+    },
+    {
+      "epoch": 0.03846631095669438,
+      "grad_norm": 2554.2158203125,
+      "learning_rate": 3.757575757575758e-06,
+      "loss": 258.9923,
+      "step": 4650
+    },
+    {
+      "epoch": 0.038549034206063613,
+      "grad_norm": 3580.094482421875,
+      "learning_rate": 3.765656565656566e-06,
+      "loss": 270.2465,
+      "step": 4660
+    },
+    {
+      "epoch": 0.03863175745543285,
+      "grad_norm": 1391.0946044921875,
+      "learning_rate": 3.773737373737374e-06,
+      "loss": 184.8681,
+      "step": 4670
+    },
+    {
+      "epoch": 0.03871448070480209,
+      "grad_norm": 1413.2205810546875,
+      "learning_rate": 3.781818181818182e-06,
+      "loss": 210.8876,
+      "step": 4680
+    },
+    {
+      "epoch": 0.03879720395417132,
+      "grad_norm": 1123.7315673828125,
+      "learning_rate": 3.7898989898989903e-06,
+      "loss": 226.6469,
+      "step": 4690
+    },
+    {
+      "epoch": 0.038879927203540554,
+      "grad_norm": 1507.7408447265625,
+      "learning_rate": 3.7979797979797984e-06,
+      "loss": 243.835,
+      "step": 4700
+    },
+    {
+      "epoch": 0.03896265045290979,
+      "grad_norm": 1154.9927978515625,
+      "learning_rate": 3.8060606060606064e-06,
+      "loss": 275.5142,
+      "step": 4710
+    },
+    {
+      "epoch": 0.03904537370227903,
+      "grad_norm": 1263.9693603515625,
+      "learning_rate": 3.8141414141414144e-06,
+      "loss": 208.4059,
+      "step": 4720
+    },
+    {
+      "epoch": 0.03912809695164826,
+      "grad_norm": 2182.939208984375,
+      "learning_rate": 3.8222222222222224e-06,
+      "loss": 225.735,
+      "step": 4730
+    },
+    {
+      "epoch": 0.039210820201017495,
+      "grad_norm": 1238.331298828125,
+      "learning_rate": 3.830303030303031e-06,
+      "loss": 215.773,
+      "step": 4740
+    },
+    {
+      "epoch": 0.03929354345038673,
+      "grad_norm": 1922.897216796875,
+      "learning_rate": 3.8383838383838385e-06,
+      "loss": 205.833,
+      "step": 4750
+    },
+    {
+      "epoch": 0.03937626669975597,
+      "grad_norm": 1248.5205078125,
+      "learning_rate": 3.846464646464647e-06,
+      "loss": 212.8604,
+      "step": 4760
+    },
+    {
+      "epoch": 0.0394589899491252,
+      "grad_norm": 1999.810791015625,
+      "learning_rate": 3.8545454545454545e-06,
+      "loss": 234.7124,
+      "step": 4770
+    },
+    {
+      "epoch": 0.039541713198494435,
+      "grad_norm": 1900.56787109375,
+      "learning_rate": 3.862626262626263e-06,
+      "loss": 206.3176,
+      "step": 4780
+    },
+    {
+      "epoch": 0.03962443644786367,
+      "grad_norm": 1776.326171875,
+      "learning_rate": 3.8707070707070706e-06,
+      "loss": 235.4771,
+      "step": 4790
+    },
+    {
+      "epoch": 0.03970715969723291,
+      "grad_norm": 1206.546630859375,
+      "learning_rate": 3.878787878787879e-06,
+      "loss": 232.6159,
+      "step": 4800
+    },
+    {
+      "epoch": 0.03978988294660214,
+      "grad_norm": 1193.0535888671875,
+      "learning_rate": 3.8868686868686875e-06,
+      "loss": 239.4891,
+      "step": 4810
+    },
+    {
+      "epoch": 0.039872606195971376,
+      "grad_norm": 1266.81103515625,
+      "learning_rate": 3.894949494949495e-06,
+      "loss": 264.1928,
+      "step": 4820
+    },
+    {
+      "epoch": 0.039955329445340616,
+      "grad_norm": 1152.444580078125,
+      "learning_rate": 3.9030303030303035e-06,
+      "loss": 197.2479,
+      "step": 4830
+    },
+    {
+      "epoch": 0.04003805269470985,
+      "grad_norm": 1796.5777587890625,
+      "learning_rate": 3.911111111111112e-06,
+      "loss": 232.1052,
+      "step": 4840
+    },
+    {
+      "epoch": 0.04012077594407908,
+      "grad_norm": 1190.642822265625,
+      "learning_rate": 3.9191919191919196e-06,
+      "loss": 246.6287,
+      "step": 4850
+    },
+    {
+      "epoch": 0.04020349919344832,
+      "grad_norm": 1134.314453125,
+      "learning_rate": 3.927272727272727e-06,
+      "loss": 268.4593,
+      "step": 4860
+    },
+    {
+      "epoch": 0.04028622244281756,
+      "grad_norm": 1696.9732666015625,
+      "learning_rate": 3.935353535353536e-06,
+      "loss": 226.5233,
+      "step": 4870
+    },
+    {
+      "epoch": 0.04036894569218679,
+      "grad_norm": 1237.491455078125,
+      "learning_rate": 3.943434343434343e-06,
+      "loss": 196.7495,
+      "step": 4880
+    },
+    {
+      "epoch": 0.040451668941556024,
+      "grad_norm": 2431.97265625,
+      "learning_rate": 3.951515151515152e-06,
+      "loss": 220.9505,
+      "step": 4890
+    },
+    {
+      "epoch": 0.04053439219092526,
+      "grad_norm": 1421.2952880859375,
+      "learning_rate": 3.95959595959596e-06,
+      "loss": 253.9248,
+      "step": 4900
+    },
+    {
+      "epoch": 0.0406171154402945,
+      "grad_norm": 1305.2431640625,
+      "learning_rate": 3.967676767676768e-06,
+      "loss": 302.1668,
+      "step": 4910
+    },
+    {
+      "epoch": 0.04069983868966373,
+      "grad_norm": 1975.3135986328125,
+      "learning_rate": 3.975757575757576e-06,
+      "loss": 282.536,
+      "step": 4920
+    },
+    {
+      "epoch": 0.040782561939032964,
+      "grad_norm": 1655.1588134765625,
+      "learning_rate": 3.983838383838385e-06,
+      "loss": 217.2957,
+      "step": 4930
+    },
+    {
+      "epoch": 0.0408652851884022,
+      "grad_norm": 1051.86083984375,
+      "learning_rate": 3.991919191919192e-06,
+      "loss": 240.2383,
+      "step": 4940
+    },
+    {
+      "epoch": 0.04094800843777144,
+      "grad_norm": 773.5546875,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 221.8524,
+      "step": 4950
+    },
+    {
+      "epoch": 0.04103073168714067,
+      "grad_norm": 1287.1944580078125,
+      "learning_rate": 4.008080808080808e-06,
+      "loss": 229.3614,
+      "step": 4960
+    },
+    {
+      "epoch": 0.041113454936509905,
+      "grad_norm": 1897.0220947265625,
+      "learning_rate": 4.016161616161616e-06,
+      "loss": 233.9596,
+      "step": 4970
+    },
+    {
+      "epoch": 0.04119617818587914,
+      "grad_norm": 1463.894287109375,
+      "learning_rate": 4.024242424242424e-06,
+      "loss": 205.995,
+      "step": 4980
+    },
+    {
+      "epoch": 0.04127890143524838,
+      "grad_norm": 981.1106567382812,
+      "learning_rate": 4.032323232323233e-06,
+      "loss": 226.2633,
+      "step": 4990
+    },
+    {
+      "epoch": 0.04136162468461761,
+      "grad_norm": 3404.24365234375,
+      "learning_rate": 4.04040404040404e-06,
+      "loss": 249.0778,
+      "step": 5000
+    },
+    {
+      "epoch": 0.041444347933986846,
+      "grad_norm": 1160.4622802734375,
+      "learning_rate": 4.048484848484849e-06,
+      "loss": 201.7348,
+      "step": 5010
+    },
+    {
+      "epoch": 0.04152707118335608,
+      "grad_norm": 1339.532470703125,
+      "learning_rate": 4.056565656565657e-06,
+      "loss": 296.6651,
+      "step": 5020
+    },
+    {
+      "epoch": 0.04160979443272532,
+      "grad_norm": 1330.253173828125,
+      "learning_rate": 4.064646464646465e-06,
+      "loss": 222.7661,
+      "step": 5030
+    },
+    {
+      "epoch": 0.04169251768209455,
+      "grad_norm": 1871.73095703125,
+      "learning_rate": 4.072727272727273e-06,
+      "loss": 185.6875,
+      "step": 5040
+    },
+    {
+      "epoch": 0.041775240931463786,
+      "grad_norm": 989.5170288085938,
+      "learning_rate": 4.080808080808081e-06,
+      "loss": 184.853,
+      "step": 5050
+    },
+    {
+      "epoch": 0.04185796418083302,
+      "grad_norm": 1325.712158203125,
+      "learning_rate": 4.088888888888889e-06,
+      "loss": 177.4656,
+      "step": 5060
+    },
+    {
+      "epoch": 0.04194068743020226,
+      "grad_norm": 1084.450927734375,
+      "learning_rate": 4.096969696969697e-06,
+      "loss": 218.0059,
+      "step": 5070
+    },
+    {
+      "epoch": 0.04202341067957149,
+      "grad_norm": 1230.160888671875,
+      "learning_rate": 4.105050505050505e-06,
+      "loss": 202.8811,
+      "step": 5080
+    },
+    {
+      "epoch": 0.04210613392894073,
+      "grad_norm": 909.71728515625,
+      "learning_rate": 4.113131313131313e-06,
+      "loss": 229.2698,
+      "step": 5090
+    },
+    {
+      "epoch": 0.04218885717830997,
+      "grad_norm": 2449.3154296875,
+      "learning_rate": 4.1212121212121215e-06,
+      "loss": 211.7621,
+      "step": 5100
+    },
+    {
+      "epoch": 0.0422715804276792,
+      "grad_norm": 578.8762817382812,
+      "learning_rate": 4.12929292929293e-06,
+      "loss": 220.8765,
+      "step": 5110
+    },
+    {
+      "epoch": 0.042354303677048434,
+      "grad_norm": 1294.6402587890625,
+      "learning_rate": 4.1373737373737375e-06,
+      "loss": 252.3842,
+      "step": 5120
+    },
+    {
+      "epoch": 0.04243702692641767,
+      "grad_norm": 911.5681762695312,
+      "learning_rate": 4.145454545454546e-06,
+      "loss": 294.7006,
+      "step": 5130
+    },
+    {
+      "epoch": 0.04251975017578691,
+      "grad_norm": 1328.2413330078125,
+      "learning_rate": 4.1535353535353536e-06,
+      "loss": 211.482,
+      "step": 5140
+    },
+    {
+      "epoch": 0.04260247342515614,
+      "grad_norm": 1352.004638671875,
+      "learning_rate": 4.161616161616162e-06,
+      "loss": 210.5752,
+      "step": 5150
+    },
+    {
+      "epoch": 0.042685196674525375,
+      "grad_norm": 1081.876953125,
+      "learning_rate": 4.1696969696969705e-06,
+      "loss": 224.4739,
+      "step": 5160
+    },
+    {
+      "epoch": 0.04276791992389461,
+      "grad_norm": 2295.4033203125,
+      "learning_rate": 4.177777777777778e-06,
+      "loss": 229.6548,
+      "step": 5170
+    },
+    {
+      "epoch": 0.04285064317326385,
+      "grad_norm": 1036.4749755859375,
+      "learning_rate": 4.185858585858586e-06,
+      "loss": 208.1364,
+      "step": 5180
+    },
+    {
+      "epoch": 0.04293336642263308,
+      "grad_norm": 1558.9718017578125,
+      "learning_rate": 4.193939393939394e-06,
+      "loss": 262.7995,
+      "step": 5190
+    },
+    {
+      "epoch": 0.043016089672002315,
+      "grad_norm": 1155.65478515625,
+      "learning_rate": 4.2020202020202026e-06,
+      "loss": 251.0647,
+      "step": 5200
+    },
+    {
+      "epoch": 0.04309881292137155,
+      "grad_norm": 1081.0423583984375,
+      "learning_rate": 4.21010101010101e-06,
+      "loss": 235.0402,
+      "step": 5210
+    },
+    {
+      "epoch": 0.04318153617074079,
+      "grad_norm": 725.3358764648438,
+      "learning_rate": 4.218181818181819e-06,
+      "loss": 310.5599,
+      "step": 5220
+    },
+    {
+      "epoch": 0.04326425942011002,
+      "grad_norm": 1182.2801513671875,
+      "learning_rate": 4.226262626262626e-06,
+      "loss": 178.3826,
+      "step": 5230
+    },
+    {
+      "epoch": 0.043346982669479256,
+      "grad_norm": 1413.02587890625,
+      "learning_rate": 4.234343434343435e-06,
+      "loss": 257.1767,
+      "step": 5240
+    },
+    {
+      "epoch": 0.04342970591884849,
+      "grad_norm": 1222.1142578125,
+      "learning_rate": 4.242424242424243e-06,
+      "loss": 218.1251,
+      "step": 5250
+    },
+    {
+      "epoch": 0.04351242916821773,
+      "grad_norm": 998.1055297851562,
+      "learning_rate": 4.250505050505051e-06,
+      "loss": 211.1936,
+      "step": 5260
+    },
+    {
+      "epoch": 0.04359515241758696,
+      "grad_norm": 844.65478515625,
+      "learning_rate": 4.258585858585859e-06,
+      "loss": 235.3665,
+      "step": 5270
+    },
+    {
+      "epoch": 0.0436778756669562,
+      "grad_norm": 691.6785278320312,
+      "learning_rate": 4.266666666666668e-06,
+      "loss": 173.3902,
+      "step": 5280
+    },
+    {
+      "epoch": 0.04376059891632543,
+      "grad_norm": 1113.0621337890625,
+      "learning_rate": 4.274747474747475e-06,
+      "loss": 169.7838,
+      "step": 5290
+    },
+    {
+      "epoch": 0.04384332216569467,
+      "grad_norm": 1622.1474609375,
+      "learning_rate": 4.282828282828283e-06,
+      "loss": 217.8731,
+      "step": 5300
+    },
+    {
+      "epoch": 0.043926045415063904,
+      "grad_norm": 1753.8427734375,
+      "learning_rate": 4.290909090909091e-06,
+      "loss": 251.789,
+      "step": 5310
+    },
+    {
+      "epoch": 0.04400876866443314,
+      "grad_norm": 918.1004638671875,
+      "learning_rate": 4.298989898989899e-06,
+      "loss": 201.8801,
+      "step": 5320
+    },
+    {
+      "epoch": 0.04409149191380238,
+      "grad_norm": 899.498779296875,
+      "learning_rate": 4.307070707070707e-06,
+      "loss": 230.2721,
+      "step": 5330
+    },
+    {
+      "epoch": 0.04417421516317161,
+      "grad_norm": 1194.7314453125,
+      "learning_rate": 4.315151515151516e-06,
+      "loss": 302.4559,
+      "step": 5340
+    },
+    {
+      "epoch": 0.044256938412540844,
+      "grad_norm": 745.4715576171875,
+      "learning_rate": 4.323232323232323e-06,
+      "loss": 217.5794,
+      "step": 5350
+    },
+    {
+      "epoch": 0.04433966166191008,
+      "grad_norm": 3024.5693359375,
+      "learning_rate": 4.331313131313132e-06,
+      "loss": 215.8062,
+      "step": 5360
+    },
+    {
+      "epoch": 0.04442238491127932,
+      "grad_norm": 692.5128173828125,
+      "learning_rate": 4.33939393939394e-06,
+      "loss": 206.9033,
+      "step": 5370
+    },
+    {
+      "epoch": 0.04450510816064855,
+      "grad_norm": 638.6357421875,
+      "learning_rate": 4.347474747474748e-06,
+      "loss": 277.6233,
+      "step": 5380
+    },
+    {
+      "epoch": 0.044587831410017785,
+      "grad_norm": 1237.0350341796875,
+      "learning_rate": 4.3555555555555555e-06,
+      "loss": 202.0422,
+      "step": 5390
+    },
+    {
+      "epoch": 0.04467055465938702,
+      "grad_norm": 1546.1956787109375,
+      "learning_rate": 4.363636363636364e-06,
+      "loss": 195.5879,
+      "step": 5400
+    },
+    {
+      "epoch": 0.04475327790875626,
+      "grad_norm": 1744.332763671875,
+      "learning_rate": 4.3717171717171715e-06,
+      "loss": 308.2887,
+      "step": 5410
+    },
+    {
+      "epoch": 0.04483600115812549,
+      "grad_norm": 1885.436767578125,
+      "learning_rate": 4.37979797979798e-06,
+      "loss": 257.4365,
+      "step": 5420
+    },
+    {
+      "epoch": 0.044918724407494726,
+      "grad_norm": 1949.8758544921875,
+      "learning_rate": 4.387878787878788e-06,
+      "loss": 288.0396,
+      "step": 5430
+    },
+    {
+      "epoch": 0.04500144765686396,
+      "grad_norm": 2583.996826171875,
+      "learning_rate": 4.395959595959596e-06,
+      "loss": 243.6479,
+      "step": 5440
+    },
+    {
+      "epoch": 0.0450841709062332,
+      "grad_norm": 940.0198974609375,
+      "learning_rate": 4.4040404040404044e-06,
+      "loss": 209.4657,
+      "step": 5450
+    },
+    {
+      "epoch": 0.04516689415560243,
+      "grad_norm": 897.3545532226562,
+      "learning_rate": 4.412121212121213e-06,
+      "loss": 244.724,
+      "step": 5460
+    },
+    {
+      "epoch": 0.045249617404971666,
+      "grad_norm": 1344.112060546875,
+      "learning_rate": 4.4202020202020205e-06,
+      "loss": 210.6839,
+      "step": 5470
+    },
+    {
+      "epoch": 0.0453323406543409,
+      "grad_norm": 2638.66357421875,
+      "learning_rate": 4.428282828282829e-06,
+      "loss": 281.9325,
+      "step": 5480
+    },
+    {
+      "epoch": 0.04541506390371014,
+      "grad_norm": 1480.95068359375,
+      "learning_rate": 4.436363636363637e-06,
+      "loss": 227.8617,
+      "step": 5490
+    },
+    {
+      "epoch": 0.04549778715307937,
+      "grad_norm": 1484.61962890625,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 307.6048,
+      "step": 5500
+    },
+    {
+      "epoch": 0.04558051040244861,
+      "grad_norm": 921.8578491210938,
+      "learning_rate": 4.452525252525253e-06,
+      "loss": 283.7532,
+      "step": 5510
+    },
+    {
+      "epoch": 0.04566323365181784,
+      "grad_norm": 1015.0638427734375,
+      "learning_rate": 4.460606060606061e-06,
+      "loss": 232.6935,
+      "step": 5520
+    },
+    {
+      "epoch": 0.04574595690118708,
+      "grad_norm": 1154.900146484375,
+      "learning_rate": 4.468686868686869e-06,
+      "loss": 237.65,
+      "step": 5530
+    },
+    {
+      "epoch": 0.045828680150556314,
+      "grad_norm": 1067.7034912109375,
+      "learning_rate": 4.476767676767677e-06,
+      "loss": 178.2383,
+      "step": 5540
+    },
+    {
+      "epoch": 0.04591140339992555,
+      "grad_norm": 1582.29541015625,
+      "learning_rate": 4.4848484848484855e-06,
+      "loss": 208.2042,
+      "step": 5550
+    },
+    {
+      "epoch": 0.04599412664929478,
+      "grad_norm": 1326.168212890625,
+      "learning_rate": 4.492929292929293e-06,
+      "loss": 206.2426,
+      "step": 5560
+    },
+    {
+      "epoch": 0.04607684989866402,
+      "grad_norm": 1166.575927734375,
+      "learning_rate": 4.501010101010102e-06,
+      "loss": 246.9012,
+      "step": 5570
+    },
+    {
+      "epoch": 0.046159573148033255,
+      "grad_norm": 1051.04345703125,
+      "learning_rate": 4.50909090909091e-06,
+      "loss": 220.1656,
+      "step": 5580
+    },
+    {
+      "epoch": 0.04624229639740249,
+      "grad_norm": 994.1414794921875,
+      "learning_rate": 4.517171717171718e-06,
+      "loss": 246.7432,
+      "step": 5590
+    },
+    {
+      "epoch": 0.04632501964677173,
+      "grad_norm": 1792.089111328125,
+      "learning_rate": 4.525252525252526e-06,
+      "loss": 195.5538,
+      "step": 5600
+    },
+    {
+      "epoch": 0.04640774289614096,
+      "grad_norm": 889.2850341796875,
+      "learning_rate": 4.533333333333334e-06,
+      "loss": 205.7484,
+      "step": 5610
+    },
+    {
+      "epoch": 0.046490466145510195,
+      "grad_norm": 986.1232299804688,
+      "learning_rate": 4.541414141414141e-06,
+      "loss": 259.9392,
+      "step": 5620
+    },
+    {
+      "epoch": 0.04657318939487943,
+      "grad_norm": 1162.5496826171875,
+      "learning_rate": 4.54949494949495e-06,
+      "loss": 215.2503,
+      "step": 5630
+    },
+    {
+      "epoch": 0.04665591264424867,
+      "grad_norm": 1568.9271240234375,
+      "learning_rate": 4.557575757575758e-06,
+      "loss": 198.5209,
+      "step": 5640
+    },
+    {
+      "epoch": 0.0467386358936179,
+      "grad_norm": 1163.0565185546875,
+      "learning_rate": 4.565656565656566e-06,
+      "loss": 199.4977,
+      "step": 5650
+    },
+    {
+      "epoch": 0.046821359142987136,
+      "grad_norm": 918.8221435546875,
+      "learning_rate": 4.573737373737374e-06,
+      "loss": 292.0952,
+      "step": 5660
+    },
+    {
+      "epoch": 0.04690408239235637,
+      "grad_norm": 1420.076171875,
+      "learning_rate": 4.581818181818183e-06,
+      "loss": 273.1453,
+      "step": 5670
+    },
+    {
+      "epoch": 0.04698680564172561,
+      "grad_norm": 751.3970947265625,
+      "learning_rate": 4.58989898989899e-06,
+      "loss": 227.5678,
+      "step": 5680
+    },
+    {
+      "epoch": 0.04706952889109484,
+      "grad_norm": 4106.84814453125,
+      "learning_rate": 4.597979797979799e-06,
+      "loss": 257.4646,
+      "step": 5690
+    },
+    {
+      "epoch": 0.04715225214046408,
+      "grad_norm": 1618.319091796875,
+      "learning_rate": 4.606060606060606e-06,
+      "loss": 242.2784,
+      "step": 5700
+    },
+    {
+      "epoch": 0.04723497538983331,
+      "grad_norm": 1682.986083984375,
+      "learning_rate": 4.614141414141414e-06,
+      "loss": 270.8784,
+      "step": 5710
+    },
+    {
+      "epoch": 0.04731769863920255,
+      "grad_norm": 2293.774658203125,
+      "learning_rate": 4.622222222222222e-06,
+      "loss": 310.2911,
+      "step": 5720
+    },
+    {
+      "epoch": 0.047400421888571784,
+      "grad_norm": 896.1268920898438,
+      "learning_rate": 4.630303030303031e-06,
+      "loss": 262.9586,
+      "step": 5730
+    },
+    {
+      "epoch": 0.04748314513794102,
+      "grad_norm": 1522.369384765625,
+      "learning_rate": 4.6383838383838384e-06,
+      "loss": 200.667,
+      "step": 5740
+    },
+    {
+      "epoch": 0.04756586838731025,
+      "grad_norm": 1271.126220703125,
+      "learning_rate": 4.646464646464647e-06,
+      "loss": 211.7462,
+      "step": 5750
+    },
+    {
+      "epoch": 0.04764859163667949,
+      "grad_norm": 1342.4813232421875,
+      "learning_rate": 4.654545454545455e-06,
+      "loss": 271.1495,
+      "step": 5760
+    },
+    {
+      "epoch": 0.047731314886048724,
+      "grad_norm": 1042.1790771484375,
+      "learning_rate": 4.662626262626263e-06,
+      "loss": 207.4777,
+      "step": 5770
+    },
+    {
+      "epoch": 0.04781403813541796,
+      "grad_norm": 1017.2898559570312,
+      "learning_rate": 4.670707070707071e-06,
+      "loss": 272.7402,
+      "step": 5780
+    },
+    {
+      "epoch": 0.04789676138478719,
+      "grad_norm": 1240.869140625,
+      "learning_rate": 4.678787878787879e-06,
+      "loss": 224.074,
+      "step": 5790
+    },
+    {
+      "epoch": 0.04797948463415643,
+      "grad_norm": 1895.0374755859375,
+      "learning_rate": 4.6868686868686874e-06,
+      "loss": 348.4789,
+      "step": 5800
+    },
+    {
+      "epoch": 0.048062207883525665,
+      "grad_norm": 1665.503662109375,
+      "learning_rate": 4.694949494949496e-06,
+      "loss": 287.3858,
+      "step": 5810
+    },
+    {
+      "epoch": 0.0481449311328949,
+      "grad_norm": 888.2938842773438,
+      "learning_rate": 4.7030303030303035e-06,
+      "loss": 186.7068,
+      "step": 5820
+    },
+    {
+      "epoch": 0.04822765438226414,
+      "grad_norm": 1077.885009765625,
+      "learning_rate": 4.711111111111111e-06,
+      "loss": 188.8208,
+      "step": 5830
+    },
+    {
+      "epoch": 0.04831037763163337,
+      "grad_norm": 1367.186279296875,
+      "learning_rate": 4.7191919191919195e-06,
+      "loss": 235.7021,
+      "step": 5840
+    },
+    {
+      "epoch": 0.048393100881002606,
+      "grad_norm": 880.5682373046875,
+      "learning_rate": 4.727272727272728e-06,
+      "loss": 150.1969,
+      "step": 5850
+    },
+    {
+      "epoch": 0.04847582413037184,
+      "grad_norm": 1563.840576171875,
+      "learning_rate": 4.735353535353536e-06,
+      "loss": 213.3092,
+      "step": 5860
+    },
+    {
+      "epoch": 0.04855854737974108,
+      "grad_norm": 1928.1456298828125,
+      "learning_rate": 4.743434343434344e-06,
+      "loss": 258.8421,
+      "step": 5870
+    },
+    {
+      "epoch": 0.04864127062911031,
+      "grad_norm": 1255.3079833984375,
+      "learning_rate": 4.751515151515152e-06,
+      "loss": 242.0784,
+      "step": 5880
+    },
+    {
+      "epoch": 0.048723993878479546,
+      "grad_norm": 1542.299072265625,
+      "learning_rate": 4.75959595959596e-06,
+      "loss": 303.837,
+      "step": 5890
+    },
+    {
+      "epoch": 0.04880671712784878,
+      "grad_norm": 2009.2406005859375,
+      "learning_rate": 4.7676767676767685e-06,
+      "loss": 155.5049,
+      "step": 5900
+    },
+    {
+      "epoch": 0.04888944037721802,
+      "grad_norm": 1701.25830078125,
+      "learning_rate": 4.775757575757576e-06,
+      "loss": 289.9675,
+      "step": 5910
+    },
+    {
+      "epoch": 0.04897216362658725,
+      "grad_norm": 1794.075439453125,
+      "learning_rate": 4.783838383838385e-06,
+      "loss": 206.6406,
+      "step": 5920
+    },
+    {
+      "epoch": 0.04905488687595649,
+      "grad_norm": 2225.670654296875,
+      "learning_rate": 4.791919191919192e-06,
+      "loss": 241.3904,
+      "step": 5930
+    },
+    {
+      "epoch": 0.04913761012532572,
+      "grad_norm": 1904.87939453125,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 245.3052,
+      "step": 5940
+    },
+    {
+      "epoch": 0.04922033337469496,
+      "grad_norm": 1126.554931640625,
+      "learning_rate": 4.808080808080808e-06,
+      "loss": 188.594,
+      "step": 5950
+    },
+    {
+      "epoch": 0.049303056624064194,
+      "grad_norm": 691.1046142578125,
+      "learning_rate": 4.816161616161617e-06,
+      "loss": 248.3712,
+      "step": 5960
+    },
+    {
+      "epoch": 0.04938577987343343,
+      "grad_norm": 1169.880859375,
+      "learning_rate": 4.824242424242424e-06,
+      "loss": 245.4409,
+      "step": 5970
+    },
+    {
+      "epoch": 0.04946850312280266,
+      "grad_norm": 1042.1029052734375,
+      "learning_rate": 4.832323232323233e-06,
+      "loss": 202.4587,
+      "step": 5980
+    },
+    {
+      "epoch": 0.0495512263721719,
+      "grad_norm": 1452.813720703125,
+      "learning_rate": 4.840404040404041e-06,
+      "loss": 257.6912,
+      "step": 5990
+    },
+    {
+      "epoch": 0.049633949621541135,
+      "grad_norm": 1247.24609375,
+      "learning_rate": 4.848484848484849e-06,
+      "loss": 210.8361,
+      "step": 6000
+    },
+    {
+      "epoch": 0.04971667287091037,
+      "grad_norm": 1635.0625,
+      "learning_rate": 4.856565656565657e-06,
+      "loss": 241.0676,
+      "step": 6010
+    },
+    {
+      "epoch": 0.0497993961202796,
+      "grad_norm": 1609.6522216796875,
+      "learning_rate": 4.864646464646466e-06,
+      "loss": 188.2872,
+      "step": 6020
+    },
+    {
+      "epoch": 0.04988211936964884,
+      "grad_norm": 1085.433837890625,
+      "learning_rate": 4.872727272727273e-06,
+      "loss": 194.121,
+      "step": 6030
+    },
+    {
+      "epoch": 0.049964842619018075,
+      "grad_norm": 1211.4505615234375,
+      "learning_rate": 4.880808080808081e-06,
+      "loss": 237.9093,
+      "step": 6040
+    },
+    {
+      "epoch": 0.05004756586838731,
+      "grad_norm": 1017.5370483398438,
+      "learning_rate": 4.888888888888889e-06,
+      "loss": 298.4579,
+      "step": 6050
+    },
+    {
+      "epoch": 0.05013028911775654,
+      "grad_norm": 1557.1414794921875,
+      "learning_rate": 4.896969696969697e-06,
+      "loss": 254.7002,
+      "step": 6060
+    },
+    {
+      "epoch": 0.05021301236712578,
+      "grad_norm": 966.5237426757812,
+      "learning_rate": 4.905050505050505e-06,
+      "loss": 218.6844,
+      "step": 6070
+    },
+    {
+      "epoch": 0.050295735616495016,
+      "grad_norm": 1119.049072265625,
+      "learning_rate": 4.913131313131314e-06,
+      "loss": 185.6555,
+      "step": 6080
+    },
+    {
+      "epoch": 0.05037845886586425,
+      "grad_norm": 813.0082397460938,
+      "learning_rate": 4.9212121212121214e-06,
+      "loss": 193.8167,
+      "step": 6090
+    },
+    {
+      "epoch": 0.05046118211523349,
+      "grad_norm": 1165.0540771484375,
+      "learning_rate": 4.92929292929293e-06,
+      "loss": 270.3936,
+      "step": 6100
+    },
+    {
+      "epoch": 0.05054390536460272,
+      "grad_norm": 774.7178344726562,
+      "learning_rate": 4.937373737373738e-06,
+      "loss": 187.8722,
+      "step": 6110
+    },
+    {
+      "epoch": 0.050626628613971957,
+      "grad_norm": 1326.367919921875,
+      "learning_rate": 4.945454545454546e-06,
+      "loss": 243.3256,
+      "step": 6120
+    },
+    {
+      "epoch": 0.05070935186334119,
+      "grad_norm": 1486.3304443359375,
+      "learning_rate": 4.953535353535354e-06,
+      "loss": 186.0064,
+      "step": 6130
+    },
+    {
+      "epoch": 0.05079207511271043,
+      "grad_norm": 1937.40234375,
+      "learning_rate": 4.961616161616162e-06,
+      "loss": 246.7979,
+      "step": 6140
+    },
+    {
+      "epoch": 0.050874798362079664,
+      "grad_norm": 1517.21923828125,
+      "learning_rate": 4.9696969696969696e-06,
+      "loss": 231.4793,
+      "step": 6150
+    },
+    {
+      "epoch": 0.0509575216114489,
+      "grad_norm": 5960.52197265625,
+      "learning_rate": 4.977777777777778e-06,
+      "loss": 197.895,
+      "step": 6160
+    },
+    {
+      "epoch": 0.05104024486081813,
+      "grad_norm": 1942.2510986328125,
+      "learning_rate": 4.9858585858585865e-06,
+      "loss": 300.5423,
+      "step": 6170
+    },
+    {
+      "epoch": 0.05112296811018737,
+      "grad_norm": 1276.1055908203125,
+      "learning_rate": 4.993939393939394e-06,
+      "loss": 288.2531,
+      "step": 6180
+    },
+    {
+      "epoch": 0.051205691359556604,
+      "grad_norm": 2154.565673828125,
+      "learning_rate": 5.0020202020202025e-06,
+      "loss": 259.9872,
+      "step": 6190
+    },
+    {
+      "epoch": 0.05128841460892584,
+      "grad_norm": 961.3928833007812,
+      "learning_rate": 5.010101010101011e-06,
+      "loss": 177.8174,
+      "step": 6200
+    },
+    {
+      "epoch": 0.05137113785829507,
+      "grad_norm": 1438.574462890625,
+      "learning_rate": 5.0181818181818186e-06,
+      "loss": 197.5066,
+      "step": 6210
+    },
+    {
+      "epoch": 0.05145386110766431,
+      "grad_norm": 766.4448852539062,
+      "learning_rate": 5.026262626262627e-06,
+      "loss": 257.3491,
+      "step": 6220
+    },
+    {
+      "epoch": 0.051536584357033545,
+      "grad_norm": 793.5451049804688,
+      "learning_rate": 5.034343434343435e-06,
+      "loss": 184.9512,
+      "step": 6230
+    },
+    {
+      "epoch": 0.05161930760640278,
+      "grad_norm": 869.6885375976562,
+      "learning_rate": 5.042424242424243e-06,
+      "loss": 253.1606,
+      "step": 6240
+    },
+    {
+      "epoch": 0.05170203085577201,
+      "grad_norm": 1229.890869140625,
+      "learning_rate": 5.0505050505050515e-06,
+      "loss": 184.6765,
+      "step": 6250
+    },
+    {
+      "epoch": 0.05178475410514125,
+      "grad_norm": 804.93994140625,
+      "learning_rate": 5.058585858585859e-06,
+      "loss": 162.1275,
+      "step": 6260
+    },
+    {
+      "epoch": 0.051867477354510486,
+      "grad_norm": 1098.615478515625,
+      "learning_rate": 5.0666666666666676e-06,
+      "loss": 174.5565,
+      "step": 6270
+    },
+    {
+      "epoch": 0.05195020060387972,
+      "grad_norm": 570.5386352539062,
+      "learning_rate": 5.074747474747476e-06,
+      "loss": 182.8037,
+      "step": 6280
+    },
+    {
+      "epoch": 0.05203292385324895,
+      "grad_norm": 3363.004150390625,
+      "learning_rate": 5.082828282828284e-06,
+      "loss": 265.7634,
+      "step": 6290
+    },
+    {
+      "epoch": 0.05211564710261819,
+      "grad_norm": 2040.7962646484375,
+      "learning_rate": 5.090909090909091e-06,
+      "loss": 240.1976,
+      "step": 6300
+    },
+    {
+      "epoch": 0.052198370351987426,
+      "grad_norm": 1316.55908203125,
+      "learning_rate": 5.098989898989899e-06,
+      "loss": 249.6734,
+      "step": 6310
+    },
+    {
+      "epoch": 0.05228109360135666,
+      "grad_norm": 2261.989013671875,
+      "learning_rate": 5.107070707070707e-06,
+      "loss": 208.5467,
+      "step": 6320
+    },
+    {
+      "epoch": 0.0523638168507259,
+      "grad_norm": 1246.7120361328125,
+      "learning_rate": 5.115151515151515e-06,
+      "loss": 207.1795,
+      "step": 6330
+    },
+    {
+      "epoch": 0.05244654010009513,
+      "grad_norm": 1235.8924560546875,
+      "learning_rate": 5.123232323232323e-06,
+      "loss": 231.5552,
+      "step": 6340
+    },
+    {
+      "epoch": 0.05252926334946437,
+      "grad_norm": 805.1367797851562,
+      "learning_rate": 5.131313131313132e-06,
+      "loss": 217.3461,
+      "step": 6350
+    },
+    {
+      "epoch": 0.0526119865988336,
+      "grad_norm": 902.8432006835938,
+      "learning_rate": 5.139393939393939e-06,
+      "loss": 250.8667,
+      "step": 6360
+    },
+    {
+      "epoch": 0.05269470984820284,
+      "grad_norm": 1314.75634765625,
+      "learning_rate": 5.147474747474748e-06,
+      "loss": 188.1431,
+      "step": 6370
+    },
+    {
+      "epoch": 0.052777433097572074,
+      "grad_norm": 1282.649169921875,
+      "learning_rate": 5.155555555555556e-06,
+      "loss": 198.1481,
+      "step": 6380
+    },
+    {
+      "epoch": 0.05286015634694131,
+      "grad_norm": 1167.904052734375,
+      "learning_rate": 5.163636363636364e-06,
+      "loss": 192.307,
+      "step": 6390
+    },
+    {
+      "epoch": 0.05294287959631054,
+      "grad_norm": 686.8750610351562,
+      "learning_rate": 5.171717171717172e-06,
+      "loss": 225.3385,
+      "step": 6400
+    },
+    {
+      "epoch": 0.05302560284567978,
+      "grad_norm": 1144.7125244140625,
+      "learning_rate": 5.17979797979798e-06,
+      "loss": 188.2769,
+      "step": 6410
+    },
+    {
+      "epoch": 0.053108326095049015,
+      "grad_norm": 1010.5237426757812,
+      "learning_rate": 5.187878787878788e-06,
+      "loss": 225.7106,
+      "step": 6420
+    },
+    {
+      "epoch": 0.05319104934441825,
+      "grad_norm": 1336.3463134765625,
+      "learning_rate": 5.195959595959597e-06,
+      "loss": 208.7707,
+      "step": 6430
+    },
+    {
+      "epoch": 0.05327377259378748,
+      "grad_norm": 1576.3480224609375,
+      "learning_rate": 5.204040404040404e-06,
+      "loss": 236.1072,
+      "step": 6440
+    },
+    {
+      "epoch": 0.05335649584315672,
+      "grad_norm": 1445.076416015625,
+      "learning_rate": 5.212121212121213e-06,
+      "loss": 271.4749,
+      "step": 6450
+    },
+    {
+      "epoch": 0.053439219092525955,
+      "grad_norm": 895.9810791015625,
+      "learning_rate": 5.220202020202021e-06,
+      "loss": 258.855,
+      "step": 6460
+    },
+    {
+      "epoch": 0.05352194234189519,
+      "grad_norm": 952.8675537109375,
+      "learning_rate": 5.228282828282829e-06,
+      "loss": 175.3387,
+      "step": 6470
+    },
+    {
+      "epoch": 0.05360466559126442,
+      "grad_norm": 1066.5716552734375,
+      "learning_rate": 5.236363636363637e-06,
+      "loss": 218.2292,
+      "step": 6480
+    },
+    {
+      "epoch": 0.05368738884063366,
+      "grad_norm": 849.3695678710938,
+      "learning_rate": 5.244444444444445e-06,
+      "loss": 221.0137,
+      "step": 6490
+    },
+    {
+      "epoch": 0.053770112090002896,
+      "grad_norm": 1675.2149658203125,
+      "learning_rate": 5.252525252525253e-06,
+      "loss": 223.2969,
+      "step": 6500
+    },
+    {
+      "epoch": 0.05385283533937213,
+      "grad_norm": 1693.541748046875,
+      "learning_rate": 5.26060606060606e-06,
+      "loss": 229.6179,
+      "step": 6510
+    },
+    {
+      "epoch": 0.05393555858874136,
+      "grad_norm": 1324.1651611328125,
+      "learning_rate": 5.268686868686869e-06,
+      "loss": 244.8779,
+      "step": 6520
+    },
+    {
+      "epoch": 0.0540182818381106,
+      "grad_norm": 1403.1368408203125,
+      "learning_rate": 5.276767676767677e-06,
+      "loss": 268.8457,
+      "step": 6530
+    },
+    {
+      "epoch": 0.054101005087479836,
+      "grad_norm": 1165.638671875,
+      "learning_rate": 5.284848484848485e-06,
+      "loss": 239.8147,
+      "step": 6540
+    },
+    {
+      "epoch": 0.05418372833684907,
+      "grad_norm": 995.2945556640625,
+      "learning_rate": 5.292929292929293e-06,
+      "loss": 229.9961,
+      "step": 6550
+    },
+    {
+      "epoch": 0.0542664515862183,
+      "grad_norm": 1505.5985107421875,
+      "learning_rate": 5.3010101010101016e-06,
+      "loss": 223.1589,
+      "step": 6560
+    },
+    {
+      "epoch": 0.054349174835587544,
+      "grad_norm": 1667.1300048828125,
+      "learning_rate": 5.309090909090909e-06,
+      "loss": 217.6828,
+      "step": 6570
+    },
+    {
+      "epoch": 0.05443189808495678,
+      "grad_norm": 1639.0364990234375,
+      "learning_rate": 5.317171717171718e-06,
+      "loss": 205.0005,
+      "step": 6580
+    },
+    {
+      "epoch": 0.05451462133432601,
+      "grad_norm": 1414.21240234375,
+      "learning_rate": 5.325252525252525e-06,
+      "loss": 215.9356,
+      "step": 6590
+    },
+    {
+      "epoch": 0.05459734458369525,
+      "grad_norm": 586.2081298828125,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 227.373,
+      "step": 6600
+    },
+    {
+      "epoch": 0.054680067833064484,
+      "grad_norm": 2097.269775390625,
+      "learning_rate": 5.341414141414142e-06,
+      "loss": 234.4245,
+      "step": 6610
+    },
+    {
+      "epoch": 0.05476279108243372,
+      "grad_norm": 1416.9189453125,
+      "learning_rate": 5.34949494949495e-06,
+      "loss": 262.4427,
+      "step": 6620
+    },
+    {
+      "epoch": 0.05484551433180295,
+      "grad_norm": 1144.749755859375,
+      "learning_rate": 5.357575757575758e-06,
+      "loss": 201.3394,
+      "step": 6630
+    },
+    {
+      "epoch": 0.05492823758117219,
+      "grad_norm": 1287.0030517578125,
+      "learning_rate": 5.365656565656567e-06,
+      "loss": 281.6339,
+      "step": 6640
+    },
+    {
+      "epoch": 0.055010960830541425,
+      "grad_norm": 1555.04150390625,
+      "learning_rate": 5.373737373737374e-06,
+      "loss": 225.9727,
+      "step": 6650
+    },
+    {
+      "epoch": 0.05509368407991066,
+      "grad_norm": 1129.1629638671875,
+      "learning_rate": 5.381818181818183e-06,
+      "loss": 159.6996,
+      "step": 6660
+    },
+    {
+      "epoch": 0.05517640732927989,
+      "grad_norm": 1159.3424072265625,
+      "learning_rate": 5.38989898989899e-06,
+      "loss": 226.1631,
+      "step": 6670
+    },
+    {
+      "epoch": 0.05525913057864913,
+      "grad_norm": 1065.8494873046875,
+      "learning_rate": 5.397979797979799e-06,
+      "loss": 213.2366,
+      "step": 6680
+    },
+    {
+      "epoch": 0.055341853828018366,
+      "grad_norm": 801.6089477539062,
+      "learning_rate": 5.406060606060607e-06,
+      "loss": 180.9512,
+      "step": 6690
+    },
+    {
+      "epoch": 0.0554245770773876,
+      "grad_norm": 1254.9515380859375,
+      "learning_rate": 5.414141414141415e-06,
+      "loss": 223.2975,
+      "step": 6700
+    },
+    {
+      "epoch": 0.05550730032675683,
+      "grad_norm": 1734.849609375,
+      "learning_rate": 5.422222222222223e-06,
+      "loss": 179.4333,
+      "step": 6710
+    },
+    {
+      "epoch": 0.05559002357612607,
+      "grad_norm": 1987.9564208984375,
+      "learning_rate": 5.430303030303032e-06,
+      "loss": 248.2462,
+      "step": 6720
+    },
+    {
+      "epoch": 0.055672746825495306,
+      "grad_norm": 1181.2037353515625,
+      "learning_rate": 5.438383838383838e-06,
+      "loss": 182.878,
+      "step": 6730
+    },
+    {
+      "epoch": 0.05575547007486454,
+      "grad_norm": 1901.66015625,
+      "learning_rate": 5.446464646464647e-06,
+      "loss": 251.8006,
+      "step": 6740
+    },
+    {
+      "epoch": 0.05583819332423377,
+      "grad_norm": 632.8333740234375,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 196.6042,
+      "step": 6750
+    },
+    {
+      "epoch": 0.05592091657360301,
+      "grad_norm": 604.5985717773438,
+      "learning_rate": 5.462626262626263e-06,
+      "loss": 177.947,
+      "step": 6760
+    },
+    {
+      "epoch": 0.05600363982297225,
+      "grad_norm": 1062.9954833984375,
+      "learning_rate": 5.4707070707070705e-06,
+      "loss": 312.2176,
+      "step": 6770
+    },
+    {
+      "epoch": 0.05608636307234148,
+      "grad_norm": 1964.1070556640625,
+      "learning_rate": 5.478787878787879e-06,
+      "loss": 192.9068,
+      "step": 6780
+    },
+    {
+      "epoch": 0.056169086321710714,
+      "grad_norm": 1671.50146484375,
+      "learning_rate": 5.486868686868687e-06,
+      "loss": 236.6793,
+      "step": 6790
+    },
+    {
+      "epoch": 0.056251809571079954,
+      "grad_norm": 1363.7489013671875,
+      "learning_rate": 5.494949494949495e-06,
+      "loss": 207.5258,
+      "step": 6800
+    },
+    {
+      "epoch": 0.05633453282044919,
+      "grad_norm": 1217.836669921875,
+      "learning_rate": 5.5030303030303034e-06,
+      "loss": 185.7013,
+      "step": 6810
+    },
+    {
+      "epoch": 0.05641725606981842,
+      "grad_norm": 1585.8988037109375,
+      "learning_rate": 5.511111111111112e-06,
+      "loss": 217.5803,
+      "step": 6820
+    },
+    {
+      "epoch": 0.056499979319187654,
+      "grad_norm": 983.3916625976562,
+      "learning_rate": 5.5191919191919195e-06,
+      "loss": 224.2803,
+      "step": 6830
+    },
+    {
+      "epoch": 0.056582702568556895,
+      "grad_norm": 1328.7664794921875,
+      "learning_rate": 5.527272727272728e-06,
+      "loss": 213.4681,
+      "step": 6840
+    },
+    {
+      "epoch": 0.05666542581792613,
+      "grad_norm": 908.5962524414062,
+      "learning_rate": 5.5353535353535355e-06,
+      "loss": 191.1099,
+      "step": 6850
+    },
+    {
+      "epoch": 0.05674814906729536,
+      "grad_norm": 965.225341796875,
+      "learning_rate": 5.543434343434344e-06,
+      "loss": 180.1947,
+      "step": 6860
+    },
+    {
+      "epoch": 0.0568308723166646,
+      "grad_norm": 1269.7061767578125,
+      "learning_rate": 5.5515151515151524e-06,
+      "loss": 235.1135,
+      "step": 6870
+    },
+    {
+      "epoch": 0.056913595566033835,
+      "grad_norm": 1158.512939453125,
+      "learning_rate": 5.55959595959596e-06,
+      "loss": 215.2743,
+      "step": 6880
+    },
+    {
+      "epoch": 0.05699631881540307,
+      "grad_norm": 1231.1844482421875,
+      "learning_rate": 5.5676767676767685e-06,
+      "loss": 327.6613,
+      "step": 6890
+    },
+    {
+      "epoch": 0.0570790420647723,
+      "grad_norm": 1180.1192626953125,
+      "learning_rate": 5.575757575757577e-06,
+      "loss": 228.0458,
+      "step": 6900
+    },
+    {
+      "epoch": 0.05716176531414154,
+      "grad_norm": 952.682861328125,
+      "learning_rate": 5.5838383838383845e-06,
+      "loss": 250.6257,
+      "step": 6910
+    },
+    {
+      "epoch": 0.057244488563510776,
+      "grad_norm": 16013.0146484375,
+      "learning_rate": 5.591919191919193e-06,
+      "loss": 250.4285,
+      "step": 6920
+    },
+    {
+      "epoch": 0.05732721181288001,
+      "grad_norm": 1266.31787109375,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 213.8897,
+      "step": 6930
+    },
+    {
+      "epoch": 0.05740993506224924,
+      "grad_norm": 1832.4088134765625,
+      "learning_rate": 5.608080808080808e-06,
+      "loss": 218.5001,
+      "step": 6940
+    },
+    {
+      "epoch": 0.05749265831161848,
+      "grad_norm": 1213.3984375,
+      "learning_rate": 5.616161616161616e-06,
+      "loss": 211.7139,
+      "step": 6950
+    },
+    {
+      "epoch": 0.057575381560987716,
+      "grad_norm": 1369.9437255859375,
+      "learning_rate": 5.624242424242424e-06,
+      "loss": 192.9413,
+      "step": 6960
+    },
+    {
+      "epoch": 0.05765810481035695,
+      "grad_norm": 1018.5819091796875,
+      "learning_rate": 5.632323232323233e-06,
+      "loss": 150.2073,
+      "step": 6970
+    },
+    {
+      "epoch": 0.05774082805972618,
+      "grad_norm": 1435.1353759765625,
+      "learning_rate": 5.64040404040404e-06,
+      "loss": 187.5981,
+      "step": 6980
+    },
+    {
+      "epoch": 0.057823551309095424,
+      "grad_norm": 1339.41650390625,
+      "learning_rate": 5.648484848484849e-06,
+      "loss": 165.7803,
+      "step": 6990
+    },
+    {
+      "epoch": 0.05790627455846466,
+      "grad_norm": 3150.48974609375,
+      "learning_rate": 5.656565656565657e-06,
+      "loss": 183.762,
+      "step": 7000
+    },
+    {
+      "epoch": 0.05798899780783389,
+      "grad_norm": 1491.4512939453125,
+      "learning_rate": 5.664646464646465e-06,
+      "loss": 205.812,
+      "step": 7010
+    },
+    {
+      "epoch": 0.058071721057203124,
+      "grad_norm": 1045.4873046875,
+      "learning_rate": 5.672727272727273e-06,
+      "loss": 222.1431,
+      "step": 7020
+    },
+    {
+      "epoch": 0.058154444306572364,
+      "grad_norm": 1604.70556640625,
+      "learning_rate": 5.680808080808081e-06,
+      "loss": 312.9387,
+      "step": 7030
+    },
+    {
+      "epoch": 0.0582371675559416,
+      "grad_norm": 794.0164184570312,
+      "learning_rate": 5.688888888888889e-06,
+      "loss": 148.5473,
+      "step": 7040
+    },
+    {
+      "epoch": 0.05831989080531083,
+      "grad_norm": 1710.36328125,
+      "learning_rate": 5.696969696969698e-06,
+      "loss": 187.5921,
+      "step": 7050
+    },
+    {
+      "epoch": 0.058402614054680065,
+      "grad_norm": 1974.2613525390625,
+      "learning_rate": 5.705050505050505e-06,
+      "loss": 192.2352,
+      "step": 7060
+    },
+    {
+      "epoch": 0.058485337304049305,
+      "grad_norm": 1533.34619140625,
+      "learning_rate": 5.713131313131314e-06,
+      "loss": 189.592,
+      "step": 7070
+    },
+    {
+      "epoch": 0.05856806055341854,
+      "grad_norm": 1127.27392578125,
+      "learning_rate": 5.721212121212122e-06,
+      "loss": 211.6275,
+      "step": 7080
+    },
+    {
+      "epoch": 0.05865078380278777,
+      "grad_norm": 1902.27001953125,
+      "learning_rate": 5.72929292929293e-06,
+      "loss": 189.9227,
+      "step": 7090
+    },
+    {
+      "epoch": 0.05873350705215701,
+      "grad_norm": 1564.533203125,
+      "learning_rate": 5.737373737373738e-06,
+      "loss": 231.5597,
+      "step": 7100
+    },
+    {
+      "epoch": 0.058816230301526246,
+      "grad_norm": 2943.590576171875,
+      "learning_rate": 5.745454545454546e-06,
+      "loss": 177.4852,
+      "step": 7110
+    },
+    {
+      "epoch": 0.05889895355089548,
+      "grad_norm": 1309.1622314453125,
+      "learning_rate": 5.753535353535354e-06,
+      "loss": 165.6322,
+      "step": 7120
+    },
+    {
+      "epoch": 0.05898167680026471,
+      "grad_norm": 1501.813720703125,
+      "learning_rate": 5.761616161616163e-06,
+      "loss": 189.5367,
+      "step": 7130
+    },
+    {
+      "epoch": 0.05906440004963395,
+      "grad_norm": 1574.177978515625,
+      "learning_rate": 5.76969696969697e-06,
+      "loss": 234.3292,
+      "step": 7140
+    },
+    {
+      "epoch": 0.059147123299003186,
+      "grad_norm": 1052.1170654296875,
+      "learning_rate": 5.777777777777778e-06,
+      "loss": 214.4601,
+      "step": 7150
+    },
+    {
+      "epoch": 0.05922984654837242,
+      "grad_norm": 1018.6898803710938,
+      "learning_rate": 5.785858585858586e-06,
+      "loss": 156.9678,
+      "step": 7160
+    },
+    {
+      "epoch": 0.05931256979774165,
+      "grad_norm": 1917.61865234375,
+      "learning_rate": 5.793939393939394e-06,
+      "loss": 233.6657,
+      "step": 7170
+    },
+    {
+      "epoch": 0.05939529304711089,
+      "grad_norm": 1654.3441162109375,
+      "learning_rate": 5.8020202020202025e-06,
+      "loss": 173.1022,
+      "step": 7180
+    },
+    {
+      "epoch": 0.05947801629648013,
+      "grad_norm": 1147.4951171875,
+      "learning_rate": 5.81010101010101e-06,
+      "loss": 209.7125,
+      "step": 7190
+    },
+    {
+      "epoch": 0.05956073954584936,
+      "grad_norm": 1505.357177734375,
+      "learning_rate": 5.8181818181818185e-06,
+      "loss": 239.9892,
+      "step": 7200
+    },
+    {
+      "epoch": 0.059643462795218594,
+      "grad_norm": 3524.7958984375,
+      "learning_rate": 5.826262626262626e-06,
+      "loss": 199.2161,
+      "step": 7210
+    },
+    {
+      "epoch": 0.059726186044587834,
+      "grad_norm": 2917.4033203125,
+      "learning_rate": 5.834343434343435e-06,
+      "loss": 282.2665,
+      "step": 7220
+    },
+    {
+      "epoch": 0.05980890929395707,
+      "grad_norm": 976.106689453125,
+      "learning_rate": 5.842424242424243e-06,
+      "loss": 220.0685,
+      "step": 7230
+    },
+    {
+      "epoch": 0.0598916325433263,
+      "grad_norm": 1532.435791015625,
+      "learning_rate": 5.850505050505051e-06,
+      "loss": 247.5289,
+      "step": 7240
+    },
+    {
+      "epoch": 0.059974355792695534,
+      "grad_norm": 868.5018920898438,
+      "learning_rate": 5.858585858585859e-06,
+      "loss": 204.2799,
+      "step": 7250
+    },
+    {
+      "epoch": 0.060057079042064775,
+      "grad_norm": 1144.6495361328125,
+      "learning_rate": 5.8666666666666675e-06,
+      "loss": 233.1151,
+      "step": 7260
+    },
+    {
+      "epoch": 0.06013980229143401,
+      "grad_norm": 1262.310302734375,
+      "learning_rate": 5.874747474747475e-06,
+      "loss": 215.0127,
+      "step": 7270
+    },
+    {
+      "epoch": 0.06022252554080324,
+      "grad_norm": 1465.82568359375,
+      "learning_rate": 5.882828282828284e-06,
+      "loss": 214.6379,
+      "step": 7280
+    },
+    {
+      "epoch": 0.060305248790172475,
+      "grad_norm": 890.1478271484375,
+      "learning_rate": 5.890909090909091e-06,
+      "loss": 212.3409,
+      "step": 7290
+    },
+    {
+      "epoch": 0.060387972039541715,
+      "grad_norm": 804.8695068359375,
+      "learning_rate": 5.8989898989899e-06,
+      "loss": 161.0226,
+      "step": 7300
+    },
+    {
+      "epoch": 0.06047069528891095,
+      "grad_norm": 1049.8092041015625,
+      "learning_rate": 5.907070707070708e-06,
+      "loss": 228.0541,
+      "step": 7310
+    },
+    {
+      "epoch": 0.06055341853828018,
+      "grad_norm": 1314.708984375,
+      "learning_rate": 5.915151515151516e-06,
+      "loss": 169.0287,
+      "step": 7320
+    },
+    {
+      "epoch": 0.060636141787649415,
+      "grad_norm": 877.56982421875,
+      "learning_rate": 5.923232323232324e-06,
+      "loss": 182.722,
+      "step": 7330
+    },
+    {
+      "epoch": 0.060718865037018656,
+      "grad_norm": 1187.692138671875,
+      "learning_rate": 5.9313131313131326e-06,
+      "loss": 216.9126,
+      "step": 7340
+    },
+    {
+      "epoch": 0.06080158828638789,
+      "grad_norm": 910.7957763671875,
+      "learning_rate": 5.93939393939394e-06,
+      "loss": 156.4217,
+      "step": 7350
+    },
+    {
+      "epoch": 0.06088431153575712,
+      "grad_norm": 937.8931274414062,
+      "learning_rate": 5.947474747474749e-06,
+      "loss": 206.2811,
+      "step": 7360
+    },
+    {
+      "epoch": 0.06096703478512636,
+      "grad_norm": 1068.356201171875,
+      "learning_rate": 5.955555555555555e-06,
+      "loss": 209.255,
+      "step": 7370
+    },
+    {
+      "epoch": 0.061049758034495596,
+      "grad_norm": 1415.2979736328125,
+      "learning_rate": 5.963636363636364e-06,
+      "loss": 180.4899,
+      "step": 7380
+    },
+    {
+      "epoch": 0.06113248128386483,
+      "grad_norm": 1216.490966796875,
+      "learning_rate": 5.9717171717171714e-06,
+      "loss": 181.2195,
+      "step": 7390
+    },
+    {
+      "epoch": 0.06121520453323406,
+      "grad_norm": 1969.770751953125,
+      "learning_rate": 5.97979797979798e-06,
+      "loss": 185.2952,
+      "step": 7400
+    },
+    {
+      "epoch": 0.061297927782603304,
+      "grad_norm": 1073.1572265625,
+      "learning_rate": 5.987878787878788e-06,
+      "loss": 194.6978,
+      "step": 7410
+    },
+    {
+      "epoch": 0.06138065103197254,
+      "grad_norm": 1329.2269287109375,
+      "learning_rate": 5.995959595959596e-06,
+      "loss": 225.0991,
+      "step": 7420
+    },
+    {
+      "epoch": 0.06146337428134177,
+      "grad_norm": 1447.0704345703125,
+      "learning_rate": 6.004040404040404e-06,
+      "loss": 222.2881,
+      "step": 7430
+    },
+    {
+      "epoch": 0.061546097530711004,
+      "grad_norm": 1847.17578125,
+      "learning_rate": 6.012121212121213e-06,
+      "loss": 250.8386,
+      "step": 7440
+    },
+    {
+      "epoch": 0.061628820780080244,
+      "grad_norm": 1011.4767456054688,
+      "learning_rate": 6.0202020202020204e-06,
+      "loss": 217.8489,
+      "step": 7450
+    },
+    {
+      "epoch": 0.06171154402944948,
+      "grad_norm": 905.0313110351562,
+      "learning_rate": 6.028282828282829e-06,
+      "loss": 176.8829,
+      "step": 7460
+    },
+    {
+      "epoch": 0.06179426727881871,
+      "grad_norm": 1627.3341064453125,
+      "learning_rate": 6.0363636363636365e-06,
+      "loss": 245.858,
+      "step": 7470
+    },
+    {
+      "epoch": 0.061876990528187945,
+      "grad_norm": 1333.339111328125,
+      "learning_rate": 6.044444444444445e-06,
+      "loss": 172.512,
+      "step": 7480
+    },
+    {
+      "epoch": 0.061959713777557185,
+      "grad_norm": 981.724609375,
+      "learning_rate": 6.052525252525253e-06,
+      "loss": 188.6218,
+      "step": 7490
+    },
+    {
+      "epoch": 0.06204243702692642,
+      "grad_norm": 963.2952880859375,
+      "learning_rate": 6.060606060606061e-06,
+      "loss": 200.9197,
+      "step": 7500
+    },
+    {
+      "epoch": 0.06212516027629565,
+      "grad_norm": 1976.6124267578125,
+      "learning_rate": 6.068686868686869e-06,
+      "loss": 206.6058,
+      "step": 7510
+    },
+    {
+      "epoch": 0.062207883525664885,
+      "grad_norm": 964.7189331054688,
+      "learning_rate": 6.076767676767678e-06,
+      "loss": 244.5135,
+      "step": 7520
+    },
+    {
+      "epoch": 0.062290606775034126,
+      "grad_norm": 940.2135009765625,
+      "learning_rate": 6.0848484848484855e-06,
+      "loss": 286.4893,
+      "step": 7530
+    },
+    {
+      "epoch": 0.06237333002440336,
+      "grad_norm": 1086.095947265625,
+      "learning_rate": 6.092929292929294e-06,
+      "loss": 166.7001,
+      "step": 7540
+    },
+    {
+      "epoch": 0.06245605327377259,
+      "grad_norm": 824.2728881835938,
+      "learning_rate": 6.1010101010101015e-06,
+      "loss": 178.9928,
+      "step": 7550
+    },
+    {
+      "epoch": 0.06253877652314183,
+      "grad_norm": 1227.130859375,
+      "learning_rate": 6.10909090909091e-06,
+      "loss": 188.424,
+      "step": 7560
+    },
+    {
+      "epoch": 0.06262149977251107,
+      "grad_norm": 943.979736328125,
+      "learning_rate": 6.117171717171718e-06,
+      "loss": 170.1881,
+      "step": 7570
+    },
+    {
+      "epoch": 0.06270422302188029,
+      "grad_norm": 893.2919921875,
+      "learning_rate": 6.125252525252525e-06,
+      "loss": 203.6881,
+      "step": 7580
+    },
+    {
+      "epoch": 0.06278694627124953,
+      "grad_norm": 1502.2542724609375,
+      "learning_rate": 6.133333333333334e-06,
+      "loss": 227.4439,
+      "step": 7590
+    },
+    {
+      "epoch": 0.06286966952061877,
+      "grad_norm": 2379.823974609375,
+      "learning_rate": 6.141414141414141e-06,
+      "loss": 216.3211,
+      "step": 7600
+    },
+    {
+      "epoch": 0.062952392769988,
+      "grad_norm": 2946.320068359375,
+      "learning_rate": 6.14949494949495e-06,
+      "loss": 181.0182,
+      "step": 7610
+    },
+    {
+      "epoch": 0.06303511601935724,
+      "grad_norm": 1127.286865234375,
+      "learning_rate": 6.157575757575758e-06,
+      "loss": 219.9298,
+      "step": 7620
+    },
+    {
+      "epoch": 0.06311783926872648,
+      "grad_norm": 2195.520751953125,
+      "learning_rate": 6.165656565656566e-06,
+      "loss": 225.9704,
+      "step": 7630
+    },
+    {
+      "epoch": 0.06320056251809571,
+      "grad_norm": 1115.85107421875,
+      "learning_rate": 6.173737373737374e-06,
+      "loss": 192.5431,
+      "step": 7640
+    },
+    {
+      "epoch": 0.06328328576746495,
+      "grad_norm": 1610.82861328125,
+      "learning_rate": 6.181818181818182e-06,
+      "loss": 234.5798,
+      "step": 7650
+    },
+    {
+      "epoch": 0.06336600901683419,
+      "grad_norm": 2228.57080078125,
+      "learning_rate": 6.18989898989899e-06,
+      "loss": 203.8422,
+      "step": 7660
+    },
+    {
+      "epoch": 0.06344873226620341,
+      "grad_norm": 1072.0853271484375,
+      "learning_rate": 6.197979797979799e-06,
+      "loss": 136.4943,
+      "step": 7670
+    },
+    {
+      "epoch": 0.06353145551557265,
+      "grad_norm": 987.9502563476562,
+      "learning_rate": 6.206060606060606e-06,
+      "loss": 176.214,
+      "step": 7680
+    },
+    {
+      "epoch": 0.06361417876494188,
+      "grad_norm": 1846.4237060546875,
+      "learning_rate": 6.214141414141415e-06,
+      "loss": 328.0838,
+      "step": 7690
+    },
+    {
+      "epoch": 0.06369690201431112,
+      "grad_norm": 1142.6070556640625,
+      "learning_rate": 6.222222222222223e-06,
+      "loss": 197.7707,
+      "step": 7700
+    },
+    {
+      "epoch": 0.06377962526368036,
+      "grad_norm": 2267.01904296875,
+      "learning_rate": 6.230303030303031e-06,
+      "loss": 246.8396,
+      "step": 7710
+    },
+    {
+      "epoch": 0.06386234851304959,
+      "grad_norm": 1803.31591796875,
+      "learning_rate": 6.238383838383839e-06,
+      "loss": 156.6524,
+      "step": 7720
+    },
+    {
+      "epoch": 0.06394507176241883,
+      "grad_norm": 1524.9769287109375,
+      "learning_rate": 6.246464646464647e-06,
+      "loss": 252.2839,
+      "step": 7730
+    },
+    {
+      "epoch": 0.06402779501178807,
+      "grad_norm": 1911.2991943359375,
+      "learning_rate": 6.254545454545455e-06,
+      "loss": 185.8043,
+      "step": 7740
+    },
+    {
+      "epoch": 0.0641105182611573,
+      "grad_norm": 1456.3634033203125,
+      "learning_rate": 6.262626262626264e-06,
+      "loss": 189.6922,
+      "step": 7750
+    },
+    {
+      "epoch": 0.06419324151052654,
+      "grad_norm": 872.6129150390625,
+      "learning_rate": 6.270707070707071e-06,
+      "loss": 211.5073,
+      "step": 7760
+    },
+    {
+      "epoch": 0.06427596475989576,
+      "grad_norm": 2161.432861328125,
+      "learning_rate": 6.27878787878788e-06,
+      "loss": 227.991,
+      "step": 7770
+    },
+    {
+      "epoch": 0.064358688009265,
+      "grad_norm": 1427.72900390625,
+      "learning_rate": 6.286868686868688e-06,
+      "loss": 187.7627,
+      "step": 7780
+    },
+    {
+      "epoch": 0.06444141125863424,
+      "grad_norm": 1048.0509033203125,
+      "learning_rate": 6.294949494949495e-06,
+      "loss": 216.9241,
+      "step": 7790
+    },
+    {
+      "epoch": 0.06452413450800347,
+      "grad_norm": 831.1154174804688,
+      "learning_rate": 6.303030303030303e-06,
+      "loss": 156.4739,
+      "step": 7800
+    },
+    {
+      "epoch": 0.06460685775737271,
+      "grad_norm": 1421.5413818359375,
+      "learning_rate": 6.311111111111111e-06,
+      "loss": 240.7841,
+      "step": 7810
+    },
+    {
+      "epoch": 0.06468958100674195,
+      "grad_norm": 1609.0704345703125,
+      "learning_rate": 6.3191919191919195e-06,
+      "loss": 202.0292,
+      "step": 7820
+    },
+    {
+      "epoch": 0.06477230425611118,
+      "grad_norm": 1056.16064453125,
+      "learning_rate": 6.327272727272727e-06,
+      "loss": 150.6357,
+      "step": 7830
+    },
+    {
+      "epoch": 0.06485502750548042,
+      "grad_norm": 3895.295166015625,
+      "learning_rate": 6.3353535353535355e-06,
+      "loss": 166.0538,
+      "step": 7840
+    },
+    {
+      "epoch": 0.06493775075484966,
+      "grad_norm": 1299.444580078125,
+      "learning_rate": 6.343434343434344e-06,
+      "loss": 160.5584,
+      "step": 7850
+    },
+    {
+      "epoch": 0.06502047400421888,
+      "grad_norm": 712.9761352539062,
+      "learning_rate": 6.3515151515151516e-06,
+      "loss": 197.3979,
+      "step": 7860
+    },
+    {
+      "epoch": 0.06510319725358812,
+      "grad_norm": 1037.716064453125,
+      "learning_rate": 6.35959595959596e-06,
+      "loss": 200.6918,
+      "step": 7870
+    },
+    {
+      "epoch": 0.06518592050295735,
+      "grad_norm": 1494.057373046875,
+      "learning_rate": 6.3676767676767685e-06,
+      "loss": 238.1615,
+      "step": 7880
+    },
+    {
+      "epoch": 0.06526864375232659,
+      "grad_norm": 1253.9837646484375,
+      "learning_rate": 6.375757575757576e-06,
+      "loss": 190.7486,
+      "step": 7890
+    },
+    {
+      "epoch": 0.06535136700169583,
+      "grad_norm": 1362.12353515625,
+      "learning_rate": 6.3838383838383845e-06,
+      "loss": 214.7477,
+      "step": 7900
+    },
+    {
+      "epoch": 0.06543409025106506,
+      "grad_norm": 890.4193725585938,
+      "learning_rate": 6.391919191919192e-06,
+      "loss": 170.1652,
+      "step": 7910
+    },
+    {
+      "epoch": 0.0655168135004343,
+      "grad_norm": 1644.9490966796875,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 175.0597,
+      "step": 7920
+    },
+    {
+      "epoch": 0.06559953674980354,
+      "grad_norm": 777.9044799804688,
+      "learning_rate": 6.408080808080809e-06,
+      "loss": 238.5318,
+      "step": 7930
+    },
+    {
+      "epoch": 0.06568225999917277,
+      "grad_norm": 1498.41455078125,
+      "learning_rate": 6.416161616161617e-06,
+      "loss": 183.5011,
+      "step": 7940
+    },
+    {
+      "epoch": 0.065764983248542,
+      "grad_norm": 1392.91748046875,
+      "learning_rate": 6.424242424242425e-06,
+      "loss": 164.5937,
+      "step": 7950
+    },
+    {
+      "epoch": 0.06584770649791123,
+      "grad_norm": 1324.7021484375,
+      "learning_rate": 6.4323232323232335e-06,
+      "loss": 230.2011,
+      "step": 7960
+    },
+    {
+      "epoch": 0.06593042974728047,
+      "grad_norm": 1371.1768798828125,
+      "learning_rate": 6.440404040404041e-06,
+      "loss": 226.2801,
+      "step": 7970
+    },
+    {
+      "epoch": 0.06601315299664971,
+      "grad_norm": 2229.455078125,
+      "learning_rate": 6.4484848484848496e-06,
+      "loss": 196.2276,
+      "step": 7980
+    },
+    {
+      "epoch": 0.06609587624601894,
+      "grad_norm": 1298.6649169921875,
+      "learning_rate": 6.456565656565658e-06,
+      "loss": 211.2707,
+      "step": 7990
+    },
+    {
+      "epoch": 0.06617859949538818,
+      "grad_norm": 784.8013305664062,
+      "learning_rate": 6.464646464646466e-06,
+      "loss": 207.776,
+      "step": 8000
+    },
+    {
+      "epoch": 0.06626132274475742,
+      "grad_norm": 911.982666015625,
+      "learning_rate": 6.472727272727272e-06,
+      "loss": 162.3888,
+      "step": 8010
+    },
+    {
+      "epoch": 0.06634404599412665,
+      "grad_norm": 1066.7635498046875,
+      "learning_rate": 6.480808080808081e-06,
+      "loss": 182.5109,
+      "step": 8020
+    },
+    {
+      "epoch": 0.06642676924349589,
+      "grad_norm": 733.5493774414062,
+      "learning_rate": 6.488888888888889e-06,
+      "loss": 140.7046,
+      "step": 8030
+    },
+    {
+      "epoch": 0.06650949249286511,
+      "grad_norm": 1074.85888671875,
+      "learning_rate": 6.496969696969697e-06,
+      "loss": 182.7351,
+      "step": 8040
+    },
+    {
+      "epoch": 0.06659221574223435,
+      "grad_norm": 1245.099365234375,
+      "learning_rate": 6.505050505050505e-06,
+      "loss": 228.4951,
+      "step": 8050
+    },
+    {
+      "epoch": 0.0666749389916036,
+      "grad_norm": 820.326171875,
+      "learning_rate": 6.513131313131314e-06,
+      "loss": 172.91,
+      "step": 8060
+    },
+    {
+      "epoch": 0.06675766224097282,
+      "grad_norm": 1347.561279296875,
+      "learning_rate": 6.521212121212121e-06,
+      "loss": 175.6294,
+      "step": 8070
+    },
+    {
+      "epoch": 0.06684038549034206,
+      "grad_norm": 2743.030517578125,
+      "learning_rate": 6.52929292929293e-06,
+      "loss": 240.5082,
+      "step": 8080
+    },
+    {
+      "epoch": 0.0669231087397113,
+      "grad_norm": 1014.7551879882812,
+      "learning_rate": 6.537373737373737e-06,
+      "loss": 181.1287,
+      "step": 8090
+    },
+    {
+      "epoch": 0.06700583198908053,
+      "grad_norm": 1014.5245971679688,
+      "learning_rate": 6.545454545454546e-06,
+      "loss": 189.7482,
+      "step": 8100
+    },
+    {
+      "epoch": 0.06708855523844977,
+      "grad_norm": 778.7886962890625,
+      "learning_rate": 6.553535353535354e-06,
+      "loss": 170.7764,
+      "step": 8110
+    },
+    {
+      "epoch": 0.06717127848781901,
+      "grad_norm": 2081.901611328125,
+      "learning_rate": 6.561616161616162e-06,
+      "loss": 245.8872,
+      "step": 8120
+    },
+    {
+      "epoch": 0.06725400173718823,
+      "grad_norm": 1881.017333984375,
+      "learning_rate": 6.56969696969697e-06,
+      "loss": 252.341,
+      "step": 8130
+    },
+    {
+      "epoch": 0.06733672498655748,
+      "grad_norm": 910.3214111328125,
+      "learning_rate": 6.577777777777779e-06,
+      "loss": 179.5906,
+      "step": 8140
+    },
+    {
+      "epoch": 0.0674194482359267,
+      "grad_norm": 1163.5965576171875,
+      "learning_rate": 6.585858585858586e-06,
+      "loss": 233.8698,
+      "step": 8150
+    },
+    {
+      "epoch": 0.06750217148529594,
+      "grad_norm": 884.6065673828125,
+      "learning_rate": 6.593939393939395e-06,
+      "loss": 139.5692,
+      "step": 8160
+    },
+    {
+      "epoch": 0.06758489473466518,
+      "grad_norm": 1140.02392578125,
+      "learning_rate": 6.602020202020203e-06,
+      "loss": 195.1031,
+      "step": 8170
+    },
+    {
+      "epoch": 0.06766761798403441,
+      "grad_norm": 1378.050537109375,
+      "learning_rate": 6.610101010101011e-06,
+      "loss": 215.5058,
+      "step": 8180
+    },
+    {
+      "epoch": 0.06775034123340365,
+      "grad_norm": 1974.1138916015625,
+      "learning_rate": 6.618181818181819e-06,
+      "loss": 294.2934,
+      "step": 8190
+    },
+    {
+      "epoch": 0.06783306448277289,
+      "grad_norm": 832.8023681640625,
+      "learning_rate": 6.626262626262627e-06,
+      "loss": 184.4631,
+      "step": 8200
+    },
+    {
+      "epoch": 0.06791578773214212,
+      "grad_norm": 856.9111938476562,
+      "learning_rate": 6.634343434343435e-06,
+      "loss": 170.4937,
+      "step": 8210
+    },
+    {
+      "epoch": 0.06799851098151136,
+      "grad_norm": 916.113525390625,
+      "learning_rate": 6.642424242424242e-06,
+      "loss": 170.3063,
+      "step": 8220
+    },
+    {
+      "epoch": 0.06808123423088058,
+      "grad_norm": 1267.204345703125,
+      "learning_rate": 6.650505050505051e-06,
+      "loss": 154.5532,
+      "step": 8230
+    },
+    {
+      "epoch": 0.06816395748024982,
+      "grad_norm": 1163.3558349609375,
+      "learning_rate": 6.658585858585859e-06,
+      "loss": 196.9229,
+      "step": 8240
+    },
+    {
+      "epoch": 0.06824668072961906,
+      "grad_norm": 1783.79541015625,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 210.5392,
+      "step": 8250
+    },
+    {
+      "epoch": 0.06832940397898829,
+      "grad_norm": 917.7781982421875,
+      "learning_rate": 6.674747474747475e-06,
+      "loss": 177.4379,
+      "step": 8260
+    },
+    {
+      "epoch": 0.06841212722835753,
+      "grad_norm": 674.7391967773438,
+      "learning_rate": 6.682828282828283e-06,
+      "loss": 186.2342,
+      "step": 8270
+    },
+    {
+      "epoch": 0.06849485047772677,
+      "grad_norm": 654.15234375,
+      "learning_rate": 6.690909090909091e-06,
+      "loss": 136.4612,
+      "step": 8280
+    },
+    {
+      "epoch": 0.068577573727096,
+      "grad_norm": 1094.0560302734375,
+      "learning_rate": 6.6989898989899e-06,
+      "loss": 170.5708,
+      "step": 8290
+    },
+    {
+      "epoch": 0.06866029697646524,
+      "grad_norm": 1474.585693359375,
+      "learning_rate": 6.707070707070707e-06,
+      "loss": 203.4748,
+      "step": 8300
+    },
+    {
+      "epoch": 0.06874302022583446,
+      "grad_norm": 1227.391845703125,
+      "learning_rate": 6.715151515151516e-06,
+      "loss": 222.0174,
+      "step": 8310
+    },
+    {
+      "epoch": 0.0688257434752037,
+      "grad_norm": 1838.2691650390625,
+      "learning_rate": 6.723232323232324e-06,
+      "loss": 224.6444,
+      "step": 8320
+    },
+    {
+      "epoch": 0.06890846672457294,
+      "grad_norm": 1169.060791015625,
+      "learning_rate": 6.731313131313132e-06,
+      "loss": 142.1164,
+      "step": 8330
+    },
+    {
+      "epoch": 0.06899118997394217,
+      "grad_norm": 547.6967163085938,
+      "learning_rate": 6.73939393939394e-06,
+      "loss": 153.1282,
+      "step": 8340
+    },
+    {
+      "epoch": 0.06907391322331141,
+      "grad_norm": 1083.063720703125,
+      "learning_rate": 6.747474747474749e-06,
+      "loss": 184.0296,
+      "step": 8350
+    },
+    {
+      "epoch": 0.06915663647268065,
+      "grad_norm": 5596.943359375,
+      "learning_rate": 6.755555555555556e-06,
+      "loss": 173.6666,
+      "step": 8360
+    },
+    {
+      "epoch": 0.06923935972204988,
+      "grad_norm": 1326.8721923828125,
+      "learning_rate": 6.763636363636365e-06,
+      "loss": 190.1673,
+      "step": 8370
+    },
+    {
+      "epoch": 0.06932208297141912,
+      "grad_norm": 1101.3780517578125,
+      "learning_rate": 6.771717171717172e-06,
+      "loss": 220.6135,
+      "step": 8380
+    },
+    {
+      "epoch": 0.06940480622078836,
+      "grad_norm": 1435.279296875,
+      "learning_rate": 6.779797979797981e-06,
+      "loss": 178.5454,
+      "step": 8390
+    },
+    {
+      "epoch": 0.06948752947015759,
+      "grad_norm": 1620.093994140625,
+      "learning_rate": 6.787878787878789e-06,
+      "loss": 151.146,
+      "step": 8400
+    },
+    {
+      "epoch": 0.06957025271952683,
+      "grad_norm": 1524.4627685546875,
+      "learning_rate": 6.795959595959597e-06,
+      "loss": 185.1478,
+      "step": 8410
+    },
+    {
+      "epoch": 0.06965297596889605,
+      "grad_norm": 1806.7978515625,
+      "learning_rate": 6.804040404040405e-06,
+      "loss": 173.0628,
+      "step": 8420
+    },
+    {
+      "epoch": 0.06973569921826529,
+      "grad_norm": 1826.8575439453125,
+      "learning_rate": 6.812121212121212e-06,
+      "loss": 167.8523,
+      "step": 8430
+    },
+    {
+      "epoch": 0.06981842246763453,
+      "grad_norm": 1353.9345703125,
+      "learning_rate": 6.82020202020202e-06,
+      "loss": 194.5669,
+      "step": 8440
+    },
+    {
+      "epoch": 0.06990114571700376,
+      "grad_norm": 2063.500244140625,
+      "learning_rate": 6.828282828282828e-06,
+      "loss": 184.4123,
+      "step": 8450
+    },
+    {
+      "epoch": 0.069983868966373,
+      "grad_norm": 1244.160400390625,
+      "learning_rate": 6.8363636363636364e-06,
+      "loss": 234.5416,
+      "step": 8460
+    },
+    {
+      "epoch": 0.07006659221574224,
+      "grad_norm": 894.0686645507812,
+      "learning_rate": 6.844444444444445e-06,
+      "loss": 157.8413,
+      "step": 8470
+    },
+    {
+      "epoch": 0.07014931546511147,
+      "grad_norm": 953.1651000976562,
+      "learning_rate": 6.8525252525252525e-06,
+      "loss": 187.5382,
+      "step": 8480
+    },
+    {
+      "epoch": 0.07023203871448071,
+      "grad_norm": 868.2008056640625,
+      "learning_rate": 6.860606060606061e-06,
+      "loss": 153.3808,
+      "step": 8490
+    },
+    {
+      "epoch": 0.07031476196384993,
+      "grad_norm": 1246.61328125,
+      "learning_rate": 6.868686868686869e-06,
+      "loss": 198.8839,
+      "step": 8500
+    },
+    {
+      "epoch": 0.07039748521321917,
+      "grad_norm": 1278.021484375,
+      "learning_rate": 6.876767676767677e-06,
+      "loss": 162.5859,
+      "step": 8510
+    },
+    {
+      "epoch": 0.07048020846258841,
+      "grad_norm": 1224.3677978515625,
+      "learning_rate": 6.8848484848484854e-06,
+      "loss": 216.0751,
+      "step": 8520
+    },
+    {
+      "epoch": 0.07056293171195764,
+      "grad_norm": 629.5072021484375,
+      "learning_rate": 6.892929292929294e-06,
+      "loss": 219.3646,
+      "step": 8530
+    },
+    {
+      "epoch": 0.07064565496132688,
+      "grad_norm": 873.804443359375,
+      "learning_rate": 6.9010101010101015e-06,
+      "loss": 176.4796,
+      "step": 8540
+    },
+    {
+      "epoch": 0.07072837821069612,
+      "grad_norm": 1832.682373046875,
+      "learning_rate": 6.90909090909091e-06,
+      "loss": 256.3848,
+      "step": 8550
+    },
+    {
+      "epoch": 0.07081110146006535,
+      "grad_norm": 1244.7239990234375,
+      "learning_rate": 6.9171717171717175e-06,
+      "loss": 215.3179,
+      "step": 8560
+    },
+    {
+      "epoch": 0.07089382470943459,
+      "grad_norm": 938.3501586914062,
+      "learning_rate": 6.925252525252526e-06,
+      "loss": 169.5242,
+      "step": 8570
+    },
+    {
+      "epoch": 0.07097654795880382,
+      "grad_norm": 2076.439697265625,
+      "learning_rate": 6.9333333333333344e-06,
+      "loss": 200.3409,
+      "step": 8580
+    },
+    {
+      "epoch": 0.07105927120817306,
+      "grad_norm": 800.4453735351562,
+      "learning_rate": 6.941414141414142e-06,
+      "loss": 192.4127,
+      "step": 8590
+    },
+    {
+      "epoch": 0.0711419944575423,
+      "grad_norm": 1373.4234619140625,
+      "learning_rate": 6.9494949494949505e-06,
+      "loss": 161.1228,
+      "step": 8600
+    },
+    {
+      "epoch": 0.07122471770691152,
+      "grad_norm": 950.77197265625,
+      "learning_rate": 6.957575757575759e-06,
+      "loss": 179.3362,
+      "step": 8610
+    },
+    {
+      "epoch": 0.07130744095628076,
+      "grad_norm": 1121.46044921875,
+      "learning_rate": 6.9656565656565665e-06,
+      "loss": 196.0917,
+      "step": 8620
+    },
+    {
+      "epoch": 0.07139016420565,
+      "grad_norm": 906.2406005859375,
+      "learning_rate": 6.973737373737375e-06,
+      "loss": 139.1432,
+      "step": 8630
+    },
+    {
+      "epoch": 0.07147288745501923,
+      "grad_norm": 2017.980712890625,
+      "learning_rate": 6.981818181818183e-06,
+      "loss": 212.8915,
+      "step": 8640
+    },
+    {
+      "epoch": 0.07155561070438847,
+      "grad_norm": 1261.973388671875,
+      "learning_rate": 6.98989898989899e-06,
+      "loss": 200.444,
+      "step": 8650
+    },
+    {
+      "epoch": 0.07163833395375771,
+      "grad_norm": 855.2569580078125,
+      "learning_rate": 6.997979797979798e-06,
+      "loss": 222.0111,
+      "step": 8660
+    },
+    {
+      "epoch": 0.07172105720312694,
+      "grad_norm": 1921.39599609375,
+      "learning_rate": 7.006060606060606e-06,
+      "loss": 222.5992,
+      "step": 8670
+    },
+    {
+      "epoch": 0.07180378045249618,
+      "grad_norm": 857.9965209960938,
+      "learning_rate": 7.014141414141415e-06,
+      "loss": 219.586,
+      "step": 8680
+    },
+    {
+      "epoch": 0.0718865037018654,
+      "grad_norm": 915.5559692382812,
+      "learning_rate": 7.022222222222222e-06,
+      "loss": 184.9689,
+      "step": 8690
+    },
+    {
+      "epoch": 0.07196922695123464,
+      "grad_norm": 1838.1370849609375,
+      "learning_rate": 7.030303030303031e-06,
+      "loss": 232.5889,
+      "step": 8700
+    },
+    {
+      "epoch": 0.07205195020060388,
+      "grad_norm": 932.5780639648438,
+      "learning_rate": 7.038383838383839e-06,
+      "loss": 171.6244,
+      "step": 8710
+    },
+    {
+      "epoch": 0.07213467344997311,
+      "grad_norm": 1225.2701416015625,
+      "learning_rate": 7.046464646464647e-06,
+      "loss": 153.8994,
+      "step": 8720
+    },
+    {
+      "epoch": 0.07221739669934235,
+      "grad_norm": 1276.4774169921875,
+      "learning_rate": 7.054545454545455e-06,
+      "loss": 195.4201,
+      "step": 8730
+    },
+    {
+      "epoch": 0.07230011994871159,
+      "grad_norm": 1169.0623779296875,
+      "learning_rate": 7.062626262626263e-06,
+      "loss": 184.115,
+      "step": 8740
+    },
+    {
+      "epoch": 0.07238284319808082,
+      "grad_norm": 1720.584716796875,
+      "learning_rate": 7.070707070707071e-06,
+      "loss": 225.3465,
+      "step": 8750
+    },
+    {
+      "epoch": 0.07246556644745006,
+      "grad_norm": 811.720947265625,
+      "learning_rate": 7.07878787878788e-06,
+      "loss": 155.7371,
+      "step": 8760
+    },
+    {
+      "epoch": 0.07254828969681928,
+      "grad_norm": 1744.4354248046875,
+      "learning_rate": 7.086868686868687e-06,
+      "loss": 193.3227,
+      "step": 8770
+    },
+    {
+      "epoch": 0.07263101294618853,
+      "grad_norm": 862.0523681640625,
+      "learning_rate": 7.094949494949496e-06,
+      "loss": 170.9686,
+      "step": 8780
+    },
+    {
+      "epoch": 0.07271373619555777,
+      "grad_norm": 2165.270751953125,
+      "learning_rate": 7.103030303030304e-06,
+      "loss": 234.9279,
+      "step": 8790
+    },
+    {
+      "epoch": 0.07279645944492699,
+      "grad_norm": 1295.5478515625,
+      "learning_rate": 7.111111111111112e-06,
+      "loss": 211.1977,
+      "step": 8800
+    },
+    {
+      "epoch": 0.07287918269429623,
+      "grad_norm": 957.7193603515625,
+      "learning_rate": 7.11919191919192e-06,
+      "loss": 191.1015,
+      "step": 8810
+    },
+    {
+      "epoch": 0.07296190594366547,
+      "grad_norm": 1650.5340576171875,
+      "learning_rate": 7.127272727272728e-06,
+      "loss": 164.5978,
+      "step": 8820
+    },
+    {
+      "epoch": 0.0730446291930347,
+      "grad_norm": 1488.514404296875,
+      "learning_rate": 7.135353535353536e-06,
+      "loss": 187.6664,
+      "step": 8830
+    },
+    {
+      "epoch": 0.07312735244240394,
+      "grad_norm": 1756.99853515625,
+      "learning_rate": 7.143434343434345e-06,
+      "loss": 219.4122,
+      "step": 8840
+    },
+    {
+      "epoch": 0.07321007569177317,
+      "grad_norm": 924.37158203125,
+      "learning_rate": 7.151515151515152e-06,
+      "loss": 157.6518,
+      "step": 8850
+    },
+    {
+      "epoch": 0.0732927989411424,
+      "grad_norm": 1157.555908203125,
+      "learning_rate": 7.15959595959596e-06,
+      "loss": 162.5851,
+      "step": 8860
+    },
+    {
+      "epoch": 0.07337552219051165,
+      "grad_norm": 668.1375732421875,
+      "learning_rate": 7.1676767676767676e-06,
+      "loss": 153.8082,
+      "step": 8870
+    },
+    {
+      "epoch": 0.07345824543988087,
+      "grad_norm": 1706.42333984375,
+      "learning_rate": 7.175757575757576e-06,
+      "loss": 145.3626,
+      "step": 8880
+    },
+    {
+      "epoch": 0.07354096868925011,
+      "grad_norm": 2563.07861328125,
+      "learning_rate": 7.1838383838383845e-06,
+      "loss": 217.706,
+      "step": 8890
+    },
+    {
+      "epoch": 0.07362369193861935,
+      "grad_norm": 1756.02392578125,
+      "learning_rate": 7.191919191919192e-06,
+      "loss": 177.5173,
+      "step": 8900
+    },
+    {
+      "epoch": 0.07370641518798858,
+      "grad_norm": 1470.858642578125,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 218.2071,
+      "step": 8910
+    },
+    {
+      "epoch": 0.07378913843735782,
+      "grad_norm": 1429.1534423828125,
+      "learning_rate": 7.208080808080808e-06,
+      "loss": 192.6579,
+      "step": 8920
+    },
+    {
+      "epoch": 0.07387186168672706,
+      "grad_norm": 1044.5311279296875,
+      "learning_rate": 7.2161616161616166e-06,
+      "loss": 175.2781,
+      "step": 8930
+    },
+    {
+      "epoch": 0.07395458493609629,
+      "grad_norm": 1081.16748046875,
+      "learning_rate": 7.224242424242425e-06,
+      "loss": 195.337,
+      "step": 8940
+    },
+    {
+      "epoch": 0.07403730818546553,
+      "grad_norm": 1816.50634765625,
+      "learning_rate": 7.232323232323233e-06,
+      "loss": 164.159,
+      "step": 8950
+    },
+    {
+      "epoch": 0.07412003143483475,
+      "grad_norm": 941.095947265625,
+      "learning_rate": 7.240404040404041e-06,
+      "loss": 167.4316,
+      "step": 8960
+    },
+    {
+      "epoch": 0.074202754684204,
+      "grad_norm": 1430.9918212890625,
+      "learning_rate": 7.2484848484848495e-06,
+      "loss": 142.7207,
+      "step": 8970
+    },
+    {
+      "epoch": 0.07428547793357324,
+      "grad_norm": 987.8670043945312,
+      "learning_rate": 7.256565656565657e-06,
+      "loss": 185.2262,
+      "step": 8980
+    },
+    {
+      "epoch": 0.07436820118294246,
+      "grad_norm": 1886.202392578125,
+      "learning_rate": 7.2646464646464656e-06,
+      "loss": 201.7182,
+      "step": 8990
+    },
+    {
+      "epoch": 0.0744509244323117,
+      "grad_norm": 6469.64501953125,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 242.3246,
+      "step": 9000
+    },
+    {
+      "epoch": 0.07453364768168094,
+      "grad_norm": 1446.23388671875,
+      "learning_rate": 7.280808080808082e-06,
+      "loss": 185.7719,
+      "step": 9010
+    },
+    {
+      "epoch": 0.07461637093105017,
+      "grad_norm": 1158.1019287109375,
+      "learning_rate": 7.28888888888889e-06,
+      "loss": 135.7039,
+      "step": 9020
+    },
+    {
+      "epoch": 0.07469909418041941,
+      "grad_norm": 1710.8179931640625,
+      "learning_rate": 7.296969696969698e-06,
+      "loss": 214.3662,
+      "step": 9030
+    },
+    {
+      "epoch": 0.07478181742978864,
+      "grad_norm": 693.0316772460938,
+      "learning_rate": 7.305050505050506e-06,
+      "loss": 168.4758,
+      "step": 9040
+    },
+    {
+      "epoch": 0.07486454067915788,
+      "grad_norm": 1307.99267578125,
+      "learning_rate": 7.3131313131313146e-06,
+      "loss": 221.4772,
+      "step": 9050
+    },
+    {
+      "epoch": 0.07494726392852712,
+      "grad_norm": 578.5460815429688,
+      "learning_rate": 7.321212121212122e-06,
+      "loss": 172.2857,
+      "step": 9060
+    },
+    {
+      "epoch": 0.07502998717789634,
+      "grad_norm": 4154.9052734375,
+      "learning_rate": 7.32929292929293e-06,
+      "loss": 200.6511,
+      "step": 9070
+    },
+    {
+      "epoch": 0.07511271042726558,
+      "grad_norm": 3818.542236328125,
+      "learning_rate": 7.337373737373737e-06,
+      "loss": 198.6826,
+      "step": 9080
+    },
+    {
+      "epoch": 0.07519543367663482,
+      "grad_norm": 630.0006713867188,
+      "learning_rate": 7.345454545454546e-06,
+      "loss": 146.3737,
+      "step": 9090
+    },
+    {
+      "epoch": 0.07527815692600405,
+      "grad_norm": 1059.06640625,
+      "learning_rate": 7.353535353535353e-06,
+      "loss": 163.5028,
+      "step": 9100
+    },
+    {
+      "epoch": 0.07536088017537329,
+      "grad_norm": 668.64111328125,
+      "learning_rate": 7.361616161616162e-06,
+      "loss": 170.1274,
+      "step": 9110
+    },
+    {
+      "epoch": 0.07544360342474253,
+      "grad_norm": 1227.2437744140625,
+      "learning_rate": 7.36969696969697e-06,
+      "loss": 167.1523,
+      "step": 9120
+    },
+    {
+      "epoch": 0.07552632667411176,
+      "grad_norm": 989.3958740234375,
+      "learning_rate": 7.377777777777778e-06,
+      "loss": 179.0444,
+      "step": 9130
+    },
+    {
+      "epoch": 0.075609049923481,
+      "grad_norm": 2812.70166015625,
+      "learning_rate": 7.385858585858586e-06,
+      "loss": 170.3505,
+      "step": 9140
+    },
+    {
+      "epoch": 0.07569177317285022,
+      "grad_norm": 1429.5191650390625,
+      "learning_rate": 7.393939393939395e-06,
+      "loss": 198.7005,
+      "step": 9150
+    },
+    {
+      "epoch": 0.07577449642221946,
+      "grad_norm": 937.4029541015625,
+      "learning_rate": 7.402020202020202e-06,
+      "loss": 149.7339,
+      "step": 9160
+    },
+    {
+      "epoch": 0.0758572196715887,
+      "grad_norm": 6428.83642578125,
+      "learning_rate": 7.410101010101011e-06,
+      "loss": 224.5054,
+      "step": 9170
+    },
+    {
+      "epoch": 0.07593994292095793,
+      "grad_norm": 1135.309814453125,
+      "learning_rate": 7.4181818181818185e-06,
+      "loss": 206.6003,
+      "step": 9180
+    },
+    {
+      "epoch": 0.07602266617032717,
+      "grad_norm": 663.953125,
+      "learning_rate": 7.426262626262627e-06,
+      "loss": 148.6399,
+      "step": 9190
+    },
+    {
+      "epoch": 0.07610538941969641,
+      "grad_norm": 1465.2401123046875,
+      "learning_rate": 7.434343434343435e-06,
+      "loss": 129.7046,
+      "step": 9200
+    },
+    {
+      "epoch": 0.07618811266906564,
+      "grad_norm": 840.3555297851562,
+      "learning_rate": 7.442424242424243e-06,
+      "loss": 212.8563,
+      "step": 9210
+    },
+    {
+      "epoch": 0.07627083591843488,
+      "grad_norm": 1056.4595947265625,
+      "learning_rate": 7.450505050505051e-06,
+      "loss": 219.987,
+      "step": 9220
+    },
+    {
+      "epoch": 0.0763535591678041,
+      "grad_norm": 1585.2525634765625,
+      "learning_rate": 7.45858585858586e-06,
+      "loss": 254.3914,
+      "step": 9230
+    },
+    {
+      "epoch": 0.07643628241717335,
+      "grad_norm": 1167.71728515625,
+      "learning_rate": 7.4666666666666675e-06,
+      "loss": 173.4369,
+      "step": 9240
+    },
+    {
+      "epoch": 0.07651900566654259,
+      "grad_norm": 1241.547119140625,
+      "learning_rate": 7.474747474747476e-06,
+      "loss": 196.5241,
+      "step": 9250
+    },
+    {
+      "epoch": 0.07660172891591181,
+      "grad_norm": 1076.36279296875,
+      "learning_rate": 7.4828282828282835e-06,
+      "loss": 141.0298,
+      "step": 9260
+    },
+    {
+      "epoch": 0.07668445216528105,
+      "grad_norm": 894.4515991210938,
+      "learning_rate": 7.490909090909092e-06,
+      "loss": 182.4499,
+      "step": 9270
+    },
+    {
+      "epoch": 0.0767671754146503,
+      "grad_norm": 1318.6171875,
+      "learning_rate": 7.4989898989899e-06,
+      "loss": 294.3835,
+      "step": 9280
+    },
+    {
+      "epoch": 0.07684989866401952,
+      "grad_norm": 1835.936279296875,
+      "learning_rate": 7.507070707070707e-06,
+      "loss": 200.7481,
+      "step": 9290
+    },
+    {
+      "epoch": 0.07693262191338876,
+      "grad_norm": 1094.5806884765625,
+      "learning_rate": 7.515151515151516e-06,
+      "loss": 155.6739,
+      "step": 9300
+    },
+    {
+      "epoch": 0.07701534516275799,
+      "grad_norm": 2438.4599609375,
+      "learning_rate": 7.523232323232323e-06,
+      "loss": 168.196,
+      "step": 9310
+    },
+    {
+      "epoch": 0.07709806841212723,
+      "grad_norm": 1409.369384765625,
+      "learning_rate": 7.531313131313132e-06,
+      "loss": 168.5339,
+      "step": 9320
+    },
+    {
+      "epoch": 0.07718079166149647,
+      "grad_norm": 792.833740234375,
+      "learning_rate": 7.53939393939394e-06,
+      "loss": 204.8236,
+      "step": 9330
+    },
+    {
+      "epoch": 0.0772635149108657,
+      "grad_norm": 4106.85107421875,
+      "learning_rate": 7.547474747474748e-06,
+      "loss": 206.0253,
+      "step": 9340
+    },
+    {
+      "epoch": 0.07734623816023493,
+      "grad_norm": 1158.9046630859375,
+      "learning_rate": 7.555555555555556e-06,
+      "loss": 200.0595,
+      "step": 9350
+    },
+    {
+      "epoch": 0.07742896140960417,
+      "grad_norm": 829.5303344726562,
+      "learning_rate": 7.563636363636364e-06,
+      "loss": 152.1526,
+      "step": 9360
+    },
+    {
+      "epoch": 0.0775116846589734,
+      "grad_norm": 1672.3824462890625,
+      "learning_rate": 7.571717171717172e-06,
+      "loss": 196.9728,
+      "step": 9370
+    },
+    {
+      "epoch": 0.07759440790834264,
+      "grad_norm": 1697.9437255859375,
+      "learning_rate": 7.579797979797981e-06,
+      "loss": 140.7818,
+      "step": 9380
+    },
+    {
+      "epoch": 0.07767713115771188,
+      "grad_norm": 1467.42138671875,
+      "learning_rate": 7.587878787878788e-06,
+      "loss": 237.8365,
+      "step": 9390
+    },
+    {
+      "epoch": 0.07775985440708111,
+      "grad_norm": 1372.5516357421875,
+      "learning_rate": 7.595959595959597e-06,
+      "loss": 201.537,
+      "step": 9400
+    },
+    {
+      "epoch": 0.07784257765645035,
+      "grad_norm": 1280.147216796875,
+      "learning_rate": 7.604040404040405e-06,
+      "loss": 168.6732,
+      "step": 9410
+    },
+    {
+      "epoch": 0.07792530090581958,
+      "grad_norm": 1045.321533203125,
+      "learning_rate": 7.612121212121213e-06,
+      "loss": 154.092,
+      "step": 9420
+    },
+    {
+      "epoch": 0.07800802415518882,
+      "grad_norm": 6423.11962890625,
+      "learning_rate": 7.620202020202021e-06,
+      "loss": 251.8613,
+      "step": 9430
+    },
+    {
+      "epoch": 0.07809074740455806,
+      "grad_norm": 1650.4332275390625,
+      "learning_rate": 7.628282828282829e-06,
+      "loss": 218.2032,
+      "step": 9440
+    },
+    {
+      "epoch": 0.07817347065392728,
+      "grad_norm": 1520.5958251953125,
+      "learning_rate": 7.636363636363638e-06,
+      "loss": 194.5603,
+      "step": 9450
+    },
+    {
+      "epoch": 0.07825619390329652,
+      "grad_norm": 1989.30126953125,
+      "learning_rate": 7.644444444444445e-06,
+      "loss": 167.2613,
+      "step": 9460
+    },
+    {
+      "epoch": 0.07833891715266576,
+      "grad_norm": 1363.1075439453125,
+      "learning_rate": 7.652525252525253e-06,
+      "loss": 193.0324,
+      "step": 9470
+    },
+    {
+      "epoch": 0.07842164040203499,
+      "grad_norm": 1044.422607421875,
+      "learning_rate": 7.660606060606062e-06,
+      "loss": 239.5535,
+      "step": 9480
+    },
+    {
+      "epoch": 0.07850436365140423,
+      "grad_norm": 1506.1378173828125,
+      "learning_rate": 7.66868686868687e-06,
+      "loss": 157.3207,
+      "step": 9490
+    },
+    {
+      "epoch": 0.07858708690077346,
+      "grad_norm": 1023.1807250976562,
+      "learning_rate": 7.676767676767677e-06,
+      "loss": 177.0086,
+      "step": 9500
+    },
+    {
+      "epoch": 0.0786698101501427,
+      "grad_norm": 766.3860473632812,
+      "learning_rate": 7.684848484848485e-06,
+      "loss": 180.8027,
+      "step": 9510
+    },
+    {
+      "epoch": 0.07875253339951194,
+      "grad_norm": 1220.765380859375,
+      "learning_rate": 7.692929292929294e-06,
+      "loss": 199.764,
+      "step": 9520
+    },
+    {
+      "epoch": 0.07883525664888116,
+      "grad_norm": 798.783447265625,
+      "learning_rate": 7.7010101010101e-06,
+      "loss": 178.7332,
+      "step": 9530
+    },
+    {
+      "epoch": 0.0789179798982504,
+      "grad_norm": 1135.2913818359375,
+      "learning_rate": 7.709090909090909e-06,
+      "loss": 138.7885,
+      "step": 9540
+    },
+    {
+      "epoch": 0.07900070314761964,
+      "grad_norm": 820.0831298828125,
+      "learning_rate": 7.717171717171717e-06,
+      "loss": 192.7479,
+      "step": 9550
+    },
+    {
+      "epoch": 0.07908342639698887,
+      "grad_norm": 1487.1383056640625,
+      "learning_rate": 7.725252525252526e-06,
+      "loss": 202.4578,
+      "step": 9560
+    },
+    {
+      "epoch": 0.07916614964635811,
+      "grad_norm": 3294.249267578125,
+      "learning_rate": 7.733333333333334e-06,
+      "loss": 154.1931,
+      "step": 9570
+    },
+    {
+      "epoch": 0.07924887289572734,
+      "grad_norm": 1593.376708984375,
+      "learning_rate": 7.741414141414141e-06,
+      "loss": 220.1035,
+      "step": 9580
+    },
+    {
+      "epoch": 0.07933159614509658,
+      "grad_norm": 1254.923095703125,
+      "learning_rate": 7.74949494949495e-06,
+      "loss": 162.3905,
+      "step": 9590
+    },
+    {
+      "epoch": 0.07941431939446582,
+      "grad_norm": 989.010009765625,
+      "learning_rate": 7.757575757575758e-06,
+      "loss": 150.6632,
+      "step": 9600
+    },
+    {
+      "epoch": 0.07949704264383504,
+      "grad_norm": 668.603759765625,
+      "learning_rate": 7.765656565656566e-06,
+      "loss": 180.0927,
+      "step": 9610
+    },
+    {
+      "epoch": 0.07957976589320429,
+      "grad_norm": 918.044677734375,
+      "learning_rate": 7.773737373737375e-06,
+      "loss": 216.1173,
+      "step": 9620
+    },
+    {
+      "epoch": 0.07966248914257353,
+      "grad_norm": 1150.803955078125,
+      "learning_rate": 7.781818181818183e-06,
+      "loss": 165.5561,
+      "step": 9630
+    },
+    {
+      "epoch": 0.07974521239194275,
+      "grad_norm": 1339.1085205078125,
+      "learning_rate": 7.78989898989899e-06,
+      "loss": 120.4266,
+      "step": 9640
+    },
+    {
+      "epoch": 0.07982793564131199,
+      "grad_norm": 1119.98388671875,
+      "learning_rate": 7.797979797979799e-06,
+      "loss": 186.139,
+      "step": 9650
+    },
+    {
+      "epoch": 0.07991065889068123,
+      "grad_norm": 845.5322875976562,
+      "learning_rate": 7.806060606060607e-06,
+      "loss": 147.2671,
+      "step": 9660
+    },
+    {
+      "epoch": 0.07999338214005046,
+      "grad_norm": 1525.1080322265625,
+      "learning_rate": 7.814141414141415e-06,
+      "loss": 227.4823,
+      "step": 9670
+    },
+    {
+      "epoch": 0.0800761053894197,
+      "grad_norm": 936.9368286132812,
+      "learning_rate": 7.822222222222224e-06,
+      "loss": 147.8492,
+      "step": 9680
+    },
+    {
+      "epoch": 0.08015882863878893,
+      "grad_norm": 1528.34375,
+      "learning_rate": 7.83030303030303e-06,
+      "loss": 176.4171,
+      "step": 9690
+    },
+    {
+      "epoch": 0.08024155188815817,
+      "grad_norm": 1028.9813232421875,
+      "learning_rate": 7.838383838383839e-06,
+      "loss": 168.1568,
+      "step": 9700
+    },
+    {
+      "epoch": 0.0803242751375274,
+      "grad_norm": 3313.676513671875,
+      "learning_rate": 7.846464646464646e-06,
+      "loss": 211.1931,
+      "step": 9710
+    },
+    {
+      "epoch": 0.08040699838689663,
+      "grad_norm": 698.055419921875,
+      "learning_rate": 7.854545454545454e-06,
+      "loss": 147.0731,
+      "step": 9720
+    },
+    {
+      "epoch": 0.08048972163626587,
+      "grad_norm": 1256.8193359375,
+      "learning_rate": 7.862626262626263e-06,
+      "loss": 141.3686,
+      "step": 9730
+    },
+    {
+      "epoch": 0.08057244488563511,
+      "grad_norm": 1118.2659912109375,
+      "learning_rate": 7.870707070707071e-06,
+      "loss": 191.6792,
+      "step": 9740
+    },
+    {
+      "epoch": 0.08065516813500434,
+      "grad_norm": 1615.0177001953125,
+      "learning_rate": 7.87878787878788e-06,
+      "loss": 171.2617,
+      "step": 9750
+    },
+    {
+      "epoch": 0.08073789138437358,
+      "grad_norm": 1359.4404296875,
+      "learning_rate": 7.886868686868686e-06,
+      "loss": 233.8179,
+      "step": 9760
+    },
+    {
+      "epoch": 0.08082061463374281,
+      "grad_norm": 1080.4814453125,
+      "learning_rate": 7.894949494949495e-06,
+      "loss": 167.5727,
+      "step": 9770
+    },
+    {
+      "epoch": 0.08090333788311205,
+      "grad_norm": 1568.5997314453125,
+      "learning_rate": 7.903030303030303e-06,
+      "loss": 180.408,
+      "step": 9780
+    },
+    {
+      "epoch": 0.08098606113248129,
+      "grad_norm": 1154.7166748046875,
+      "learning_rate": 7.911111111111112e-06,
+      "loss": 141.3988,
+      "step": 9790
+    },
+    {
+      "epoch": 0.08106878438185051,
+      "grad_norm": 1465.5701904296875,
+      "learning_rate": 7.91919191919192e-06,
+      "loss": 235.8258,
+      "step": 9800
+    },
+    {
+      "epoch": 0.08115150763121975,
+      "grad_norm": 1399.992919921875,
+      "learning_rate": 7.927272727272729e-06,
+      "loss": 197.8094,
+      "step": 9810
+    },
+    {
+      "epoch": 0.081234230880589,
+      "grad_norm": 1609.233154296875,
+      "learning_rate": 7.935353535353535e-06,
+      "loss": 169.6143,
+      "step": 9820
+    },
+    {
+      "epoch": 0.08131695412995822,
+      "grad_norm": 771.2155151367188,
+      "learning_rate": 7.943434343434344e-06,
+      "loss": 174.0228,
+      "step": 9830
+    },
+    {
+      "epoch": 0.08139967737932746,
+      "grad_norm": 1397.8433837890625,
+      "learning_rate": 7.951515151515152e-06,
+      "loss": 195.8236,
+      "step": 9840
+    },
+    {
+      "epoch": 0.08148240062869669,
+      "grad_norm": 1942.0560302734375,
+      "learning_rate": 7.95959595959596e-06,
+      "loss": 174.9514,
+      "step": 9850
+    },
+    {
+      "epoch": 0.08156512387806593,
+      "grad_norm": 881.597900390625,
+      "learning_rate": 7.96767676767677e-06,
+      "loss": 155.3664,
+      "step": 9860
+    },
+    {
+      "epoch": 0.08164784712743517,
+      "grad_norm": 982.1563110351562,
+      "learning_rate": 7.975757575757576e-06,
+      "loss": 166.2256,
+      "step": 9870
+    },
+    {
+      "epoch": 0.0817305703768044,
+      "grad_norm": 1786.5938720703125,
+      "learning_rate": 7.983838383838384e-06,
+      "loss": 179.014,
+      "step": 9880
+    },
+    {
+      "epoch": 0.08181329362617364,
+      "grad_norm": 1422.998046875,
+      "learning_rate": 7.991919191919193e-06,
+      "loss": 185.8182,
+      "step": 9890
+    },
+    {
+      "epoch": 0.08189601687554288,
+      "grad_norm": 1009.948486328125,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 153.7416,
+      "step": 9900
+    },
+    {
+      "epoch": 0.0819787401249121,
+      "grad_norm": 929.018798828125,
+      "learning_rate": 8.00808080808081e-06,
+      "loss": 178.9259,
+      "step": 9910
+    },
+    {
+      "epoch": 0.08206146337428134,
+      "grad_norm": 654.3829345703125,
+      "learning_rate": 8.016161616161617e-06,
+      "loss": 130.9901,
+      "step": 9920
+    },
+    {
+      "epoch": 0.08214418662365058,
+      "grad_norm": 1169.20068359375,
+      "learning_rate": 8.024242424242425e-06,
+      "loss": 125.0886,
+      "step": 9930
+    },
+    {
+      "epoch": 0.08222690987301981,
+      "grad_norm": 1203.2630615234375,
+      "learning_rate": 8.032323232323232e-06,
+      "loss": 191.309,
+      "step": 9940
+    },
+    {
+      "epoch": 0.08230963312238905,
+      "grad_norm": 1353.7027587890625,
+      "learning_rate": 8.04040404040404e-06,
+      "loss": 239.4487,
+      "step": 9950
+    },
+    {
+      "epoch": 0.08239235637175828,
+      "grad_norm": 1199.4378662109375,
+      "learning_rate": 8.048484848484849e-06,
+      "loss": 173.5184,
+      "step": 9960
+    },
+    {
+      "epoch": 0.08247507962112752,
+      "grad_norm": 1027.1614990234375,
+      "learning_rate": 8.056565656565657e-06,
+      "loss": 236.1041,
+      "step": 9970
+    },
+    {
+      "epoch": 0.08255780287049676,
+      "grad_norm": 2071.078125,
+      "learning_rate": 8.064646464646466e-06,
+      "loss": 210.167,
+      "step": 9980
+    },
+    {
+      "epoch": 0.08264052611986598,
+      "grad_norm": 931.7273559570312,
+      "learning_rate": 8.072727272727274e-06,
+      "loss": 169.0378,
+      "step": 9990
+    },
+    {
+      "epoch": 0.08272324936923522,
+      "grad_norm": 1780.4796142578125,
+      "learning_rate": 8.08080808080808e-06,
+      "loss": 182.981,
+      "step": 10000
+    },
+    {
+      "epoch": 0.08280597261860446,
+      "grad_norm": 1484.16845703125,
+      "learning_rate": 8.08888888888889e-06,
+      "loss": 190.9732,
+      "step": 10010
+    },
+    {
+      "epoch": 0.08288869586797369,
+      "grad_norm": 1517.7740478515625,
+      "learning_rate": 8.096969696969698e-06,
+      "loss": 214.248,
+      "step": 10020
+    },
+    {
+      "epoch": 0.08297141911734293,
+      "grad_norm": 4507.2001953125,
+      "learning_rate": 8.105050505050506e-06,
+      "loss": 149.1075,
+      "step": 10030
+    },
+    {
+      "epoch": 0.08305414236671216,
+      "grad_norm": 1202.4232177734375,
+      "learning_rate": 8.113131313131315e-06,
+      "loss": 167.8956,
+      "step": 10040
+    },
+    {
+      "epoch": 0.0831368656160814,
+      "grad_norm": 3145.95556640625,
+      "learning_rate": 8.121212121212121e-06,
+      "loss": 218.0878,
+      "step": 10050
+    },
+    {
+      "epoch": 0.08321958886545064,
+      "grad_norm": 833.5780029296875,
+      "learning_rate": 8.12929292929293e-06,
+      "loss": 183.5963,
+      "step": 10060
+    },
+    {
+      "epoch": 0.08330231211481987,
+      "grad_norm": 1200.9332275390625,
+      "learning_rate": 8.137373737373738e-06,
+      "loss": 200.6742,
+      "step": 10070
+    },
+    {
+      "epoch": 0.0833850353641891,
+      "grad_norm": 2536.378662109375,
+      "learning_rate": 8.145454545454547e-06,
+      "loss": 244.585,
+      "step": 10080
+    },
+    {
+      "epoch": 0.08346775861355835,
+      "grad_norm": 0.0,
+      "learning_rate": 8.153535353535355e-06,
+      "loss": 151.4828,
+      "step": 10090
+    },
+    {
+      "epoch": 0.08355048186292757,
+      "grad_norm": 1108.849853515625,
+      "learning_rate": 8.161616161616162e-06,
+      "loss": 178.3399,
+      "step": 10100
+    },
+    {
+      "epoch": 0.08363320511229681,
+      "grad_norm": 809.9747314453125,
+      "learning_rate": 8.16969696969697e-06,
+      "loss": 201.301,
+      "step": 10110
+    },
+    {
+      "epoch": 0.08371592836166604,
+      "grad_norm": 1331.628173828125,
+      "learning_rate": 8.177777777777779e-06,
+      "loss": 141.3669,
+      "step": 10120
+    },
+    {
+      "epoch": 0.08379865161103528,
+      "grad_norm": 990.9824829101562,
+      "learning_rate": 8.185858585858587e-06,
+      "loss": 169.3207,
+      "step": 10130
+    },
+    {
+      "epoch": 0.08388137486040452,
+      "grad_norm": 1191.9073486328125,
+      "learning_rate": 8.193939393939394e-06,
+      "loss": 136.0156,
+      "step": 10140
+    },
+    {
+      "epoch": 0.08396409810977375,
+      "grad_norm": 829.3964233398438,
+      "learning_rate": 8.202020202020202e-06,
+      "loss": 165.8068,
+      "step": 10150
+    },
+    {
+      "epoch": 0.08404682135914299,
+      "grad_norm": 781.4263305664062,
+      "learning_rate": 8.21010101010101e-06,
+      "loss": 186.8697,
+      "step": 10160
+    },
+    {
+      "epoch": 0.08412954460851223,
+      "grad_norm": 815.2606811523438,
+      "learning_rate": 8.21818181818182e-06,
+      "loss": 257.8631,
+      "step": 10170
+    },
+    {
+      "epoch": 0.08421226785788145,
+      "grad_norm": 1860.0218505859375,
+      "learning_rate": 8.226262626262626e-06,
+      "loss": 187.696,
+      "step": 10180
+    },
+    {
+      "epoch": 0.0842949911072507,
+      "grad_norm": 2099.6728515625,
+      "learning_rate": 8.234343434343434e-06,
+      "loss": 205.9569,
+      "step": 10190
+    },
+    {
+      "epoch": 0.08437771435661993,
+      "grad_norm": 1526.5322265625,
+      "learning_rate": 8.242424242424243e-06,
+      "loss": 200.3001,
+      "step": 10200
+    },
+    {
+      "epoch": 0.08446043760598916,
+      "grad_norm": 1122.8441162109375,
+      "learning_rate": 8.250505050505051e-06,
+      "loss": 132.1188,
+      "step": 10210
+    },
+    {
+      "epoch": 0.0845431608553584,
+      "grad_norm": 952.8866577148438,
+      "learning_rate": 8.25858585858586e-06,
+      "loss": 180.2814,
+      "step": 10220
+    },
+    {
+      "epoch": 0.08462588410472763,
+      "grad_norm": 1098.6842041015625,
+      "learning_rate": 8.266666666666667e-06,
+      "loss": 186.863,
+      "step": 10230
+    },
+    {
+      "epoch": 0.08470860735409687,
+      "grad_norm": 1201.576904296875,
+      "learning_rate": 8.274747474747475e-06,
+      "loss": 166.658,
+      "step": 10240
+    },
+    {
+      "epoch": 0.08479133060346611,
+      "grad_norm": 3628.583984375,
+      "learning_rate": 8.282828282828283e-06,
+      "loss": 280.9522,
+      "step": 10250
+    },
+    {
+      "epoch": 0.08487405385283533,
+      "grad_norm": 956.4686279296875,
+      "learning_rate": 8.290909090909092e-06,
+      "loss": 177.9082,
+      "step": 10260
+    },
+    {
+      "epoch": 0.08495677710220458,
+      "grad_norm": 1213.8653564453125,
+      "learning_rate": 8.2989898989899e-06,
+      "loss": 176.0302,
+      "step": 10270
+    },
+    {
+      "epoch": 0.08503950035157382,
+      "grad_norm": 1263.095458984375,
+      "learning_rate": 8.307070707070707e-06,
+      "loss": 171.6865,
+      "step": 10280
+    },
+    {
+      "epoch": 0.08512222360094304,
+      "grad_norm": 1094.4754638671875,
+      "learning_rate": 8.315151515151516e-06,
+      "loss": 158.2062,
+      "step": 10290
+    },
+    {
+      "epoch": 0.08520494685031228,
+      "grad_norm": 957.9495239257812,
+      "learning_rate": 8.323232323232324e-06,
+      "loss": 169.5506,
+      "step": 10300
+    },
+    {
+      "epoch": 0.08528767009968151,
+      "grad_norm": 1073.7796630859375,
+      "learning_rate": 8.331313131313132e-06,
+      "loss": 147.6098,
+      "step": 10310
+    },
+    {
+      "epoch": 0.08537039334905075,
+      "grad_norm": 1453.0513916015625,
+      "learning_rate": 8.339393939393941e-06,
+      "loss": 189.7265,
+      "step": 10320
+    },
+    {
+      "epoch": 0.08545311659841999,
+      "grad_norm": 791.720703125,
+      "learning_rate": 8.34747474747475e-06,
+      "loss": 214.0726,
+      "step": 10330
+    },
+    {
+      "epoch": 0.08553583984778922,
+      "grad_norm": 1202.690673828125,
+      "learning_rate": 8.355555555555556e-06,
+      "loss": 197.4867,
+      "step": 10340
+    },
+    {
+      "epoch": 0.08561856309715846,
+      "grad_norm": 1175.0545654296875,
+      "learning_rate": 8.363636363636365e-06,
+      "loss": 153.0685,
+      "step": 10350
+    },
+    {
+      "epoch": 0.0857012863465277,
+      "grad_norm": 745.9451293945312,
+      "learning_rate": 8.371717171717171e-06,
+      "loss": 170.8535,
+      "step": 10360
+    },
+    {
+      "epoch": 0.08578400959589692,
+      "grad_norm": 1357.7061767578125,
+      "learning_rate": 8.37979797979798e-06,
+      "loss": 162.8026,
+      "step": 10370
+    },
+    {
+      "epoch": 0.08586673284526616,
+      "grad_norm": 1293.4013671875,
+      "learning_rate": 8.387878787878788e-06,
+      "loss": 152.5479,
+      "step": 10380
+    },
+    {
+      "epoch": 0.0859494560946354,
+      "grad_norm": 1522.9151611328125,
+      "learning_rate": 8.395959595959597e-06,
+      "loss": 180.1763,
+      "step": 10390
+    },
+    {
+      "epoch": 0.08603217934400463,
+      "grad_norm": 2097.494140625,
+      "learning_rate": 8.404040404040405e-06,
+      "loss": 168.4439,
+      "step": 10400
+    },
+    {
+      "epoch": 0.08611490259337387,
+      "grad_norm": 1112.0291748046875,
+      "learning_rate": 8.412121212121212e-06,
+      "loss": 193.0542,
+      "step": 10410
+    },
+    {
+      "epoch": 0.0861976258427431,
+      "grad_norm": 927.9266357421875,
+      "learning_rate": 8.42020202020202e-06,
+      "loss": 211.676,
+      "step": 10420
+    },
+    {
+      "epoch": 0.08628034909211234,
+      "grad_norm": 1505.004150390625,
+      "learning_rate": 8.428282828282829e-06,
+      "loss": 203.2798,
+      "step": 10430
+    },
+    {
+      "epoch": 0.08636307234148158,
+      "grad_norm": 1962.483154296875,
+      "learning_rate": 8.436363636363637e-06,
+      "loss": 217.3728,
+      "step": 10440
+    },
+    {
+      "epoch": 0.0864457955908508,
+      "grad_norm": 2074.048583984375,
+      "learning_rate": 8.444444444444446e-06,
+      "loss": 224.9844,
+      "step": 10450
+    },
+    {
+      "epoch": 0.08652851884022005,
+      "grad_norm": 1316.5950927734375,
+      "learning_rate": 8.452525252525252e-06,
+      "loss": 165.5474,
+      "step": 10460
+    },
+    {
+      "epoch": 0.08661124208958929,
+      "grad_norm": 1096.7896728515625,
+      "learning_rate": 8.460606060606061e-06,
+      "loss": 138.2374,
+      "step": 10470
+    },
+    {
+      "epoch": 0.08669396533895851,
+      "grad_norm": 1451.124755859375,
+      "learning_rate": 8.46868686868687e-06,
+      "loss": 141.8237,
+      "step": 10480
+    },
+    {
+      "epoch": 0.08677668858832775,
+      "grad_norm": 1239.64697265625,
+      "learning_rate": 8.476767676767678e-06,
+      "loss": 156.3808,
+      "step": 10490
+    },
+    {
+      "epoch": 0.08685941183769698,
+      "grad_norm": 689.178955078125,
+      "learning_rate": 8.484848484848486e-06,
+      "loss": 143.1666,
+      "step": 10500
+    },
+    {
+      "epoch": 0.08694213508706622,
+      "grad_norm": 684.1348266601562,
+      "learning_rate": 8.492929292929295e-06,
+      "loss": 182.6362,
+      "step": 10510
+    },
+    {
+      "epoch": 0.08702485833643546,
+      "grad_norm": 1299.8443603515625,
+      "learning_rate": 8.501010101010101e-06,
+      "loss": 160.0256,
+      "step": 10520
+    },
+    {
+      "epoch": 0.08710758158580469,
+      "grad_norm": 845.9354858398438,
+      "learning_rate": 8.50909090909091e-06,
+      "loss": 184.4307,
+      "step": 10530
+    },
+    {
+      "epoch": 0.08719030483517393,
+      "grad_norm": 1035.8701171875,
+      "learning_rate": 8.517171717171718e-06,
+      "loss": 228.6084,
+      "step": 10540
+    },
+    {
+      "epoch": 0.08727302808454317,
+      "grad_norm": 668.9830932617188,
+      "learning_rate": 8.525252525252527e-06,
+      "loss": 200.758,
+      "step": 10550
+    },
+    {
+      "epoch": 0.0873557513339124,
+      "grad_norm": 1058.86181640625,
+      "learning_rate": 8.533333333333335e-06,
+      "loss": 204.7663,
+      "step": 10560
+    },
+    {
+      "epoch": 0.08743847458328163,
+      "grad_norm": 1521.737548828125,
+      "learning_rate": 8.541414141414142e-06,
+      "loss": 192.2928,
+      "step": 10570
+    },
+    {
+      "epoch": 0.08752119783265086,
+      "grad_norm": 1117.1220703125,
+      "learning_rate": 8.54949494949495e-06,
+      "loss": 142.4138,
+      "step": 10580
+    },
+    {
+      "epoch": 0.0876039210820201,
+      "grad_norm": 1011.63525390625,
+      "learning_rate": 8.557575757575757e-06,
+      "loss": 193.8034,
+      "step": 10590
+    },
+    {
+      "epoch": 0.08768664433138934,
+      "grad_norm": 2114.847900390625,
+      "learning_rate": 8.565656565656566e-06,
+      "loss": 184.3933,
+      "step": 10600
+    },
+    {
+      "epoch": 0.08776936758075857,
+      "grad_norm": 823.4564819335938,
+      "learning_rate": 8.573737373737374e-06,
+      "loss": 186.983,
+      "step": 10610
+    },
+    {
+      "epoch": 0.08785209083012781,
+      "grad_norm": 1262.8677978515625,
+      "learning_rate": 8.581818181818183e-06,
+      "loss": 145.9637,
+      "step": 10620
+    },
+    {
+      "epoch": 0.08793481407949705,
+      "grad_norm": 1084.4609375,
+      "learning_rate": 8.589898989898991e-06,
+      "loss": 199.8638,
+      "step": 10630
+    },
+    {
+      "epoch": 0.08801753732886627,
+      "grad_norm": 1483.7017822265625,
+      "learning_rate": 8.597979797979798e-06,
+      "loss": 183.0729,
+      "step": 10640
+    },
+    {
+      "epoch": 0.08810026057823551,
+      "grad_norm": 1132.1685791015625,
+      "learning_rate": 8.606060606060606e-06,
+      "loss": 160.7905,
+      "step": 10650
+    },
+    {
+      "epoch": 0.08818298382760476,
+      "grad_norm": 1912.802978515625,
+      "learning_rate": 8.614141414141415e-06,
+      "loss": 235.4324,
+      "step": 10660
+    },
+    {
+      "epoch": 0.08826570707697398,
+      "grad_norm": 1315.4351806640625,
+      "learning_rate": 8.622222222222223e-06,
+      "loss": 186.3054,
+      "step": 10670
+    },
+    {
+      "epoch": 0.08834843032634322,
+      "grad_norm": 884.4146118164062,
+      "learning_rate": 8.630303030303032e-06,
+      "loss": 157.2276,
+      "step": 10680
+    },
+    {
+      "epoch": 0.08843115357571245,
+      "grad_norm": 1087.9329833984375,
+      "learning_rate": 8.63838383838384e-06,
+      "loss": 154.3768,
+      "step": 10690
+    },
+    {
+      "epoch": 0.08851387682508169,
+      "grad_norm": 913.88818359375,
+      "learning_rate": 8.646464646464647e-06,
+      "loss": 185.6447,
+      "step": 10700
+    },
+    {
+      "epoch": 0.08859660007445093,
+      "grad_norm": 2494.311279296875,
+      "learning_rate": 8.654545454545455e-06,
+      "loss": 195.7933,
+      "step": 10710
+    },
+    {
+      "epoch": 0.08867932332382016,
+      "grad_norm": 1353.1029052734375,
+      "learning_rate": 8.662626262626264e-06,
+      "loss": 179.2149,
+      "step": 10720
+    },
+    {
+      "epoch": 0.0887620465731894,
+      "grad_norm": 1235.08056640625,
+      "learning_rate": 8.670707070707072e-06,
+      "loss": 140.4072,
+      "step": 10730
+    },
+    {
+      "epoch": 0.08884476982255864,
+      "grad_norm": 1513.452880859375,
+      "learning_rate": 8.67878787878788e-06,
+      "loss": 188.1772,
+      "step": 10740
+    },
+    {
+      "epoch": 0.08892749307192786,
+      "grad_norm": 576.0892333984375,
+      "learning_rate": 8.686868686868687e-06,
+      "loss": 149.6942,
+      "step": 10750
+    },
+    {
+      "epoch": 0.0890102163212971,
+      "grad_norm": 1103.5166015625,
+      "learning_rate": 8.694949494949496e-06,
+      "loss": 165.9619,
+      "step": 10760
+    },
+    {
+      "epoch": 0.08909293957066633,
+      "grad_norm": 964.4625244140625,
+      "learning_rate": 8.703030303030304e-06,
+      "loss": 138.5367,
+      "step": 10770
+    },
+    {
+      "epoch": 0.08917566282003557,
+      "grad_norm": 1614.7874755859375,
+      "learning_rate": 8.711111111111111e-06,
+      "loss": 175.2976,
+      "step": 10780
+    },
+    {
+      "epoch": 0.08925838606940481,
+      "grad_norm": 1648.361328125,
+      "learning_rate": 8.71919191919192e-06,
+      "loss": 187.3021,
+      "step": 10790
+    },
+    {
+      "epoch": 0.08934110931877404,
+      "grad_norm": 1707.2655029296875,
+      "learning_rate": 8.727272727272728e-06,
+      "loss": 258.8253,
+      "step": 10800
+    },
+    {
+      "epoch": 0.08942383256814328,
+      "grad_norm": 1160.01806640625,
+      "learning_rate": 8.735353535353536e-06,
+      "loss": 194.5743,
+      "step": 10810
+    },
+    {
+      "epoch": 0.08950655581751252,
+      "grad_norm": 1146.39990234375,
+      "learning_rate": 8.743434343434343e-06,
+      "loss": 203.5324,
+      "step": 10820
+    },
+    {
+      "epoch": 0.08958927906688174,
+      "grad_norm": 3626.928955078125,
+      "learning_rate": 8.751515151515151e-06,
+      "loss": 253.2095,
+      "step": 10830
+    },
+    {
+      "epoch": 0.08967200231625098,
+      "grad_norm": 728.5478515625,
+      "learning_rate": 8.75959595959596e-06,
+      "loss": 145.7956,
+      "step": 10840
+    },
+    {
+      "epoch": 0.08975472556562021,
+      "grad_norm": 1002.9689331054688,
+      "learning_rate": 8.767676767676768e-06,
+      "loss": 181.9694,
+      "step": 10850
+    },
+    {
+      "epoch": 0.08983744881498945,
+      "grad_norm": 1103.3865966796875,
+      "learning_rate": 8.775757575757577e-06,
+      "loss": 177.4934,
+      "step": 10860
+    },
+    {
+      "epoch": 0.08992017206435869,
+      "grad_norm": 1987.7679443359375,
+      "learning_rate": 8.783838383838385e-06,
+      "loss": 190.5272,
+      "step": 10870
+    },
+    {
+      "epoch": 0.09000289531372792,
+      "grad_norm": 1918.5572509765625,
+      "learning_rate": 8.791919191919192e-06,
+      "loss": 176.4213,
+      "step": 10880
+    },
+    {
+      "epoch": 0.09008561856309716,
+      "grad_norm": 979.5933837890625,
+      "learning_rate": 8.8e-06,
+      "loss": 208.7932,
+      "step": 10890
+    },
+    {
+      "epoch": 0.0901683418124664,
+      "grad_norm": 1254.18603515625,
+      "learning_rate": 8.808080808080809e-06,
+      "loss": 179.7351,
+      "step": 10900
+    },
+    {
+      "epoch": 0.09025106506183563,
+      "grad_norm": 1393.46044921875,
+      "learning_rate": 8.816161616161617e-06,
+      "loss": 239.5334,
+      "step": 10910
+    },
+    {
+      "epoch": 0.09033378831120487,
+      "grad_norm": 932.0426025390625,
+      "learning_rate": 8.824242424242426e-06,
+      "loss": 159.2398,
+      "step": 10920
+    },
+    {
+      "epoch": 0.0904165115605741,
+      "grad_norm": 1175.749755859375,
+      "learning_rate": 8.832323232323233e-06,
+      "loss": 196.0824,
+      "step": 10930
+    },
+    {
+      "epoch": 0.09049923480994333,
+      "grad_norm": 669.9056396484375,
+      "learning_rate": 8.840404040404041e-06,
+      "loss": 195.3154,
+      "step": 10940
+    },
+    {
+      "epoch": 0.09058195805931257,
+      "grad_norm": 2161.71826171875,
+      "learning_rate": 8.84848484848485e-06,
+      "loss": 201.2208,
+      "step": 10950
+    },
+    {
+      "epoch": 0.0906646813086818,
+      "grad_norm": 1083.76318359375,
+      "learning_rate": 8.856565656565658e-06,
+      "loss": 181.3205,
+      "step": 10960
+    },
+    {
+      "epoch": 0.09074740455805104,
+      "grad_norm": 1360.2738037109375,
+      "learning_rate": 8.864646464646466e-06,
+      "loss": 173.5026,
+      "step": 10970
+    },
+    {
+      "epoch": 0.09083012780742028,
+      "grad_norm": 1274.3690185546875,
+      "learning_rate": 8.872727272727275e-06,
+      "loss": 191.3013,
+      "step": 10980
+    },
+    {
+      "epoch": 0.0909128510567895,
+      "grad_norm": 1275.59814453125,
+      "learning_rate": 8.880808080808082e-06,
+      "loss": 180.379,
+      "step": 10990
+    },
+    {
+      "epoch": 0.09099557430615875,
+      "grad_norm": 1908.28125,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 226.928,
+      "step": 11000
+    },
+    {
+      "epoch": 0.09107829755552799,
+      "grad_norm": 1636.098388671875,
+      "learning_rate": 8.896969696969697e-06,
+      "loss": 166.5137,
+      "step": 11010
+    },
+    {
+      "epoch": 0.09116102080489721,
+      "grad_norm": 1362.2508544921875,
+      "learning_rate": 8.905050505050505e-06,
+      "loss": 162.6263,
+      "step": 11020
+    },
+    {
+      "epoch": 0.09124374405426645,
+      "grad_norm": 958.8033447265625,
+      "learning_rate": 8.913131313131314e-06,
+      "loss": 126.4212,
+      "step": 11030
+    },
+    {
+      "epoch": 0.09132646730363568,
+      "grad_norm": 4022.8896484375,
+      "learning_rate": 8.921212121212122e-06,
+      "loss": 171.8223,
+      "step": 11040
+    },
+    {
+      "epoch": 0.09140919055300492,
+      "grad_norm": 1437.0355224609375,
+      "learning_rate": 8.92929292929293e-06,
+      "loss": 150.3858,
+      "step": 11050
+    },
+    {
+      "epoch": 0.09149191380237416,
+      "grad_norm": 1457.9029541015625,
+      "learning_rate": 8.937373737373737e-06,
+      "loss": 191.1608,
+      "step": 11060
+    },
+    {
+      "epoch": 0.09157463705174339,
+      "grad_norm": 1079.6673583984375,
+      "learning_rate": 8.945454545454546e-06,
+      "loss": 169.316,
+      "step": 11070
+    },
+    {
+      "epoch": 0.09165736030111263,
+      "grad_norm": 1028.742431640625,
+      "learning_rate": 8.953535353535354e-06,
+      "loss": 141.8136,
+      "step": 11080
+    },
+    {
+      "epoch": 0.09174008355048187,
+      "grad_norm": 1539.916259765625,
+      "learning_rate": 8.961616161616163e-06,
+      "loss": 196.1172,
+      "step": 11090
+    },
+    {
+      "epoch": 0.0918228067998511,
+      "grad_norm": 1310.006591796875,
+      "learning_rate": 8.969696969696971e-06,
+      "loss": 153.3139,
+      "step": 11100
+    },
+    {
+      "epoch": 0.09190553004922034,
+      "grad_norm": 3632.826416015625,
+      "learning_rate": 8.977777777777778e-06,
+      "loss": 274.3921,
+      "step": 11110
+    },
+    {
+      "epoch": 0.09198825329858956,
+      "grad_norm": 1360.5455322265625,
+      "learning_rate": 8.985858585858586e-06,
+      "loss": 199.4904,
+      "step": 11120
+    },
+    {
+      "epoch": 0.0920709765479588,
+      "grad_norm": 1219.487548828125,
+      "learning_rate": 8.993939393939395e-06,
+      "loss": 138.8559,
+      "step": 11130
+    },
+    {
+      "epoch": 0.09215369979732804,
+      "grad_norm": 1320.4920654296875,
+      "learning_rate": 9.002020202020203e-06,
+      "loss": 179.0958,
+      "step": 11140
+    },
+    {
+      "epoch": 0.09223642304669727,
+      "grad_norm": 754.0531616210938,
+      "learning_rate": 9.010101010101012e-06,
+      "loss": 183.8919,
+      "step": 11150
+    },
+    {
+      "epoch": 0.09231914629606651,
+      "grad_norm": 1132.8018798828125,
+      "learning_rate": 9.01818181818182e-06,
+      "loss": 138.0276,
+      "step": 11160
+    },
+    {
+      "epoch": 0.09240186954543575,
+      "grad_norm": 1015.0302124023438,
+      "learning_rate": 9.026262626262627e-06,
+      "loss": 169.833,
+      "step": 11170
+    },
+    {
+      "epoch": 0.09248459279480498,
+      "grad_norm": 1256.841552734375,
+      "learning_rate": 9.034343434343435e-06,
+      "loss": 192.5117,
+      "step": 11180
+    },
+    {
+      "epoch": 0.09256731604417422,
+      "grad_norm": 637.6029663085938,
+      "learning_rate": 9.042424242424244e-06,
+      "loss": 136.4471,
+      "step": 11190
+    },
+    {
+      "epoch": 0.09265003929354346,
+      "grad_norm": 2063.305419921875,
+      "learning_rate": 9.050505050505052e-06,
+      "loss": 187.3035,
+      "step": 11200
+    },
+    {
+      "epoch": 0.09273276254291268,
+      "grad_norm": 893.4026489257812,
+      "learning_rate": 9.058585858585859e-06,
+      "loss": 197.6645,
+      "step": 11210
+    },
+    {
+      "epoch": 0.09281548579228192,
+      "grad_norm": 11727.92578125,
+      "learning_rate": 9.066666666666667e-06,
+      "loss": 184.4137,
+      "step": 11220
+    },
+    {
+      "epoch": 0.09289820904165115,
+      "grad_norm": 939.4194946289062,
+      "learning_rate": 9.074747474747476e-06,
+      "loss": 163.7697,
+      "step": 11230
+    },
+    {
+      "epoch": 0.09298093229102039,
+      "grad_norm": 1211.7532958984375,
+      "learning_rate": 9.082828282828283e-06,
+      "loss": 150.1024,
+      "step": 11240
+    },
+    {
+      "epoch": 0.09306365554038963,
+      "grad_norm": 1020.1021118164062,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 176.2725,
+      "step": 11250
+    },
+    {
+      "epoch": 0.09314637878975886,
+      "grad_norm": 1597.582763671875,
+      "learning_rate": 9.0989898989899e-06,
+      "loss": 161.5863,
+      "step": 11260
+    },
+    {
+      "epoch": 0.0932291020391281,
+      "grad_norm": 1205.4434814453125,
+      "learning_rate": 9.107070707070708e-06,
+      "loss": 191.3436,
+      "step": 11270
+    },
+    {
+      "epoch": 0.09331182528849734,
+      "grad_norm": 1347.427001953125,
+      "learning_rate": 9.115151515151516e-06,
+      "loss": 147.056,
+      "step": 11280
+    },
+    {
+      "epoch": 0.09339454853786656,
+      "grad_norm": 1340.0323486328125,
+      "learning_rate": 9.123232323232323e-06,
+      "loss": 166.0659,
+      "step": 11290
+    },
+    {
+      "epoch": 0.0934772717872358,
+      "grad_norm": 1243.469482421875,
+      "learning_rate": 9.131313131313132e-06,
+      "loss": 160.2771,
+      "step": 11300
+    },
+    {
+      "epoch": 0.09355999503660503,
+      "grad_norm": 1073.79638671875,
+      "learning_rate": 9.13939393939394e-06,
+      "loss": 189.4957,
+      "step": 11310
+    },
+    {
+      "epoch": 0.09364271828597427,
+      "grad_norm": 1185.810546875,
+      "learning_rate": 9.147474747474748e-06,
+      "loss": 169.7764,
+      "step": 11320
+    },
+    {
+      "epoch": 0.09372544153534351,
+      "grad_norm": 1002.9879760742188,
+      "learning_rate": 9.155555555555557e-06,
+      "loss": 117.5662,
+      "step": 11330
+    },
+    {
+      "epoch": 0.09380816478471274,
+      "grad_norm": 1063.5313720703125,
+      "learning_rate": 9.163636363636365e-06,
+      "loss": 199.0167,
+      "step": 11340
+    },
+    {
+      "epoch": 0.09389088803408198,
+      "grad_norm": 975.7847290039062,
+      "learning_rate": 9.171717171717172e-06,
+      "loss": 178.3596,
+      "step": 11350
+    },
+    {
+      "epoch": 0.09397361128345122,
+      "grad_norm": 1078.6475830078125,
+      "learning_rate": 9.17979797979798e-06,
+      "loss": 167.8156,
+      "step": 11360
+    },
+    {
+      "epoch": 0.09405633453282045,
+      "grad_norm": 2150.571044921875,
+      "learning_rate": 9.187878787878789e-06,
+      "loss": 151.6369,
+      "step": 11370
+    },
+    {
+      "epoch": 0.09413905778218969,
+      "grad_norm": 1107.6488037109375,
+      "learning_rate": 9.195959595959597e-06,
+      "loss": 138.8753,
+      "step": 11380
+    },
+    {
+      "epoch": 0.09422178103155891,
+      "grad_norm": 2013.7427978515625,
+      "learning_rate": 9.204040404040406e-06,
+      "loss": 168.8607,
+      "step": 11390
+    },
+    {
+      "epoch": 0.09430450428092815,
+      "grad_norm": 926.5513916015625,
+      "learning_rate": 9.212121212121213e-06,
+      "loss": 155.2859,
+      "step": 11400
+    },
+    {
+      "epoch": 0.0943872275302974,
+      "grad_norm": 640.766357421875,
+      "learning_rate": 9.220202020202021e-06,
+      "loss": 170.4931,
+      "step": 11410
+    },
+    {
+      "epoch": 0.09446995077966662,
+      "grad_norm": 2521.883544921875,
+      "learning_rate": 9.228282828282828e-06,
+      "loss": 163.8456,
+      "step": 11420
+    },
+    {
+      "epoch": 0.09455267402903586,
+      "grad_norm": 2574.5419921875,
+      "learning_rate": 9.236363636363636e-06,
+      "loss": 192.8077,
+      "step": 11430
+    },
+    {
+      "epoch": 0.0946353972784051,
+      "grad_norm": 1514.116943359375,
+      "learning_rate": 9.244444444444445e-06,
+      "loss": 190.4653,
+      "step": 11440
+    },
+    {
+      "epoch": 0.09471812052777433,
+      "grad_norm": 1116.5655517578125,
+      "learning_rate": 9.252525252525253e-06,
+      "loss": 150.5008,
+      "step": 11450
+    },
+    {
+      "epoch": 0.09480084377714357,
+      "grad_norm": 7808.6015625,
+      "learning_rate": 9.260606060606062e-06,
+      "loss": 282.1488,
+      "step": 11460
+    },
+    {
+      "epoch": 0.09488356702651281,
+      "grad_norm": 1106.9825439453125,
+      "learning_rate": 9.268686868686868e-06,
+      "loss": 195.7642,
+      "step": 11470
+    },
+    {
+      "epoch": 0.09496629027588203,
+      "grad_norm": 3581.72900390625,
+      "learning_rate": 9.276767676767677e-06,
+      "loss": 179.7523,
+      "step": 11480
+    },
+    {
+      "epoch": 0.09504901352525127,
+      "grad_norm": 1101.411865234375,
+      "learning_rate": 9.284848484848485e-06,
+      "loss": 199.7394,
+      "step": 11490
+    },
+    {
+      "epoch": 0.0951317367746205,
+      "grad_norm": 1929.9052734375,
+      "learning_rate": 9.292929292929294e-06,
+      "loss": 174.8564,
+      "step": 11500
+    },
+    {
+      "epoch": 0.09521446002398974,
+      "grad_norm": 1121.5028076171875,
+      "learning_rate": 9.301010101010102e-06,
+      "loss": 134.3296,
+      "step": 11510
+    },
+    {
+      "epoch": 0.09529718327335898,
+      "grad_norm": 1080.38671875,
+      "learning_rate": 9.30909090909091e-06,
+      "loss": 173.9918,
+      "step": 11520
+    },
+    {
+      "epoch": 0.09537990652272821,
+      "grad_norm": 1371.1961669921875,
+      "learning_rate": 9.317171717171717e-06,
+      "loss": 213.5175,
+      "step": 11530
+    },
+    {
+      "epoch": 0.09546262977209745,
+      "grad_norm": 1480.9495849609375,
+      "learning_rate": 9.325252525252526e-06,
+      "loss": 136.4908,
+      "step": 11540
+    },
+    {
+      "epoch": 0.09554535302146669,
+      "grad_norm": 912.1797485351562,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 215.026,
+      "step": 11550
+    },
+    {
+      "epoch": 0.09562807627083592,
+      "grad_norm": 2468.00341796875,
+      "learning_rate": 9.341414141414143e-06,
+      "loss": 175.5986,
+      "step": 11560
+    },
+    {
+      "epoch": 0.09571079952020516,
+      "grad_norm": 1296.7786865234375,
+      "learning_rate": 9.349494949494951e-06,
+      "loss": 209.1408,
+      "step": 11570
+    },
+    {
+      "epoch": 0.09579352276957438,
+      "grad_norm": 1264.5654296875,
+      "learning_rate": 9.357575757575758e-06,
+      "loss": 200.8274,
+      "step": 11580
+    },
+    {
+      "epoch": 0.09587624601894362,
+      "grad_norm": 1155.8309326171875,
+      "learning_rate": 9.365656565656566e-06,
+      "loss": 164.7913,
+      "step": 11590
+    },
+    {
+      "epoch": 0.09595896926831286,
+      "grad_norm": 1020.3081665039062,
+      "learning_rate": 9.373737373737375e-06,
+      "loss": 155.3867,
+      "step": 11600
+    },
+    {
+      "epoch": 0.09604169251768209,
+      "grad_norm": 1648.6768798828125,
+      "learning_rate": 9.381818181818183e-06,
+      "loss": 150.0429,
+      "step": 11610
+    },
+    {
+      "epoch": 0.09612441576705133,
+      "grad_norm": 1153.5174560546875,
+      "learning_rate": 9.389898989898992e-06,
+      "loss": 183.8605,
+      "step": 11620
+    },
+    {
+      "epoch": 0.09620713901642057,
+      "grad_norm": 1894.8492431640625,
+      "learning_rate": 9.397979797979799e-06,
+      "loss": 178.0279,
+      "step": 11630
+    },
+    {
+      "epoch": 0.0962898622657898,
+      "grad_norm": 1127.750244140625,
+      "learning_rate": 9.406060606060607e-06,
+      "loss": 162.6299,
+      "step": 11640
+    },
+    {
+      "epoch": 0.09637258551515904,
+      "grad_norm": 1268.74267578125,
+      "learning_rate": 9.414141414141414e-06,
+      "loss": 155.4865,
+      "step": 11650
+    },
+    {
+      "epoch": 0.09645530876452828,
+      "grad_norm": 1007.6961669921875,
+      "learning_rate": 9.422222222222222e-06,
+      "loss": 209.233,
+      "step": 11660
+    },
+    {
+      "epoch": 0.0965380320138975,
+      "grad_norm": 1457.4285888671875,
+      "learning_rate": 9.43030303030303e-06,
+      "loss": 163.3144,
+      "step": 11670
+    },
+    {
+      "epoch": 0.09662075526326674,
+      "grad_norm": 1042.603515625,
+      "learning_rate": 9.438383838383839e-06,
+      "loss": 182.3341,
+      "step": 11680
+    },
+    {
+      "epoch": 0.09670347851263597,
+      "grad_norm": 1047.1702880859375,
+      "learning_rate": 9.446464646464648e-06,
+      "loss": 167.2329,
+      "step": 11690
+    },
+    {
+      "epoch": 0.09678620176200521,
+      "grad_norm": 1840.010009765625,
+      "learning_rate": 9.454545454545456e-06,
+      "loss": 204.4755,
+      "step": 11700
+    },
+    {
+      "epoch": 0.09686892501137445,
+      "grad_norm": 573.9185180664062,
+      "learning_rate": 9.462626262626263e-06,
+      "loss": 106.0682,
+      "step": 11710
+    },
+    {
+      "epoch": 0.09695164826074368,
+      "grad_norm": 1605.5093994140625,
+      "learning_rate": 9.470707070707071e-06,
+      "loss": 214.7734,
+      "step": 11720
+    },
+    {
+      "epoch": 0.09703437151011292,
+      "grad_norm": 2460.1376953125,
+      "learning_rate": 9.47878787878788e-06,
+      "loss": 180.7382,
+      "step": 11730
+    },
+    {
+      "epoch": 0.09711709475948216,
+      "grad_norm": 1100.57080078125,
+      "learning_rate": 9.486868686868688e-06,
+      "loss": 220.1118,
+      "step": 11740
+    },
+    {
+      "epoch": 0.09719981800885139,
+      "grad_norm": 937.7584838867188,
+      "learning_rate": 9.494949494949497e-06,
+      "loss": 176.8332,
+      "step": 11750
+    },
+    {
+      "epoch": 0.09728254125822063,
+      "grad_norm": 1534.452880859375,
+      "learning_rate": 9.503030303030303e-06,
+      "loss": 173.9935,
+      "step": 11760
+    },
+    {
+      "epoch": 0.09736526450758985,
+      "grad_norm": 1466.6376953125,
+      "learning_rate": 9.511111111111112e-06,
+      "loss": 196.6087,
+      "step": 11770
+    },
+    {
+      "epoch": 0.09744798775695909,
+      "grad_norm": 1731.87841796875,
+      "learning_rate": 9.51919191919192e-06,
+      "loss": 179.0696,
+      "step": 11780
+    },
+    {
+      "epoch": 0.09753071100632833,
+      "grad_norm": 1235.0560302734375,
+      "learning_rate": 9.527272727272729e-06,
+      "loss": 164.9756,
+      "step": 11790
+    },
+    {
+      "epoch": 0.09761343425569756,
+      "grad_norm": 662.07568359375,
+      "learning_rate": 9.535353535353537e-06,
+      "loss": 155.2859,
+      "step": 11800
+    },
+    {
+      "epoch": 0.0976961575050668,
+      "grad_norm": 642.6212158203125,
+      "learning_rate": 9.543434343434344e-06,
+      "loss": 183.2646,
+      "step": 11810
+    },
+    {
+      "epoch": 0.09777888075443604,
+      "grad_norm": 861.9931030273438,
+      "learning_rate": 9.551515151515152e-06,
+      "loss": 140.6058,
+      "step": 11820
+    },
+    {
+      "epoch": 0.09786160400380527,
+      "grad_norm": 2064.848388671875,
+      "learning_rate": 9.55959595959596e-06,
+      "loss": 175.8456,
+      "step": 11830
+    },
+    {
+      "epoch": 0.0979443272531745,
+      "grad_norm": 1402.7391357421875,
+      "learning_rate": 9.56767676767677e-06,
+      "loss": 168.1851,
+      "step": 11840
+    },
+    {
+      "epoch": 0.09802705050254373,
+      "grad_norm": 1274.8916015625,
+      "learning_rate": 9.575757575757576e-06,
+      "loss": 177.3613,
+      "step": 11850
+    },
+    {
+      "epoch": 0.09810977375191297,
+      "grad_norm": 1175.4793701171875,
+      "learning_rate": 9.583838383838384e-06,
+      "loss": 155.9946,
+      "step": 11860
+    },
+    {
+      "epoch": 0.09819249700128221,
+      "grad_norm": 1113.3656005859375,
+      "learning_rate": 9.591919191919193e-06,
+      "loss": 174.7684,
+      "step": 11870
+    },
+    {
+      "epoch": 0.09827522025065144,
+      "grad_norm": 1227.9544677734375,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 173.7199,
+      "step": 11880
+    },
+    {
+      "epoch": 0.09835794350002068,
+      "grad_norm": 1131.5029296875,
+      "learning_rate": 9.608080808080808e-06,
+      "loss": 179.6841,
+      "step": 11890
+    },
+    {
+      "epoch": 0.09844066674938992,
+      "grad_norm": 1879.9422607421875,
+      "learning_rate": 9.616161616161616e-06,
+      "loss": 201.8642,
+      "step": 11900
+    },
+    {
+      "epoch": 0.09852338999875915,
+      "grad_norm": 1383.9271240234375,
+      "learning_rate": 9.624242424242425e-06,
+      "loss": 155.471,
+      "step": 11910
+    },
+    {
+      "epoch": 0.09860611324812839,
+      "grad_norm": 1202.215576171875,
+      "learning_rate": 9.632323232323233e-06,
+      "loss": 176.2691,
+      "step": 11920
+    },
+    {
+      "epoch": 0.09868883649749763,
+      "grad_norm": 713.47216796875,
+      "learning_rate": 9.640404040404042e-06,
+      "loss": 129.0435,
+      "step": 11930
+    },
+    {
+      "epoch": 0.09877155974686685,
+      "grad_norm": 1676.49658203125,
+      "learning_rate": 9.648484848484849e-06,
+      "loss": 158.0141,
+      "step": 11940
+    },
+    {
+      "epoch": 0.0988542829962361,
+      "grad_norm": 822.8709716796875,
+      "learning_rate": 9.656565656565657e-06,
+      "loss": 128.7683,
+      "step": 11950
+    },
+    {
+      "epoch": 0.09893700624560532,
+      "grad_norm": 2505.596923828125,
+      "learning_rate": 9.664646464646465e-06,
+      "loss": 173.4101,
+      "step": 11960
+    },
+    {
+      "epoch": 0.09901972949497456,
+      "grad_norm": 764.0625610351562,
+      "learning_rate": 9.672727272727274e-06,
+      "loss": 186.7536,
+      "step": 11970
+    },
+    {
+      "epoch": 0.0991024527443438,
+      "grad_norm": 1434.5181884765625,
+      "learning_rate": 9.680808080808082e-06,
+      "loss": 195.2917,
+      "step": 11980
+    },
+    {
+      "epoch": 0.09918517599371303,
+      "grad_norm": 916.0157470703125,
+      "learning_rate": 9.688888888888889e-06,
+      "loss": 151.0252,
+      "step": 11990
+    },
+    {
+      "epoch": 0.09926789924308227,
+      "grad_norm": 1034.74072265625,
+      "learning_rate": 9.696969696969698e-06,
+      "loss": 163.2282,
+      "step": 12000
+    },
+    {
+      "epoch": 0.09935062249245151,
+      "grad_norm": 1380.8863525390625,
+      "learning_rate": 9.705050505050506e-06,
+      "loss": 139.9053,
+      "step": 12010
+    },
+    {
+      "epoch": 0.09943334574182074,
+      "grad_norm": 911.8162231445312,
+      "learning_rate": 9.713131313131314e-06,
+      "loss": 159.4434,
+      "step": 12020
+    },
+    {
+      "epoch": 0.09951606899118998,
+      "grad_norm": 1710.5338134765625,
+      "learning_rate": 9.721212121212123e-06,
+      "loss": 152.8905,
+      "step": 12030
+    },
+    {
+      "epoch": 0.0995987922405592,
+      "grad_norm": 1827.9671630859375,
+      "learning_rate": 9.729292929292931e-06,
+      "loss": 195.7561,
+      "step": 12040
+    },
+    {
+      "epoch": 0.09968151548992844,
+      "grad_norm": 1594.53759765625,
+      "learning_rate": 9.737373737373738e-06,
+      "loss": 172.4959,
+      "step": 12050
+    },
+    {
+      "epoch": 0.09976423873929768,
+      "grad_norm": 985.9189453125,
+      "learning_rate": 9.745454545454547e-06,
+      "loss": 167.3074,
+      "step": 12060
+    },
+    {
+      "epoch": 0.09984696198866691,
+      "grad_norm": 1075.2579345703125,
+      "learning_rate": 9.753535353535353e-06,
+      "loss": 163.0599,
+      "step": 12070
+    },
+    {
+      "epoch": 0.09992968523803615,
+      "grad_norm": 2030.679443359375,
+      "learning_rate": 9.761616161616162e-06,
+      "loss": 163.872,
+      "step": 12080
+    },
+    {
+      "epoch": 0.10001240848740539,
+      "grad_norm": 1216.58984375,
+      "learning_rate": 9.76969696969697e-06,
+      "loss": 158.4474,
+      "step": 12090
+    },
+    {
+      "epoch": 0.10009513173677462,
+      "grad_norm": 1842.97998046875,
+      "learning_rate": 9.777777777777779e-06,
+      "loss": 225.2362,
+      "step": 12100
+    },
+    {
+      "epoch": 0.10017785498614386,
+      "grad_norm": 1359.1461181640625,
+      "learning_rate": 9.785858585858587e-06,
+      "loss": 172.9057,
+      "step": 12110
+    },
+    {
+      "epoch": 0.10026057823551308,
+      "grad_norm": 1055.8148193359375,
+      "learning_rate": 9.793939393939394e-06,
+      "loss": 153.7417,
+      "step": 12120
+    },
+    {
+      "epoch": 0.10034330148488232,
+      "grad_norm": 1045.6163330078125,
+      "learning_rate": 9.802020202020202e-06,
+      "loss": 151.8674,
+      "step": 12130
+    },
+    {
+      "epoch": 0.10042602473425156,
+      "grad_norm": 702.3483276367188,
+      "learning_rate": 9.81010101010101e-06,
+      "loss": 154.9376,
+      "step": 12140
+    },
+    {
+      "epoch": 0.10050874798362079,
+      "grad_norm": 582.2290649414062,
+      "learning_rate": 9.81818181818182e-06,
+      "loss": 191.1368,
+      "step": 12150
+    },
+    {
+      "epoch": 0.10059147123299003,
+      "grad_norm": 1067.549072265625,
+      "learning_rate": 9.826262626262628e-06,
+      "loss": 215.4936,
+      "step": 12160
+    },
+    {
+      "epoch": 0.10067419448235927,
+      "grad_norm": 1477.08251953125,
+      "learning_rate": 9.834343434343434e-06,
+      "loss": 179.9154,
+      "step": 12170
+    },
+    {
+      "epoch": 0.1007569177317285,
+      "grad_norm": 1615.7445068359375,
+      "learning_rate": 9.842424242424243e-06,
+      "loss": 151.6433,
+      "step": 12180
+    },
+    {
+      "epoch": 0.10083964098109774,
+      "grad_norm": 1283.7108154296875,
+      "learning_rate": 9.850505050505051e-06,
+      "loss": 176.6817,
+      "step": 12190
+    },
+    {
+      "epoch": 0.10092236423046698,
+      "grad_norm": 1741.5172119140625,
+      "learning_rate": 9.85858585858586e-06,
+      "loss": 148.4978,
+      "step": 12200
+    },
+    {
+      "epoch": 0.1010050874798362,
+      "grad_norm": 956.2932739257812,
+      "learning_rate": 9.866666666666668e-06,
+      "loss": 178.0466,
+      "step": 12210
+    },
+    {
+      "epoch": 0.10108781072920545,
+      "grad_norm": 771.7984008789062,
+      "learning_rate": 9.874747474747477e-06,
+      "loss": 177.3261,
+      "step": 12220
+    },
+    {
+      "epoch": 0.10117053397857467,
+      "grad_norm": 703.4427490234375,
+      "learning_rate": 9.882828282828283e-06,
+      "loss": 175.3549,
+      "step": 12230
+    },
+    {
+      "epoch": 0.10125325722794391,
+      "grad_norm": 931.7167358398438,
+      "learning_rate": 9.890909090909092e-06,
+      "loss": 158.6873,
+      "step": 12240
+    },
+    {
+      "epoch": 0.10133598047731315,
+      "grad_norm": 1080.5570068359375,
+      "learning_rate": 9.8989898989899e-06,
+      "loss": 109.4355,
+      "step": 12250
+    },
+    {
+      "epoch": 0.10141870372668238,
+      "grad_norm": 1163.09423828125,
+      "learning_rate": 9.907070707070709e-06,
+      "loss": 164.6523,
+      "step": 12260
+    },
+    {
+      "epoch": 0.10150142697605162,
+      "grad_norm": 870.4569091796875,
+      "learning_rate": 9.915151515151515e-06,
+      "loss": 162.8174,
+      "step": 12270
+    },
+    {
+      "epoch": 0.10158415022542086,
+      "grad_norm": 976.0827026367188,
+      "learning_rate": 9.923232323232324e-06,
+      "loss": 149.6927,
+      "step": 12280
+    },
+    {
+      "epoch": 0.10166687347479009,
+      "grad_norm": 881.8626098632812,
+      "learning_rate": 9.931313131313132e-06,
+      "loss": 128.7362,
+      "step": 12290
+    },
+    {
+      "epoch": 0.10174959672415933,
+      "grad_norm": 612.5805053710938,
+      "learning_rate": 9.939393939393939e-06,
+      "loss": 129.0437,
+      "step": 12300
+    },
+    {
+      "epoch": 0.10183231997352855,
+      "grad_norm": 1372.064453125,
+      "learning_rate": 9.947474747474748e-06,
+      "loss": 188.9469,
+      "step": 12310
+    },
+    {
+      "epoch": 0.1019150432228978,
+      "grad_norm": 1295.59375,
+      "learning_rate": 9.955555555555556e-06,
+      "loss": 167.0848,
+      "step": 12320
+    },
+    {
+      "epoch": 0.10199776647226703,
+      "grad_norm": 1037.228515625,
+      "learning_rate": 9.963636363636364e-06,
+      "loss": 176.8787,
+      "step": 12330
+    },
+    {
+      "epoch": 0.10208048972163626,
+      "grad_norm": 1227.047607421875,
+      "learning_rate": 9.971717171717173e-06,
+      "loss": 198.2504,
+      "step": 12340
+    },
+    {
+      "epoch": 0.1021632129710055,
+      "grad_norm": 1416.220458984375,
+      "learning_rate": 9.97979797979798e-06,
+      "loss": 179.9958,
+      "step": 12350
+    },
+    {
+      "epoch": 0.10224593622037474,
+      "grad_norm": 1424.654052734375,
+      "learning_rate": 9.987878787878788e-06,
+      "loss": 191.8388,
+      "step": 12360
+    },
+    {
+      "epoch": 0.10232865946974397,
+      "grad_norm": 1317.3236083984375,
+      "learning_rate": 9.995959595959597e-06,
+      "loss": 167.6917,
+      "step": 12370
+    },
+    {
+      "epoch": 0.10241138271911321,
+      "grad_norm": 868.8121337890625,
+      "learning_rate": 9.99999995027162e-06,
+      "loss": 156.1878,
+      "step": 12380
+    },
+    {
+      "epoch": 0.10249410596848244,
+      "grad_norm": 889.690185546875,
+      "learning_rate": 9.99999955244457e-06,
+      "loss": 188.0309,
+      "step": 12390
+    },
+    {
+      "epoch": 0.10257682921785168,
+      "grad_norm": 964.9615478515625,
+      "learning_rate": 9.999998756790503e-06,
+      "loss": 162.0736,
+      "step": 12400
+    },
+    {
+      "epoch": 0.10265955246722092,
+      "grad_norm": 2292.847900390625,
+      "learning_rate": 9.999997563309483e-06,
+      "loss": 138.4502,
+      "step": 12410
+    },
+    {
+      "epoch": 0.10274227571659014,
+      "grad_norm": 1073.1864013671875,
+      "learning_rate": 9.999995972001602e-06,
+      "loss": 183.7101,
+      "step": 12420
+    },
+    {
+      "epoch": 0.10282499896595938,
+      "grad_norm": 1411.09521484375,
+      "learning_rate": 9.99999398286699e-06,
+      "loss": 172.5208,
+      "step": 12430
+    },
+    {
+      "epoch": 0.10290772221532862,
+      "grad_norm": 1048.4619140625,
+      "learning_rate": 9.999991595905803e-06,
+      "loss": 169.999,
+      "step": 12440
+    },
+    {
+      "epoch": 0.10299044546469785,
+      "grad_norm": 986.3778686523438,
+      "learning_rate": 9.999988811118232e-06,
+      "loss": 146.7089,
+      "step": 12450
+    },
+    {
+      "epoch": 0.10307316871406709,
+      "grad_norm": 795.2116088867188,
+      "learning_rate": 9.999985628504498e-06,
+      "loss": 159.4188,
+      "step": 12460
+    },
+    {
+      "epoch": 0.10315589196343633,
+      "grad_norm": 919.0172119140625,
+      "learning_rate": 9.999982048064854e-06,
+      "loss": 144.6598,
+      "step": 12470
+    },
+    {
+      "epoch": 0.10323861521280556,
+      "grad_norm": 1297.25732421875,
+      "learning_rate": 9.999978069799585e-06,
+      "loss": 171.9091,
+      "step": 12480
+    },
+    {
+      "epoch": 0.1033213384621748,
+      "grad_norm": 1352.2923583984375,
+      "learning_rate": 9.999973693709008e-06,
+      "loss": 151.0232,
+      "step": 12490
+    },
+    {
+      "epoch": 0.10340406171154402,
+      "grad_norm": 1723.31640625,
+      "learning_rate": 9.99996891979347e-06,
+      "loss": 131.0583,
+      "step": 12500
+    },
+    {
+      "epoch": 0.10348678496091326,
+      "grad_norm": 1077.4090576171875,
+      "learning_rate": 9.999963748053354e-06,
+      "loss": 155.0065,
+      "step": 12510
+    },
+    {
+      "epoch": 0.1035695082102825,
+      "grad_norm": 881.5154418945312,
+      "learning_rate": 9.999958178489069e-06,
+      "loss": 150.2538,
+      "step": 12520
+    },
+    {
+      "epoch": 0.10365223145965173,
+      "grad_norm": 1021.1143188476562,
+      "learning_rate": 9.999952211101056e-06,
+      "loss": 154.2661,
+      "step": 12530
+    },
+    {
+      "epoch": 0.10373495470902097,
+      "grad_norm": 1220.9776611328125,
+      "learning_rate": 9.999945845889795e-06,
+      "loss": 154.2683,
+      "step": 12540
+    },
+    {
+      "epoch": 0.10381767795839021,
+      "grad_norm": 1148.5760498046875,
+      "learning_rate": 9.999939082855788e-06,
+      "loss": 137.0497,
+      "step": 12550
+    },
+    {
+      "epoch": 0.10390040120775944,
+      "grad_norm": 1300.0975341796875,
+      "learning_rate": 9.999931921999575e-06,
+      "loss": 169.4536,
+      "step": 12560
+    },
+    {
+      "epoch": 0.10398312445712868,
+      "grad_norm": 1183.8826904296875,
+      "learning_rate": 9.999924363321726e-06,
+      "loss": 166.6497,
+      "step": 12570
+    },
+    {
+      "epoch": 0.1040658477064979,
+      "grad_norm": 718.7395629882812,
+      "learning_rate": 9.999916406822843e-06,
+      "loss": 109.411,
+      "step": 12580
+    },
+    {
+      "epoch": 0.10414857095586715,
+      "grad_norm": 841.6725463867188,
+      "learning_rate": 9.999908052503557e-06,
+      "loss": 146.8344,
+      "step": 12590
+    },
+    {
+      "epoch": 0.10423129420523639,
+      "grad_norm": 918.6251831054688,
+      "learning_rate": 9.999899300364534e-06,
+      "loss": 162.9566,
+      "step": 12600
+    },
+    {
+      "epoch": 0.10431401745460561,
+      "grad_norm": 1467.325927734375,
+      "learning_rate": 9.99989015040647e-06,
+      "loss": 189.8623,
+      "step": 12610
+    },
+    {
+      "epoch": 0.10439674070397485,
+      "grad_norm": 2947.025146484375,
+      "learning_rate": 9.999880602630092e-06,
+      "loss": 141.2983,
+      "step": 12620
+    },
+    {
+      "epoch": 0.10447946395334409,
+      "grad_norm": 1483.4896240234375,
+      "learning_rate": 9.999870657036161e-06,
+      "loss": 162.2035,
+      "step": 12630
+    },
+    {
+      "epoch": 0.10456218720271332,
+      "grad_norm": 1846.94580078125,
+      "learning_rate": 9.99986031362547e-06,
+      "loss": 150.0733,
+      "step": 12640
+    },
+    {
+      "epoch": 0.10464491045208256,
+      "grad_norm": 1538.964111328125,
+      "learning_rate": 9.99984957239884e-06,
+      "loss": 206.5021,
+      "step": 12650
+    },
+    {
+      "epoch": 0.1047276337014518,
+      "grad_norm": 1100.305908203125,
+      "learning_rate": 9.999838433357124e-06,
+      "loss": 197.57,
+      "step": 12660
+    },
+    {
+      "epoch": 0.10481035695082103,
+      "grad_norm": 594.6061401367188,
+      "learning_rate": 9.99982689650121e-06,
+      "loss": 179.5414,
+      "step": 12670
+    },
+    {
+      "epoch": 0.10489308020019027,
+      "grad_norm": 737.6015014648438,
+      "learning_rate": 9.999814961832018e-06,
+      "loss": 170.4644,
+      "step": 12680
+    },
+    {
+      "epoch": 0.1049758034495595,
+      "grad_norm": 1346.4503173828125,
+      "learning_rate": 9.999802629350492e-06,
+      "loss": 202.0369,
+      "step": 12690
+    },
+    {
+      "epoch": 0.10505852669892873,
+      "grad_norm": 2559.85546875,
+      "learning_rate": 9.99978989905762e-06,
+      "loss": 221.5045,
+      "step": 12700
+    },
+    {
+      "epoch": 0.10514124994829797,
+      "grad_norm": 1314.4476318359375,
+      "learning_rate": 9.999776770954411e-06,
+      "loss": 173.8742,
+      "step": 12710
+    },
+    {
+      "epoch": 0.1052239731976672,
+      "grad_norm": 3551.19677734375,
+      "learning_rate": 9.99976324504191e-06,
+      "loss": 149.5013,
+      "step": 12720
+    },
+    {
+      "epoch": 0.10530669644703644,
+      "grad_norm": 1428.34814453125,
+      "learning_rate": 9.999749321321192e-06,
+      "loss": 251.0179,
+      "step": 12730
+    },
+    {
+      "epoch": 0.10538941969640568,
+      "grad_norm": 753.5355224609375,
+      "learning_rate": 9.999734999793369e-06,
+      "loss": 163.6853,
+      "step": 12740
+    },
+    {
+      "epoch": 0.10547214294577491,
+      "grad_norm": 1578.79541015625,
+      "learning_rate": 9.999720280459576e-06,
+      "loss": 154.6416,
+      "step": 12750
+    },
+    {
+      "epoch": 0.10555486619514415,
+      "grad_norm": 1111.734130859375,
+      "learning_rate": 9.999705163320987e-06,
+      "loss": 177.3941,
+      "step": 12760
+    },
+    {
+      "epoch": 0.10563758944451337,
+      "grad_norm": 1722.9654541015625,
+      "learning_rate": 9.999689648378801e-06,
+      "loss": 179.8888,
+      "step": 12770
+    },
+    {
+      "epoch": 0.10572031269388261,
+      "grad_norm": 1038.706298828125,
+      "learning_rate": 9.999673735634259e-06,
+      "loss": 127.1906,
+      "step": 12780
+    },
+    {
+      "epoch": 0.10580303594325186,
+      "grad_norm": 11962.8779296875,
+      "learning_rate": 9.99965742508862e-06,
+      "loss": 213.939,
+      "step": 12790
+    },
+    {
+      "epoch": 0.10588575919262108,
+      "grad_norm": 1482.868408203125,
+      "learning_rate": 9.999640716743186e-06,
+      "loss": 133.2547,
+      "step": 12800
+    },
+    {
+      "epoch": 0.10596848244199032,
+      "grad_norm": 3282.25390625,
+      "learning_rate": 9.999623610599287e-06,
+      "loss": 160.013,
+      "step": 12810
+    },
+    {
+      "epoch": 0.10605120569135956,
+      "grad_norm": 1836.9112548828125,
+      "learning_rate": 9.999606106658282e-06,
+      "loss": 201.4363,
+      "step": 12820
+    },
+    {
+      "epoch": 0.10613392894072879,
+      "grad_norm": 816.5084838867188,
+      "learning_rate": 9.999588204921562e-06,
+      "loss": 174.8686,
+      "step": 12830
+    },
+    {
+      "epoch": 0.10621665219009803,
+      "grad_norm": 1352.0396728515625,
+      "learning_rate": 9.999569905390556e-06,
+      "loss": 167.4276,
+      "step": 12840
+    },
+    {
+      "epoch": 0.10629937543946726,
+      "grad_norm": 533.4453125,
+      "learning_rate": 9.999551208066716e-06,
+      "loss": 100.8425,
+      "step": 12850
+    },
+    {
+      "epoch": 0.1063820986888365,
+      "grad_norm": 1191.3555908203125,
+      "learning_rate": 9.99953211295153e-06,
+      "loss": 147.7438,
+      "step": 12860
+    },
+    {
+      "epoch": 0.10646482193820574,
+      "grad_norm": 1642.895263671875,
+      "learning_rate": 9.999512620046523e-06,
+      "loss": 170.0533,
+      "step": 12870
+    },
+    {
+      "epoch": 0.10654754518757496,
+      "grad_norm": 1362.4990234375,
+      "learning_rate": 9.999492729353238e-06,
+      "loss": 177.1871,
+      "step": 12880
+    },
+    {
+      "epoch": 0.1066302684369442,
+      "grad_norm": 879.8888549804688,
+      "learning_rate": 9.999472440873261e-06,
+      "loss": 160.7778,
+      "step": 12890
+    },
+    {
+      "epoch": 0.10671299168631344,
+      "grad_norm": 1243.377685546875,
+      "learning_rate": 9.999451754608208e-06,
+      "loss": 150.4305,
+      "step": 12900
+    },
+    {
+      "epoch": 0.10679571493568267,
+      "grad_norm": 984.1234130859375,
+      "learning_rate": 9.999430670559723e-06,
+      "loss": 139.6723,
+      "step": 12910
+    },
+    {
+      "epoch": 0.10687843818505191,
+      "grad_norm": 852.8709716796875,
+      "learning_rate": 9.999409188729484e-06,
+      "loss": 134.8206,
+      "step": 12920
+    },
+    {
+      "epoch": 0.10696116143442115,
+      "grad_norm": 886.4750366210938,
+      "learning_rate": 9.999387309119198e-06,
+      "loss": 150.4601,
+      "step": 12930
+    },
+    {
+      "epoch": 0.10704388468379038,
+      "grad_norm": 1050.5213623046875,
+      "learning_rate": 9.999365031730609e-06,
+      "loss": 162.7591,
+      "step": 12940
+    },
+    {
+      "epoch": 0.10712660793315962,
+      "grad_norm": 1679.367919921875,
+      "learning_rate": 9.99934235656549e-06,
+      "loss": 146.1199,
+      "step": 12950
+    },
+    {
+      "epoch": 0.10720933118252884,
+      "grad_norm": 1546.12939453125,
+      "learning_rate": 9.999319283625641e-06,
+      "loss": 207.9114,
+      "step": 12960
+    },
+    {
+      "epoch": 0.10729205443189808,
+      "grad_norm": 856.313720703125,
+      "learning_rate": 9.999295812912902e-06,
+      "loss": 168.0912,
+      "step": 12970
+    },
+    {
+      "epoch": 0.10737477768126732,
+      "grad_norm": 1466.169677734375,
+      "learning_rate": 9.999271944429139e-06,
+      "loss": 202.6795,
+      "step": 12980
+    },
+    {
+      "epoch": 0.10745750093063655,
+      "grad_norm": 1105.8492431640625,
+      "learning_rate": 9.99924767817625e-06,
+      "loss": 139.1426,
+      "step": 12990
+    },
+    {
+      "epoch": 0.10754022418000579,
+      "grad_norm": 1072.4610595703125,
+      "learning_rate": 9.999223014156167e-06,
+      "loss": 196.2584,
+      "step": 13000
+    },
+    {
+      "epoch": 0.10762294742937503,
+      "grad_norm": 820.6765747070312,
+      "learning_rate": 9.999197952370851e-06,
+      "loss": 177.7073,
+      "step": 13010
+    },
+    {
+      "epoch": 0.10770567067874426,
+      "grad_norm": 788.7813720703125,
+      "learning_rate": 9.9991724928223e-06,
+      "loss": 157.0522,
+      "step": 13020
+    },
+    {
+      "epoch": 0.1077883939281135,
+      "grad_norm": 1179.322265625,
+      "learning_rate": 9.999146635512535e-06,
+      "loss": 189.2783,
+      "step": 13030
+    },
+    {
+      "epoch": 0.10787111717748273,
+      "grad_norm": 1035.255615234375,
+      "learning_rate": 9.999120380443614e-06,
+      "loss": 145.2693,
+      "step": 13040
+    },
+    {
+      "epoch": 0.10795384042685197,
+      "grad_norm": 1015.682861328125,
+      "learning_rate": 9.99909372761763e-06,
+      "loss": 143.6347,
+      "step": 13050
+    },
+    {
+      "epoch": 0.1080365636762212,
+      "grad_norm": 2936.619873046875,
+      "learning_rate": 9.9990666770367e-06,
+      "loss": 179.5399,
+      "step": 13060
+    },
+    {
+      "epoch": 0.10811928692559043,
+      "grad_norm": 1349.863037109375,
+      "learning_rate": 9.999039228702975e-06,
+      "loss": 176.5338,
+      "step": 13070
+    },
+    {
+      "epoch": 0.10820201017495967,
+      "grad_norm": 1154.5147705078125,
+      "learning_rate": 9.999011382618644e-06,
+      "loss": 166.4444,
+      "step": 13080
+    },
+    {
+      "epoch": 0.10828473342432891,
+      "grad_norm": 1495.1844482421875,
+      "learning_rate": 9.998983138785919e-06,
+      "loss": 149.649,
+      "step": 13090
+    },
+    {
+      "epoch": 0.10836745667369814,
+      "grad_norm": 1207.58984375,
+      "learning_rate": 9.998954497207045e-06,
+      "loss": 149.596,
+      "step": 13100
+    },
+    {
+      "epoch": 0.10845017992306738,
+      "grad_norm": 1018.6047973632812,
+      "learning_rate": 9.998925457884307e-06,
+      "loss": 154.7165,
+      "step": 13110
+    },
+    {
+      "epoch": 0.1085329031724366,
+      "grad_norm": 1084.2874755859375,
+      "learning_rate": 9.99889602082001e-06,
+      "loss": 145.5709,
+      "step": 13120
+    },
+    {
+      "epoch": 0.10861562642180585,
+      "grad_norm": 1162.242919921875,
+      "learning_rate": 9.998866186016501e-06,
+      "loss": 168.6215,
+      "step": 13130
+    },
+    {
+      "epoch": 0.10869834967117509,
+      "grad_norm": 1217.2177734375,
+      "learning_rate": 9.99883595347615e-06,
+      "loss": 150.9769,
+      "step": 13140
+    },
+    {
+      "epoch": 0.10878107292054431,
+      "grad_norm": 884.3536376953125,
+      "learning_rate": 9.998805323201364e-06,
+      "loss": 161.5837,
+      "step": 13150
+    },
+    {
+      "epoch": 0.10886379616991355,
+      "grad_norm": 1069.2374267578125,
+      "learning_rate": 9.998774295194579e-06,
+      "loss": 167.3784,
+      "step": 13160
+    },
+    {
+      "epoch": 0.1089465194192828,
+      "grad_norm": 749.6088256835938,
+      "learning_rate": 9.998742869458264e-06,
+      "loss": 115.7084,
+      "step": 13170
+    },
+    {
+      "epoch": 0.10902924266865202,
+      "grad_norm": 703.9533081054688,
+      "learning_rate": 9.998711045994922e-06,
+      "loss": 162.0743,
+      "step": 13180
+    },
+    {
+      "epoch": 0.10911196591802126,
+      "grad_norm": 1521.7666015625,
+      "learning_rate": 9.998678824807082e-06,
+      "loss": 189.686,
+      "step": 13190
+    },
+    {
+      "epoch": 0.1091946891673905,
+      "grad_norm": 1742.9715576171875,
+      "learning_rate": 9.99864620589731e-06,
+      "loss": 163.9251,
+      "step": 13200
+    },
+    {
+      "epoch": 0.10927741241675973,
+      "grad_norm": 999.13623046875,
+      "learning_rate": 9.998613189268197e-06,
+      "loss": 142.0809,
+      "step": 13210
+    },
+    {
+      "epoch": 0.10936013566612897,
+      "grad_norm": 1409.1549072265625,
+      "learning_rate": 9.998579774922377e-06,
+      "loss": 149.1874,
+      "step": 13220
+    },
+    {
+      "epoch": 0.1094428589154982,
+      "grad_norm": 675.3158569335938,
+      "learning_rate": 9.998545962862503e-06,
+      "loss": 117.2426,
+      "step": 13230
+    },
+    {
+      "epoch": 0.10952558216486744,
+      "grad_norm": 1737.0997314453125,
+      "learning_rate": 9.998511753091267e-06,
+      "loss": 155.6872,
+      "step": 13240
+    },
+    {
+      "epoch": 0.10960830541423668,
+      "grad_norm": 4492.49853515625,
+      "learning_rate": 9.998477145611389e-06,
+      "loss": 195.1054,
+      "step": 13250
+    },
+    {
+      "epoch": 0.1096910286636059,
+      "grad_norm": 1431.2005615234375,
+      "learning_rate": 9.998442140425625e-06,
+      "loss": 189.3633,
+      "step": 13260
+    },
+    {
+      "epoch": 0.10977375191297514,
+      "grad_norm": 1214.457763671875,
+      "learning_rate": 9.998406737536761e-06,
+      "loss": 167.9152,
+      "step": 13270
+    },
+    {
+      "epoch": 0.10985647516234438,
+      "grad_norm": 553.1201171875,
+      "learning_rate": 9.998370936947614e-06,
+      "loss": 154.5592,
+      "step": 13280
+    },
+    {
+      "epoch": 0.10993919841171361,
+      "grad_norm": 1713.546875,
+      "learning_rate": 9.998334738661028e-06,
+      "loss": 178.8647,
+      "step": 13290
+    },
+    {
+      "epoch": 0.11002192166108285,
+      "grad_norm": 1338.5965576171875,
+      "learning_rate": 9.998298142679888e-06,
+      "loss": 209.809,
+      "step": 13300
+    },
+    {
+      "epoch": 0.11010464491045208,
+      "grad_norm": 1447.3448486328125,
+      "learning_rate": 9.998261149007104e-06,
+      "loss": 151.2987,
+      "step": 13310
+    },
+    {
+      "epoch": 0.11018736815982132,
+      "grad_norm": 1176.512939453125,
+      "learning_rate": 9.998223757645618e-06,
+      "loss": 154.0995,
+      "step": 13320
+    },
+    {
+      "epoch": 0.11027009140919056,
+      "grad_norm": 1491.28466796875,
+      "learning_rate": 9.998185968598407e-06,
+      "loss": 172.6219,
+      "step": 13330
+    },
+    {
+      "epoch": 0.11035281465855978,
+      "grad_norm": 872.3587036132812,
+      "learning_rate": 9.998147781868477e-06,
+      "loss": 136.3148,
+      "step": 13340
+    },
+    {
+      "epoch": 0.11043553790792902,
+      "grad_norm": 1718.0472412109375,
+      "learning_rate": 9.998109197458865e-06,
+      "loss": 147.2434,
+      "step": 13350
+    },
+    {
+      "epoch": 0.11051826115729826,
+      "grad_norm": 1724.40966796875,
+      "learning_rate": 9.998070215372645e-06,
+      "loss": 155.5677,
+      "step": 13360
+    },
+    {
+      "epoch": 0.11060098440666749,
+      "grad_norm": 1238.861572265625,
+      "learning_rate": 9.998030835612914e-06,
+      "loss": 177.9599,
+      "step": 13370
+    },
+    {
+      "epoch": 0.11068370765603673,
+      "grad_norm": 1163.747802734375,
+      "learning_rate": 9.997991058182807e-06,
+      "loss": 159.718,
+      "step": 13380
+    },
+    {
+      "epoch": 0.11076643090540596,
+      "grad_norm": 1202.32763671875,
+      "learning_rate": 9.997950883085492e-06,
+      "loss": 161.9838,
+      "step": 13390
+    },
+    {
+      "epoch": 0.1108491541547752,
+      "grad_norm": 1815.2415771484375,
+      "learning_rate": 9.99791031032416e-06,
+      "loss": 187.6479,
+      "step": 13400
+    },
+    {
+      "epoch": 0.11093187740414444,
+      "grad_norm": 1177.880126953125,
+      "learning_rate": 9.997869339902043e-06,
+      "loss": 180.2671,
+      "step": 13410
+    },
+    {
+      "epoch": 0.11101460065351366,
+      "grad_norm": 1133.1861572265625,
+      "learning_rate": 9.9978279718224e-06,
+      "loss": 161.116,
+      "step": 13420
+    },
+    {
+      "epoch": 0.1110973239028829,
+      "grad_norm": 1010.6656494140625,
+      "learning_rate": 9.99778620608852e-06,
+      "loss": 178.742,
+      "step": 13430
+    },
+    {
+      "epoch": 0.11118004715225215,
+      "grad_norm": 1107.10986328125,
+      "learning_rate": 9.997744042703731e-06,
+      "loss": 139.692,
+      "step": 13440
+    },
+    {
+      "epoch": 0.11126277040162137,
+      "grad_norm": 1005.2382202148438,
+      "learning_rate": 9.997701481671384e-06,
+      "loss": 178.2121,
+      "step": 13450
+    },
+    {
+      "epoch": 0.11134549365099061,
+      "grad_norm": 2284.540283203125,
+      "learning_rate": 9.997658522994867e-06,
+      "loss": 156.5188,
+      "step": 13460
+    },
+    {
+      "epoch": 0.11142821690035985,
+      "grad_norm": 5658.24658203125,
+      "learning_rate": 9.997615166677597e-06,
+      "loss": 146.769,
+      "step": 13470
+    },
+    {
+      "epoch": 0.11151094014972908,
+      "grad_norm": 457.560302734375,
+      "learning_rate": 9.997571412723024e-06,
+      "loss": 119.6845,
+      "step": 13480
+    },
+    {
+      "epoch": 0.11159366339909832,
+      "grad_norm": 1155.69677734375,
+      "learning_rate": 9.99752726113463e-06,
+      "loss": 126.1926,
+      "step": 13490
+    },
+    {
+      "epoch": 0.11167638664846755,
+      "grad_norm": 1057.7230224609375,
+      "learning_rate": 9.997482711915926e-06,
+      "loss": 145.3562,
+      "step": 13500
+    },
+    {
+      "epoch": 0.11175910989783679,
+      "grad_norm": 1107.236328125,
+      "learning_rate": 9.99743776507046e-06,
+      "loss": 189.204,
+      "step": 13510
+    },
+    {
+      "epoch": 0.11184183314720603,
+      "grad_norm": 1068.25634765625,
+      "learning_rate": 9.997392420601804e-06,
+      "loss": 135.3788,
+      "step": 13520
+    },
+    {
+      "epoch": 0.11192455639657525,
+      "grad_norm": 875.785888671875,
+      "learning_rate": 9.99734667851357e-06,
+      "loss": 152.927,
+      "step": 13530
+    },
+    {
+      "epoch": 0.1120072796459445,
+      "grad_norm": 1772.181396484375,
+      "learning_rate": 9.997300538809394e-06,
+      "loss": 222.4996,
+      "step": 13540
+    },
+    {
+      "epoch": 0.11209000289531373,
+      "grad_norm": 1264.271240234375,
+      "learning_rate": 9.99725400149295e-06,
+      "loss": 145.0414,
+      "step": 13550
+    },
+    {
+      "epoch": 0.11217272614468296,
+      "grad_norm": 2198.46630859375,
+      "learning_rate": 9.997207066567939e-06,
+      "loss": 194.5429,
+      "step": 13560
+    },
+    {
+      "epoch": 0.1122554493940522,
+      "grad_norm": 1130.449951171875,
+      "learning_rate": 9.997159734038096e-06,
+      "loss": 166.257,
+      "step": 13570
+    },
+    {
+      "epoch": 0.11233817264342143,
+      "grad_norm": 1828.01513671875,
+      "learning_rate": 9.997112003907186e-06,
+      "loss": 152.8911,
+      "step": 13580
+    },
+    {
+      "epoch": 0.11242089589279067,
+      "grad_norm": 1196.669677734375,
+      "learning_rate": 9.997063876179007e-06,
+      "loss": 129.7313,
+      "step": 13590
+    },
+    {
+      "epoch": 0.11250361914215991,
+      "grad_norm": 1254.476806640625,
+      "learning_rate": 9.997015350857391e-06,
+      "loss": 169.0213,
+      "step": 13600
+    },
+    {
+      "epoch": 0.11258634239152913,
+      "grad_norm": 1430.6507568359375,
+      "learning_rate": 9.996966427946195e-06,
+      "loss": 150.1627,
+      "step": 13610
+    },
+    {
+      "epoch": 0.11266906564089837,
+      "grad_norm": 906.8787231445312,
+      "learning_rate": 9.996917107449313e-06,
+      "loss": 174.3134,
+      "step": 13620
+    },
+    {
+      "epoch": 0.11275178889026762,
+      "grad_norm": 812.057373046875,
+      "learning_rate": 9.99686738937067e-06,
+      "loss": 138.5191,
+      "step": 13630
+    },
+    {
+      "epoch": 0.11283451213963684,
+      "grad_norm": 1879.1136474609375,
+      "learning_rate": 9.996817273714222e-06,
+      "loss": 174.5974,
+      "step": 13640
+    },
+    {
+      "epoch": 0.11291723538900608,
+      "grad_norm": 945.3467407226562,
+      "learning_rate": 9.996766760483955e-06,
+      "loss": 161.973,
+      "step": 13650
+    },
+    {
+      "epoch": 0.11299995863837531,
+      "grad_norm": 862.3170776367188,
+      "learning_rate": 9.996715849683889e-06,
+      "loss": 137.5633,
+      "step": 13660
+    },
+    {
+      "epoch": 0.11308268188774455,
+      "grad_norm": 690.8417358398438,
+      "learning_rate": 9.996664541318076e-06,
+      "loss": 141.2179,
+      "step": 13670
+    },
+    {
+      "epoch": 0.11316540513711379,
+      "grad_norm": 972.900634765625,
+      "learning_rate": 9.996612835390596e-06,
+      "loss": 115.8736,
+      "step": 13680
+    },
+    {
+      "epoch": 0.11324812838648302,
+      "grad_norm": 1904.7318115234375,
+      "learning_rate": 9.996560731905565e-06,
+      "loss": 154.7887,
+      "step": 13690
+    },
+    {
+      "epoch": 0.11333085163585226,
+      "grad_norm": 851.8038940429688,
+      "learning_rate": 9.996508230867126e-06,
+      "loss": 137.4024,
+      "step": 13700
+    },
+    {
+      "epoch": 0.1134135748852215,
+      "grad_norm": 1190.90380859375,
+      "learning_rate": 9.996455332279458e-06,
+      "loss": 153.5202,
+      "step": 13710
+    },
+    {
+      "epoch": 0.11349629813459072,
+      "grad_norm": 2241.255126953125,
+      "learning_rate": 9.99640203614677e-06,
+      "loss": 173.7966,
+      "step": 13720
+    },
+    {
+      "epoch": 0.11357902138395996,
+      "grad_norm": 1084.945068359375,
+      "learning_rate": 9.996348342473304e-06,
+      "loss": 159.4762,
+      "step": 13730
+    },
+    {
+      "epoch": 0.1136617446333292,
+      "grad_norm": 1753.5206298828125,
+      "learning_rate": 9.99629425126333e-06,
+      "loss": 218.3281,
+      "step": 13740
+    },
+    {
+      "epoch": 0.11374446788269843,
+      "grad_norm": 802.15576171875,
+      "learning_rate": 9.996239762521152e-06,
+      "loss": 153.0149,
+      "step": 13750
+    },
+    {
+      "epoch": 0.11382719113206767,
+      "grad_norm": 2257.760986328125,
+      "learning_rate": 9.996184876251105e-06,
+      "loss": 134.656,
+      "step": 13760
+    },
+    {
+      "epoch": 0.1139099143814369,
+      "grad_norm": 1485.7083740234375,
+      "learning_rate": 9.996129592457558e-06,
+      "loss": 119.6472,
+      "step": 13770
+    },
+    {
+      "epoch": 0.11399263763080614,
+      "grad_norm": 1323.489990234375,
+      "learning_rate": 9.996073911144907e-06,
+      "loss": 135.8627,
+      "step": 13780
+    },
+    {
+      "epoch": 0.11407536088017538,
+      "grad_norm": 1439.948974609375,
+      "learning_rate": 9.996017832317583e-06,
+      "loss": 105.2017,
+      "step": 13790
+    },
+    {
+      "epoch": 0.1141580841295446,
+      "grad_norm": 755.0234375,
+      "learning_rate": 9.995961355980052e-06,
+      "loss": 119.7319,
+      "step": 13800
+    },
+    {
+      "epoch": 0.11424080737891384,
+      "grad_norm": 1265.96826171875,
+      "learning_rate": 9.995904482136803e-06,
+      "loss": 144.368,
+      "step": 13810
+    },
+    {
+      "epoch": 0.11432353062828308,
+      "grad_norm": 1014.328857421875,
+      "learning_rate": 9.99584721079236e-06,
+      "loss": 180.5596,
+      "step": 13820
+    },
+    {
+      "epoch": 0.11440625387765231,
+      "grad_norm": 1355.7919921875,
+      "learning_rate": 9.995789541951287e-06,
+      "loss": 169.1609,
+      "step": 13830
+    },
+    {
+      "epoch": 0.11448897712702155,
+      "grad_norm": 1709.610107421875,
+      "learning_rate": 9.995731475618163e-06,
+      "loss": 152.0147,
+      "step": 13840
+    },
+    {
+      "epoch": 0.11457170037639078,
+      "grad_norm": 1426.723388671875,
+      "learning_rate": 9.995673011797615e-06,
+      "loss": 142.7122,
+      "step": 13850
+    },
+    {
+      "epoch": 0.11465442362576002,
+      "grad_norm": 3322.473876953125,
+      "learning_rate": 9.995614150494293e-06,
+      "loss": 192.3159,
+      "step": 13860
+    },
+    {
+      "epoch": 0.11473714687512926,
+      "grad_norm": 852.6958618164062,
+      "learning_rate": 9.995554891712879e-06,
+      "loss": 221.4455,
+      "step": 13870
+    },
+    {
+      "epoch": 0.11481987012449849,
+      "grad_norm": 679.8392944335938,
+      "learning_rate": 9.995495235458087e-06,
+      "loss": 205.1969,
+      "step": 13880
+    },
+    {
+      "epoch": 0.11490259337386773,
+      "grad_norm": 1818.237548828125,
+      "learning_rate": 9.99543518173467e-06,
+      "loss": 157.8227,
+      "step": 13890
+    },
+    {
+      "epoch": 0.11498531662323697,
+      "grad_norm": 911.2511596679688,
+      "learning_rate": 9.995374730547397e-06,
+      "loss": 213.6541,
+      "step": 13900
+    },
+    {
+      "epoch": 0.11506803987260619,
+      "grad_norm": 1822.821044921875,
+      "learning_rate": 9.995313881901085e-06,
+      "loss": 198.7188,
+      "step": 13910
+    },
+    {
+      "epoch": 0.11515076312197543,
+      "grad_norm": 653.8102416992188,
+      "learning_rate": 9.995252635800572e-06,
+      "loss": 127.7723,
+      "step": 13920
+    },
+    {
+      "epoch": 0.11523348637134467,
+      "grad_norm": 1713.5816650390625,
+      "learning_rate": 9.995190992250732e-06,
+      "loss": 225.4239,
+      "step": 13930
+    },
+    {
+      "epoch": 0.1153162096207139,
+      "grad_norm": 1139.3336181640625,
+      "learning_rate": 9.995128951256469e-06,
+      "loss": 140.1807,
+      "step": 13940
+    },
+    {
+      "epoch": 0.11539893287008314,
+      "grad_norm": 1414.932861328125,
+      "learning_rate": 9.99506651282272e-06,
+      "loss": 161.69,
+      "step": 13950
+    },
+    {
+      "epoch": 0.11548165611945237,
+      "grad_norm": 1176.623291015625,
+      "learning_rate": 9.995003676954454e-06,
+      "loss": 151.0156,
+      "step": 13960
+    },
+    {
+      "epoch": 0.11556437936882161,
+      "grad_norm": 784.6156616210938,
+      "learning_rate": 9.994940443656668e-06,
+      "loss": 198.3028,
+      "step": 13970
+    },
+    {
+      "epoch": 0.11564710261819085,
+      "grad_norm": 862.885986328125,
+      "learning_rate": 9.994876812934395e-06,
+      "loss": 153.6012,
+      "step": 13980
+    },
+    {
+      "epoch": 0.11572982586756007,
+      "grad_norm": 2193.662109375,
+      "learning_rate": 9.994812784792698e-06,
+      "loss": 165.4299,
+      "step": 13990
+    },
+    {
+      "epoch": 0.11581254911692931,
+      "grad_norm": 1005.8052978515625,
+      "learning_rate": 9.99474835923667e-06,
+      "loss": 117.067,
+      "step": 14000
+    },
+    {
+      "epoch": 0.11589527236629855,
+      "grad_norm": 870.8054809570312,
+      "learning_rate": 9.994683536271437e-06,
+      "loss": 177.513,
+      "step": 14010
+    },
+    {
+      "epoch": 0.11597799561566778,
+      "grad_norm": 1938.4117431640625,
+      "learning_rate": 9.994618315902161e-06,
+      "loss": 147.8295,
+      "step": 14020
+    },
+    {
+      "epoch": 0.11606071886503702,
+      "grad_norm": 1145.3619384765625,
+      "learning_rate": 9.994552698134023e-06,
+      "loss": 126.2492,
+      "step": 14030
+    },
+    {
+      "epoch": 0.11614344211440625,
+      "grad_norm": 704.4757080078125,
+      "learning_rate": 9.994486682972253e-06,
+      "loss": 183.3489,
+      "step": 14040
+    },
+    {
+      "epoch": 0.11622616536377549,
+      "grad_norm": 799.8057250976562,
+      "learning_rate": 9.994420270422096e-06,
+      "loss": 155.8286,
+      "step": 14050
+    },
+    {
+      "epoch": 0.11630888861314473,
+      "grad_norm": 2168.39794921875,
+      "learning_rate": 9.994353460488842e-06,
+      "loss": 165.6206,
+      "step": 14060
+    },
+    {
+      "epoch": 0.11639161186251396,
+      "grad_norm": 1048.3438720703125,
+      "learning_rate": 9.994286253177803e-06,
+      "loss": 196.4472,
+      "step": 14070
+    },
+    {
+      "epoch": 0.1164743351118832,
+      "grad_norm": 1240.358642578125,
+      "learning_rate": 9.994218648494327e-06,
+      "loss": 169.1644,
+      "step": 14080
+    },
+    {
+      "epoch": 0.11655705836125244,
+      "grad_norm": 1450.99169921875,
+      "learning_rate": 9.994150646443793e-06,
+      "loss": 119.286,
+      "step": 14090
+    },
+    {
+      "epoch": 0.11663978161062166,
+      "grad_norm": 1026.7149658203125,
+      "learning_rate": 9.994082247031613e-06,
+      "loss": 166.7578,
+      "step": 14100
+    },
+    {
+      "epoch": 0.1167225048599909,
+      "grad_norm": 1081.6656494140625,
+      "learning_rate": 9.99401345026323e-06,
+      "loss": 141.0757,
+      "step": 14110
+    },
+    {
+      "epoch": 0.11680522810936013,
+      "grad_norm": 746.9979248046875,
+      "learning_rate": 9.993944256144115e-06,
+      "loss": 124.9759,
+      "step": 14120
+    },
+    {
+      "epoch": 0.11688795135872937,
+      "grad_norm": 891.9210815429688,
+      "learning_rate": 9.993874664679774e-06,
+      "loss": 150.3685,
+      "step": 14130
+    },
+    {
+      "epoch": 0.11697067460809861,
+      "grad_norm": 1401.4002685546875,
+      "learning_rate": 9.993804675875744e-06,
+      "loss": 168.8493,
+      "step": 14140
+    },
+    {
+      "epoch": 0.11705339785746784,
+      "grad_norm": 1588.7640380859375,
+      "learning_rate": 9.993734289737596e-06,
+      "loss": 141.4464,
+      "step": 14150
+    },
+    {
+      "epoch": 0.11713612110683708,
+      "grad_norm": 1859.66552734375,
+      "learning_rate": 9.993663506270928e-06,
+      "loss": 162.024,
+      "step": 14160
+    },
+    {
+      "epoch": 0.11721884435620632,
+      "grad_norm": 1133.1839599609375,
+      "learning_rate": 9.993592325481373e-06,
+      "loss": 166.6096,
+      "step": 14170
+    },
+    {
+      "epoch": 0.11730156760557554,
+      "grad_norm": 1811.849365234375,
+      "learning_rate": 9.993520747374594e-06,
+      "loss": 127.2197,
+      "step": 14180
+    },
+    {
+      "epoch": 0.11738429085494478,
+      "grad_norm": 909.2362060546875,
+      "learning_rate": 9.993448771956285e-06,
+      "loss": 189.4919,
+      "step": 14190
+    },
+    {
+      "epoch": 0.11746701410431402,
+      "grad_norm": 1350.44140625,
+      "learning_rate": 9.993376399232175e-06,
+      "loss": 142.4382,
+      "step": 14200
+    },
+    {
+      "epoch": 0.11754973735368325,
+      "grad_norm": 1765.2679443359375,
+      "learning_rate": 9.993303629208023e-06,
+      "loss": 148.8411,
+      "step": 14210
+    },
+    {
+      "epoch": 0.11763246060305249,
+      "grad_norm": 2343.818359375,
+      "learning_rate": 9.993230461889616e-06,
+      "loss": 212.7168,
+      "step": 14220
+    },
+    {
+      "epoch": 0.11771518385242172,
+      "grad_norm": 683.275146484375,
+      "learning_rate": 9.993156897282776e-06,
+      "loss": 148.4446,
+      "step": 14230
+    },
+    {
+      "epoch": 0.11779790710179096,
+      "grad_norm": 998.6349487304688,
+      "learning_rate": 9.99308293539336e-06,
+      "loss": 117.4103,
+      "step": 14240
+    },
+    {
+      "epoch": 0.1178806303511602,
+      "grad_norm": 823.5191040039062,
+      "learning_rate": 9.993008576227248e-06,
+      "loss": 130.8048,
+      "step": 14250
+    },
+    {
+      "epoch": 0.11796335360052942,
+      "grad_norm": 1420.8035888671875,
+      "learning_rate": 9.992933819790358e-06,
+      "loss": 163.5295,
+      "step": 14260
+    },
+    {
+      "epoch": 0.11804607684989867,
+      "grad_norm": 1167.037109375,
+      "learning_rate": 9.992858666088638e-06,
+      "loss": 164.6194,
+      "step": 14270
+    },
+    {
+      "epoch": 0.1181288000992679,
+      "grad_norm": 1515.985107421875,
+      "learning_rate": 9.992783115128072e-06,
+      "loss": 163.406,
+      "step": 14280
+    },
+    {
+      "epoch": 0.11821152334863713,
+      "grad_norm": 1549.8917236328125,
+      "learning_rate": 9.992707166914662e-06,
+      "loss": 168.7726,
+      "step": 14290
+    },
+    {
+      "epoch": 0.11829424659800637,
+      "grad_norm": 1190.7861328125,
+      "learning_rate": 9.992630821454458e-06,
+      "loss": 140.9276,
+      "step": 14300
+    },
+    {
+      "epoch": 0.1183769698473756,
+      "grad_norm": 1568.7037353515625,
+      "learning_rate": 9.992554078753534e-06,
+      "loss": 147.5554,
+      "step": 14310
+    },
+    {
+      "epoch": 0.11845969309674484,
+      "grad_norm": 1000.02880859375,
+      "learning_rate": 9.992476938817994e-06,
+      "loss": 180.1213,
+      "step": 14320
+    },
+    {
+      "epoch": 0.11854241634611408,
+      "grad_norm": 1296.5947265625,
+      "learning_rate": 9.992399401653976e-06,
+      "loss": 137.781,
+      "step": 14330
+    },
+    {
+      "epoch": 0.1186251395954833,
+      "grad_norm": 1144.0504150390625,
+      "learning_rate": 9.99232146726765e-06,
+      "loss": 140.0204,
+      "step": 14340
+    },
+    {
+      "epoch": 0.11870786284485255,
+      "grad_norm": 1281.876708984375,
+      "learning_rate": 9.992243135665217e-06,
+      "loss": 154.8919,
+      "step": 14350
+    },
+    {
+      "epoch": 0.11879058609422179,
+      "grad_norm": 871.8610229492188,
+      "learning_rate": 9.992164406852908e-06,
+      "loss": 186.1516,
+      "step": 14360
+    },
+    {
+      "epoch": 0.11887330934359101,
+      "grad_norm": 1534.3536376953125,
+      "learning_rate": 9.992085280836988e-06,
+      "loss": 160.6092,
+      "step": 14370
+    },
+    {
+      "epoch": 0.11895603259296025,
+      "grad_norm": 988.0948486328125,
+      "learning_rate": 9.992005757623753e-06,
+      "loss": 203.4977,
+      "step": 14380
+    },
+    {
+      "epoch": 0.11903875584232948,
+      "grad_norm": 1357.1668701171875,
+      "learning_rate": 9.991925837219532e-06,
+      "loss": 160.5042,
+      "step": 14390
+    },
+    {
+      "epoch": 0.11912147909169872,
+      "grad_norm": 636.4329223632812,
+      "learning_rate": 9.991845519630679e-06,
+      "loss": 137.9073,
+      "step": 14400
+    },
+    {
+      "epoch": 0.11920420234106796,
+      "grad_norm": 1027.252197265625,
+      "learning_rate": 9.991764804863588e-06,
+      "loss": 122.9028,
+      "step": 14410
+    },
+    {
+      "epoch": 0.11928692559043719,
+      "grad_norm": 2356.058837890625,
+      "learning_rate": 9.991683692924682e-06,
+      "loss": 155.1582,
+      "step": 14420
+    },
+    {
+      "epoch": 0.11936964883980643,
+      "grad_norm": 743.142578125,
+      "learning_rate": 9.991602183820412e-06,
+      "loss": 142.9998,
+      "step": 14430
+    },
+    {
+      "epoch": 0.11945237208917567,
+      "grad_norm": 1039.77978515625,
+      "learning_rate": 9.991520277557266e-06,
+      "loss": 156.9646,
+      "step": 14440
+    },
+    {
+      "epoch": 0.1195350953385449,
+      "grad_norm": 1624.250732421875,
+      "learning_rate": 9.991437974141759e-06,
+      "loss": 165.5059,
+      "step": 14450
+    },
+    {
+      "epoch": 0.11961781858791413,
+      "grad_norm": 589.4429321289062,
+      "learning_rate": 9.99135527358044e-06,
+      "loss": 119.63,
+      "step": 14460
+    },
+    {
+      "epoch": 0.11970054183728338,
+      "grad_norm": 1199.8302001953125,
+      "learning_rate": 9.991272175879888e-06,
+      "loss": 175.7935,
+      "step": 14470
+    },
+    {
+      "epoch": 0.1197832650866526,
+      "grad_norm": 1233.6771240234375,
+      "learning_rate": 9.991188681046718e-06,
+      "loss": 192.3081,
+      "step": 14480
+    },
+    {
+      "epoch": 0.11986598833602184,
+      "grad_norm": 1209.5980224609375,
+      "learning_rate": 9.991104789087568e-06,
+      "loss": 139.1357,
+      "step": 14490
+    },
+    {
+      "epoch": 0.11994871158539107,
+      "grad_norm": 1937.0008544921875,
+      "learning_rate": 9.991020500009118e-06,
+      "loss": 138.8762,
+      "step": 14500
+    },
+    {
+      "epoch": 0.12003143483476031,
+      "grad_norm": 545.6674194335938,
+      "learning_rate": 9.990935813818073e-06,
+      "loss": 139.6014,
+      "step": 14510
+    },
+    {
+      "epoch": 0.12011415808412955,
+      "grad_norm": 650.5000610351562,
+      "learning_rate": 9.99085073052117e-06,
+      "loss": 195.5592,
+      "step": 14520
+    },
+    {
+      "epoch": 0.12019688133349878,
+      "grad_norm": 566.500732421875,
+      "learning_rate": 9.990765250125179e-06,
+      "loss": 154.2413,
+      "step": 14530
+    },
+    {
+      "epoch": 0.12027960458286802,
+      "grad_norm": 1750.378662109375,
+      "learning_rate": 9.990679372636902e-06,
+      "loss": 161.8778,
+      "step": 14540
+    },
+    {
+      "epoch": 0.12036232783223726,
+      "grad_norm": 1349.2432861328125,
+      "learning_rate": 9.99059309806317e-06,
+      "loss": 172.9573,
+      "step": 14550
+    },
+    {
+      "epoch": 0.12044505108160648,
+      "grad_norm": 975.1179809570312,
+      "learning_rate": 9.990506426410851e-06,
+      "loss": 128.1013,
+      "step": 14560
+    },
+    {
+      "epoch": 0.12052777433097572,
+      "grad_norm": 3406.892333984375,
+      "learning_rate": 9.990419357686839e-06,
+      "loss": 177.5993,
+      "step": 14570
+    },
+    {
+      "epoch": 0.12061049758034495,
+      "grad_norm": 428.1646423339844,
+      "learning_rate": 9.99033189189806e-06,
+      "loss": 128.9519,
+      "step": 14580
+    },
+    {
+      "epoch": 0.12069322082971419,
+      "grad_norm": 996.7293701171875,
+      "learning_rate": 9.990244029051475e-06,
+      "loss": 149.3833,
+      "step": 14590
+    },
+    {
+      "epoch": 0.12077594407908343,
+      "grad_norm": 1458.9307861328125,
+      "learning_rate": 9.990155769154077e-06,
+      "loss": 160.3518,
+      "step": 14600
+    },
+    {
+      "epoch": 0.12085866732845266,
+      "grad_norm": 714.3155517578125,
+      "learning_rate": 9.990067112212884e-06,
+      "loss": 128.5733,
+      "step": 14610
+    },
+    {
+      "epoch": 0.1209413905778219,
+      "grad_norm": 629.7801513671875,
+      "learning_rate": 9.989978058234952e-06,
+      "loss": 141.1231,
+      "step": 14620
+    },
+    {
+      "epoch": 0.12102411382719114,
+      "grad_norm": 1056.1544189453125,
+      "learning_rate": 9.989888607227369e-06,
+      "loss": 173.8705,
+      "step": 14630
+    },
+    {
+      "epoch": 0.12110683707656036,
+      "grad_norm": 1272.472412109375,
+      "learning_rate": 9.989798759197247e-06,
+      "loss": 146.9385,
+      "step": 14640
+    },
+    {
+      "epoch": 0.1211895603259296,
+      "grad_norm": 1054.3629150390625,
+      "learning_rate": 9.989708514151739e-06,
+      "loss": 164.1719,
+      "step": 14650
+    },
+    {
+      "epoch": 0.12127228357529883,
+      "grad_norm": 659.7613525390625,
+      "learning_rate": 9.989617872098026e-06,
+      "loss": 149.6539,
+      "step": 14660
+    },
+    {
+      "epoch": 0.12135500682466807,
+      "grad_norm": 815.8479614257812,
+      "learning_rate": 9.989526833043316e-06,
+      "loss": 140.1702,
+      "step": 14670
+    },
+    {
+      "epoch": 0.12143773007403731,
+      "grad_norm": 1148.9129638671875,
+      "learning_rate": 9.989435396994856e-06,
+      "loss": 125.2471,
+      "step": 14680
+    },
+    {
+      "epoch": 0.12152045332340654,
+      "grad_norm": 3006.224609375,
+      "learning_rate": 9.989343563959919e-06,
+      "loss": 150.3076,
+      "step": 14690
+    },
+    {
+      "epoch": 0.12160317657277578,
+      "grad_norm": 1365.89892578125,
+      "learning_rate": 9.989251333945813e-06,
+      "loss": 179.2145,
+      "step": 14700
+    },
+    {
+      "epoch": 0.12168589982214502,
+      "grad_norm": 982.8682861328125,
+      "learning_rate": 9.989158706959875e-06,
+      "loss": 137.9394,
+      "step": 14710
+    },
+    {
+      "epoch": 0.12176862307151425,
+      "grad_norm": 1103.151123046875,
+      "learning_rate": 9.989065683009477e-06,
+      "loss": 150.3043,
+      "step": 14720
+    },
+    {
+      "epoch": 0.12185134632088349,
+      "grad_norm": 1068.439208984375,
+      "learning_rate": 9.988972262102018e-06,
+      "loss": 115.0475,
+      "step": 14730
+    },
+    {
+      "epoch": 0.12193406957025273,
+      "grad_norm": 1475.9112548828125,
+      "learning_rate": 9.988878444244937e-06,
+      "loss": 162.3183,
+      "step": 14740
+    },
+    {
+      "epoch": 0.12201679281962195,
+      "grad_norm": 1475.7916259765625,
+      "learning_rate": 9.988784229445689e-06,
+      "loss": 132.9056,
+      "step": 14750
+    },
+    {
+      "epoch": 0.12209951606899119,
+      "grad_norm": 1580.335205078125,
+      "learning_rate": 9.988689617711777e-06,
+      "loss": 180.2133,
+      "step": 14760
+    },
+    {
+      "epoch": 0.12218223931836042,
+      "grad_norm": 1742.8638916015625,
+      "learning_rate": 9.988594609050726e-06,
+      "loss": 170.6644,
+      "step": 14770
+    },
+    {
+      "epoch": 0.12226496256772966,
+      "grad_norm": 778.4093017578125,
+      "learning_rate": 9.988499203470097e-06,
+      "loss": 163.6835,
+      "step": 14780
+    },
+    {
+      "epoch": 0.1223476858170989,
+      "grad_norm": 908.4758911132812,
+      "learning_rate": 9.988403400977482e-06,
+      "loss": 143.079,
+      "step": 14790
+    },
+    {
+      "epoch": 0.12243040906646813,
+      "grad_norm": 1540.624755859375,
+      "learning_rate": 9.9883072015805e-06,
+      "loss": 160.3763,
+      "step": 14800
+    },
+    {
+      "epoch": 0.12251313231583737,
+      "grad_norm": 919.294677734375,
+      "learning_rate": 9.98821060528681e-06,
+      "loss": 165.4283,
+      "step": 14810
+    },
+    {
+      "epoch": 0.12259585556520661,
+      "grad_norm": 865.2339477539062,
+      "learning_rate": 9.988113612104093e-06,
+      "loss": 128.951,
+      "step": 14820
+    },
+    {
+      "epoch": 0.12267857881457583,
+      "grad_norm": 2098.492919921875,
+      "learning_rate": 9.988016222040067e-06,
+      "loss": 151.1649,
+      "step": 14830
+    },
+    {
+      "epoch": 0.12276130206394507,
+      "grad_norm": 847.621337890625,
+      "learning_rate": 9.987918435102484e-06,
+      "loss": 121.6645,
+      "step": 14840
+    },
+    {
+      "epoch": 0.1228440253133143,
+      "grad_norm": 1472.7208251953125,
+      "learning_rate": 9.987820251299121e-06,
+      "loss": 140.8588,
+      "step": 14850
+    },
+    {
+      "epoch": 0.12292674856268354,
+      "grad_norm": 1310.5726318359375,
+      "learning_rate": 9.987721670637794e-06,
+      "loss": 132.6207,
+      "step": 14860
+    },
+    {
+      "epoch": 0.12300947181205278,
+      "grad_norm": 788.9578247070312,
+      "learning_rate": 9.987622693126342e-06,
+      "loss": 139.2334,
+      "step": 14870
+    },
+    {
+      "epoch": 0.12309219506142201,
+      "grad_norm": 1761.3287353515625,
+      "learning_rate": 9.987523318772644e-06,
+      "loss": 156.3363,
+      "step": 14880
+    },
+    {
+      "epoch": 0.12317491831079125,
+      "grad_norm": 1024.44140625,
+      "learning_rate": 9.987423547584605e-06,
+      "loss": 167.0266,
+      "step": 14890
+    },
+    {
+      "epoch": 0.12325764156016049,
+      "grad_norm": 891.505126953125,
+      "learning_rate": 9.987323379570161e-06,
+      "loss": 144.4436,
+      "step": 14900
+    },
+    {
+      "epoch": 0.12334036480952972,
+      "grad_norm": 1373.43359375,
+      "learning_rate": 9.987222814737287e-06,
+      "loss": 139.1032,
+      "step": 14910
+    },
+    {
+      "epoch": 0.12342308805889896,
+      "grad_norm": 2005.6689453125,
+      "learning_rate": 9.987121853093982e-06,
+      "loss": 179.2018,
+      "step": 14920
+    },
+    {
+      "epoch": 0.12350581130826818,
+      "grad_norm": 1276.0216064453125,
+      "learning_rate": 9.987020494648279e-06,
+      "loss": 157.102,
+      "step": 14930
+    },
+    {
+      "epoch": 0.12358853455763742,
+      "grad_norm": 1294.2474365234375,
+      "learning_rate": 9.986918739408241e-06,
+      "loss": 176.7196,
+      "step": 14940
+    },
+    {
+      "epoch": 0.12367125780700666,
+      "grad_norm": 1715.0291748046875,
+      "learning_rate": 9.986816587381966e-06,
+      "loss": 150.2139,
+      "step": 14950
+    },
+    {
+      "epoch": 0.12375398105637589,
+      "grad_norm": 1020.9821166992188,
+      "learning_rate": 9.986714038577582e-06,
+      "loss": 145.5669,
+      "step": 14960
+    },
+    {
+      "epoch": 0.12383670430574513,
+      "grad_norm": 969.8027954101562,
+      "learning_rate": 9.986611093003249e-06,
+      "loss": 129.0563,
+      "step": 14970
+    },
+    {
+      "epoch": 0.12391942755511437,
+      "grad_norm": 1188.594482421875,
+      "learning_rate": 9.986507750667157e-06,
+      "loss": 130.371,
+      "step": 14980
+    },
+    {
+      "epoch": 0.1240021508044836,
+      "grad_norm": 2070.416015625,
+      "learning_rate": 9.986404011577525e-06,
+      "loss": 164.5774,
+      "step": 14990
+    },
+    {
+      "epoch": 0.12408487405385284,
+      "grad_norm": 1122.52587890625,
+      "learning_rate": 9.986299875742612e-06,
+      "loss": 187.0694,
+      "step": 15000
+    },
+    {
+      "epoch": 0.12416759730322208,
+      "grad_norm": 1119.6961669921875,
+      "learning_rate": 9.986195343170703e-06,
+      "loss": 180.5289,
+      "step": 15010
+    },
+    {
+      "epoch": 0.1242503205525913,
+      "grad_norm": 1574.2568359375,
+      "learning_rate": 9.986090413870114e-06,
+      "loss": 144.7522,
+      "step": 15020
+    },
+    {
+      "epoch": 0.12433304380196054,
+      "grad_norm": 1006.7045288085938,
+      "learning_rate": 9.985985087849193e-06,
+      "loss": 143.7221,
+      "step": 15030
+    },
+    {
+      "epoch": 0.12441576705132977,
+      "grad_norm": 1300.2181396484375,
+      "learning_rate": 9.98587936511632e-06,
+      "loss": 150.1932,
+      "step": 15040
+    },
+    {
+      "epoch": 0.12449849030069901,
+      "grad_norm": 821.7041625976562,
+      "learning_rate": 9.98577324567991e-06,
+      "loss": 139.0086,
+      "step": 15050
+    },
+    {
+      "epoch": 0.12458121355006825,
+      "grad_norm": 830.7269287109375,
+      "learning_rate": 9.985666729548404e-06,
+      "loss": 146.4651,
+      "step": 15060
+    },
+    {
+      "epoch": 0.12466393679943748,
+      "grad_norm": 1310.355224609375,
+      "learning_rate": 9.985559816730277e-06,
+      "loss": 141.5489,
+      "step": 15070
+    },
+    {
+      "epoch": 0.12474666004880672,
+      "grad_norm": 1190.0335693359375,
+      "learning_rate": 9.985452507234037e-06,
+      "loss": 144.9001,
+      "step": 15080
+    },
+    {
+      "epoch": 0.12482938329817596,
+      "grad_norm": 2714.714599609375,
+      "learning_rate": 9.98534480106822e-06,
+      "loss": 154.9118,
+      "step": 15090
+    },
+    {
+      "epoch": 0.12491210654754518,
+      "grad_norm": 792.223388671875,
+      "learning_rate": 9.985236698241396e-06,
+      "loss": 149.7406,
+      "step": 15100
+    },
+    {
+      "epoch": 0.12499482979691443,
+      "grad_norm": 1287.8345947265625,
+      "learning_rate": 9.985128198762168e-06,
+      "loss": 171.4261,
+      "step": 15110
+    },
+    {
+      "epoch": 0.12507755304628365,
+      "grad_norm": 957.4619140625,
+      "learning_rate": 9.98501930263917e-06,
+      "loss": 234.8733,
+      "step": 15120
+    },
+    {
+      "epoch": 0.1251602762956529,
+      "grad_norm": 862.0460205078125,
+      "learning_rate": 9.984910009881062e-06,
+      "loss": 112.6332,
+      "step": 15130
+    },
+    {
+      "epoch": 0.12524299954502213,
+      "grad_norm": 1100.07080078125,
+      "learning_rate": 9.984800320496542e-06,
+      "loss": 139.7673,
+      "step": 15140
+    },
+    {
+      "epoch": 0.12532572279439136,
+      "grad_norm": 1111.9737548828125,
+      "learning_rate": 9.984690234494338e-06,
+      "loss": 106.7051,
+      "step": 15150
+    },
+    {
+      "epoch": 0.12540844604376059,
+      "grad_norm": 744.8794555664062,
+      "learning_rate": 9.98457975188321e-06,
+      "loss": 142.1312,
+      "step": 15160
+    },
+    {
+      "epoch": 0.12549116929312984,
+      "grad_norm": 923.189697265625,
+      "learning_rate": 9.984468872671945e-06,
+      "loss": 139.3656,
+      "step": 15170
+    },
+    {
+      "epoch": 0.12557389254249907,
+      "grad_norm": 1322.687255859375,
+      "learning_rate": 9.984357596869369e-06,
+      "loss": 148.6495,
+      "step": 15180
+    },
+    {
+      "epoch": 0.1256566157918683,
+      "grad_norm": 882.3487548828125,
+      "learning_rate": 9.984245924484334e-06,
+      "loss": 141.7766,
+      "step": 15190
+    },
+    {
+      "epoch": 0.12573933904123755,
+      "grad_norm": 725.9840698242188,
+      "learning_rate": 9.984133855525723e-06,
+      "loss": 138.5364,
+      "step": 15200
+    },
+    {
+      "epoch": 0.12582206229060677,
+      "grad_norm": 1547.522705078125,
+      "learning_rate": 9.984021390002458e-06,
+      "loss": 136.4458,
+      "step": 15210
+    },
+    {
+      "epoch": 0.125904785539976,
+      "grad_norm": 1425.56494140625,
+      "learning_rate": 9.983908527923486e-06,
+      "loss": 222.0387,
+      "step": 15220
+    },
+    {
+      "epoch": 0.12598750878934525,
+      "grad_norm": 911.7035522460938,
+      "learning_rate": 9.983795269297782e-06,
+      "loss": 169.7902,
+      "step": 15230
+    },
+    {
+      "epoch": 0.12607023203871448,
+      "grad_norm": 1066.0264892578125,
+      "learning_rate": 9.983681614134363e-06,
+      "loss": 122.5573,
+      "step": 15240
+    },
+    {
+      "epoch": 0.1261529552880837,
+      "grad_norm": 1829.2509765625,
+      "learning_rate": 9.98356756244227e-06,
+      "loss": 154.7958,
+      "step": 15250
+    },
+    {
+      "epoch": 0.12623567853745296,
+      "grad_norm": 1402.93408203125,
+      "learning_rate": 9.983453114230575e-06,
+      "loss": 145.442,
+      "step": 15260
+    },
+    {
+      "epoch": 0.1263184017868222,
+      "grad_norm": 990.7800903320312,
+      "learning_rate": 9.98333826950839e-06,
+      "loss": 138.916,
+      "step": 15270
+    },
+    {
+      "epoch": 0.12640112503619141,
+      "grad_norm": 861.1292724609375,
+      "learning_rate": 9.983223028284847e-06,
+      "loss": 152.3527,
+      "step": 15280
+    },
+    {
+      "epoch": 0.12648384828556067,
+      "grad_norm": 887.3511962890625,
+      "learning_rate": 9.983107390569118e-06,
+      "loss": 129.7973,
+      "step": 15290
+    },
+    {
+      "epoch": 0.1265665715349299,
+      "grad_norm": 1043.2373046875,
+      "learning_rate": 9.982991356370404e-06,
+      "loss": 116.1451,
+      "step": 15300
+    },
+    {
+      "epoch": 0.12664929478429912,
+      "grad_norm": 1244.5079345703125,
+      "learning_rate": 9.982874925697937e-06,
+      "loss": 221.0664,
+      "step": 15310
+    },
+    {
+      "epoch": 0.12673201803366838,
+      "grad_norm": 1715.1995849609375,
+      "learning_rate": 9.982758098560978e-06,
+      "loss": 186.7455,
+      "step": 15320
+    },
+    {
+      "epoch": 0.1268147412830376,
+      "grad_norm": 679.9988403320312,
+      "learning_rate": 9.982640874968827e-06,
+      "loss": 171.8672,
+      "step": 15330
+    },
+    {
+      "epoch": 0.12689746453240683,
+      "grad_norm": 595.40625,
+      "learning_rate": 9.98252325493081e-06,
+      "loss": 130.3511,
+      "step": 15340
+    },
+    {
+      "epoch": 0.12698018778177605,
+      "grad_norm": 915.3275146484375,
+      "learning_rate": 9.982405238456281e-06,
+      "loss": 153.7831,
+      "step": 15350
+    },
+    {
+      "epoch": 0.1270629110311453,
+      "grad_norm": 1383.0423583984375,
+      "learning_rate": 9.982286825554636e-06,
+      "loss": 155.1486,
+      "step": 15360
+    },
+    {
+      "epoch": 0.12714563428051454,
+      "grad_norm": 1527.0670166015625,
+      "learning_rate": 9.982168016235292e-06,
+      "loss": 235.3831,
+      "step": 15370
+    },
+    {
+      "epoch": 0.12722835752988376,
+      "grad_norm": 1168.0416259765625,
+      "learning_rate": 9.982048810507706e-06,
+      "loss": 175.3166,
+      "step": 15380
+    },
+    {
+      "epoch": 0.12731108077925302,
+      "grad_norm": 1577.9107666015625,
+      "learning_rate": 9.98192920838136e-06,
+      "loss": 136.4098,
+      "step": 15390
+    },
+    {
+      "epoch": 0.12739380402862224,
+      "grad_norm": 2239.125244140625,
+      "learning_rate": 9.98180920986577e-06,
+      "loss": 162.4811,
+      "step": 15400
+    },
+    {
+      "epoch": 0.12747652727799147,
+      "grad_norm": 1140.6561279296875,
+      "learning_rate": 9.981688814970485e-06,
+      "loss": 159.3877,
+      "step": 15410
+    },
+    {
+      "epoch": 0.12755925052736072,
+      "grad_norm": 929.8948974609375,
+      "learning_rate": 9.981568023705085e-06,
+      "loss": 113.0717,
+      "step": 15420
+    },
+    {
+      "epoch": 0.12764197377672995,
+      "grad_norm": 1146.56396484375,
+      "learning_rate": 9.981446836079178e-06,
+      "loss": 121.9914,
+      "step": 15430
+    },
+    {
+      "epoch": 0.12772469702609918,
+      "grad_norm": 877.8550415039062,
+      "learning_rate": 9.981325252102408e-06,
+      "loss": 173.6141,
+      "step": 15440
+    },
+    {
+      "epoch": 0.12780742027546843,
+      "grad_norm": 969.0079956054688,
+      "learning_rate": 9.98120327178445e-06,
+      "loss": 178.5706,
+      "step": 15450
+    },
+    {
+      "epoch": 0.12789014352483766,
+      "grad_norm": 1263.2391357421875,
+      "learning_rate": 9.981080895135007e-06,
+      "loss": 180.7431,
+      "step": 15460
+    },
+    {
+      "epoch": 0.12797286677420688,
+      "grad_norm": 742.5184326171875,
+      "learning_rate": 9.980958122163818e-06,
+      "loss": 111.0224,
+      "step": 15470
+    },
+    {
+      "epoch": 0.12805559002357614,
+      "grad_norm": 1423.19873046875,
+      "learning_rate": 9.980834952880652e-06,
+      "loss": 128.3473,
+      "step": 15480
+    },
+    {
+      "epoch": 0.12813831327294536,
+      "grad_norm": 1583.2940673828125,
+      "learning_rate": 9.980711387295306e-06,
+      "loss": 149.955,
+      "step": 15490
+    },
+    {
+      "epoch": 0.1282210365223146,
+      "grad_norm": 1052.7265625,
+      "learning_rate": 9.980587425417612e-06,
+      "loss": 159.9205,
+      "step": 15500
+    },
+    {
+      "epoch": 0.12830375977168385,
+      "grad_norm": 2138.17724609375,
+      "learning_rate": 9.980463067257437e-06,
+      "loss": 169.7366,
+      "step": 15510
+    },
+    {
+      "epoch": 0.12838648302105307,
+      "grad_norm": 1006.1878662109375,
+      "learning_rate": 9.980338312824672e-06,
+      "loss": 193.1612,
+      "step": 15520
+    },
+    {
+      "epoch": 0.1284692062704223,
+      "grad_norm": 1047.7593994140625,
+      "learning_rate": 9.980213162129244e-06,
+      "loss": 175.5892,
+      "step": 15530
+    },
+    {
+      "epoch": 0.12855192951979152,
+      "grad_norm": 1267.4644775390625,
+      "learning_rate": 9.980087615181111e-06,
+      "loss": 149.4357,
+      "step": 15540
+    },
+    {
+      "epoch": 0.12863465276916078,
+      "grad_norm": 1171.0859375,
+      "learning_rate": 9.979961671990263e-06,
+      "loss": 165.6414,
+      "step": 15550
+    },
+    {
+      "epoch": 0.12871737601853,
+      "grad_norm": 911.0418701171875,
+      "learning_rate": 9.979835332566719e-06,
+      "loss": 155.2462,
+      "step": 15560
+    },
+    {
+      "epoch": 0.12880009926789923,
+      "grad_norm": 1016.1674194335938,
+      "learning_rate": 9.97970859692053e-06,
+      "loss": 142.4974,
+      "step": 15570
+    },
+    {
+      "epoch": 0.12888282251726849,
+      "grad_norm": 653.232421875,
+      "learning_rate": 9.979581465061784e-06,
+      "loss": 155.5012,
+      "step": 15580
+    },
+    {
+      "epoch": 0.1289655457666377,
+      "grad_norm": 1058.5008544921875,
+      "learning_rate": 9.979453937000594e-06,
+      "loss": 101.9423,
+      "step": 15590
+    },
+    {
+      "epoch": 0.12904826901600694,
+      "grad_norm": 820.3455200195312,
+      "learning_rate": 9.979326012747106e-06,
+      "loss": 117.5258,
+      "step": 15600
+    },
+    {
+      "epoch": 0.1291309922653762,
+      "grad_norm": 1352.7105712890625,
+      "learning_rate": 9.9791976923115e-06,
+      "loss": 99.8209,
+      "step": 15610
+    },
+    {
+      "epoch": 0.12921371551474542,
+      "grad_norm": 1026.9713134765625,
+      "learning_rate": 9.979068975703984e-06,
+      "loss": 166.7305,
+      "step": 15620
+    },
+    {
+      "epoch": 0.12929643876411465,
+      "grad_norm": 1245.904296875,
+      "learning_rate": 9.978939862934802e-06,
+      "loss": 126.3938,
+      "step": 15630
+    },
+    {
+      "epoch": 0.1293791620134839,
+      "grad_norm": 1658.7640380859375,
+      "learning_rate": 9.978810354014223e-06,
+      "loss": 135.5493,
+      "step": 15640
+    },
+    {
+      "epoch": 0.12946188526285313,
+      "grad_norm": 1517.82373046875,
+      "learning_rate": 9.978680448952556e-06,
+      "loss": 139.2036,
+      "step": 15650
+    },
+    {
+      "epoch": 0.12954460851222235,
+      "grad_norm": 1112.2469482421875,
+      "learning_rate": 9.978550147760133e-06,
+      "loss": 127.4167,
+      "step": 15660
+    },
+    {
+      "epoch": 0.1296273317615916,
+      "grad_norm": 995.6966552734375,
+      "learning_rate": 9.978419450447325e-06,
+      "loss": 128.1456,
+      "step": 15670
+    },
+    {
+      "epoch": 0.12971005501096083,
+      "grad_norm": 785.5421142578125,
+      "learning_rate": 9.978288357024527e-06,
+      "loss": 142.7447,
+      "step": 15680
+    },
+    {
+      "epoch": 0.12979277826033006,
+      "grad_norm": 3418.008544921875,
+      "learning_rate": 9.978156867502173e-06,
+      "loss": 161.2918,
+      "step": 15690
+    },
+    {
+      "epoch": 0.12987550150969931,
+      "grad_norm": 1545.32568359375,
+      "learning_rate": 9.978024981890724e-06,
+      "loss": 107.0028,
+      "step": 15700
+    },
+    {
+      "epoch": 0.12995822475906854,
+      "grad_norm": 963.7340087890625,
+      "learning_rate": 9.977892700200673e-06,
+      "loss": 140.569,
+      "step": 15710
+    },
+    {
+      "epoch": 0.13004094800843777,
+      "grad_norm": 692.7611694335938,
+      "learning_rate": 9.977760022442545e-06,
+      "loss": 110.664,
+      "step": 15720
+    },
+    {
+      "epoch": 0.130123671257807,
+      "grad_norm": 1015.7322998046875,
+      "learning_rate": 9.977626948626897e-06,
+      "loss": 158.9243,
+      "step": 15730
+    },
+    {
+      "epoch": 0.13020639450717625,
+      "grad_norm": 1334.6917724609375,
+      "learning_rate": 9.977493478764316e-06,
+      "loss": 152.3215,
+      "step": 15740
+    },
+    {
+      "epoch": 0.13028911775654548,
+      "grad_norm": 963.1575927734375,
+      "learning_rate": 9.977359612865424e-06,
+      "loss": 137.1868,
+      "step": 15750
+    },
+    {
+      "epoch": 0.1303718410059147,
+      "grad_norm": 1332.0909423828125,
+      "learning_rate": 9.97722535094087e-06,
+      "loss": 130.2153,
+      "step": 15760
+    },
+    {
+      "epoch": 0.13045456425528396,
+      "grad_norm": 864.9472045898438,
+      "learning_rate": 9.977090693001336e-06,
+      "loss": 142.6017,
+      "step": 15770
+    },
+    {
+      "epoch": 0.13053728750465318,
+      "grad_norm": 1091.3128662109375,
+      "learning_rate": 9.976955639057539e-06,
+      "loss": 126.0693,
+      "step": 15780
+    },
+    {
+      "epoch": 0.1306200107540224,
+      "grad_norm": 1137.7115478515625,
+      "learning_rate": 9.976820189120223e-06,
+      "loss": 147.4185,
+      "step": 15790
+    },
+    {
+      "epoch": 0.13070273400339166,
+      "grad_norm": 1658.2982177734375,
+      "learning_rate": 9.976684343200164e-06,
+      "loss": 135.441,
+      "step": 15800
+    },
+    {
+      "epoch": 0.1307854572527609,
+      "grad_norm": 1823.3642578125,
+      "learning_rate": 9.976548101308173e-06,
+      "loss": 138.8229,
+      "step": 15810
+    },
+    {
+      "epoch": 0.13086818050213012,
+      "grad_norm": 1511.6375732421875,
+      "learning_rate": 9.976411463455088e-06,
+      "loss": 140.3549,
+      "step": 15820
+    },
+    {
+      "epoch": 0.13095090375149937,
+      "grad_norm": 970.9559326171875,
+      "learning_rate": 9.976274429651783e-06,
+      "loss": 188.4605,
+      "step": 15830
+    },
+    {
+      "epoch": 0.1310336270008686,
+      "grad_norm": 1540.7110595703125,
+      "learning_rate": 9.976136999909156e-06,
+      "loss": 106.2589,
+      "step": 15840
+    },
+    {
+      "epoch": 0.13111635025023782,
+      "grad_norm": 829.7328491210938,
+      "learning_rate": 9.97599917423815e-06,
+      "loss": 166.2885,
+      "step": 15850
+    },
+    {
+      "epoch": 0.13119907349960708,
+      "grad_norm": 0.0,
+      "learning_rate": 9.975860952649724e-06,
+      "loss": 180.9173,
+      "step": 15860
+    },
+    {
+      "epoch": 0.1312817967489763,
+      "grad_norm": 1011.6902465820312,
+      "learning_rate": 9.975722335154876e-06,
+      "loss": 161.2201,
+      "step": 15870
+    },
+    {
+      "epoch": 0.13136451999834553,
+      "grad_norm": 1166.7960205078125,
+      "learning_rate": 9.975583321764638e-06,
+      "loss": 144.3113,
+      "step": 15880
+    },
+    {
+      "epoch": 0.13144724324771476,
+      "grad_norm": 1041.4771728515625,
+      "learning_rate": 9.975443912490073e-06,
+      "loss": 149.5042,
+      "step": 15890
+    },
+    {
+      "epoch": 0.131529966497084,
+      "grad_norm": 2316.5087890625,
+      "learning_rate": 9.975304107342268e-06,
+      "loss": 179.2303,
+      "step": 15900
+    },
+    {
+      "epoch": 0.13161268974645324,
+      "grad_norm": 692.1578369140625,
+      "learning_rate": 9.97516390633235e-06,
+      "loss": 133.4318,
+      "step": 15910
+    },
+    {
+      "epoch": 0.13169541299582246,
+      "grad_norm": 1125.5006103515625,
+      "learning_rate": 9.975023309471473e-06,
+      "loss": 156.1001,
+      "step": 15920
+    },
+    {
+      "epoch": 0.13177813624519172,
+      "grad_norm": 1064.159423828125,
+      "learning_rate": 9.974882316770823e-06,
+      "loss": 147.876,
+      "step": 15930
+    },
+    {
+      "epoch": 0.13186085949456094,
+      "grad_norm": 1456.9761962890625,
+      "learning_rate": 9.974740928241617e-06,
+      "loss": 146.098,
+      "step": 15940
+    },
+    {
+      "epoch": 0.13194358274393017,
+      "grad_norm": 1191.022705078125,
+      "learning_rate": 9.974599143895107e-06,
+      "loss": 139.6693,
+      "step": 15950
+    },
+    {
+      "epoch": 0.13202630599329943,
+      "grad_norm": 2010.4088134765625,
+      "learning_rate": 9.974456963742573e-06,
+      "loss": 152.4677,
+      "step": 15960
+    },
+    {
+      "epoch": 0.13210902924266865,
+      "grad_norm": 1077.85205078125,
+      "learning_rate": 9.97431438779533e-06,
+      "loss": 199.6097,
+      "step": 15970
+    },
+    {
+      "epoch": 0.13219175249203788,
+      "grad_norm": 975.5093994140625,
+      "learning_rate": 9.974171416064719e-06,
+      "loss": 110.029,
+      "step": 15980
+    },
+    {
+      "epoch": 0.13227447574140713,
+      "grad_norm": 1180.7437744140625,
+      "learning_rate": 9.974028048562118e-06,
+      "loss": 136.7102,
+      "step": 15990
+    },
+    {
+      "epoch": 0.13235719899077636,
+      "grad_norm": 1241.7110595703125,
+      "learning_rate": 9.973884285298932e-06,
+      "loss": 154.7749,
+      "step": 16000
+    },
+    {
+      "epoch": 0.13243992224014559,
+      "grad_norm": 1181.715576171875,
+      "learning_rate": 9.9737401262866e-06,
+      "loss": 163.9049,
+      "step": 16010
+    },
+    {
+      "epoch": 0.13252264548951484,
+      "grad_norm": 1441.7060546875,
+      "learning_rate": 9.973595571536593e-06,
+      "loss": 131.6654,
+      "step": 16020
+    },
+    {
+      "epoch": 0.13260536873888407,
+      "grad_norm": 1810.145751953125,
+      "learning_rate": 9.973450621060412e-06,
+      "loss": 155.4361,
+      "step": 16030
+    },
+    {
+      "epoch": 0.1326880919882533,
+      "grad_norm": 1024.084716796875,
+      "learning_rate": 9.97330527486959e-06,
+      "loss": 130.6234,
+      "step": 16040
+    },
+    {
+      "epoch": 0.13277081523762255,
+      "grad_norm": 1294.561279296875,
+      "learning_rate": 9.973159532975691e-06,
+      "loss": 122.6079,
+      "step": 16050
+    },
+    {
+      "epoch": 0.13285353848699177,
+      "grad_norm": 1282.573486328125,
+      "learning_rate": 9.973013395390314e-06,
+      "loss": 173.6021,
+      "step": 16060
+    },
+    {
+      "epoch": 0.132936261736361,
+      "grad_norm": 1436.6795654296875,
+      "learning_rate": 9.972866862125083e-06,
+      "loss": 201.6667,
+      "step": 16070
+    },
+    {
+      "epoch": 0.13301898498573023,
+      "grad_norm": 880.5997924804688,
+      "learning_rate": 9.972719933191657e-06,
+      "loss": 121.1312,
+      "step": 16080
+    },
+    {
+      "epoch": 0.13310170823509948,
+      "grad_norm": 720.5911254882812,
+      "learning_rate": 9.97257260860173e-06,
+      "loss": 117.1484,
+      "step": 16090
+    },
+    {
+      "epoch": 0.1331844314844687,
+      "grad_norm": 1505.3927001953125,
+      "learning_rate": 9.972424888367019e-06,
+      "loss": 146.7309,
+      "step": 16100
+    },
+    {
+      "epoch": 0.13326715473383793,
+      "grad_norm": 958.6402587890625,
+      "learning_rate": 9.972276772499281e-06,
+      "loss": 156.9766,
+      "step": 16110
+    },
+    {
+      "epoch": 0.1333498779832072,
+      "grad_norm": 877.50244140625,
+      "learning_rate": 9.9721282610103e-06,
+      "loss": 191.0899,
+      "step": 16120
+    },
+    {
+      "epoch": 0.13343260123257641,
+      "grad_norm": 1021.2138671875,
+      "learning_rate": 9.971979353911891e-06,
+      "loss": 133.9165,
+      "step": 16130
+    },
+    {
+      "epoch": 0.13351532448194564,
+      "grad_norm": 847.0870971679688,
+      "learning_rate": 9.971830051215905e-06,
+      "loss": 101.3374,
+      "step": 16140
+    },
+    {
+      "epoch": 0.1335980477313149,
+      "grad_norm": 2785.597412109375,
+      "learning_rate": 9.97168035293422e-06,
+      "loss": 267.7292,
+      "step": 16150
+    },
+    {
+      "epoch": 0.13368077098068412,
+      "grad_norm": 801.3421020507812,
+      "learning_rate": 9.971530259078743e-06,
+      "loss": 111.4734,
+      "step": 16160
+    },
+    {
+      "epoch": 0.13376349423005335,
+      "grad_norm": 768.2542114257812,
+      "learning_rate": 9.971379769661422e-06,
+      "loss": 149.4196,
+      "step": 16170
+    },
+    {
+      "epoch": 0.1338462174794226,
+      "grad_norm": 893.0291748046875,
+      "learning_rate": 9.971228884694228e-06,
+      "loss": 122.37,
+      "step": 16180
+    },
+    {
+      "epoch": 0.13392894072879183,
+      "grad_norm": 1295.072509765625,
+      "learning_rate": 9.971077604189166e-06,
+      "loss": 156.3286,
+      "step": 16190
+    },
+    {
+      "epoch": 0.13401166397816106,
+      "grad_norm": 998.2085571289062,
+      "learning_rate": 9.970925928158275e-06,
+      "loss": 122.403,
+      "step": 16200
+    },
+    {
+      "epoch": 0.1340943872275303,
+      "grad_norm": 701.3370361328125,
+      "learning_rate": 9.970773856613617e-06,
+      "loss": 140.6802,
+      "step": 16210
+    },
+    {
+      "epoch": 0.13417711047689954,
+      "grad_norm": 971.6983032226562,
+      "learning_rate": 9.970621389567301e-06,
+      "loss": 178.1052,
+      "step": 16220
+    },
+    {
+      "epoch": 0.13425983372626876,
+      "grad_norm": 2665.119384765625,
+      "learning_rate": 9.97046852703145e-06,
+      "loss": 138.6044,
+      "step": 16230
+    },
+    {
+      "epoch": 0.13434255697563802,
+      "grad_norm": 2127.31884765625,
+      "learning_rate": 9.970315269018231e-06,
+      "loss": 157.2493,
+      "step": 16240
+    },
+    {
+      "epoch": 0.13442528022500724,
+      "grad_norm": 1778.2391357421875,
+      "learning_rate": 9.970161615539837e-06,
+      "loss": 134.0471,
+      "step": 16250
+    },
+    {
+      "epoch": 0.13450800347437647,
+      "grad_norm": 993.4716796875,
+      "learning_rate": 9.970007566608492e-06,
+      "loss": 146.2506,
+      "step": 16260
+    },
+    {
+      "epoch": 0.1345907267237457,
+      "grad_norm": 798.2664184570312,
+      "learning_rate": 9.969853122236455e-06,
+      "loss": 114.1296,
+      "step": 16270
+    },
+    {
+      "epoch": 0.13467344997311495,
+      "grad_norm": 703.0869750976562,
+      "learning_rate": 9.969698282436013e-06,
+      "loss": 120.5299,
+      "step": 16280
+    },
+    {
+      "epoch": 0.13475617322248418,
+      "grad_norm": 1201.6317138671875,
+      "learning_rate": 9.969543047219487e-06,
+      "loss": 125.8007,
+      "step": 16290
+    },
+    {
+      "epoch": 0.1348388964718534,
+      "grad_norm": 1785.0177001953125,
+      "learning_rate": 9.969387416599227e-06,
+      "loss": 144.5029,
+      "step": 16300
+    },
+    {
+      "epoch": 0.13492161972122266,
+      "grad_norm": 1228.9619140625,
+      "learning_rate": 9.969231390587618e-06,
+      "loss": 164.9693,
+      "step": 16310
+    },
+    {
+      "epoch": 0.13500434297059188,
+      "grad_norm": 864.3604736328125,
+      "learning_rate": 9.969074969197072e-06,
+      "loss": 168.7043,
+      "step": 16320
+    },
+    {
+      "epoch": 0.1350870662199611,
+      "grad_norm": 1214.023681640625,
+      "learning_rate": 9.968918152440036e-06,
+      "loss": 172.751,
+      "step": 16330
+    },
+    {
+      "epoch": 0.13516978946933036,
+      "grad_norm": 928.501220703125,
+      "learning_rate": 9.968760940328987e-06,
+      "loss": 131.5311,
+      "step": 16340
+    },
+    {
+      "epoch": 0.1352525127186996,
+      "grad_norm": 510.1147155761719,
+      "learning_rate": 9.968603332876435e-06,
+      "loss": 171.1721,
+      "step": 16350
+    },
+    {
+      "epoch": 0.13533523596806882,
+      "grad_norm": 1110.3807373046875,
+      "learning_rate": 9.968445330094915e-06,
+      "loss": 169.255,
+      "step": 16360
+    },
+    {
+      "epoch": 0.13541795921743807,
+      "grad_norm": 1672.8614501953125,
+      "learning_rate": 9.968286931997004e-06,
+      "loss": 112.5926,
+      "step": 16370
+    },
+    {
+      "epoch": 0.1355006824668073,
+      "grad_norm": 1014.0128784179688,
+      "learning_rate": 9.968128138595304e-06,
+      "loss": 100.9882,
+      "step": 16380
+    },
+    {
+      "epoch": 0.13558340571617652,
+      "grad_norm": 1446.147216796875,
+      "learning_rate": 9.967968949902448e-06,
+      "loss": 185.0402,
+      "step": 16390
+    },
+    {
+      "epoch": 0.13566612896554578,
+      "grad_norm": 753.0343627929688,
+      "learning_rate": 9.967809365931102e-06,
+      "loss": 148.759,
+      "step": 16400
+    },
+    {
+      "epoch": 0.135748852214915,
+      "grad_norm": 909.8871459960938,
+      "learning_rate": 9.967649386693964e-06,
+      "loss": 123.6662,
+      "step": 16410
+    },
+    {
+      "epoch": 0.13583157546428423,
+      "grad_norm": 1223.7244873046875,
+      "learning_rate": 9.967489012203765e-06,
+      "loss": 132.6178,
+      "step": 16420
+    },
+    {
+      "epoch": 0.13591429871365346,
+      "grad_norm": 1106.0858154296875,
+      "learning_rate": 9.967328242473261e-06,
+      "loss": 146.9553,
+      "step": 16430
+    },
+    {
+      "epoch": 0.1359970219630227,
+      "grad_norm": 1789.8829345703125,
+      "learning_rate": 9.967167077515246e-06,
+      "loss": 133.0784,
+      "step": 16440
+    },
+    {
+      "epoch": 0.13607974521239194,
+      "grad_norm": 741.1414184570312,
+      "learning_rate": 9.967005517342544e-06,
+      "loss": 143.1583,
+      "step": 16450
+    },
+    {
+      "epoch": 0.13616246846176117,
+      "grad_norm": 1324.021240234375,
+      "learning_rate": 9.966843561968005e-06,
+      "loss": 108.1861,
+      "step": 16460
+    },
+    {
+      "epoch": 0.13624519171113042,
+      "grad_norm": 866.0011596679688,
+      "learning_rate": 9.966681211404521e-06,
+      "loss": 138.6324,
+      "step": 16470
+    },
+    {
+      "epoch": 0.13632791496049965,
+      "grad_norm": 520.3377685546875,
+      "learning_rate": 9.966518465665007e-06,
+      "loss": 113.3134,
+      "step": 16480
+    },
+    {
+      "epoch": 0.13641063820986887,
+      "grad_norm": 883.1153564453125,
+      "learning_rate": 9.966355324762412e-06,
+      "loss": 163.313,
+      "step": 16490
+    },
+    {
+      "epoch": 0.13649336145923813,
+      "grad_norm": 1007.1843872070312,
+      "learning_rate": 9.966191788709716e-06,
+      "loss": 140.2184,
+      "step": 16500
+    },
+    {
+      "epoch": 0.13657608470860735,
+      "grad_norm": 1669.2816162109375,
+      "learning_rate": 9.966027857519931e-06,
+      "loss": 188.2176,
+      "step": 16510
+    },
+    {
+      "epoch": 0.13665880795797658,
+      "grad_norm": 772.6116943359375,
+      "learning_rate": 9.9658635312061e-06,
+      "loss": 163.7544,
+      "step": 16520
+    },
+    {
+      "epoch": 0.13674153120734583,
+      "grad_norm": 706.4850463867188,
+      "learning_rate": 9.965698809781298e-06,
+      "loss": 121.3989,
+      "step": 16530
+    },
+    {
+      "epoch": 0.13682425445671506,
+      "grad_norm": 766.0828247070312,
+      "learning_rate": 9.965533693258632e-06,
+      "loss": 213.4713,
+      "step": 16540
+    },
+    {
+      "epoch": 0.1369069777060843,
+      "grad_norm": 957.917724609375,
+      "learning_rate": 9.965368181651239e-06,
+      "loss": 183.1273,
+      "step": 16550
+    },
+    {
+      "epoch": 0.13698970095545354,
+      "grad_norm": 696.8062744140625,
+      "learning_rate": 9.965202274972288e-06,
+      "loss": 112.6891,
+      "step": 16560
+    },
+    {
+      "epoch": 0.13707242420482277,
+      "grad_norm": 902.3621215820312,
+      "learning_rate": 9.965035973234977e-06,
+      "loss": 113.6838,
+      "step": 16570
+    },
+    {
+      "epoch": 0.137155147454192,
+      "grad_norm": 1020.390625,
+      "learning_rate": 9.964869276452542e-06,
+      "loss": 106.0109,
+      "step": 16580
+    },
+    {
+      "epoch": 0.13723787070356125,
+      "grad_norm": 1181.8326416015625,
+      "learning_rate": 9.964702184638244e-06,
+      "loss": 139.7021,
+      "step": 16590
+    },
+    {
+      "epoch": 0.13732059395293048,
+      "grad_norm": 629.9285278320312,
+      "learning_rate": 9.964534697805377e-06,
+      "loss": 193.1732,
+      "step": 16600
+    },
+    {
+      "epoch": 0.1374033172022997,
+      "grad_norm": 1531.7962646484375,
+      "learning_rate": 9.96436681596727e-06,
+      "loss": 154.7776,
+      "step": 16610
+    },
+    {
+      "epoch": 0.13748604045166893,
+      "grad_norm": 1220.1796875,
+      "learning_rate": 9.964198539137277e-06,
+      "loss": 191.2195,
+      "step": 16620
+    },
+    {
+      "epoch": 0.13756876370103818,
+      "grad_norm": 0.0,
+      "learning_rate": 9.964029867328791e-06,
+      "loss": 112.8693,
+      "step": 16630
+    },
+    {
+      "epoch": 0.1376514869504074,
+      "grad_norm": 1105.817626953125,
+      "learning_rate": 9.963860800555228e-06,
+      "loss": 103.0777,
+      "step": 16640
+    },
+    {
+      "epoch": 0.13773421019977664,
+      "grad_norm": 472.4584655761719,
+      "learning_rate": 9.963691338830045e-06,
+      "loss": 123.1952,
+      "step": 16650
+    },
+    {
+      "epoch": 0.1378169334491459,
+      "grad_norm": 990.940673828125,
+      "learning_rate": 9.963521482166718e-06,
+      "loss": 136.4567,
+      "step": 16660
+    },
+    {
+      "epoch": 0.13789965669851512,
+      "grad_norm": 1503.9461669921875,
+      "learning_rate": 9.96335123057877e-06,
+      "loss": 136.2858,
+      "step": 16670
+    },
+    {
+      "epoch": 0.13798237994788434,
+      "grad_norm": 1348.58740234375,
+      "learning_rate": 9.963180584079741e-06,
+      "loss": 137.5341,
+      "step": 16680
+    },
+    {
+      "epoch": 0.1380651031972536,
+      "grad_norm": 1100.0037841796875,
+      "learning_rate": 9.963009542683214e-06,
+      "loss": 199.9709,
+      "step": 16690
+    },
+    {
+      "epoch": 0.13814782644662282,
+      "grad_norm": 718.8609619140625,
+      "learning_rate": 9.962838106402791e-06,
+      "loss": 184.6782,
+      "step": 16700
+    },
+    {
+      "epoch": 0.13823054969599205,
+      "grad_norm": 865.5576782226562,
+      "learning_rate": 9.962666275252117e-06,
+      "loss": 104.1854,
+      "step": 16710
+    },
+    {
+      "epoch": 0.1383132729453613,
+      "grad_norm": 1161.63525390625,
+      "learning_rate": 9.962494049244866e-06,
+      "loss": 169.3983,
+      "step": 16720
+    },
+    {
+      "epoch": 0.13839599619473053,
+      "grad_norm": 589.0774536132812,
+      "learning_rate": 9.962321428394735e-06,
+      "loss": 165.776,
+      "step": 16730
+    },
+    {
+      "epoch": 0.13847871944409976,
+      "grad_norm": 2693.160888671875,
+      "learning_rate": 9.962148412715464e-06,
+      "loss": 154.1448,
+      "step": 16740
+    },
+    {
+      "epoch": 0.138561442693469,
+      "grad_norm": 1310.2269287109375,
+      "learning_rate": 9.961975002220816e-06,
+      "loss": 166.3599,
+      "step": 16750
+    },
+    {
+      "epoch": 0.13864416594283824,
+      "grad_norm": 1167.153076171875,
+      "learning_rate": 9.96180119692459e-06,
+      "loss": 171.0495,
+      "step": 16760
+    },
+    {
+      "epoch": 0.13872688919220746,
+      "grad_norm": 1377.29833984375,
+      "learning_rate": 9.961626996840613e-06,
+      "loss": 102.7167,
+      "step": 16770
+    },
+    {
+      "epoch": 0.13880961244157672,
+      "grad_norm": 977.5831909179688,
+      "learning_rate": 9.961452401982748e-06,
+      "loss": 136.4004,
+      "step": 16780
+    },
+    {
+      "epoch": 0.13889233569094595,
+      "grad_norm": 1010.1982421875,
+      "learning_rate": 9.961277412364884e-06,
+      "loss": 146.971,
+      "step": 16790
+    },
+    {
+      "epoch": 0.13897505894031517,
+      "grad_norm": 814.7576293945312,
+      "learning_rate": 9.961102028000948e-06,
+      "loss": 213.2676,
+      "step": 16800
+    },
+    {
+      "epoch": 0.1390577821896844,
+      "grad_norm": 881.7014770507812,
+      "learning_rate": 9.96092624890489e-06,
+      "loss": 91.0271,
+      "step": 16810
+    },
+    {
+      "epoch": 0.13914050543905365,
+      "grad_norm": 4899.205078125,
+      "learning_rate": 9.960750075090698e-06,
+      "loss": 166.8467,
+      "step": 16820
+    },
+    {
+      "epoch": 0.13922322868842288,
+      "grad_norm": 1270.030029296875,
+      "learning_rate": 9.960573506572391e-06,
+      "loss": 186.535,
+      "step": 16830
+    },
+    {
+      "epoch": 0.1393059519377921,
+      "grad_norm": 1338.3089599609375,
+      "learning_rate": 9.960396543364013e-06,
+      "loss": 192.4324,
+      "step": 16840
+    },
+    {
+      "epoch": 0.13938867518716136,
+      "grad_norm": 1512.3917236328125,
+      "learning_rate": 9.96021918547965e-06,
+      "loss": 124.9194,
+      "step": 16850
+    },
+    {
+      "epoch": 0.13947139843653059,
+      "grad_norm": 1637.7535400390625,
+      "learning_rate": 9.96004143293341e-06,
+      "loss": 131.2566,
+      "step": 16860
+    },
+    {
+      "epoch": 0.1395541216858998,
+      "grad_norm": 1564.211669921875,
+      "learning_rate": 9.959863285739436e-06,
+      "loss": 124.8255,
+      "step": 16870
+    },
+    {
+      "epoch": 0.13963684493526907,
+      "grad_norm": 720.8834228515625,
+      "learning_rate": 9.959684743911904e-06,
+      "loss": 140.7759,
+      "step": 16880
+    },
+    {
+      "epoch": 0.1397195681846383,
+      "grad_norm": 796.6300659179688,
+      "learning_rate": 9.959505807465018e-06,
+      "loss": 120.1176,
+      "step": 16890
+    },
+    {
+      "epoch": 0.13980229143400752,
+      "grad_norm": 1232.4276123046875,
+      "learning_rate": 9.959326476413016e-06,
+      "loss": 130.2664,
+      "step": 16900
+    },
+    {
+      "epoch": 0.13988501468337677,
+      "grad_norm": 457.3919677734375,
+      "learning_rate": 9.959146750770167e-06,
+      "loss": 124.8512,
+      "step": 16910
+    },
+    {
+      "epoch": 0.139967737932746,
+      "grad_norm": 708.2092895507812,
+      "learning_rate": 9.95896663055077e-06,
+      "loss": 120.5444,
+      "step": 16920
+    },
+    {
+      "epoch": 0.14005046118211523,
+      "grad_norm": 995.7003784179688,
+      "learning_rate": 9.958786115769157e-06,
+      "loss": 114.9213,
+      "step": 16930
+    },
+    {
+      "epoch": 0.14013318443148448,
+      "grad_norm": 1515.4827880859375,
+      "learning_rate": 9.958605206439692e-06,
+      "loss": 146.7894,
+      "step": 16940
+    },
+    {
+      "epoch": 0.1402159076808537,
+      "grad_norm": 814.6317138671875,
+      "learning_rate": 9.958423902576764e-06,
+      "loss": 99.1024,
+      "step": 16950
+    },
+    {
+      "epoch": 0.14029863093022293,
+      "grad_norm": 760.5602416992188,
+      "learning_rate": 9.958242204194804e-06,
+      "loss": 160.827,
+      "step": 16960
+    },
+    {
+      "epoch": 0.1403813541795922,
+      "grad_norm": 738.33349609375,
+      "learning_rate": 9.958060111308267e-06,
+      "loss": 136.0457,
+      "step": 16970
+    },
+    {
+      "epoch": 0.14046407742896141,
+      "grad_norm": 1149.28857421875,
+      "learning_rate": 9.957877623931642e-06,
+      "loss": 151.1577,
+      "step": 16980
+    },
+    {
+      "epoch": 0.14054680067833064,
+      "grad_norm": 1362.2108154296875,
+      "learning_rate": 9.95769474207945e-06,
+      "loss": 173.5694,
+      "step": 16990
+    },
+    {
+      "epoch": 0.14062952392769987,
+      "grad_norm": 1314.1846923828125,
+      "learning_rate": 9.957511465766236e-06,
+      "loss": 169.4035,
+      "step": 17000
+    },
+    {
+      "epoch": 0.14071224717706912,
+      "grad_norm": 1065.3922119140625,
+      "learning_rate": 9.957327795006589e-06,
+      "loss": 169.1779,
+      "step": 17010
+    },
+    {
+      "epoch": 0.14079497042643835,
+      "grad_norm": 1382.638427734375,
+      "learning_rate": 9.95714372981512e-06,
+      "loss": 145.6161,
+      "step": 17020
+    },
+    {
+      "epoch": 0.14087769367580757,
+      "grad_norm": 1485.4481201171875,
+      "learning_rate": 9.956959270206474e-06,
+      "loss": 131.7884,
+      "step": 17030
+    },
+    {
+      "epoch": 0.14096041692517683,
+      "grad_norm": 901.7747192382812,
+      "learning_rate": 9.956774416195329e-06,
+      "loss": 129.2612,
+      "step": 17040
+    },
+    {
+      "epoch": 0.14104314017454606,
+      "grad_norm": 1346.950439453125,
+      "learning_rate": 9.956589167796392e-06,
+      "loss": 108.1172,
+      "step": 17050
+    },
+    {
+      "epoch": 0.14112586342391528,
+      "grad_norm": 857.8418579101562,
+      "learning_rate": 9.956403525024402e-06,
+      "loss": 132.697,
+      "step": 17060
+    },
+    {
+      "epoch": 0.14120858667328454,
+      "grad_norm": 1938.1868896484375,
+      "learning_rate": 9.956217487894131e-06,
+      "loss": 165.6452,
+      "step": 17070
+    },
+    {
+      "epoch": 0.14129130992265376,
+      "grad_norm": 749.3518676757812,
+      "learning_rate": 9.95603105642038e-06,
+      "loss": 212.4321,
+      "step": 17080
+    },
+    {
+      "epoch": 0.141374033172023,
+      "grad_norm": 709.408447265625,
+      "learning_rate": 9.955844230617985e-06,
+      "loss": 156.41,
+      "step": 17090
+    },
+    {
+      "epoch": 0.14145675642139224,
+      "grad_norm": 1008.6261596679688,
+      "learning_rate": 9.955657010501807e-06,
+      "loss": 118.0272,
+      "step": 17100
+    },
+    {
+      "epoch": 0.14153947967076147,
+      "grad_norm": 828.895751953125,
+      "learning_rate": 9.955469396086743e-06,
+      "loss": 138.8411,
+      "step": 17110
+    },
+    {
+      "epoch": 0.1416222029201307,
+      "grad_norm": 1362.32421875,
+      "learning_rate": 9.955281387387724e-06,
+      "loss": 145.7589,
+      "step": 17120
+    },
+    {
+      "epoch": 0.14170492616949995,
+      "grad_norm": 1597.079345703125,
+      "learning_rate": 9.955092984419705e-06,
+      "loss": 170.475,
+      "step": 17130
+    },
+    {
+      "epoch": 0.14178764941886918,
+      "grad_norm": 1059.4306640625,
+      "learning_rate": 9.954904187197679e-06,
+      "loss": 158.0434,
+      "step": 17140
+    },
+    {
+      "epoch": 0.1418703726682384,
+      "grad_norm": 694.0506591796875,
+      "learning_rate": 9.954714995736667e-06,
+      "loss": 142.6755,
+      "step": 17150
+    },
+    {
+      "epoch": 0.14195309591760763,
+      "grad_norm": 1392.7862548828125,
+      "learning_rate": 9.95452541005172e-06,
+      "loss": 192.9698,
+      "step": 17160
+    },
+    {
+      "epoch": 0.14203581916697688,
+      "grad_norm": 1239.712646484375,
+      "learning_rate": 9.954335430157926e-06,
+      "loss": 126.2119,
+      "step": 17170
+    },
+    {
+      "epoch": 0.1421185424163461,
+      "grad_norm": 949.230712890625,
+      "learning_rate": 9.9541450560704e-06,
+      "loss": 76.8772,
+      "step": 17180
+    },
+    {
+      "epoch": 0.14220126566571534,
+      "grad_norm": 1190.4364013671875,
+      "learning_rate": 9.953954287804286e-06,
+      "loss": 156.7768,
+      "step": 17190
+    },
+    {
+      "epoch": 0.1422839889150846,
+      "grad_norm": 1422.4742431640625,
+      "learning_rate": 9.953763125374767e-06,
+      "loss": 107.7513,
+      "step": 17200
+    },
+    {
+      "epoch": 0.14236671216445382,
+      "grad_norm": 1076.408935546875,
+      "learning_rate": 9.953571568797049e-06,
+      "loss": 136.0641,
+      "step": 17210
+    },
+    {
+      "epoch": 0.14244943541382304,
+      "grad_norm": 930.828125,
+      "learning_rate": 9.953379618086377e-06,
+      "loss": 143.9599,
+      "step": 17220
+    },
+    {
+      "epoch": 0.1425321586631923,
+      "grad_norm": 1367.8873291015625,
+      "learning_rate": 9.95318727325802e-06,
+      "loss": 128.7768,
+      "step": 17230
+    },
+    {
+      "epoch": 0.14261488191256153,
+      "grad_norm": 1150.171875,
+      "learning_rate": 9.952994534327283e-06,
+      "loss": 124.427,
+      "step": 17240
+    },
+    {
+      "epoch": 0.14269760516193075,
+      "grad_norm": 821.237548828125,
+      "learning_rate": 9.952801401309504e-06,
+      "loss": 137.096,
+      "step": 17250
+    },
+    {
+      "epoch": 0.1427803284113,
+      "grad_norm": 1357.8616943359375,
+      "learning_rate": 9.952607874220048e-06,
+      "loss": 201.047,
+      "step": 17260
+    },
+    {
+      "epoch": 0.14286305166066923,
+      "grad_norm": 1452.91650390625,
+      "learning_rate": 9.952413953074312e-06,
+      "loss": 199.8793,
+      "step": 17270
+    },
+    {
+      "epoch": 0.14294577491003846,
+      "grad_norm": 965.8828125,
+      "learning_rate": 9.952219637887725e-06,
+      "loss": 129.7407,
+      "step": 17280
+    },
+    {
+      "epoch": 0.1430284981594077,
+      "grad_norm": 1721.4344482421875,
+      "learning_rate": 9.952024928675752e-06,
+      "loss": 177.8543,
+      "step": 17290
+    },
+    {
+      "epoch": 0.14311122140877694,
+      "grad_norm": 3541.317626953125,
+      "learning_rate": 9.951829825453881e-06,
+      "loss": 167.7698,
+      "step": 17300
+    },
+    {
+      "epoch": 0.14319394465814617,
+      "grad_norm": 2036.2423095703125,
+      "learning_rate": 9.951634328237635e-06,
+      "loss": 141.8449,
+      "step": 17310
+    },
+    {
+      "epoch": 0.14327666790751542,
+      "grad_norm": 880.5416870117188,
+      "learning_rate": 9.951438437042572e-06,
+      "loss": 198.8033,
+      "step": 17320
+    },
+    {
+      "epoch": 0.14335939115688465,
+      "grad_norm": 807.236572265625,
+      "learning_rate": 9.951242151884275e-06,
+      "loss": 112.0078,
+      "step": 17330
+    },
+    {
+      "epoch": 0.14344211440625387,
+      "grad_norm": 1530.7301025390625,
+      "learning_rate": 9.951045472778365e-06,
+      "loss": 133.3953,
+      "step": 17340
+    },
+    {
+      "epoch": 0.1435248376556231,
+      "grad_norm": 1775.3485107421875,
+      "learning_rate": 9.950848399740488e-06,
+      "loss": 132.5112,
+      "step": 17350
+    },
+    {
+      "epoch": 0.14360756090499235,
+      "grad_norm": 1216.1314697265625,
+      "learning_rate": 9.950650932786325e-06,
+      "loss": 150.7454,
+      "step": 17360
+    },
+    {
+      "epoch": 0.14369028415436158,
+      "grad_norm": 756.1212158203125,
+      "learning_rate": 9.95045307193159e-06,
+      "loss": 114.4585,
+      "step": 17370
+    },
+    {
+      "epoch": 0.1437730074037308,
+      "grad_norm": 987.248779296875,
+      "learning_rate": 9.95025481719202e-06,
+      "loss": 140.8504,
+      "step": 17380
+    },
+    {
+      "epoch": 0.14385573065310006,
+      "grad_norm": 1126.249267578125,
+      "learning_rate": 9.950056168583395e-06,
+      "loss": 225.9696,
+      "step": 17390
+    },
+    {
+      "epoch": 0.1439384539024693,
+      "grad_norm": 706.3463745117188,
+      "learning_rate": 9.949857126121519e-06,
+      "loss": 113.696,
+      "step": 17400
+    },
+    {
+      "epoch": 0.14402117715183851,
+      "grad_norm": 892.3402099609375,
+      "learning_rate": 9.949657689822226e-06,
+      "loss": 162.9231,
+      "step": 17410
+    },
+    {
+      "epoch": 0.14410390040120777,
+      "grad_norm": 856.6466674804688,
+      "learning_rate": 9.949457859701388e-06,
+      "loss": 99.4635,
+      "step": 17420
+    },
+    {
+      "epoch": 0.144186623650577,
+      "grad_norm": 775.4996948242188,
+      "learning_rate": 9.949257635774903e-06,
+      "loss": 152.7363,
+      "step": 17430
+    },
+    {
+      "epoch": 0.14426934689994622,
+      "grad_norm": 842.1768798828125,
+      "learning_rate": 9.9490570180587e-06,
+      "loss": 85.8346,
+      "step": 17440
+    },
+    {
+      "epoch": 0.14435207014931548,
+      "grad_norm": 1798.95849609375,
+      "learning_rate": 9.948856006568746e-06,
+      "loss": 197.5757,
+      "step": 17450
+    },
+    {
+      "epoch": 0.1444347933986847,
+      "grad_norm": 1381.5155029296875,
+      "learning_rate": 9.94865460132103e-06,
+      "loss": 150.2531,
+      "step": 17460
+    },
+    {
+      "epoch": 0.14451751664805393,
+      "grad_norm": 997.7630004882812,
+      "learning_rate": 9.948452802331578e-06,
+      "loss": 133.1603,
+      "step": 17470
+    },
+    {
+      "epoch": 0.14460023989742318,
+      "grad_norm": 1275.1690673828125,
+      "learning_rate": 9.948250609616449e-06,
+      "loss": 168.5733,
+      "step": 17480
+    },
+    {
+      "epoch": 0.1446829631467924,
+      "grad_norm": 1112.8721923828125,
+      "learning_rate": 9.948048023191728e-06,
+      "loss": 182.301,
+      "step": 17490
+    },
+    {
+      "epoch": 0.14476568639616164,
+      "grad_norm": 950.4414672851562,
+      "learning_rate": 9.947845043073533e-06,
+      "loss": 149.5477,
+      "step": 17500
+    },
+    {
+      "epoch": 0.1448484096455309,
+      "grad_norm": 1122.95751953125,
+      "learning_rate": 9.947641669278016e-06,
+      "loss": 123.1119,
+      "step": 17510
+    },
+    {
+      "epoch": 0.14493113289490012,
+      "grad_norm": 1148.9334716796875,
+      "learning_rate": 9.947437901821358e-06,
+      "loss": 128.3063,
+      "step": 17520
+    },
+    {
+      "epoch": 0.14501385614426934,
+      "grad_norm": 1392.179443359375,
+      "learning_rate": 9.947233740719772e-06,
+      "loss": 139.3278,
+      "step": 17530
+    },
+    {
+      "epoch": 0.14509657939363857,
+      "grad_norm": 604.5231323242188,
+      "learning_rate": 9.947029185989501e-06,
+      "loss": 163.2896,
+      "step": 17540
+    },
+    {
+      "epoch": 0.14517930264300782,
+      "grad_norm": 1102.948486328125,
+      "learning_rate": 9.946824237646823e-06,
+      "loss": 153.8839,
+      "step": 17550
+    },
+    {
+      "epoch": 0.14526202589237705,
+      "grad_norm": 2167.79638671875,
+      "learning_rate": 9.946618895708043e-06,
+      "loss": 172.0367,
+      "step": 17560
+    },
+    {
+      "epoch": 0.14534474914174628,
+      "grad_norm": 1476.7362060546875,
+      "learning_rate": 9.946413160189498e-06,
+      "loss": 138.3295,
+      "step": 17570
+    },
+    {
+      "epoch": 0.14542747239111553,
+      "grad_norm": 882.1810913085938,
+      "learning_rate": 9.946207031107562e-06,
+      "loss": 186.2194,
+      "step": 17580
+    },
+    {
+      "epoch": 0.14551019564048476,
+      "grad_norm": 2111.673095703125,
+      "learning_rate": 9.94600050847863e-06,
+      "loss": 170.1872,
+      "step": 17590
+    },
+    {
+      "epoch": 0.14559291888985398,
+      "grad_norm": 1099.032958984375,
+      "learning_rate": 9.945793592319137e-06,
+      "loss": 128.6498,
+      "step": 17600
+    },
+    {
+      "epoch": 0.14567564213922324,
+      "grad_norm": 1059.4005126953125,
+      "learning_rate": 9.945586282645545e-06,
+      "loss": 134.5357,
+      "step": 17610
+    },
+    {
+      "epoch": 0.14575836538859246,
+      "grad_norm": 1566.564208984375,
+      "learning_rate": 9.945378579474351e-06,
+      "loss": 164.359,
+      "step": 17620
+    },
+    {
+      "epoch": 0.1458410886379617,
+      "grad_norm": 782.61279296875,
+      "learning_rate": 9.945170482822079e-06,
+      "loss": 106.899,
+      "step": 17630
+    },
+    {
+      "epoch": 0.14592381188733095,
+      "grad_norm": 1026.7816162109375,
+      "learning_rate": 9.944961992705288e-06,
+      "loss": 142.0462,
+      "step": 17640
+    },
+    {
+      "epoch": 0.14600653513670017,
+      "grad_norm": 817.039306640625,
+      "learning_rate": 9.944753109140564e-06,
+      "loss": 166.4367,
+      "step": 17650
+    },
+    {
+      "epoch": 0.1460892583860694,
+      "grad_norm": 856.3842163085938,
+      "learning_rate": 9.94454383214453e-06,
+      "loss": 131.3289,
+      "step": 17660
+    },
+    {
+      "epoch": 0.14617198163543865,
+      "grad_norm": 1656.999755859375,
+      "learning_rate": 9.944334161733835e-06,
+      "loss": 129.1978,
+      "step": 17670
+    },
+    {
+      "epoch": 0.14625470488480788,
+      "grad_norm": 1338.8382568359375,
+      "learning_rate": 9.944124097925161e-06,
+      "loss": 184.4288,
+      "step": 17680
+    },
+    {
+      "epoch": 0.1463374281341771,
+      "grad_norm": 1011.0686645507812,
+      "learning_rate": 9.943913640735224e-06,
+      "loss": 127.4451,
+      "step": 17690
+    },
+    {
+      "epoch": 0.14642015138354633,
+      "grad_norm": 923.1884765625,
+      "learning_rate": 9.94370279018077e-06,
+      "loss": 120.5529,
+      "step": 17700
+    },
+    {
+      "epoch": 0.1465028746329156,
+      "grad_norm": 1001.093505859375,
+      "learning_rate": 9.94349154627857e-06,
+      "loss": 129.3988,
+      "step": 17710
+    },
+    {
+      "epoch": 0.1465855978822848,
+      "grad_norm": 1354.6356201171875,
+      "learning_rate": 9.943279909045438e-06,
+      "loss": 122.9835,
+      "step": 17720
+    },
+    {
+      "epoch": 0.14666832113165404,
+      "grad_norm": 1260.7392578125,
+      "learning_rate": 9.94306787849821e-06,
+      "loss": 101.6319,
+      "step": 17730
+    },
+    {
+      "epoch": 0.1467510443810233,
+      "grad_norm": 1424.63330078125,
+      "learning_rate": 9.942855454653755e-06,
+      "loss": 179.1118,
+      "step": 17740
+    },
+    {
+      "epoch": 0.14683376763039252,
+      "grad_norm": 1053.8809814453125,
+      "learning_rate": 9.942642637528977e-06,
+      "loss": 167.5939,
+      "step": 17750
+    },
+    {
+      "epoch": 0.14691649087976175,
+      "grad_norm": 936.3515014648438,
+      "learning_rate": 9.942429427140807e-06,
+      "loss": 154.7948,
+      "step": 17760
+    },
+    {
+      "epoch": 0.146999214129131,
+      "grad_norm": 876.3916015625,
+      "learning_rate": 9.942215823506211e-06,
+      "loss": 114.5722,
+      "step": 17770
+    },
+    {
+      "epoch": 0.14708193737850023,
+      "grad_norm": 1172.0423583984375,
+      "learning_rate": 9.942001826642184e-06,
+      "loss": 142.9646,
+      "step": 17780
+    },
+    {
+      "epoch": 0.14716466062786945,
+      "grad_norm": 1635.97802734375,
+      "learning_rate": 9.941787436565751e-06,
+      "loss": 150.69,
+      "step": 17790
+    },
+    {
+      "epoch": 0.1472473838772387,
+      "grad_norm": 728.3792724609375,
+      "learning_rate": 9.941572653293974e-06,
+      "loss": 97.5937,
+      "step": 17800
+    },
+    {
+      "epoch": 0.14733010712660793,
+      "grad_norm": 935.0343627929688,
+      "learning_rate": 9.941357476843938e-06,
+      "loss": 135.0443,
+      "step": 17810
+    },
+    {
+      "epoch": 0.14741283037597716,
+      "grad_norm": 583.3887329101562,
+      "learning_rate": 9.941141907232766e-06,
+      "loss": 134.4311,
+      "step": 17820
+    },
+    {
+      "epoch": 0.14749555362534642,
+      "grad_norm": 1191.19677734375,
+      "learning_rate": 9.940925944477608e-06,
+      "loss": 129.727,
+      "step": 17830
+    },
+    {
+      "epoch": 0.14757827687471564,
+      "grad_norm": 1111.1417236328125,
+      "learning_rate": 9.940709588595649e-06,
+      "loss": 171.4274,
+      "step": 17840
+    },
+    {
+      "epoch": 0.14766100012408487,
+      "grad_norm": 2006.4134521484375,
+      "learning_rate": 9.940492839604103e-06,
+      "loss": 152.9817,
+      "step": 17850
+    },
+    {
+      "epoch": 0.14774372337345412,
+      "grad_norm": 1163.596923828125,
+      "learning_rate": 9.940275697520216e-06,
+      "loss": 169.9584,
+      "step": 17860
+    },
+    {
+      "epoch": 0.14782644662282335,
+      "grad_norm": 1189.015869140625,
+      "learning_rate": 9.940058162361264e-06,
+      "loss": 152.1794,
+      "step": 17870
+    },
+    {
+      "epoch": 0.14790916987219258,
+      "grad_norm": 998.8855590820312,
+      "learning_rate": 9.939840234144556e-06,
+      "loss": 129.5204,
+      "step": 17880
+    },
+    {
+      "epoch": 0.1479918931215618,
+      "grad_norm": 442.9149475097656,
+      "learning_rate": 9.939621912887431e-06,
+      "loss": 106.7805,
+      "step": 17890
+    },
+    {
+      "epoch": 0.14807461637093106,
+      "grad_norm": 830.00927734375,
+      "learning_rate": 9.93940319860726e-06,
+      "loss": 139.6457,
+      "step": 17900
+    },
+    {
+      "epoch": 0.14815733962030028,
+      "grad_norm": 1069.5220947265625,
+      "learning_rate": 9.939184091321445e-06,
+      "loss": 129.1493,
+      "step": 17910
+    },
+    {
+      "epoch": 0.1482400628696695,
+      "grad_norm": 1180.868896484375,
+      "learning_rate": 9.938964591047421e-06,
+      "loss": 108.2578,
+      "step": 17920
+    },
+    {
+      "epoch": 0.14832278611903876,
+      "grad_norm": 1095.6793212890625,
+      "learning_rate": 9.938744697802651e-06,
+      "loss": 145.4649,
+      "step": 17930
+    },
+    {
+      "epoch": 0.148405509368408,
+      "grad_norm": 1292.62744140625,
+      "learning_rate": 9.938524411604631e-06,
+      "loss": 145.161,
+      "step": 17940
+    },
+    {
+      "epoch": 0.14848823261777722,
+      "grad_norm": 1319.2213134765625,
+      "learning_rate": 9.938303732470888e-06,
+      "loss": 129.5037,
+      "step": 17950
+    },
+    {
+      "epoch": 0.14857095586714647,
+      "grad_norm": 697.8318481445312,
+      "learning_rate": 9.938082660418981e-06,
+      "loss": 103.5571,
+      "step": 17960
+    },
+    {
+      "epoch": 0.1486536791165157,
+      "grad_norm": 784.6300659179688,
+      "learning_rate": 9.937861195466498e-06,
+      "loss": 133.7046,
+      "step": 17970
+    },
+    {
+      "epoch": 0.14873640236588492,
+      "grad_norm": 966.1806030273438,
+      "learning_rate": 9.937639337631064e-06,
+      "loss": 170.2544,
+      "step": 17980
+    },
+    {
+      "epoch": 0.14881912561525418,
+      "grad_norm": 862.203857421875,
+      "learning_rate": 9.937417086930328e-06,
+      "loss": 129.5846,
+      "step": 17990
+    },
+    {
+      "epoch": 0.1489018488646234,
+      "grad_norm": 3391.59716796875,
+      "learning_rate": 9.937194443381972e-06,
+      "loss": 195.0929,
+      "step": 18000
+    },
+    {
+      "epoch": 0.14898457211399263,
+      "grad_norm": 737.4010009765625,
+      "learning_rate": 9.936971407003714e-06,
+      "loss": 110.9804,
+      "step": 18010
+    },
+    {
+      "epoch": 0.14906729536336188,
+      "grad_norm": 1527.3822021484375,
+      "learning_rate": 9.936747977813299e-06,
+      "loss": 124.5241,
+      "step": 18020
+    },
+    {
+      "epoch": 0.1491500186127311,
+      "grad_norm": 1383.10986328125,
+      "learning_rate": 9.936524155828503e-06,
+      "loss": 138.0007,
+      "step": 18030
+    },
+    {
+      "epoch": 0.14923274186210034,
+      "grad_norm": 850.4631958007812,
+      "learning_rate": 9.936299941067137e-06,
+      "loss": 131.9197,
+      "step": 18040
+    },
+    {
+      "epoch": 0.1493154651114696,
+      "grad_norm": 736.7586059570312,
+      "learning_rate": 9.93607533354704e-06,
+      "loss": 116.8003,
+      "step": 18050
+    },
+    {
+      "epoch": 0.14939818836083882,
+      "grad_norm": 3558.953857421875,
+      "learning_rate": 9.935850333286081e-06,
+      "loss": 236.4352,
+      "step": 18060
+    },
+    {
+      "epoch": 0.14948091161020804,
+      "grad_norm": 2170.6923828125,
+      "learning_rate": 9.935624940302165e-06,
+      "loss": 162.2385,
+      "step": 18070
+    },
+    {
+      "epoch": 0.14956363485957727,
+      "grad_norm": 890.1776123046875,
+      "learning_rate": 9.93539915461322e-06,
+      "loss": 152.4946,
+      "step": 18080
+    },
+    {
+      "epoch": 0.14964635810894653,
+      "grad_norm": 1055.08447265625,
+      "learning_rate": 9.935172976237218e-06,
+      "loss": 179.4581,
+      "step": 18090
+    },
+    {
+      "epoch": 0.14972908135831575,
+      "grad_norm": 1069.735595703125,
+      "learning_rate": 9.934946405192152e-06,
+      "loss": 109.1896,
+      "step": 18100
+    },
+    {
+      "epoch": 0.14981180460768498,
+      "grad_norm": 936.50048828125,
+      "learning_rate": 9.934719441496048e-06,
+      "loss": 170.2172,
+      "step": 18110
+    },
+    {
+      "epoch": 0.14989452785705423,
+      "grad_norm": 1424.594970703125,
+      "learning_rate": 9.934492085166965e-06,
+      "loss": 120.5943,
+      "step": 18120
+    },
+    {
+      "epoch": 0.14997725110642346,
+      "grad_norm": 1259.637939453125,
+      "learning_rate": 9.934264336222992e-06,
+      "loss": 141.8418,
+      "step": 18130
+    },
+    {
+      "epoch": 0.15005997435579269,
+      "grad_norm": 1107.58447265625,
+      "learning_rate": 9.934036194682253e-06,
+      "loss": 132.8073,
+      "step": 18140
+    },
+    {
+      "epoch": 0.15014269760516194,
+      "grad_norm": 1031.5169677734375,
+      "learning_rate": 9.933807660562898e-06,
+      "loss": 122.9906,
+      "step": 18150
+    },
+    {
+      "epoch": 0.15022542085453117,
+      "grad_norm": 683.3692626953125,
+      "learning_rate": 9.933578733883109e-06,
+      "loss": 175.373,
+      "step": 18160
+    },
+    {
+      "epoch": 0.1503081441039004,
+      "grad_norm": 841.3174438476562,
+      "learning_rate": 9.933349414661103e-06,
+      "loss": 143.8702,
+      "step": 18170
+    },
+    {
+      "epoch": 0.15039086735326965,
+      "grad_norm": 1086.541015625,
+      "learning_rate": 9.933119702915125e-06,
+      "loss": 149.0898,
+      "step": 18180
+    },
+    {
+      "epoch": 0.15047359060263887,
+      "grad_norm": 1380.8690185546875,
+      "learning_rate": 9.932889598663452e-06,
+      "loss": 142.0298,
+      "step": 18190
+    },
+    {
+      "epoch": 0.1505563138520081,
+      "grad_norm": 800.4336547851562,
+      "learning_rate": 9.932659101924393e-06,
+      "loss": 169.3204,
+      "step": 18200
+    },
+    {
+      "epoch": 0.15063903710137735,
+      "grad_norm": 883.3157348632812,
+      "learning_rate": 9.932428212716287e-06,
+      "loss": 183.8594,
+      "step": 18210
+    },
+    {
+      "epoch": 0.15072176035074658,
+      "grad_norm": 922.6904907226562,
+      "learning_rate": 9.932196931057505e-06,
+      "loss": 157.8369,
+      "step": 18220
+    },
+    {
+      "epoch": 0.1508044836001158,
+      "grad_norm": 1918.9375,
+      "learning_rate": 9.931965256966449e-06,
+      "loss": 143.9471,
+      "step": 18230
+    },
+    {
+      "epoch": 0.15088720684948506,
+      "grad_norm": 1153.788818359375,
+      "learning_rate": 9.931733190461552e-06,
+      "loss": 167.4599,
+      "step": 18240
+    },
+    {
+      "epoch": 0.1509699300988543,
+      "grad_norm": 1510.7779541015625,
+      "learning_rate": 9.931500731561279e-06,
+      "loss": 123.1982,
+      "step": 18250
+    },
+    {
+      "epoch": 0.15105265334822351,
+      "grad_norm": 939.08447265625,
+      "learning_rate": 9.931267880284124e-06,
+      "loss": 128.6788,
+      "step": 18260
+    },
+    {
+      "epoch": 0.15113537659759274,
+      "grad_norm": 521.10693359375,
+      "learning_rate": 9.931034636648616e-06,
+      "loss": 110.3548,
+      "step": 18270
+    },
+    {
+      "epoch": 0.151218099846962,
+      "grad_norm": 1793.3514404296875,
+      "learning_rate": 9.930801000673314e-06,
+      "loss": 226.4601,
+      "step": 18280
+    },
+    {
+      "epoch": 0.15130082309633122,
+      "grad_norm": 1056.244384765625,
+      "learning_rate": 9.930566972376803e-06,
+      "loss": 137.7991,
+      "step": 18290
+    },
+    {
+      "epoch": 0.15138354634570045,
+      "grad_norm": 1053.8623046875,
+      "learning_rate": 9.930332551777709e-06,
+      "loss": 126.393,
+      "step": 18300
+    },
+    {
+      "epoch": 0.1514662695950697,
+      "grad_norm": 466.3129577636719,
+      "learning_rate": 9.930097738894679e-06,
+      "loss": 142.9212,
+      "step": 18310
+    },
+    {
+      "epoch": 0.15154899284443893,
+      "grad_norm": 1002.7549438476562,
+      "learning_rate": 9.929862533746398e-06,
+      "loss": 142.7721,
+      "step": 18320
+    },
+    {
+      "epoch": 0.15163171609380816,
+      "grad_norm": 758.2431030273438,
+      "learning_rate": 9.92962693635158e-06,
+      "loss": 138.2474,
+      "step": 18330
+    },
+    {
+      "epoch": 0.1517144393431774,
+      "grad_norm": 640.2601318359375,
+      "learning_rate": 9.929390946728972e-06,
+      "loss": 127.6863,
+      "step": 18340
+    },
+    {
+      "epoch": 0.15179716259254664,
+      "grad_norm": 974.4703979492188,
+      "learning_rate": 9.929154564897347e-06,
+      "loss": 127.8559,
+      "step": 18350
+    },
+    {
+      "epoch": 0.15187988584191586,
+      "grad_norm": 989.0883178710938,
+      "learning_rate": 9.928917790875519e-06,
+      "loss": 146.4885,
+      "step": 18360
+    },
+    {
+      "epoch": 0.15196260909128512,
+      "grad_norm": 926.8894653320312,
+      "learning_rate": 9.92868062468232e-06,
+      "loss": 140.2718,
+      "step": 18370
+    },
+    {
+      "epoch": 0.15204533234065434,
+      "grad_norm": 1093.3875732421875,
+      "learning_rate": 9.928443066336624e-06,
+      "loss": 156.4275,
+      "step": 18380
+    },
+    {
+      "epoch": 0.15212805559002357,
+      "grad_norm": 1011.3756713867188,
+      "learning_rate": 9.92820511585733e-06,
+      "loss": 110.9518,
+      "step": 18390
+    },
+    {
+      "epoch": 0.15221077883939282,
+      "grad_norm": 963.0042114257812,
+      "learning_rate": 9.927966773263375e-06,
+      "loss": 126.9806,
+      "step": 18400
+    },
+    {
+      "epoch": 0.15229350208876205,
+      "grad_norm": 1338.420654296875,
+      "learning_rate": 9.92772803857372e-06,
+      "loss": 151.1979,
+      "step": 18410
+    },
+    {
+      "epoch": 0.15237622533813128,
+      "grad_norm": 2231.09326171875,
+      "learning_rate": 9.927488911807359e-06,
+      "loss": 126.2641,
+      "step": 18420
+    },
+    {
+      "epoch": 0.1524589485875005,
+      "grad_norm": 740.568603515625,
+      "learning_rate": 9.927249392983319e-06,
+      "loss": 161.8315,
+      "step": 18430
+    },
+    {
+      "epoch": 0.15254167183686976,
+      "grad_norm": 1194.8526611328125,
+      "learning_rate": 9.927009482120658e-06,
+      "loss": 147.5258,
+      "step": 18440
+    },
+    {
+      "epoch": 0.15262439508623898,
+      "grad_norm": 575.281005859375,
+      "learning_rate": 9.926769179238467e-06,
+      "loss": 123.2295,
+      "step": 18450
+    },
+    {
+      "epoch": 0.1527071183356082,
+      "grad_norm": 1439.9266357421875,
+      "learning_rate": 9.926528484355859e-06,
+      "loss": 131.7167,
+      "step": 18460
+    },
+    {
+      "epoch": 0.15278984158497746,
+      "grad_norm": 1190.1434326171875,
+      "learning_rate": 9.926287397491992e-06,
+      "loss": 147.6172,
+      "step": 18470
+    },
+    {
+      "epoch": 0.1528725648343467,
+      "grad_norm": 1017.2939453125,
+      "learning_rate": 9.926045918666045e-06,
+      "loss": 144.2414,
+      "step": 18480
+    },
+    {
+      "epoch": 0.15295528808371592,
+      "grad_norm": 1303.34814453125,
+      "learning_rate": 9.925804047897231e-06,
+      "loss": 202.542,
+      "step": 18490
+    },
+    {
+      "epoch": 0.15303801133308517,
+      "grad_norm": 1302.6015625,
+      "learning_rate": 9.925561785204797e-06,
+      "loss": 150.7994,
+      "step": 18500
+    },
+    {
+      "epoch": 0.1531207345824544,
+      "grad_norm": 1482.454345703125,
+      "learning_rate": 9.925319130608015e-06,
+      "loss": 160.8186,
+      "step": 18510
+    },
+    {
+      "epoch": 0.15320345783182363,
+      "grad_norm": 1089.9215087890625,
+      "learning_rate": 9.925076084126194e-06,
+      "loss": 140.3311,
+      "step": 18520
+    },
+    {
+      "epoch": 0.15328618108119288,
+      "grad_norm": 2586.4873046875,
+      "learning_rate": 9.924832645778674e-06,
+      "loss": 105.1053,
+      "step": 18530
+    },
+    {
+      "epoch": 0.1533689043305621,
+      "grad_norm": 986.44775390625,
+      "learning_rate": 9.924588815584822e-06,
+      "loss": 146.6998,
+      "step": 18540
+    },
+    {
+      "epoch": 0.15345162757993133,
+      "grad_norm": 1076.886474609375,
+      "learning_rate": 9.924344593564038e-06,
+      "loss": 167.1004,
+      "step": 18550
+    },
+    {
+      "epoch": 0.1535343508293006,
+      "grad_norm": 576.464599609375,
+      "learning_rate": 9.924099979735754e-06,
+      "loss": 109.8678,
+      "step": 18560
+    },
+    {
+      "epoch": 0.1536170740786698,
+      "grad_norm": 722.518310546875,
+      "learning_rate": 9.923854974119434e-06,
+      "loss": 125.0473,
+      "step": 18570
+    },
+    {
+      "epoch": 0.15369979732803904,
+      "grad_norm": 1820.7373046875,
+      "learning_rate": 9.92360957673457e-06,
+      "loss": 141.7229,
+      "step": 18580
+    },
+    {
+      "epoch": 0.1537825205774083,
+      "grad_norm": 1114.6781005859375,
+      "learning_rate": 9.923363787600688e-06,
+      "loss": 141.2934,
+      "step": 18590
+    },
+    {
+      "epoch": 0.15386524382677752,
+      "grad_norm": 1381.0604248046875,
+      "learning_rate": 9.923117606737347e-06,
+      "loss": 116.5776,
+      "step": 18600
+    },
+    {
+      "epoch": 0.15394796707614675,
+      "grad_norm": 1086.9346923828125,
+      "learning_rate": 9.92287103416413e-06,
+      "loss": 191.1018,
+      "step": 18610
+    },
+    {
+      "epoch": 0.15403069032551597,
+      "grad_norm": 1572.17529296875,
+      "learning_rate": 9.922624069900658e-06,
+      "loss": 155.7499,
+      "step": 18620
+    },
+    {
+      "epoch": 0.15411341357488523,
+      "grad_norm": 1132.931884765625,
+      "learning_rate": 9.922376713966581e-06,
+      "loss": 152.9908,
+      "step": 18630
+    },
+    {
+      "epoch": 0.15419613682425445,
+      "grad_norm": 585.323486328125,
+      "learning_rate": 9.92212896638158e-06,
+      "loss": 152.2068,
+      "step": 18640
+    },
+    {
+      "epoch": 0.15427886007362368,
+      "grad_norm": 595.2325439453125,
+      "learning_rate": 9.921880827165367e-06,
+      "loss": 118.7037,
+      "step": 18650
+    },
+    {
+      "epoch": 0.15436158332299293,
+      "grad_norm": 4941.7626953125,
+      "learning_rate": 9.921632296337683e-06,
+      "loss": 153.8302,
+      "step": 18660
+    },
+    {
+      "epoch": 0.15444430657236216,
+      "grad_norm": 1101.675048828125,
+      "learning_rate": 9.921383373918305e-06,
+      "loss": 180.7743,
+      "step": 18670
+    },
+    {
+      "epoch": 0.1545270298217314,
+      "grad_norm": 1755.6380615234375,
+      "learning_rate": 9.92113405992704e-06,
+      "loss": 180.545,
+      "step": 18680
+    },
+    {
+      "epoch": 0.15460975307110064,
+      "grad_norm": 925.7059326171875,
+      "learning_rate": 9.92088435438372e-06,
+      "loss": 134.731,
+      "step": 18690
+    },
+    {
+      "epoch": 0.15469247632046987,
+      "grad_norm": 1003.4811401367188,
+      "learning_rate": 9.920634257308217e-06,
+      "loss": 123.1074,
+      "step": 18700
+    },
+    {
+      "epoch": 0.1547751995698391,
+      "grad_norm": 853.6227416992188,
+      "learning_rate": 9.920383768720429e-06,
+      "loss": 150.079,
+      "step": 18710
+    },
+    {
+      "epoch": 0.15485792281920835,
+      "grad_norm": 1113.7686767578125,
+      "learning_rate": 9.920132888640286e-06,
+      "loss": 155.0464,
+      "step": 18720
+    },
+    {
+      "epoch": 0.15494064606857758,
+      "grad_norm": 1343.3956298828125,
+      "learning_rate": 9.91988161708775e-06,
+      "loss": 158.199,
+      "step": 18730
+    },
+    {
+      "epoch": 0.1550233693179468,
+      "grad_norm": 704.8764038085938,
+      "learning_rate": 9.919629954082813e-06,
+      "loss": 153.2144,
+      "step": 18740
+    },
+    {
+      "epoch": 0.15510609256731606,
+      "grad_norm": 550.6301879882812,
+      "learning_rate": 9.919377899645497e-06,
+      "loss": 141.5231,
+      "step": 18750
+    },
+    {
+      "epoch": 0.15518881581668528,
+      "grad_norm": 658.1661376953125,
+      "learning_rate": 9.91912545379586e-06,
+      "loss": 138.1113,
+      "step": 18760
+    },
+    {
+      "epoch": 0.1552715390660545,
+      "grad_norm": 1573.263916015625,
+      "learning_rate": 9.918872616553986e-06,
+      "loss": 129.6509,
+      "step": 18770
+    },
+    {
+      "epoch": 0.15535426231542376,
+      "grad_norm": 1778.827392578125,
+      "learning_rate": 9.918619387939991e-06,
+      "loss": 155.8357,
+      "step": 18780
+    },
+    {
+      "epoch": 0.155436985564793,
+      "grad_norm": 3056.092041015625,
+      "learning_rate": 9.918365767974025e-06,
+      "loss": 187.3279,
+      "step": 18790
+    },
+    {
+      "epoch": 0.15551970881416222,
+      "grad_norm": 905.0897827148438,
+      "learning_rate": 9.91811175667627e-06,
+      "loss": 198.9183,
+      "step": 18800
+    },
+    {
+      "epoch": 0.15560243206353144,
+      "grad_norm": 854.0653076171875,
+      "learning_rate": 9.91785735406693e-06,
+      "loss": 111.8965,
+      "step": 18810
+    },
+    {
+      "epoch": 0.1556851553129007,
+      "grad_norm": 1693.5703125,
+      "learning_rate": 9.917602560166253e-06,
+      "loss": 138.9856,
+      "step": 18820
+    },
+    {
+      "epoch": 0.15576787856226992,
+      "grad_norm": 1688.492431640625,
+      "learning_rate": 9.917347374994507e-06,
+      "loss": 118.313,
+      "step": 18830
+    },
+    {
+      "epoch": 0.15585060181163915,
+      "grad_norm": 1094.8780517578125,
+      "learning_rate": 9.917091798571998e-06,
+      "loss": 122.0171,
+      "step": 18840
+    },
+    {
+      "epoch": 0.1559333250610084,
+      "grad_norm": 1383.755859375,
+      "learning_rate": 9.916835830919062e-06,
+      "loss": 149.7231,
+      "step": 18850
+    },
+    {
+      "epoch": 0.15601604831037763,
+      "grad_norm": 4172.68603515625,
+      "learning_rate": 9.916579472056064e-06,
+      "loss": 164.1563,
+      "step": 18860
+    },
+    {
+      "epoch": 0.15609877155974686,
+      "grad_norm": 1112.282958984375,
+      "learning_rate": 9.916322722003402e-06,
+      "loss": 140.6031,
+      "step": 18870
+    },
+    {
+      "epoch": 0.1561814948091161,
+      "grad_norm": 1334.9544677734375,
+      "learning_rate": 9.916065580781504e-06,
+      "loss": 125.8786,
+      "step": 18880
+    },
+    {
+      "epoch": 0.15626421805848534,
+      "grad_norm": 759.095703125,
+      "learning_rate": 9.91580804841083e-06,
+      "loss": 123.02,
+      "step": 18890
+    },
+    {
+      "epoch": 0.15634694130785456,
+      "grad_norm": 923.3683471679688,
+      "learning_rate": 9.915550124911866e-06,
+      "loss": 111.5178,
+      "step": 18900
+    },
+    {
+      "epoch": 0.15642966455722382,
+      "grad_norm": 1883.232177734375,
+      "learning_rate": 9.915291810305141e-06,
+      "loss": 153.0945,
+      "step": 18910
+    },
+    {
+      "epoch": 0.15651238780659305,
+      "grad_norm": 1571.4327392578125,
+      "learning_rate": 9.915033104611204e-06,
+      "loss": 152.6783,
+      "step": 18920
+    },
+    {
+      "epoch": 0.15659511105596227,
+      "grad_norm": 842.9326171875,
+      "learning_rate": 9.914774007850641e-06,
+      "loss": 154.6972,
+      "step": 18930
+    },
+    {
+      "epoch": 0.15667783430533153,
+      "grad_norm": 1248.7547607421875,
+      "learning_rate": 9.914514520044065e-06,
+      "loss": 169.4783,
+      "step": 18940
+    },
+    {
+      "epoch": 0.15676055755470075,
+      "grad_norm": 976.1325073242188,
+      "learning_rate": 9.914254641212124e-06,
+      "loss": 114.911,
+      "step": 18950
+    },
+    {
+      "epoch": 0.15684328080406998,
+      "grad_norm": 893.714111328125,
+      "learning_rate": 9.913994371375494e-06,
+      "loss": 81.8798,
+      "step": 18960
+    },
+    {
+      "epoch": 0.1569260040534392,
+      "grad_norm": 1223.4085693359375,
+      "learning_rate": 9.913733710554886e-06,
+      "loss": 138.9431,
+      "step": 18970
+    },
+    {
+      "epoch": 0.15700872730280846,
+      "grad_norm": 2076.80712890625,
+      "learning_rate": 9.913472658771034e-06,
+      "loss": 113.3516,
+      "step": 18980
+    },
+    {
+      "epoch": 0.15709145055217769,
+      "grad_norm": 1146.1357421875,
+      "learning_rate": 9.913211216044715e-06,
+      "loss": 162.9254,
+      "step": 18990
+    },
+    {
+      "epoch": 0.1571741738015469,
+      "grad_norm": 1230.30224609375,
+      "learning_rate": 9.912949382396728e-06,
+      "loss": 197.952,
+      "step": 19000
+    },
+    {
+      "epoch": 0.15725689705091617,
+      "grad_norm": 1353.6494140625,
+      "learning_rate": 9.912687157847905e-06,
+      "loss": 137.2512,
+      "step": 19010
+    },
+    {
+      "epoch": 0.1573396203002854,
+      "grad_norm": 1508.077392578125,
+      "learning_rate": 9.91242454241911e-06,
+      "loss": 133.2853,
+      "step": 19020
+    },
+    {
+      "epoch": 0.15742234354965462,
+      "grad_norm": 935.4700927734375,
+      "learning_rate": 9.912161536131242e-06,
+      "loss": 126.4163,
+      "step": 19030
+    },
+    {
+      "epoch": 0.15750506679902387,
+      "grad_norm": 920.0881958007812,
+      "learning_rate": 9.911898139005222e-06,
+      "loss": 106.0859,
+      "step": 19040
+    },
+    {
+      "epoch": 0.1575877900483931,
+      "grad_norm": 1098.602783203125,
+      "learning_rate": 9.91163435106201e-06,
+      "loss": 113.0203,
+      "step": 19050
+    },
+    {
+      "epoch": 0.15767051329776233,
+      "grad_norm": 529.7808227539062,
+      "learning_rate": 9.911370172322595e-06,
+      "loss": 100.4977,
+      "step": 19060
+    },
+    {
+      "epoch": 0.15775323654713158,
+      "grad_norm": 1122.334228515625,
+      "learning_rate": 9.911105602807996e-06,
+      "loss": 147.1685,
+      "step": 19070
+    },
+    {
+      "epoch": 0.1578359597965008,
+      "grad_norm": 1302.458740234375,
+      "learning_rate": 9.910840642539261e-06,
+      "loss": 138.237,
+      "step": 19080
+    },
+    {
+      "epoch": 0.15791868304587003,
+      "grad_norm": 1324.593505859375,
+      "learning_rate": 9.910575291537476e-06,
+      "loss": 182.2281,
+      "step": 19090
+    },
+    {
+      "epoch": 0.1580014062952393,
+      "grad_norm": 746.2387084960938,
+      "learning_rate": 9.91030954982375e-06,
+      "loss": 111.4794,
+      "step": 19100
+    },
+    {
+      "epoch": 0.15808412954460851,
+      "grad_norm": 1335.113525390625,
+      "learning_rate": 9.910043417419228e-06,
+      "loss": 148.9087,
+      "step": 19110
+    },
+    {
+      "epoch": 0.15816685279397774,
+      "grad_norm": 688.1320190429688,
+      "learning_rate": 9.909776894345086e-06,
+      "loss": 141.3004,
+      "step": 19120
+    },
+    {
+      "epoch": 0.158249576043347,
+      "grad_norm": 1126.445068359375,
+      "learning_rate": 9.909509980622532e-06,
+      "loss": 112.016,
+      "step": 19130
+    },
+    {
+      "epoch": 0.15833229929271622,
+      "grad_norm": 600.5185546875,
+      "learning_rate": 9.909242676272797e-06,
+      "loss": 114.159,
+      "step": 19140
+    },
+    {
+      "epoch": 0.15841502254208545,
+      "grad_norm": 1174.3468017578125,
+      "learning_rate": 9.908974981317155e-06,
+      "loss": 171.2533,
+      "step": 19150
+    },
+    {
+      "epoch": 0.15849774579145468,
+      "grad_norm": 795.6885986328125,
+      "learning_rate": 9.9087068957769e-06,
+      "loss": 136.8247,
+      "step": 19160
+    },
+    {
+      "epoch": 0.15858046904082393,
+      "grad_norm": 1241.0509033203125,
+      "learning_rate": 9.908438419673367e-06,
+      "loss": 137.2768,
+      "step": 19170
+    },
+    {
+      "epoch": 0.15866319229019316,
+      "grad_norm": 761.6776123046875,
+      "learning_rate": 9.908169553027916e-06,
+      "loss": 165.4491,
+      "step": 19180
+    },
+    {
+      "epoch": 0.15874591553956238,
+      "grad_norm": 1572.6136474609375,
+      "learning_rate": 9.90790029586194e-06,
+      "loss": 124.7587,
+      "step": 19190
+    },
+    {
+      "epoch": 0.15882863878893164,
+      "grad_norm": 732.7517700195312,
+      "learning_rate": 9.907630648196857e-06,
+      "loss": 142.462,
+      "step": 19200
+    },
+    {
+      "epoch": 0.15891136203830086,
+      "grad_norm": 957.8698120117188,
+      "learning_rate": 9.907360610054132e-06,
+      "loss": 145.4445,
+      "step": 19210
+    },
+    {
+      "epoch": 0.1589940852876701,
+      "grad_norm": 1933.4423828125,
+      "learning_rate": 9.907090181455241e-06,
+      "loss": 126.4228,
+      "step": 19220
+    },
+    {
+      "epoch": 0.15907680853703934,
+      "grad_norm": 1341.31591796875,
+      "learning_rate": 9.906819362421707e-06,
+      "loss": 127.2506,
+      "step": 19230
+    },
+    {
+      "epoch": 0.15915953178640857,
+      "grad_norm": 1401.2039794921875,
+      "learning_rate": 9.906548152975076e-06,
+      "loss": 142.5762,
+      "step": 19240
+    },
+    {
+      "epoch": 0.1592422550357778,
+      "grad_norm": 1090.92578125,
+      "learning_rate": 9.906276553136924e-06,
+      "loss": 133.9682,
+      "step": 19250
+    },
+    {
+      "epoch": 0.15932497828514705,
+      "grad_norm": 627.4381713867188,
+      "learning_rate": 9.906004562928865e-06,
+      "loss": 123.456,
+      "step": 19260
+    },
+    {
+      "epoch": 0.15940770153451628,
+      "grad_norm": 1303.6290283203125,
+      "learning_rate": 9.905732182372538e-06,
+      "loss": 176.1459,
+      "step": 19270
+    },
+    {
+      "epoch": 0.1594904247838855,
+      "grad_norm": 1230.2550048828125,
+      "learning_rate": 9.905459411489617e-06,
+      "loss": 150.8253,
+      "step": 19280
+    },
+    {
+      "epoch": 0.15957314803325476,
+      "grad_norm": 557.597900390625,
+      "learning_rate": 9.905186250301802e-06,
+      "loss": 128.1924,
+      "step": 19290
+    },
+    {
+      "epoch": 0.15965587128262398,
+      "grad_norm": 1013.2421875,
+      "learning_rate": 9.904912698830828e-06,
+      "loss": 148.6797,
+      "step": 19300
+    },
+    {
+      "epoch": 0.1597385945319932,
+      "grad_norm": 1238.0384521484375,
+      "learning_rate": 9.904638757098464e-06,
+      "loss": 143.5567,
+      "step": 19310
+    },
+    {
+      "epoch": 0.15982131778136247,
+      "grad_norm": 1117.36962890625,
+      "learning_rate": 9.9043644251265e-06,
+      "loss": 129.1499,
+      "step": 19320
+    },
+    {
+      "epoch": 0.1599040410307317,
+      "grad_norm": 1082.633544921875,
+      "learning_rate": 9.90408970293677e-06,
+      "loss": 99.3771,
+      "step": 19330
+    },
+    {
+      "epoch": 0.15998676428010092,
+      "grad_norm": 1239.34326171875,
+      "learning_rate": 9.903814590551127e-06,
+      "loss": 152.1191,
+      "step": 19340
+    },
+    {
+      "epoch": 0.16006948752947014,
+      "grad_norm": 1026.2008056640625,
+      "learning_rate": 9.903539087991462e-06,
+      "loss": 138.5603,
+      "step": 19350
+    },
+    {
+      "epoch": 0.1601522107788394,
+      "grad_norm": 716.8160400390625,
+      "learning_rate": 9.903263195279698e-06,
+      "loss": 121.6254,
+      "step": 19360
+    },
+    {
+      "epoch": 0.16023493402820863,
+      "grad_norm": 1355.804931640625,
+      "learning_rate": 9.902986912437784e-06,
+      "loss": 121.697,
+      "step": 19370
+    },
+    {
+      "epoch": 0.16031765727757785,
+      "grad_norm": 1001.9555053710938,
+      "learning_rate": 9.902710239487702e-06,
+      "loss": 123.8956,
+      "step": 19380
+    },
+    {
+      "epoch": 0.1604003805269471,
+      "grad_norm": 929.7130737304688,
+      "learning_rate": 9.902433176451466e-06,
+      "loss": 108.8211,
+      "step": 19390
+    },
+    {
+      "epoch": 0.16048310377631633,
+      "grad_norm": 1794.6314697265625,
+      "learning_rate": 9.902155723351124e-06,
+      "loss": 119.6667,
+      "step": 19400
+    },
+    {
+      "epoch": 0.16056582702568556,
+      "grad_norm": 981.8839721679688,
+      "learning_rate": 9.901877880208747e-06,
+      "loss": 123.9001,
+      "step": 19410
+    },
+    {
+      "epoch": 0.1606485502750548,
+      "grad_norm": 1454.4476318359375,
+      "learning_rate": 9.901599647046443e-06,
+      "loss": 131.0193,
+      "step": 19420
+    },
+    {
+      "epoch": 0.16073127352442404,
+      "grad_norm": 1284.30224609375,
+      "learning_rate": 9.901321023886351e-06,
+      "loss": 169.1719,
+      "step": 19430
+    },
+    {
+      "epoch": 0.16081399677379327,
+      "grad_norm": 1159.77783203125,
+      "learning_rate": 9.901042010750641e-06,
+      "loss": 100.9739,
+      "step": 19440
+    },
+    {
+      "epoch": 0.16089672002316252,
+      "grad_norm": 1844.110107421875,
+      "learning_rate": 9.900762607661509e-06,
+      "loss": 153.9659,
+      "step": 19450
+    },
+    {
+      "epoch": 0.16097944327253175,
+      "grad_norm": 701.9683227539062,
+      "learning_rate": 9.900482814641188e-06,
+      "loss": 109.9286,
+      "step": 19460
+    },
+    {
+      "epoch": 0.16106216652190097,
+      "grad_norm": 1962.1533203125,
+      "learning_rate": 9.90020263171194e-06,
+      "loss": 158.6689,
+      "step": 19470
+    },
+    {
+      "epoch": 0.16114488977127023,
+      "grad_norm": 1527.3931884765625,
+      "learning_rate": 9.899922058896058e-06,
+      "loss": 129.6219,
+      "step": 19480
+    },
+    {
+      "epoch": 0.16122761302063945,
+      "grad_norm": 935.3746337890625,
+      "learning_rate": 9.899641096215865e-06,
+      "loss": 187.1026,
+      "step": 19490
+    },
+    {
+      "epoch": 0.16131033627000868,
+      "grad_norm": 1236.936279296875,
+      "learning_rate": 9.899359743693715e-06,
+      "loss": 194.122,
+      "step": 19500
+    },
+    {
+      "epoch": 0.16139305951937793,
+      "grad_norm": 2321.50439453125,
+      "learning_rate": 9.899078001351996e-06,
+      "loss": 164.5937,
+      "step": 19510
+    },
+    {
+      "epoch": 0.16147578276874716,
+      "grad_norm": 1305.2666015625,
+      "learning_rate": 9.898795869213125e-06,
+      "loss": 149.4349,
+      "step": 19520
+    },
+    {
+      "epoch": 0.1615585060181164,
+      "grad_norm": 1175.187744140625,
+      "learning_rate": 9.898513347299549e-06,
+      "loss": 142.7042,
+      "step": 19530
+    },
+    {
+      "epoch": 0.16164122926748561,
+      "grad_norm": 1031.960693359375,
+      "learning_rate": 9.898230435633747e-06,
+      "loss": 151.7943,
+      "step": 19540
+    },
+    {
+      "epoch": 0.16172395251685487,
+      "grad_norm": 1225.9884033203125,
+      "learning_rate": 9.897947134238228e-06,
+      "loss": 162.5945,
+      "step": 19550
+    },
+    {
+      "epoch": 0.1618066757662241,
+      "grad_norm": 1006.8120727539062,
+      "learning_rate": 9.897663443135534e-06,
+      "loss": 127.9761,
+      "step": 19560
+    },
+    {
+      "epoch": 0.16188939901559332,
+      "grad_norm": 2127.416748046875,
+      "learning_rate": 9.897379362348239e-06,
+      "loss": 171.9894,
+      "step": 19570
+    },
+    {
+      "epoch": 0.16197212226496258,
+      "grad_norm": 745.5164794921875,
+      "learning_rate": 9.897094891898942e-06,
+      "loss": 150.5477,
+      "step": 19580
+    },
+    {
+      "epoch": 0.1620548455143318,
+      "grad_norm": 2045.43896484375,
+      "learning_rate": 9.89681003181028e-06,
+      "loss": 143.4611,
+      "step": 19590
+    },
+    {
+      "epoch": 0.16213756876370103,
+      "grad_norm": 1801.7623291015625,
+      "learning_rate": 9.896524782104917e-06,
+      "loss": 136.5524,
+      "step": 19600
+    },
+    {
+      "epoch": 0.16222029201307028,
+      "grad_norm": 865.27001953125,
+      "learning_rate": 9.89623914280555e-06,
+      "loss": 135.0138,
+      "step": 19610
+    },
+    {
+      "epoch": 0.1623030152624395,
+      "grad_norm": 1108.943359375,
+      "learning_rate": 9.895953113934904e-06,
+      "loss": 131.2855,
+      "step": 19620
+    },
+    {
+      "epoch": 0.16238573851180874,
+      "grad_norm": 1284.2874755859375,
+      "learning_rate": 9.895666695515739e-06,
+      "loss": 158.5307,
+      "step": 19630
+    },
+    {
+      "epoch": 0.162468461761178,
+      "grad_norm": 1160.2169189453125,
+      "learning_rate": 9.895379887570842e-06,
+      "loss": 146.816,
+      "step": 19640
+    },
+    {
+      "epoch": 0.16255118501054722,
+      "grad_norm": 1387.5469970703125,
+      "learning_rate": 9.895092690123036e-06,
+      "loss": 130.8016,
+      "step": 19650
+    },
+    {
+      "epoch": 0.16263390825991644,
+      "grad_norm": 1327.9295654296875,
+      "learning_rate": 9.894805103195168e-06,
+      "loss": 131.4063,
+      "step": 19660
+    },
+    {
+      "epoch": 0.1627166315092857,
+      "grad_norm": 1514.6529541015625,
+      "learning_rate": 9.894517126810122e-06,
+      "loss": 209.5621,
+      "step": 19670
+    },
+    {
+      "epoch": 0.16279935475865492,
+      "grad_norm": 2022.22021484375,
+      "learning_rate": 9.894228760990811e-06,
+      "loss": 152.1554,
+      "step": 19680
+    },
+    {
+      "epoch": 0.16288207800802415,
+      "grad_norm": 1740.86767578125,
+      "learning_rate": 9.893940005760181e-06,
+      "loss": 154.0035,
+      "step": 19690
+    },
+    {
+      "epoch": 0.16296480125739338,
+      "grad_norm": 1580.342529296875,
+      "learning_rate": 9.893650861141204e-06,
+      "loss": 157.6928,
+      "step": 19700
+    },
+    {
+      "epoch": 0.16304752450676263,
+      "grad_norm": 1006.531494140625,
+      "learning_rate": 9.893361327156887e-06,
+      "loss": 127.0846,
+      "step": 19710
+    },
+    {
+      "epoch": 0.16313024775613186,
+      "grad_norm": 1319.7847900390625,
+      "learning_rate": 9.893071403830265e-06,
+      "loss": 136.8425,
+      "step": 19720
+    },
+    {
+      "epoch": 0.16321297100550108,
+      "grad_norm": 830.5723876953125,
+      "learning_rate": 9.892781091184409e-06,
+      "loss": 136.2878,
+      "step": 19730
+    },
+    {
+      "epoch": 0.16329569425487034,
+      "grad_norm": 759.5004272460938,
+      "learning_rate": 9.892490389242417e-06,
+      "loss": 120.3061,
+      "step": 19740
+    },
+    {
+      "epoch": 0.16337841750423956,
+      "grad_norm": 2213.310546875,
+      "learning_rate": 9.892199298027416e-06,
+      "loss": 143.1016,
+      "step": 19750
+    },
+    {
+      "epoch": 0.1634611407536088,
+      "grad_norm": 1020.21337890625,
+      "learning_rate": 9.891907817562572e-06,
+      "loss": 116.1548,
+      "step": 19760
+    },
+    {
+      "epoch": 0.16354386400297805,
+      "grad_norm": 810.1748657226562,
+      "learning_rate": 9.891615947871072e-06,
+      "loss": 141.0581,
+      "step": 19770
+    },
+    {
+      "epoch": 0.16362658725234727,
+      "grad_norm": 810.9425048828125,
+      "learning_rate": 9.89132368897614e-06,
+      "loss": 130.3051,
+      "step": 19780
+    },
+    {
+      "epoch": 0.1637093105017165,
+      "grad_norm": 1321.05908203125,
+      "learning_rate": 9.891031040901031e-06,
+      "loss": 154.2215,
+      "step": 19790
+    },
+    {
+      "epoch": 0.16379203375108575,
+      "grad_norm": 1216.8099365234375,
+      "learning_rate": 9.890738003669029e-06,
+      "loss": 164.4314,
+      "step": 19800
+    },
+    {
+      "epoch": 0.16387475700045498,
+      "grad_norm": 2961.08447265625,
+      "learning_rate": 9.890444577303448e-06,
+      "loss": 184.4128,
+      "step": 19810
+    },
+    {
+      "epoch": 0.1639574802498242,
+      "grad_norm": 702.2813110351562,
+      "learning_rate": 9.890150761827639e-06,
+      "loss": 118.5094,
+      "step": 19820
+    },
+    {
+      "epoch": 0.16404020349919346,
+      "grad_norm": 1224.574951171875,
+      "learning_rate": 9.889856557264975e-06,
+      "loss": 164.4189,
+      "step": 19830
+    },
+    {
+      "epoch": 0.1641229267485627,
+      "grad_norm": 1510.7064208984375,
+      "learning_rate": 9.889561963638866e-06,
+      "loss": 168.8556,
+      "step": 19840
+    },
+    {
+      "epoch": 0.1642056499979319,
+      "grad_norm": 1308.8349609375,
+      "learning_rate": 9.889266980972752e-06,
+      "loss": 157.53,
+      "step": 19850
+    },
+    {
+      "epoch": 0.16428837324730117,
+      "grad_norm": 554.7841796875,
+      "learning_rate": 9.888971609290103e-06,
+      "loss": 123.5679,
+      "step": 19860
+    },
+    {
+      "epoch": 0.1643710964966704,
+      "grad_norm": 1066.4405517578125,
+      "learning_rate": 9.88867584861442e-06,
+      "loss": 115.5183,
+      "step": 19870
+    },
+    {
+      "epoch": 0.16445381974603962,
+      "grad_norm": 823.9727172851562,
+      "learning_rate": 9.888379698969236e-06,
+      "loss": 127.0505,
+      "step": 19880
+    },
+    {
+      "epoch": 0.16453654299540885,
+      "grad_norm": 2073.93017578125,
+      "learning_rate": 9.888083160378114e-06,
+      "loss": 142.6533,
+      "step": 19890
+    },
+    {
+      "epoch": 0.1646192662447781,
+      "grad_norm": 2434.051513671875,
+      "learning_rate": 9.887786232864648e-06,
+      "loss": 147.1622,
+      "step": 19900
+    },
+    {
+      "epoch": 0.16470198949414733,
+      "grad_norm": 724.6605224609375,
+      "learning_rate": 9.887488916452463e-06,
+      "loss": 121.8898,
+      "step": 19910
+    },
+    {
+      "epoch": 0.16478471274351655,
+      "grad_norm": 1281.83203125,
+      "learning_rate": 9.887191211165217e-06,
+      "loss": 151.6535,
+      "step": 19920
+    },
+    {
+      "epoch": 0.1648674359928858,
+      "grad_norm": 1131.1641845703125,
+      "learning_rate": 9.886893117026593e-06,
+      "loss": 149.5577,
+      "step": 19930
+    },
+    {
+      "epoch": 0.16495015924225503,
+      "grad_norm": 1077.4385986328125,
+      "learning_rate": 9.886594634060314e-06,
+      "loss": 203.1148,
+      "step": 19940
+    },
+    {
+      "epoch": 0.16503288249162426,
+      "grad_norm": 1563.59228515625,
+      "learning_rate": 9.886295762290125e-06,
+      "loss": 156.8315,
+      "step": 19950
+    },
+    {
+      "epoch": 0.16511560574099352,
+      "grad_norm": 728.146240234375,
+      "learning_rate": 9.885996501739808e-06,
+      "loss": 123.6347,
+      "step": 19960
+    },
+    {
+      "epoch": 0.16519832899036274,
+      "grad_norm": 1174.09521484375,
+      "learning_rate": 9.885696852433174e-06,
+      "loss": 171.4022,
+      "step": 19970
+    },
+    {
+      "epoch": 0.16528105223973197,
+      "grad_norm": 2437.55908203125,
+      "learning_rate": 9.885396814394062e-06,
+      "loss": 166.2973,
+      "step": 19980
+    },
+    {
+      "epoch": 0.16536377548910122,
+      "grad_norm": 756.368896484375,
+      "learning_rate": 9.885096387646346e-06,
+      "loss": 102.5183,
+      "step": 19990
+    },
+    {
+      "epoch": 0.16544649873847045,
+      "grad_norm": 1208.659423828125,
+      "learning_rate": 9.88479557221393e-06,
+      "loss": 146.3919,
+      "step": 20000
+    },
+    {
+      "epoch": 0.16552922198783968,
+      "grad_norm": 793.578857421875,
+      "learning_rate": 9.88449436812075e-06,
+      "loss": 151.3374,
+      "step": 20010
+    },
+    {
+      "epoch": 0.16561194523720893,
+      "grad_norm": 3023.392333984375,
+      "learning_rate": 9.88419277539077e-06,
+      "loss": 147.2389,
+      "step": 20020
+    },
+    {
+      "epoch": 0.16569466848657816,
+      "grad_norm": 1020.14404296875,
+      "learning_rate": 9.883890794047985e-06,
+      "loss": 133.7473,
+      "step": 20030
+    },
+    {
+      "epoch": 0.16577739173594738,
+      "grad_norm": 925.8684692382812,
+      "learning_rate": 9.883588424116424e-06,
+      "loss": 145.2095,
+      "step": 20040
+    },
+    {
+      "epoch": 0.16586011498531664,
+      "grad_norm": 1657.950927734375,
+      "learning_rate": 9.883285665620145e-06,
+      "loss": 131.4692,
+      "step": 20050
+    },
+    {
+      "epoch": 0.16594283823468586,
+      "grad_norm": 1638.5106201171875,
+      "learning_rate": 9.882982518583238e-06,
+      "loss": 120.6384,
+      "step": 20060
+    },
+    {
+      "epoch": 0.1660255614840551,
+      "grad_norm": 994.8275146484375,
+      "learning_rate": 9.882678983029819e-06,
+      "loss": 191.7884,
+      "step": 20070
+    },
+    {
+      "epoch": 0.16610828473342432,
+      "grad_norm": 1858.4609375,
+      "learning_rate": 9.882375058984044e-06,
+      "loss": 145.8128,
+      "step": 20080
+    },
+    {
+      "epoch": 0.16619100798279357,
+      "grad_norm": 640.3125,
+      "learning_rate": 9.882070746470092e-06,
+      "loss": 113.2083,
+      "step": 20090
+    },
+    {
+      "epoch": 0.1662737312321628,
+      "grad_norm": 1469.511474609375,
+      "learning_rate": 9.881766045512176e-06,
+      "loss": 189.0106,
+      "step": 20100
+    },
+    {
+      "epoch": 0.16635645448153202,
+      "grad_norm": 740.4965209960938,
+      "learning_rate": 9.88146095613454e-06,
+      "loss": 130.8047,
+      "step": 20110
+    },
+    {
+      "epoch": 0.16643917773090128,
+      "grad_norm": 683.9896240234375,
+      "learning_rate": 9.881155478361459e-06,
+      "loss": 175.5372,
+      "step": 20120
+    },
+    {
+      "epoch": 0.1665219009802705,
+      "grad_norm": 938.4227905273438,
+      "learning_rate": 9.880849612217238e-06,
+      "loss": 108.5235,
+      "step": 20130
+    },
+    {
+      "epoch": 0.16660462422963973,
+      "grad_norm": 841.4732666015625,
+      "learning_rate": 9.880543357726214e-06,
+      "loss": 142.9208,
+      "step": 20140
+    },
+    {
+      "epoch": 0.16668734747900898,
+      "grad_norm": 764.7952880859375,
+      "learning_rate": 9.880236714912754e-06,
+      "loss": 136.3933,
+      "step": 20150
+    },
+    {
+      "epoch": 0.1667700707283782,
+      "grad_norm": 1391.5673828125,
+      "learning_rate": 9.879929683801254e-06,
+      "loss": 138.4007,
+      "step": 20160
+    },
+    {
+      "epoch": 0.16685279397774744,
+      "grad_norm": 1550.94873046875,
+      "learning_rate": 9.879622264416147e-06,
+      "loss": 147.9795,
+      "step": 20170
+    },
+    {
+      "epoch": 0.1669355172271167,
+      "grad_norm": 879.2140502929688,
+      "learning_rate": 9.87931445678189e-06,
+      "loss": 155.5872,
+      "step": 20180
+    },
+    {
+      "epoch": 0.16701824047648592,
+      "grad_norm": 584.1538696289062,
+      "learning_rate": 9.879006260922975e-06,
+      "loss": 98.7441,
+      "step": 20190
+    },
+    {
+      "epoch": 0.16710096372585515,
+      "grad_norm": 719.3741455078125,
+      "learning_rate": 9.878697676863922e-06,
+      "loss": 126.3837,
+      "step": 20200
+    },
+    {
+      "epoch": 0.1671836869752244,
+      "grad_norm": 1768.941162109375,
+      "learning_rate": 9.878388704629286e-06,
+      "loss": 181.439,
+      "step": 20210
+    },
+    {
+      "epoch": 0.16726641022459363,
+      "grad_norm": 1062.1995849609375,
+      "learning_rate": 9.87807934424365e-06,
+      "loss": 155.1683,
+      "step": 20220
+    },
+    {
+      "epoch": 0.16734913347396285,
+      "grad_norm": 1074.44482421875,
+      "learning_rate": 9.877769595731629e-06,
+      "loss": 149.4426,
+      "step": 20230
+    },
+    {
+      "epoch": 0.16743185672333208,
+      "grad_norm": 766.1312866210938,
+      "learning_rate": 9.877459459117864e-06,
+      "loss": 133.7859,
+      "step": 20240
+    },
+    {
+      "epoch": 0.16751457997270133,
+      "grad_norm": 1180.6207275390625,
+      "learning_rate": 9.877148934427037e-06,
+      "loss": 185.559,
+      "step": 20250
+    },
+    {
+      "epoch": 0.16759730322207056,
+      "grad_norm": 936.2619018554688,
+      "learning_rate": 9.87683802168385e-06,
+      "loss": 153.8027,
+      "step": 20260
+    },
+    {
+      "epoch": 0.16768002647143979,
+      "grad_norm": 1289.442626953125,
+      "learning_rate": 9.876526720913045e-06,
+      "loss": 146.8949,
+      "step": 20270
+    },
+    {
+      "epoch": 0.16776274972080904,
+      "grad_norm": 1198.4373779296875,
+      "learning_rate": 9.87621503213939e-06,
+      "loss": 101.3234,
+      "step": 20280
+    },
+    {
+      "epoch": 0.16784547297017827,
+      "grad_norm": 1139.7901611328125,
+      "learning_rate": 9.875902955387682e-06,
+      "loss": 105.7266,
+      "step": 20290
+    },
+    {
+      "epoch": 0.1679281962195475,
+      "grad_norm": 885.1135864257812,
+      "learning_rate": 9.875590490682754e-06,
+      "loss": 139.6578,
+      "step": 20300
+    },
+    {
+      "epoch": 0.16801091946891675,
+      "grad_norm": 1269.400146484375,
+      "learning_rate": 9.875277638049466e-06,
+      "loss": 148.561,
+      "step": 20310
+    },
+    {
+      "epoch": 0.16809364271828597,
+      "grad_norm": 1671.2281494140625,
+      "learning_rate": 9.87496439751271e-06,
+      "loss": 157.8267,
+      "step": 20320
+    },
+    {
+      "epoch": 0.1681763659676552,
+      "grad_norm": 913.9154663085938,
+      "learning_rate": 9.87465076909741e-06,
+      "loss": 141.2827,
+      "step": 20330
+    },
+    {
+      "epoch": 0.16825908921702445,
+      "grad_norm": 1082.798583984375,
+      "learning_rate": 9.874336752828523e-06,
+      "loss": 150.5321,
+      "step": 20340
+    },
+    {
+      "epoch": 0.16834181246639368,
+      "grad_norm": 967.8886108398438,
+      "learning_rate": 9.87402234873103e-06,
+      "loss": 142.4086,
+      "step": 20350
+    },
+    {
+      "epoch": 0.1684245357157629,
+      "grad_norm": 1056.4305419921875,
+      "learning_rate": 9.873707556829945e-06,
+      "loss": 97.3924,
+      "step": 20360
+    },
+    {
+      "epoch": 0.16850725896513216,
+      "grad_norm": 837.5071411132812,
+      "learning_rate": 9.873392377150318e-06,
+      "loss": 123.7359,
+      "step": 20370
+    },
+    {
+      "epoch": 0.1685899822145014,
+      "grad_norm": 874.9818115234375,
+      "learning_rate": 9.873076809717226e-06,
+      "loss": 109.1823,
+      "step": 20380
+    },
+    {
+      "epoch": 0.16867270546387061,
+      "grad_norm": 727.0296020507812,
+      "learning_rate": 9.872760854555776e-06,
+      "loss": 143.6749,
+      "step": 20390
+    },
+    {
+      "epoch": 0.16875542871323987,
+      "grad_norm": 1192.26904296875,
+      "learning_rate": 9.872444511691108e-06,
+      "loss": 128.2298,
+      "step": 20400
+    },
+    {
+      "epoch": 0.1688381519626091,
+      "grad_norm": 765.3291015625,
+      "learning_rate": 9.872127781148392e-06,
+      "loss": 140.1519,
+      "step": 20410
+    },
+    {
+      "epoch": 0.16892087521197832,
+      "grad_norm": 1048.41064453125,
+      "learning_rate": 9.871810662952828e-06,
+      "loss": 153.647,
+      "step": 20420
+    },
+    {
+      "epoch": 0.16900359846134755,
+      "grad_norm": 1050.804931640625,
+      "learning_rate": 9.87149315712965e-06,
+      "loss": 160.4528,
+      "step": 20430
+    },
+    {
+      "epoch": 0.1690863217107168,
+      "grad_norm": 1345.5584716796875,
+      "learning_rate": 9.871175263704116e-06,
+      "loss": 119.833,
+      "step": 20440
+    },
+    {
+      "epoch": 0.16916904496008603,
+      "grad_norm": 1076.057861328125,
+      "learning_rate": 9.870856982701522e-06,
+      "loss": 123.713,
+      "step": 20450
+    },
+    {
+      "epoch": 0.16925176820945526,
+      "grad_norm": 1091.8779296875,
+      "learning_rate": 9.870538314147194e-06,
+      "loss": 157.7124,
+      "step": 20460
+    },
+    {
+      "epoch": 0.1693344914588245,
+      "grad_norm": 1105.4384765625,
+      "learning_rate": 9.870219258066485e-06,
+      "loss": 165.7277,
+      "step": 20470
+    },
+    {
+      "epoch": 0.16941721470819374,
+      "grad_norm": 1101.884033203125,
+      "learning_rate": 9.86989981448478e-06,
+      "loss": 93.6696,
+      "step": 20480
+    },
+    {
+      "epoch": 0.16949993795756296,
+      "grad_norm": 1253.4547119140625,
+      "learning_rate": 9.869579983427497e-06,
+      "loss": 156.5219,
+      "step": 20490
+    },
+    {
+      "epoch": 0.16958266120693222,
+      "grad_norm": 571.4030151367188,
+      "learning_rate": 9.869259764920081e-06,
+      "loss": 108.441,
+      "step": 20500
+    },
+    {
+      "epoch": 0.16966538445630144,
+      "grad_norm": 1125.4053955078125,
+      "learning_rate": 9.868939158988016e-06,
+      "loss": 111.0379,
+      "step": 20510
+    },
+    {
+      "epoch": 0.16974810770567067,
+      "grad_norm": 1643.4638671875,
+      "learning_rate": 9.868618165656805e-06,
+      "loss": 174.5399,
+      "step": 20520
+    },
+    {
+      "epoch": 0.16983083095503992,
+      "grad_norm": 1219.7093505859375,
+      "learning_rate": 9.868296784951992e-06,
+      "loss": 143.0416,
+      "step": 20530
+    },
+    {
+      "epoch": 0.16991355420440915,
+      "grad_norm": 716.170654296875,
+      "learning_rate": 9.867975016899145e-06,
+      "loss": 129.3315,
+      "step": 20540
+    },
+    {
+      "epoch": 0.16999627745377838,
+      "grad_norm": 722.2046508789062,
+      "learning_rate": 9.867652861523866e-06,
+      "loss": 110.3743,
+      "step": 20550
+    },
+    {
+      "epoch": 0.17007900070314763,
+      "grad_norm": 8400.3681640625,
+      "learning_rate": 9.86733031885179e-06,
+      "loss": 293.1446,
+      "step": 20560
+    },
+    {
+      "epoch": 0.17016172395251686,
+      "grad_norm": 1231.3631591796875,
+      "learning_rate": 9.867007388908579e-06,
+      "loss": 158.7573,
+      "step": 20570
+    },
+    {
+      "epoch": 0.17024444720188608,
+      "grad_norm": 813.4060668945312,
+      "learning_rate": 9.866684071719926e-06,
+      "loss": 114.0276,
+      "step": 20580
+    },
+    {
+      "epoch": 0.17032717045125534,
+      "grad_norm": 691.1693115234375,
+      "learning_rate": 9.866360367311557e-06,
+      "loss": 124.2069,
+      "step": 20590
+    },
+    {
+      "epoch": 0.17040989370062457,
+      "grad_norm": 1228.2000732421875,
+      "learning_rate": 9.866036275709226e-06,
+      "loss": 113.3982,
+      "step": 20600
+    },
+    {
+      "epoch": 0.1704926169499938,
+      "grad_norm": 1183.33935546875,
+      "learning_rate": 9.86571179693872e-06,
+      "loss": 124.6813,
+      "step": 20610
+    },
+    {
+      "epoch": 0.17057534019936302,
+      "grad_norm": 1132.4244384765625,
+      "learning_rate": 9.865386931025858e-06,
+      "loss": 109.3566,
+      "step": 20620
+    },
+    {
+      "epoch": 0.17065806344873227,
+      "grad_norm": 536.39453125,
+      "learning_rate": 9.865061677996487e-06,
+      "loss": 94.1924,
+      "step": 20630
+    },
+    {
+      "epoch": 0.1707407866981015,
+      "grad_norm": 1061.803955078125,
+      "learning_rate": 9.864736037876487e-06,
+      "loss": 96.9849,
+      "step": 20640
+    },
+    {
+      "epoch": 0.17082350994747073,
+      "grad_norm": 1086.843505859375,
+      "learning_rate": 9.864410010691766e-06,
+      "loss": 158.8272,
+      "step": 20650
+    },
+    {
+      "epoch": 0.17090623319683998,
+      "grad_norm": 1095.378662109375,
+      "learning_rate": 9.864083596468263e-06,
+      "loss": 165.8036,
+      "step": 20660
+    },
+    {
+      "epoch": 0.1709889564462092,
+      "grad_norm": 773.133544921875,
+      "learning_rate": 9.863756795231953e-06,
+      "loss": 107.4877,
+      "step": 20670
+    },
+    {
+      "epoch": 0.17107167969557843,
+      "grad_norm": 922.110107421875,
+      "learning_rate": 9.863429607008837e-06,
+      "loss": 152.2869,
+      "step": 20680
+    },
+    {
+      "epoch": 0.1711544029449477,
+      "grad_norm": 828.44677734375,
+      "learning_rate": 9.863102031824946e-06,
+      "loss": 103.4225,
+      "step": 20690
+    },
+    {
+      "epoch": 0.1712371261943169,
+      "grad_norm": 1600.447021484375,
+      "learning_rate": 9.862774069706346e-06,
+      "loss": 116.9802,
+      "step": 20700
+    },
+    {
+      "epoch": 0.17131984944368614,
+      "grad_norm": 1819.4007568359375,
+      "learning_rate": 9.86244572067913e-06,
+      "loss": 193.74,
+      "step": 20710
+    },
+    {
+      "epoch": 0.1714025726930554,
+      "grad_norm": 1109.385009765625,
+      "learning_rate": 9.862116984769424e-06,
+      "loss": 155.2172,
+      "step": 20720
+    },
+    {
+      "epoch": 0.17148529594242462,
+      "grad_norm": 1088.8150634765625,
+      "learning_rate": 9.861787862003384e-06,
+      "loss": 102.3083,
+      "step": 20730
+    },
+    {
+      "epoch": 0.17156801919179385,
+      "grad_norm": 863.4269409179688,
+      "learning_rate": 9.861458352407196e-06,
+      "loss": 163.3116,
+      "step": 20740
+    },
+    {
+      "epoch": 0.1716507424411631,
+      "grad_norm": 985.338623046875,
+      "learning_rate": 9.861128456007076e-06,
+      "loss": 128.2076,
+      "step": 20750
+    },
+    {
+      "epoch": 0.17173346569053233,
+      "grad_norm": 2616.19189453125,
+      "learning_rate": 9.860798172829277e-06,
+      "loss": 128.8503,
+      "step": 20760
+    },
+    {
+      "epoch": 0.17181618893990155,
+      "grad_norm": 821.7667846679688,
+      "learning_rate": 9.860467502900076e-06,
+      "loss": 139.1303,
+      "step": 20770
+    },
+    {
+      "epoch": 0.1718989121892708,
+      "grad_norm": 1093.295654296875,
+      "learning_rate": 9.860136446245779e-06,
+      "loss": 158.2073,
+      "step": 20780
+    },
+    {
+      "epoch": 0.17198163543864003,
+      "grad_norm": 2161.525390625,
+      "learning_rate": 9.859805002892733e-06,
+      "loss": 159.4854,
+      "step": 20790
+    },
+    {
+      "epoch": 0.17206435868800926,
+      "grad_norm": 936.6251831054688,
+      "learning_rate": 9.859473172867304e-06,
+      "loss": 150.7438,
+      "step": 20800
+    },
+    {
+      "epoch": 0.1721470819373785,
+      "grad_norm": 735.8521118164062,
+      "learning_rate": 9.859140956195898e-06,
+      "loss": 166.0925,
+      "step": 20810
+    },
+    {
+      "epoch": 0.17222980518674774,
+      "grad_norm": 1382.367431640625,
+      "learning_rate": 9.858808352904946e-06,
+      "loss": 163.85,
+      "step": 20820
+    },
+    {
+      "epoch": 0.17231252843611697,
+      "grad_norm": 1131.4239501953125,
+      "learning_rate": 9.858475363020913e-06,
+      "loss": 138.0052,
+      "step": 20830
+    },
+    {
+      "epoch": 0.1723952516854862,
+      "grad_norm": 1062.67431640625,
+      "learning_rate": 9.858141986570294e-06,
+      "loss": 127.1865,
+      "step": 20840
+    },
+    {
+      "epoch": 0.17247797493485545,
+      "grad_norm": 790.3716430664062,
+      "learning_rate": 9.85780822357961e-06,
+      "loss": 99.19,
+      "step": 20850
+    },
+    {
+      "epoch": 0.17256069818422468,
+      "grad_norm": 1042.801025390625,
+      "learning_rate": 9.857474074075422e-06,
+      "loss": 152.4688,
+      "step": 20860
+    },
+    {
+      "epoch": 0.1726434214335939,
+      "grad_norm": 872.380126953125,
+      "learning_rate": 9.857139538084313e-06,
+      "loss": 180.6786,
+      "step": 20870
+    },
+    {
+      "epoch": 0.17272614468296316,
+      "grad_norm": 873.7167358398438,
+      "learning_rate": 9.856804615632904e-06,
+      "loss": 153.8926,
+      "step": 20880
+    },
+    {
+      "epoch": 0.17280886793233238,
+      "grad_norm": 1261.3304443359375,
+      "learning_rate": 9.85646930674784e-06,
+      "loss": 138.4761,
+      "step": 20890
+    },
+    {
+      "epoch": 0.1728915911817016,
+      "grad_norm": 1224.0684814453125,
+      "learning_rate": 9.856133611455802e-06,
+      "loss": 116.0446,
+      "step": 20900
+    },
+    {
+      "epoch": 0.17297431443107086,
+      "grad_norm": 1034.66552734375,
+      "learning_rate": 9.855797529783499e-06,
+      "loss": 143.5475,
+      "step": 20910
+    },
+    {
+      "epoch": 0.1730570376804401,
+      "grad_norm": 1428.3978271484375,
+      "learning_rate": 9.855461061757673e-06,
+      "loss": 202.7229,
+      "step": 20920
+    },
+    {
+      "epoch": 0.17313976092980932,
+      "grad_norm": 1692.571533203125,
+      "learning_rate": 9.855124207405093e-06,
+      "loss": 129.3957,
+      "step": 20930
+    },
+    {
+      "epoch": 0.17322248417917857,
+      "grad_norm": 901.4539184570312,
+      "learning_rate": 9.854786966752561e-06,
+      "loss": 114.9998,
+      "step": 20940
+    },
+    {
+      "epoch": 0.1733052074285478,
+      "grad_norm": 1734.4339599609375,
+      "learning_rate": 9.854449339826912e-06,
+      "loss": 126.6563,
+      "step": 20950
+    },
+    {
+      "epoch": 0.17338793067791702,
+      "grad_norm": 936.6329956054688,
+      "learning_rate": 9.854111326655006e-06,
+      "loss": 148.8187,
+      "step": 20960
+    },
+    {
+      "epoch": 0.17347065392728625,
+      "grad_norm": 792.8075561523438,
+      "learning_rate": 9.85377292726374e-06,
+      "loss": 140.2986,
+      "step": 20970
+    },
+    {
+      "epoch": 0.1735533771766555,
+      "grad_norm": 852.9043579101562,
+      "learning_rate": 9.85343414168004e-06,
+      "loss": 142.9736,
+      "step": 20980
+    },
+    {
+      "epoch": 0.17363610042602473,
+      "grad_norm": 615.2127685546875,
+      "learning_rate": 9.853094969930857e-06,
+      "loss": 131.7546,
+      "step": 20990
+    },
+    {
+      "epoch": 0.17371882367539396,
+      "grad_norm": 1290.294921875,
+      "learning_rate": 9.85275541204318e-06,
+      "loss": 157.3864,
+      "step": 21000
+    },
+    {
+      "epoch": 0.1738015469247632,
+      "grad_norm": 428.7129821777344,
+      "learning_rate": 9.852415468044027e-06,
+      "loss": 117.3043,
+      "step": 21010
+    },
+    {
+      "epoch": 0.17388427017413244,
+      "grad_norm": 1278.5567626953125,
+      "learning_rate": 9.852075137960446e-06,
+      "loss": 136.686,
+      "step": 21020
+    },
+    {
+      "epoch": 0.17396699342350166,
+      "grad_norm": 1619.6778564453125,
+      "learning_rate": 9.851734421819511e-06,
+      "loss": 136.3727,
+      "step": 21030
+    },
+    {
+      "epoch": 0.17404971667287092,
+      "grad_norm": 1088.0845947265625,
+      "learning_rate": 9.851393319648338e-06,
+      "loss": 114.8988,
+      "step": 21040
+    },
+    {
+      "epoch": 0.17413243992224015,
+      "grad_norm": 738.9354858398438,
+      "learning_rate": 9.851051831474062e-06,
+      "loss": 138.9047,
+      "step": 21050
+    },
+    {
+      "epoch": 0.17421516317160937,
+      "grad_norm": 926.2405395507812,
+      "learning_rate": 9.850709957323855e-06,
+      "loss": 127.3761,
+      "step": 21060
+    },
+    {
+      "epoch": 0.17429788642097863,
+      "grad_norm": 911.6777954101562,
+      "learning_rate": 9.85036769722492e-06,
+      "loss": 163.0244,
+      "step": 21070
+    },
+    {
+      "epoch": 0.17438060967034785,
+      "grad_norm": 868.7709350585938,
+      "learning_rate": 9.850025051204484e-06,
+      "loss": 139.7337,
+      "step": 21080
+    },
+    {
+      "epoch": 0.17446333291971708,
+      "grad_norm": 810.2794189453125,
+      "learning_rate": 9.849682019289816e-06,
+      "loss": 129.7191,
+      "step": 21090
+    },
+    {
+      "epoch": 0.17454605616908633,
+      "grad_norm": 862.4880981445312,
+      "learning_rate": 9.849338601508204e-06,
+      "loss": 110.1159,
+      "step": 21100
+    },
+    {
+      "epoch": 0.17462877941845556,
+      "grad_norm": 1751.4161376953125,
+      "learning_rate": 9.848994797886978e-06,
+      "loss": 162.2478,
+      "step": 21110
+    },
+    {
+      "epoch": 0.1747115026678248,
+      "grad_norm": 961.6451416015625,
+      "learning_rate": 9.84865060845349e-06,
+      "loss": 151.0574,
+      "step": 21120
+    },
+    {
+      "epoch": 0.17479422591719404,
+      "grad_norm": 984.0369873046875,
+      "learning_rate": 9.848306033235123e-06,
+      "loss": 114.3529,
+      "step": 21130
+    },
+    {
+      "epoch": 0.17487694916656327,
+      "grad_norm": 1510.6654052734375,
+      "learning_rate": 9.847961072259298e-06,
+      "loss": 142.4745,
+      "step": 21140
+    },
+    {
+      "epoch": 0.1749596724159325,
+      "grad_norm": 578.0482177734375,
+      "learning_rate": 9.847615725553457e-06,
+      "loss": 149.1546,
+      "step": 21150
+    },
+    {
+      "epoch": 0.17504239566530172,
+      "grad_norm": 2596.45654296875,
+      "learning_rate": 9.847269993145082e-06,
+      "loss": 140.9354,
+      "step": 21160
+    },
+    {
+      "epoch": 0.17512511891467097,
+      "grad_norm": 1056.8643798828125,
+      "learning_rate": 9.84692387506168e-06,
+      "loss": 138.6333,
+      "step": 21170
+    },
+    {
+      "epoch": 0.1752078421640402,
+      "grad_norm": 543.871826171875,
+      "learning_rate": 9.846577371330788e-06,
+      "loss": 100.0174,
+      "step": 21180
+    },
+    {
+      "epoch": 0.17529056541340943,
+      "grad_norm": 505.1315002441406,
+      "learning_rate": 9.846230481979978e-06,
+      "loss": 126.8892,
+      "step": 21190
+    },
+    {
+      "epoch": 0.17537328866277868,
+      "grad_norm": 1152.1531982421875,
+      "learning_rate": 9.84588320703685e-06,
+      "loss": 137.2241,
+      "step": 21200
+    },
+    {
+      "epoch": 0.1754560119121479,
+      "grad_norm": 2514.837646484375,
+      "learning_rate": 9.845535546529036e-06,
+      "loss": 150.1807,
+      "step": 21210
+    },
+    {
+      "epoch": 0.17553873516151713,
+      "grad_norm": 1177.4676513671875,
+      "learning_rate": 9.845187500484194e-06,
+      "loss": 136.4538,
+      "step": 21220
+    },
+    {
+      "epoch": 0.1756214584108864,
+      "grad_norm": 1041.1258544921875,
+      "learning_rate": 9.844839068930021e-06,
+      "loss": 131.5591,
+      "step": 21230
+    },
+    {
+      "epoch": 0.17570418166025562,
+      "grad_norm": 907.1217651367188,
+      "learning_rate": 9.844490251894237e-06,
+      "loss": 129.923,
+      "step": 21240
+    },
+    {
+      "epoch": 0.17578690490962484,
+      "grad_norm": 917.8026123046875,
+      "learning_rate": 9.844141049404598e-06,
+      "loss": 112.846,
+      "step": 21250
+    },
+    {
+      "epoch": 0.1758696281589941,
+      "grad_norm": 1224.1368408203125,
+      "learning_rate": 9.843791461488887e-06,
+      "loss": 142.4482,
+      "step": 21260
+    },
+    {
+      "epoch": 0.17595235140836332,
+      "grad_norm": 1229.176513671875,
+      "learning_rate": 9.843441488174918e-06,
+      "loss": 103.1861,
+      "step": 21270
+    },
+    {
+      "epoch": 0.17603507465773255,
+      "grad_norm": 850.4046630859375,
+      "learning_rate": 9.843091129490539e-06,
+      "loss": 127.6695,
+      "step": 21280
+    },
+    {
+      "epoch": 0.1761177979071018,
+      "grad_norm": 842.6237182617188,
+      "learning_rate": 9.842740385463628e-06,
+      "loss": 150.564,
+      "step": 21290
+    },
+    {
+      "epoch": 0.17620052115647103,
+      "grad_norm": 1639.1712646484375,
+      "learning_rate": 9.842389256122086e-06,
+      "loss": 154.4747,
+      "step": 21300
+    },
+    {
+      "epoch": 0.17628324440584026,
+      "grad_norm": 1027.3468017578125,
+      "learning_rate": 9.842037741493856e-06,
+      "loss": 138.2844,
+      "step": 21310
+    },
+    {
+      "epoch": 0.1763659676552095,
+      "grad_norm": 1297.7607421875,
+      "learning_rate": 9.841685841606905e-06,
+      "loss": 171.9979,
+      "step": 21320
+    },
+    {
+      "epoch": 0.17644869090457874,
+      "grad_norm": 650.8685302734375,
+      "learning_rate": 9.841333556489232e-06,
+      "loss": 138.841,
+      "step": 21330
+    },
+    {
+      "epoch": 0.17653141415394796,
+      "grad_norm": 1357.2320556640625,
+      "learning_rate": 9.840980886168866e-06,
+      "loss": 151.9759,
+      "step": 21340
+    },
+    {
+      "epoch": 0.1766141374033172,
+      "grad_norm": 938.8858032226562,
+      "learning_rate": 9.840627830673867e-06,
+      "loss": 141.0563,
+      "step": 21350
+    },
+    {
+      "epoch": 0.17669686065268644,
+      "grad_norm": 1711.847412109375,
+      "learning_rate": 9.84027439003233e-06,
+      "loss": 172.4341,
+      "step": 21360
+    },
+    {
+      "epoch": 0.17677958390205567,
+      "grad_norm": 854.195068359375,
+      "learning_rate": 9.839920564272372e-06,
+      "loss": 137.6896,
+      "step": 21370
+    },
+    {
+      "epoch": 0.1768623071514249,
+      "grad_norm": 1087.6898193359375,
+      "learning_rate": 9.839566353422148e-06,
+      "loss": 117.6248,
+      "step": 21380
+    },
+    {
+      "epoch": 0.17694503040079415,
+      "grad_norm": 2788.3251953125,
+      "learning_rate": 9.839211757509838e-06,
+      "loss": 144.8487,
+      "step": 21390
+    },
+    {
+      "epoch": 0.17702775365016338,
+      "grad_norm": 1420.71875,
+      "learning_rate": 9.83885677656366e-06,
+      "loss": 122.336,
+      "step": 21400
+    },
+    {
+      "epoch": 0.1771104768995326,
+      "grad_norm": 761.9530029296875,
+      "learning_rate": 9.838501410611852e-06,
+      "loss": 132.9662,
+      "step": 21410
+    },
+    {
+      "epoch": 0.17719320014890186,
+      "grad_norm": 2090.32666015625,
+      "learning_rate": 9.838145659682695e-06,
+      "loss": 183.7629,
+      "step": 21420
+    },
+    {
+      "epoch": 0.17727592339827108,
+      "grad_norm": 3058.696044921875,
+      "learning_rate": 9.837789523804491e-06,
+      "loss": 142.1503,
+      "step": 21430
+    },
+    {
+      "epoch": 0.1773586466476403,
+      "grad_norm": 1059.2083740234375,
+      "learning_rate": 9.837433003005578e-06,
+      "loss": 137.1923,
+      "step": 21440
+    },
+    {
+      "epoch": 0.17744136989700957,
+      "grad_norm": 1151.251708984375,
+      "learning_rate": 9.83707609731432e-06,
+      "loss": 89.3938,
+      "step": 21450
+    },
+    {
+      "epoch": 0.1775240931463788,
+      "grad_norm": 749.4866333007812,
+      "learning_rate": 9.836718806759119e-06,
+      "loss": 145.9112,
+      "step": 21460
+    },
+    {
+      "epoch": 0.17760681639574802,
+      "grad_norm": 1220.1741943359375,
+      "learning_rate": 9.836361131368398e-06,
+      "loss": 179.5797,
+      "step": 21470
+    },
+    {
+      "epoch": 0.17768953964511727,
+      "grad_norm": 446.74859619140625,
+      "learning_rate": 9.836003071170617e-06,
+      "loss": 156.5806,
+      "step": 21480
+    },
+    {
+      "epoch": 0.1777722628944865,
+      "grad_norm": 872.1563720703125,
+      "learning_rate": 9.835644626194268e-06,
+      "loss": 141.7276,
+      "step": 21490
+    },
+    {
+      "epoch": 0.17785498614385573,
+      "grad_norm": 1689.736328125,
+      "learning_rate": 9.835285796467867e-06,
+      "loss": 133.7971,
+      "step": 21500
+    },
+    {
+      "epoch": 0.17793770939322495,
+      "grad_norm": 1551.1575927734375,
+      "learning_rate": 9.834926582019968e-06,
+      "loss": 156.3386,
+      "step": 21510
+    },
+    {
+      "epoch": 0.1780204326425942,
+      "grad_norm": 1045.95849609375,
+      "learning_rate": 9.834566982879149e-06,
+      "loss": 118.8335,
+      "step": 21520
+    },
+    {
+      "epoch": 0.17810315589196343,
+      "grad_norm": 1034.3304443359375,
+      "learning_rate": 9.83420699907402e-06,
+      "loss": 136.8709,
+      "step": 21530
+    },
+    {
+      "epoch": 0.17818587914133266,
+      "grad_norm": 870.5877075195312,
+      "learning_rate": 9.83384663063323e-06,
+      "loss": 125.3036,
+      "step": 21540
+    },
+    {
+      "epoch": 0.1782686023907019,
+      "grad_norm": 3151.6611328125,
+      "learning_rate": 9.833485877585447e-06,
+      "loss": 230.3187,
+      "step": 21550
+    },
+    {
+      "epoch": 0.17835132564007114,
+      "grad_norm": 1180.687255859375,
+      "learning_rate": 9.833124739959375e-06,
+      "loss": 128.1897,
+      "step": 21560
+    },
+    {
+      "epoch": 0.17843404888944037,
+      "grad_norm": 1331.4376220703125,
+      "learning_rate": 9.83276321778375e-06,
+      "loss": 161.2605,
+      "step": 21570
+    },
+    {
+      "epoch": 0.17851677213880962,
+      "grad_norm": 347.8481750488281,
+      "learning_rate": 9.832401311087334e-06,
+      "loss": 137.9425,
+      "step": 21580
+    },
+    {
+      "epoch": 0.17859949538817885,
+      "grad_norm": 1399.0616455078125,
+      "learning_rate": 9.832039019898922e-06,
+      "loss": 141.6175,
+      "step": 21590
+    },
+    {
+      "epoch": 0.17868221863754807,
+      "grad_norm": 1321.5462646484375,
+      "learning_rate": 9.831676344247343e-06,
+      "loss": 145.0367,
+      "step": 21600
+    },
+    {
+      "epoch": 0.17876494188691733,
+      "grad_norm": 2181.245361328125,
+      "learning_rate": 9.831313284161452e-06,
+      "loss": 127.1829,
+      "step": 21610
+    },
+    {
+      "epoch": 0.17884766513628655,
+      "grad_norm": 980.81201171875,
+      "learning_rate": 9.830949839670134e-06,
+      "loss": 160.9928,
+      "step": 21620
+    },
+    {
+      "epoch": 0.17893038838565578,
+      "grad_norm": 1042.4224853515625,
+      "learning_rate": 9.83058601080231e-06,
+      "loss": 143.5391,
+      "step": 21630
+    },
+    {
+      "epoch": 0.17901311163502504,
+      "grad_norm": 1248.8353271484375,
+      "learning_rate": 9.830221797586925e-06,
+      "loss": 125.7723,
+      "step": 21640
+    },
+    {
+      "epoch": 0.17909583488439426,
+      "grad_norm": 765.4020385742188,
+      "learning_rate": 9.829857200052961e-06,
+      "loss": 145.4247,
+      "step": 21650
+    },
+    {
+      "epoch": 0.1791785581337635,
+      "grad_norm": 1584.0721435546875,
+      "learning_rate": 9.829492218229426e-06,
+      "loss": 129.5774,
+      "step": 21660
+    },
+    {
+      "epoch": 0.17926128138313274,
+      "grad_norm": 640.2371215820312,
+      "learning_rate": 9.829126852145357e-06,
+      "loss": 175.8682,
+      "step": 21670
+    },
+    {
+      "epoch": 0.17934400463250197,
+      "grad_norm": 879.1619262695312,
+      "learning_rate": 9.82876110182983e-06,
+      "loss": 148.2943,
+      "step": 21680
+    },
+    {
+      "epoch": 0.1794267278818712,
+      "grad_norm": 1808.1944580078125,
+      "learning_rate": 9.82839496731194e-06,
+      "loss": 161.7926,
+      "step": 21690
+    },
+    {
+      "epoch": 0.17950945113124042,
+      "grad_norm": 1331.4281005859375,
+      "learning_rate": 9.828028448620824e-06,
+      "loss": 144.7912,
+      "step": 21700
+    },
+    {
+      "epoch": 0.17959217438060968,
+      "grad_norm": 916.88134765625,
+      "learning_rate": 9.827661545785641e-06,
+      "loss": 111.8015,
+      "step": 21710
+    },
+    {
+      "epoch": 0.1796748976299789,
+      "grad_norm": 945.744140625,
+      "learning_rate": 9.827294258835584e-06,
+      "loss": 135.7884,
+      "step": 21720
+    },
+    {
+      "epoch": 0.17975762087934813,
+      "grad_norm": 843.4868774414062,
+      "learning_rate": 9.82692658779988e-06,
+      "loss": 166.0405,
+      "step": 21730
+    },
+    {
+      "epoch": 0.17984034412871738,
+      "grad_norm": 866.8135375976562,
+      "learning_rate": 9.826558532707777e-06,
+      "loss": 131.8315,
+      "step": 21740
+    },
+    {
+      "epoch": 0.1799230673780866,
+      "grad_norm": 1258.517822265625,
+      "learning_rate": 9.826190093588564e-06,
+      "loss": 192.2693,
+      "step": 21750
+    },
+    {
+      "epoch": 0.18000579062745584,
+      "grad_norm": 755.1361694335938,
+      "learning_rate": 9.825821270471555e-06,
+      "loss": 110.0884,
+      "step": 21760
+    },
+    {
+      "epoch": 0.1800885138768251,
+      "grad_norm": 505.46856689453125,
+      "learning_rate": 9.825452063386094e-06,
+      "loss": 117.8567,
+      "step": 21770
+    },
+    {
+      "epoch": 0.18017123712619432,
+      "grad_norm": 1369.44287109375,
+      "learning_rate": 9.825082472361558e-06,
+      "loss": 135.8278,
+      "step": 21780
+    },
+    {
+      "epoch": 0.18025396037556354,
+      "grad_norm": 917.852783203125,
+      "learning_rate": 9.824712497427354e-06,
+      "loss": 101.3896,
+      "step": 21790
+    },
+    {
+      "epoch": 0.1803366836249328,
+      "grad_norm": 987.9921264648438,
+      "learning_rate": 9.824342138612918e-06,
+      "loss": 140.2833,
+      "step": 21800
+    },
+    {
+      "epoch": 0.18041940687430202,
+      "grad_norm": 902.6442260742188,
+      "learning_rate": 9.823971395947723e-06,
+      "loss": 147.3234,
+      "step": 21810
+    },
+    {
+      "epoch": 0.18050213012367125,
+      "grad_norm": 767.7576293945312,
+      "learning_rate": 9.823600269461259e-06,
+      "loss": 148.7868,
+      "step": 21820
+    },
+    {
+      "epoch": 0.1805848533730405,
+      "grad_norm": 766.7611694335938,
+      "learning_rate": 9.823228759183058e-06,
+      "loss": 124.8973,
+      "step": 21830
+    },
+    {
+      "epoch": 0.18066757662240973,
+      "grad_norm": 836.1171264648438,
+      "learning_rate": 9.822856865142683e-06,
+      "loss": 117.0898,
+      "step": 21840
+    },
+    {
+      "epoch": 0.18075029987177896,
+      "grad_norm": 1208.75146484375,
+      "learning_rate": 9.822484587369721e-06,
+      "loss": 125.5648,
+      "step": 21850
+    },
+    {
+      "epoch": 0.1808330231211482,
+      "grad_norm": 1158.348388671875,
+      "learning_rate": 9.822111925893792e-06,
+      "loss": 139.8755,
+      "step": 21860
+    },
+    {
+      "epoch": 0.18091574637051744,
+      "grad_norm": 1267.1917724609375,
+      "learning_rate": 9.821738880744549e-06,
+      "loss": 135.5524,
+      "step": 21870
+    },
+    {
+      "epoch": 0.18099846961988666,
+      "grad_norm": 2137.44580078125,
+      "learning_rate": 9.82136545195167e-06,
+      "loss": 159.7926,
+      "step": 21880
+    },
+    {
+      "epoch": 0.1810811928692559,
+      "grad_norm": 955.8012084960938,
+      "learning_rate": 9.82099163954487e-06,
+      "loss": 121.1186,
+      "step": 21890
+    },
+    {
+      "epoch": 0.18116391611862515,
+      "grad_norm": 1042.67724609375,
+      "learning_rate": 9.820617443553889e-06,
+      "loss": 132.7085,
+      "step": 21900
+    },
+    {
+      "epoch": 0.18124663936799437,
+      "grad_norm": 988.7931518554688,
+      "learning_rate": 9.820242864008503e-06,
+      "loss": 141.3183,
+      "step": 21910
+    },
+    {
+      "epoch": 0.1813293626173636,
+      "grad_norm": 1212.5155029296875,
+      "learning_rate": 9.819867900938514e-06,
+      "loss": 139.713,
+      "step": 21920
+    },
+    {
+      "epoch": 0.18141208586673285,
+      "grad_norm": 714.6464233398438,
+      "learning_rate": 9.819492554373758e-06,
+      "loss": 122.3466,
+      "step": 21930
+    },
+    {
+      "epoch": 0.18149480911610208,
+      "grad_norm": 1552.1221923828125,
+      "learning_rate": 9.819116824344095e-06,
+      "loss": 127.3137,
+      "step": 21940
+    },
+    {
+      "epoch": 0.1815775323654713,
+      "grad_norm": 577.6259155273438,
+      "learning_rate": 9.818740710879424e-06,
+      "loss": 87.6241,
+      "step": 21950
+    },
+    {
+      "epoch": 0.18166025561484056,
+      "grad_norm": 947.9591064453125,
+      "learning_rate": 9.81836421400967e-06,
+      "loss": 109.6651,
+      "step": 21960
+    },
+    {
+      "epoch": 0.1817429788642098,
+      "grad_norm": 915.073486328125,
+      "learning_rate": 9.81798733376479e-06,
+      "loss": 112.4667,
+      "step": 21970
+    },
+    {
+      "epoch": 0.181825702113579,
+      "grad_norm": 971.461181640625,
+      "learning_rate": 9.817610070174768e-06,
+      "loss": 159.3875,
+      "step": 21980
+    },
+    {
+      "epoch": 0.18190842536294827,
+      "grad_norm": 839.3546142578125,
+      "learning_rate": 9.817232423269622e-06,
+      "loss": 141.2685,
+      "step": 21990
+    },
+    {
+      "epoch": 0.1819911486123175,
+      "grad_norm": 901.919189453125,
+      "learning_rate": 9.816854393079402e-06,
+      "loss": 150.3302,
+      "step": 22000
+    },
+    {
+      "epoch": 0.18207387186168672,
+      "grad_norm": 1099.916259765625,
+      "learning_rate": 9.816475979634183e-06,
+      "loss": 105.6784,
+      "step": 22010
+    },
+    {
+      "epoch": 0.18215659511105597,
+      "grad_norm": 1539.3607177734375,
+      "learning_rate": 9.816097182964076e-06,
+      "loss": 150.3366,
+      "step": 22020
+    },
+    {
+      "epoch": 0.1822393183604252,
+      "grad_norm": 1262.76513671875,
+      "learning_rate": 9.81571800309922e-06,
+      "loss": 205.2206,
+      "step": 22030
+    },
+    {
+      "epoch": 0.18232204160979443,
+      "grad_norm": 1119.6072998046875,
+      "learning_rate": 9.815338440069782e-06,
+      "loss": 97.6272,
+      "step": 22040
+    },
+    {
+      "epoch": 0.18240476485916368,
+      "grad_norm": 977.2727661132812,
+      "learning_rate": 9.814958493905962e-06,
+      "loss": 154.7452,
+      "step": 22050
+    },
+    {
+      "epoch": 0.1824874881085329,
+      "grad_norm": 1467.2486572265625,
+      "learning_rate": 9.814578164637996e-06,
+      "loss": 116.3554,
+      "step": 22060
+    },
+    {
+      "epoch": 0.18257021135790213,
+      "grad_norm": 1132.6796875,
+      "learning_rate": 9.81419745229614e-06,
+      "loss": 146.8583,
+      "step": 22070
+    },
+    {
+      "epoch": 0.18265293460727136,
+      "grad_norm": 1301.7706298828125,
+      "learning_rate": 9.813816356910685e-06,
+      "loss": 150.6081,
+      "step": 22080
+    },
+    {
+      "epoch": 0.18273565785664062,
+      "grad_norm": 786.4771728515625,
+      "learning_rate": 9.813434878511956e-06,
+      "loss": 98.5915,
+      "step": 22090
+    },
+    {
+      "epoch": 0.18281838110600984,
+      "grad_norm": 1082.4107666015625,
+      "learning_rate": 9.813053017130305e-06,
+      "loss": 106.2249,
+      "step": 22100
+    },
+    {
+      "epoch": 0.18290110435537907,
+      "grad_norm": 844.4359130859375,
+      "learning_rate": 9.812670772796113e-06,
+      "loss": 108.4317,
+      "step": 22110
+    },
+    {
+      "epoch": 0.18298382760474832,
+      "grad_norm": 1356.7515869140625,
+      "learning_rate": 9.812288145539796e-06,
+      "loss": 156.1458,
+      "step": 22120
+    },
+    {
+      "epoch": 0.18306655085411755,
+      "grad_norm": 571.9811401367188,
+      "learning_rate": 9.811905135391796e-06,
+      "loss": 128.8242,
+      "step": 22130
+    },
+    {
+      "epoch": 0.18314927410348678,
+      "grad_norm": 996.9983520507812,
+      "learning_rate": 9.81152174238259e-06,
+      "loss": 121.8907,
+      "step": 22140
+    },
+    {
+      "epoch": 0.18323199735285603,
+      "grad_norm": 1005.370361328125,
+      "learning_rate": 9.81113796654268e-06,
+      "loss": 130.7526,
+      "step": 22150
+    },
+    {
+      "epoch": 0.18331472060222526,
+      "grad_norm": 522.5587158203125,
+      "learning_rate": 9.810753807902603e-06,
+      "loss": 167.8494,
+      "step": 22160
+    },
+    {
+      "epoch": 0.18339744385159448,
+      "grad_norm": 1476.099853515625,
+      "learning_rate": 9.81036926649292e-06,
+      "loss": 125.2975,
+      "step": 22170
+    },
+    {
+      "epoch": 0.18348016710096374,
+      "grad_norm": 864.9098510742188,
+      "learning_rate": 9.809984342344234e-06,
+      "loss": 114.3755,
+      "step": 22180
+    },
+    {
+      "epoch": 0.18356289035033296,
+      "grad_norm": 1131.15966796875,
+      "learning_rate": 9.80959903548717e-06,
+      "loss": 130.3137,
+      "step": 22190
+    },
+    {
+      "epoch": 0.1836456135997022,
+      "grad_norm": 2763.490234375,
+      "learning_rate": 9.80921334595238e-06,
+      "loss": 159.7871,
+      "step": 22200
+    },
+    {
+      "epoch": 0.18372833684907144,
+      "grad_norm": 815.5933227539062,
+      "learning_rate": 9.808827273770558e-06,
+      "loss": 103.766,
+      "step": 22210
+    },
+    {
+      "epoch": 0.18381106009844067,
+      "grad_norm": 1680.257568359375,
+      "learning_rate": 9.80844081897242e-06,
+      "loss": 162.1204,
+      "step": 22220
+    },
+    {
+      "epoch": 0.1838937833478099,
+      "grad_norm": 934.7627563476562,
+      "learning_rate": 9.808053981588712e-06,
+      "loss": 144.453,
+      "step": 22230
+    },
+    {
+      "epoch": 0.18397650659717912,
+      "grad_norm": 1144.36767578125,
+      "learning_rate": 9.807666761650215e-06,
+      "loss": 121.5437,
+      "step": 22240
+    },
+    {
+      "epoch": 0.18405922984654838,
+      "grad_norm": 2383.975341796875,
+      "learning_rate": 9.80727915918774e-06,
+      "loss": 178.4958,
+      "step": 22250
+    },
+    {
+      "epoch": 0.1841419530959176,
+      "grad_norm": 1073.094482421875,
+      "learning_rate": 9.806891174232122e-06,
+      "loss": 149.8745,
+      "step": 22260
+    },
+    {
+      "epoch": 0.18422467634528683,
+      "grad_norm": 1233.2926025390625,
+      "learning_rate": 9.806502806814236e-06,
+      "loss": 128.9702,
+      "step": 22270
+    },
+    {
+      "epoch": 0.18430739959465609,
+      "grad_norm": 837.9373168945312,
+      "learning_rate": 9.806114056964977e-06,
+      "loss": 139.9306,
+      "step": 22280
+    },
+    {
+      "epoch": 0.1843901228440253,
+      "grad_norm": 628.1285400390625,
+      "learning_rate": 9.805724924715283e-06,
+      "loss": 123.0449,
+      "step": 22290
+    },
+    {
+      "epoch": 0.18447284609339454,
+      "grad_norm": 561.8240356445312,
+      "learning_rate": 9.80533541009611e-06,
+      "loss": 105.3535,
+      "step": 22300
+    },
+    {
+      "epoch": 0.1845555693427638,
+      "grad_norm": 1421.851806640625,
+      "learning_rate": 9.804945513138454e-06,
+      "loss": 219.4902,
+      "step": 22310
+    },
+    {
+      "epoch": 0.18463829259213302,
+      "grad_norm": 825.07080078125,
+      "learning_rate": 9.804555233873335e-06,
+      "loss": 135.7106,
+      "step": 22320
+    },
+    {
+      "epoch": 0.18472101584150225,
+      "grad_norm": 2163.25439453125,
+      "learning_rate": 9.804164572331804e-06,
+      "loss": 173.7582,
+      "step": 22330
+    },
+    {
+      "epoch": 0.1848037390908715,
+      "grad_norm": 1122.6065673828125,
+      "learning_rate": 9.80377352854495e-06,
+      "loss": 91.7703,
+      "step": 22340
+    },
+    {
+      "epoch": 0.18488646234024073,
+      "grad_norm": 909.8526000976562,
+      "learning_rate": 9.80338210254388e-06,
+      "loss": 124.849,
+      "step": 22350
+    },
+    {
+      "epoch": 0.18496918558960995,
+      "grad_norm": 784.2378540039062,
+      "learning_rate": 9.80299029435974e-06,
+      "loss": 148.4874,
+      "step": 22360
+    },
+    {
+      "epoch": 0.1850519088389792,
+      "grad_norm": 999.297119140625,
+      "learning_rate": 9.802598104023706e-06,
+      "loss": 159.1145,
+      "step": 22370
+    },
+    {
+      "epoch": 0.18513463208834843,
+      "grad_norm": 753.5262451171875,
+      "learning_rate": 9.80220553156698e-06,
+      "loss": 113.2028,
+      "step": 22380
+    },
+    {
+      "epoch": 0.18521735533771766,
+      "grad_norm": 1254.95947265625,
+      "learning_rate": 9.801812577020802e-06,
+      "loss": 139.2906,
+      "step": 22390
+    },
+    {
+      "epoch": 0.1853000785870869,
+      "grad_norm": 1023.6962890625,
+      "learning_rate": 9.801419240416432e-06,
+      "loss": 144.157,
+      "step": 22400
+    },
+    {
+      "epoch": 0.18538280183645614,
+      "grad_norm": 937.4168701171875,
+      "learning_rate": 9.80102552178517e-06,
+      "loss": 124.0214,
+      "step": 22410
+    },
+    {
+      "epoch": 0.18546552508582537,
+      "grad_norm": 1038.947265625,
+      "learning_rate": 9.800631421158341e-06,
+      "loss": 134.8847,
+      "step": 22420
+    },
+    {
+      "epoch": 0.1855482483351946,
+      "grad_norm": 839.5892333984375,
+      "learning_rate": 9.800236938567302e-06,
+      "loss": 114.9243,
+      "step": 22430
+    },
+    {
+      "epoch": 0.18563097158456385,
+      "grad_norm": 1064.1595458984375,
+      "learning_rate": 9.799842074043438e-06,
+      "loss": 132.5262,
+      "step": 22440
+    },
+    {
+      "epoch": 0.18571369483393307,
+      "grad_norm": 868.5515747070312,
+      "learning_rate": 9.799446827618172e-06,
+      "loss": 123.4177,
+      "step": 22450
+    },
+    {
+      "epoch": 0.1857964180833023,
+      "grad_norm": 1418.9991455078125,
+      "learning_rate": 9.799051199322944e-06,
+      "loss": 134.293,
+      "step": 22460
+    },
+    {
+      "epoch": 0.18587914133267155,
+      "grad_norm": 1264.387939453125,
+      "learning_rate": 9.798655189189239e-06,
+      "loss": 155.6345,
+      "step": 22470
+    },
+    {
+      "epoch": 0.18596186458204078,
+      "grad_norm": 1341.15185546875,
+      "learning_rate": 9.798258797248563e-06,
+      "loss": 151.904,
+      "step": 22480
+    },
+    {
+      "epoch": 0.18604458783141,
+      "grad_norm": 1191.142578125,
+      "learning_rate": 9.797862023532457e-06,
+      "loss": 143.8828,
+      "step": 22490
+    },
+    {
+      "epoch": 0.18612731108077926,
+      "grad_norm": 1029.2672119140625,
+      "learning_rate": 9.797464868072489e-06,
+      "loss": 111.6996,
+      "step": 22500
+    },
+    {
+      "epoch": 0.1862100343301485,
+      "grad_norm": 898.6533203125,
+      "learning_rate": 9.797067330900256e-06,
+      "loss": 123.9605,
+      "step": 22510
+    },
+    {
+      "epoch": 0.18629275757951771,
+      "grad_norm": 1148.5517578125,
+      "learning_rate": 9.796669412047392e-06,
+      "loss": 107.9766,
+      "step": 22520
+    },
+    {
+      "epoch": 0.18637548082888697,
+      "grad_norm": 1347.3740234375,
+      "learning_rate": 9.796271111545559e-06,
+      "loss": 132.5673,
+      "step": 22530
+    },
+    {
+      "epoch": 0.1864582040782562,
+      "grad_norm": 732.031982421875,
+      "learning_rate": 9.795872429426443e-06,
+      "loss": 113.1994,
+      "step": 22540
+    },
+    {
+      "epoch": 0.18654092732762542,
+      "grad_norm": 847.0791625976562,
+      "learning_rate": 9.79547336572177e-06,
+      "loss": 115.746,
+      "step": 22550
+    },
+    {
+      "epoch": 0.18662365057699468,
+      "grad_norm": 897.2965698242188,
+      "learning_rate": 9.795073920463289e-06,
+      "loss": 123.8876,
+      "step": 22560
+    },
+    {
+      "epoch": 0.1867063738263639,
+      "grad_norm": 1148.21875,
+      "learning_rate": 9.794674093682781e-06,
+      "loss": 133.7378,
+      "step": 22570
+    },
+    {
+      "epoch": 0.18678909707573313,
+      "grad_norm": 1180.834716796875,
+      "learning_rate": 9.79427388541206e-06,
+      "loss": 100.9495,
+      "step": 22580
+    },
+    {
+      "epoch": 0.18687182032510238,
+      "grad_norm": 681.8207397460938,
+      "learning_rate": 9.79387329568297e-06,
+      "loss": 108.7203,
+      "step": 22590
+    },
+    {
+      "epoch": 0.1869545435744716,
+      "grad_norm": 859.6419067382812,
+      "learning_rate": 9.793472324527383e-06,
+      "loss": 179.532,
+      "step": 22600
+    },
+    {
+      "epoch": 0.18703726682384084,
+      "grad_norm": 615.40380859375,
+      "learning_rate": 9.793070971977203e-06,
+      "loss": 103.2244,
+      "step": 22610
+    },
+    {
+      "epoch": 0.18711999007321006,
+      "grad_norm": 896.5668334960938,
+      "learning_rate": 9.79266923806436e-06,
+      "loss": 135.8797,
+      "step": 22620
+    },
+    {
+      "epoch": 0.18720271332257932,
+      "grad_norm": 685.52685546875,
+      "learning_rate": 9.792267122820823e-06,
+      "loss": 102.7591,
+      "step": 22630
+    },
+    {
+      "epoch": 0.18728543657194854,
+      "grad_norm": 656.8187255859375,
+      "learning_rate": 9.791864626278584e-06,
+      "loss": 134.8737,
+      "step": 22640
+    },
+    {
+      "epoch": 0.18736815982131777,
+      "grad_norm": 406.96820068359375,
+      "learning_rate": 9.791461748469669e-06,
+      "loss": 129.8552,
+      "step": 22650
+    },
+    {
+      "epoch": 0.18745088307068702,
+      "grad_norm": 1129.7803955078125,
+      "learning_rate": 9.791058489426134e-06,
+      "loss": 133.4618,
+      "step": 22660
+    },
+    {
+      "epoch": 0.18753360632005625,
+      "grad_norm": 1034.3385009765625,
+      "learning_rate": 9.790654849180059e-06,
+      "loss": 107.0448,
+      "step": 22670
+    },
+    {
+      "epoch": 0.18761632956942548,
+      "grad_norm": 1405.0714111328125,
+      "learning_rate": 9.790250827763566e-06,
+      "loss": 121.3317,
+      "step": 22680
+    },
+    {
+      "epoch": 0.18769905281879473,
+      "grad_norm": 1143.3116455078125,
+      "learning_rate": 9.7898464252088e-06,
+      "loss": 93.2796,
+      "step": 22690
+    },
+    {
+      "epoch": 0.18778177606816396,
+      "grad_norm": 1161.796630859375,
+      "learning_rate": 9.789441641547935e-06,
+      "loss": 112.5664,
+      "step": 22700
+    },
+    {
+      "epoch": 0.18786449931753318,
+      "grad_norm": 1665.2811279296875,
+      "learning_rate": 9.789036476813178e-06,
+      "loss": 162.8835,
+      "step": 22710
+    },
+    {
+      "epoch": 0.18794722256690244,
+      "grad_norm": 885.5316772460938,
+      "learning_rate": 9.788630931036769e-06,
+      "loss": 127.6729,
+      "step": 22720
+    },
+    {
+      "epoch": 0.18802994581627167,
+      "grad_norm": 1863.15673828125,
+      "learning_rate": 9.788225004250974e-06,
+      "loss": 136.1164,
+      "step": 22730
+    },
+    {
+      "epoch": 0.1881126690656409,
+      "grad_norm": 770.5360717773438,
+      "learning_rate": 9.78781869648809e-06,
+      "loss": 145.4966,
+      "step": 22740
+    },
+    {
+      "epoch": 0.18819539231501015,
+      "grad_norm": 1846.6016845703125,
+      "learning_rate": 9.787412007780445e-06,
+      "loss": 119.8534,
+      "step": 22750
+    },
+    {
+      "epoch": 0.18827811556437937,
+      "grad_norm": 930.654052734375,
+      "learning_rate": 9.787004938160398e-06,
+      "loss": 112.2845,
+      "step": 22760
+    },
+    {
+      "epoch": 0.1883608388137486,
+      "grad_norm": 3968.6298828125,
+      "learning_rate": 9.786597487660336e-06,
+      "loss": 166.1542,
+      "step": 22770
+    },
+    {
+      "epoch": 0.18844356206311783,
+      "grad_norm": 687.5050659179688,
+      "learning_rate": 9.78618965631268e-06,
+      "loss": 118.5364,
+      "step": 22780
+    },
+    {
+      "epoch": 0.18852628531248708,
+      "grad_norm": 1421.8592529296875,
+      "learning_rate": 9.785781444149883e-06,
+      "loss": 102.7248,
+      "step": 22790
+    },
+    {
+      "epoch": 0.1886090085618563,
+      "grad_norm": 1723.80029296875,
+      "learning_rate": 9.785372851204415e-06,
+      "loss": 136.1481,
+      "step": 22800
+    },
+    {
+      "epoch": 0.18869173181122553,
+      "grad_norm": 1364.1827392578125,
+      "learning_rate": 9.784963877508794e-06,
+      "loss": 135.9108,
+      "step": 22810
+    },
+    {
+      "epoch": 0.1887744550605948,
+      "grad_norm": 2061.134765625,
+      "learning_rate": 9.784554523095554e-06,
+      "loss": 201.537,
+      "step": 22820
+    },
+    {
+      "epoch": 0.188857178309964,
+      "grad_norm": 958.0075073242188,
+      "learning_rate": 9.784144787997272e-06,
+      "loss": 96.8345,
+      "step": 22830
+    },
+    {
+      "epoch": 0.18893990155933324,
+      "grad_norm": 717.226318359375,
+      "learning_rate": 9.783734672246545e-06,
+      "loss": 130.6322,
+      "step": 22840
+    },
+    {
+      "epoch": 0.1890226248087025,
+      "grad_norm": 716.5567016601562,
+      "learning_rate": 9.783324175876004e-06,
+      "loss": 105.1676,
+      "step": 22850
+    },
+    {
+      "epoch": 0.18910534805807172,
+      "grad_norm": 1025.28173828125,
+      "learning_rate": 9.782913298918311e-06,
+      "loss": 151.713,
+      "step": 22860
+    },
+    {
+      "epoch": 0.18918807130744095,
+      "grad_norm": 649.5341186523438,
+      "learning_rate": 9.782502041406157e-06,
+      "loss": 135.4802,
+      "step": 22870
+    },
+    {
+      "epoch": 0.1892707945568102,
+      "grad_norm": 1117.214599609375,
+      "learning_rate": 9.782090403372263e-06,
+      "loss": 148.4325,
+      "step": 22880
+    },
+    {
+      "epoch": 0.18935351780617943,
+      "grad_norm": 1014.591552734375,
+      "learning_rate": 9.781678384849385e-06,
+      "loss": 111.5626,
+      "step": 22890
+    },
+    {
+      "epoch": 0.18943624105554865,
+      "grad_norm": 2505.06982421875,
+      "learning_rate": 9.7812659858703e-06,
+      "loss": 142.1111,
+      "step": 22900
+    },
+    {
+      "epoch": 0.1895189643049179,
+      "grad_norm": 1313.8260498046875,
+      "learning_rate": 9.780853206467826e-06,
+      "loss": 133.5671,
+      "step": 22910
+    },
+    {
+      "epoch": 0.18960168755428713,
+      "grad_norm": 984.4515380859375,
+      "learning_rate": 9.780440046674803e-06,
+      "loss": 167.4173,
+      "step": 22920
+    },
+    {
+      "epoch": 0.18968441080365636,
+      "grad_norm": 893.4295043945312,
+      "learning_rate": 9.780026506524106e-06,
+      "loss": 159.9201,
+      "step": 22930
+    },
+    {
+      "epoch": 0.18976713405302562,
+      "grad_norm": 818.2760620117188,
+      "learning_rate": 9.779612586048635e-06,
+      "loss": 132.297,
+      "step": 22940
+    },
+    {
+      "epoch": 0.18984985730239484,
+      "grad_norm": 1028.792236328125,
+      "learning_rate": 9.779198285281326e-06,
+      "loss": 175.7389,
+      "step": 22950
+    },
+    {
+      "epoch": 0.18993258055176407,
+      "grad_norm": 1137.0174560546875,
+      "learning_rate": 9.778783604255145e-06,
+      "loss": 120.6092,
+      "step": 22960
+    },
+    {
+      "epoch": 0.1900153038011333,
+      "grad_norm": 2498.3369140625,
+      "learning_rate": 9.778368543003083e-06,
+      "loss": 144.8033,
+      "step": 22970
+    },
+    {
+      "epoch": 0.19009802705050255,
+      "grad_norm": 1361.688232421875,
+      "learning_rate": 9.777953101558164e-06,
+      "loss": 133.7076,
+      "step": 22980
+    },
+    {
+      "epoch": 0.19018075029987178,
+      "grad_norm": 1047.64453125,
+      "learning_rate": 9.777537279953448e-06,
+      "loss": 120.5423,
+      "step": 22990
+    },
+    {
+      "epoch": 0.190263473549241,
+      "grad_norm": 982.462158203125,
+      "learning_rate": 9.777121078222015e-06,
+      "loss": 108.502,
+      "step": 23000
+    },
+    {
+      "epoch": 0.19034619679861026,
+      "grad_norm": 1133.6990966796875,
+      "learning_rate": 9.77670449639698e-06,
+      "loss": 129.8804,
+      "step": 23010
+    },
+    {
+      "epoch": 0.19042892004797948,
+      "grad_norm": 671.9098510742188,
+      "learning_rate": 9.776287534511492e-06,
+      "loss": 130.2723,
+      "step": 23020
+    },
+    {
+      "epoch": 0.1905116432973487,
+      "grad_norm": 1220.64892578125,
+      "learning_rate": 9.775870192598726e-06,
+      "loss": 122.8446,
+      "step": 23030
+    },
+    {
+      "epoch": 0.19059436654671796,
+      "grad_norm": 2241.37548828125,
+      "learning_rate": 9.775452470691886e-06,
+      "loss": 133.7913,
+      "step": 23040
+    },
+    {
+      "epoch": 0.1906770897960872,
+      "grad_norm": 1148.03662109375,
+      "learning_rate": 9.77503436882421e-06,
+      "loss": 138.1318,
+      "step": 23050
+    },
+    {
+      "epoch": 0.19075981304545642,
+      "grad_norm": 980.452880859375,
+      "learning_rate": 9.774615887028964e-06,
+      "loss": 147.464,
+      "step": 23060
+    },
+    {
+      "epoch": 0.19084253629482567,
+      "grad_norm": 2212.347412109375,
+      "learning_rate": 9.774197025339442e-06,
+      "loss": 102.5226,
+      "step": 23070
+    },
+    {
+      "epoch": 0.1909252595441949,
+      "grad_norm": 928.7798461914062,
+      "learning_rate": 9.773777783788976e-06,
+      "loss": 113.3667,
+      "step": 23080
+    },
+    {
+      "epoch": 0.19100798279356412,
+      "grad_norm": 1282.0994873046875,
+      "learning_rate": 9.77335816241092e-06,
+      "loss": 144.8801,
+      "step": 23090
+    },
+    {
+      "epoch": 0.19109070604293338,
+      "grad_norm": 837.4090576171875,
+      "learning_rate": 9.77293816123866e-06,
+      "loss": 104.0558,
+      "step": 23100
+    },
+    {
+      "epoch": 0.1911734292923026,
+      "grad_norm": 881.4822998046875,
+      "learning_rate": 9.772517780305618e-06,
+      "loss": 121.4086,
+      "step": 23110
+    },
+    {
+      "epoch": 0.19125615254167183,
+      "grad_norm": 880.7931518554688,
+      "learning_rate": 9.772097019645236e-06,
+      "loss": 139.5056,
+      "step": 23120
+    },
+    {
+      "epoch": 0.19133887579104109,
+      "grad_norm": 776.1262817382812,
+      "learning_rate": 9.771675879290998e-06,
+      "loss": 146.6072,
+      "step": 23130
+    },
+    {
+      "epoch": 0.1914215990404103,
+      "grad_norm": 904.765380859375,
+      "learning_rate": 9.771254359276407e-06,
+      "loss": 116.9324,
+      "step": 23140
+    },
+    {
+      "epoch": 0.19150432228977954,
+      "grad_norm": 730.8990478515625,
+      "learning_rate": 9.770832459635004e-06,
+      "loss": 133.2764,
+      "step": 23150
+    },
+    {
+      "epoch": 0.19158704553914876,
+      "grad_norm": 690.1975708007812,
+      "learning_rate": 9.77041018040036e-06,
+      "loss": 175.1648,
+      "step": 23160
+    },
+    {
+      "epoch": 0.19166976878851802,
+      "grad_norm": 907.7679443359375,
+      "learning_rate": 9.769987521606068e-06,
+      "loss": 105.3016,
+      "step": 23170
+    },
+    {
+      "epoch": 0.19175249203788725,
+      "grad_norm": 693.336181640625,
+      "learning_rate": 9.769564483285761e-06,
+      "loss": 192.6149,
+      "step": 23180
+    },
+    {
+      "epoch": 0.19183521528725647,
+      "grad_norm": 557.3772583007812,
+      "learning_rate": 9.769141065473099e-06,
+      "loss": 115.1971,
+      "step": 23190
+    },
+    {
+      "epoch": 0.19191793853662573,
+      "grad_norm": 858.6890869140625,
+      "learning_rate": 9.768717268201768e-06,
+      "loss": 145.7314,
+      "step": 23200
+    },
+    {
+      "epoch": 0.19200066178599495,
+      "grad_norm": 2752.160888671875,
+      "learning_rate": 9.768293091505491e-06,
+      "loss": 128.7382,
+      "step": 23210
+    },
+    {
+      "epoch": 0.19208338503536418,
+      "grad_norm": 971.1498413085938,
+      "learning_rate": 9.767868535418014e-06,
+      "loss": 126.5057,
+      "step": 23220
+    },
+    {
+      "epoch": 0.19216610828473343,
+      "grad_norm": 793.6380615234375,
+      "learning_rate": 9.767443599973122e-06,
+      "loss": 108.5758,
+      "step": 23230
+    },
+    {
+      "epoch": 0.19224883153410266,
+      "grad_norm": 1140.6827392578125,
+      "learning_rate": 9.76701828520462e-06,
+      "loss": 117.5244,
+      "step": 23240
+    },
+    {
+      "epoch": 0.1923315547834719,
+      "grad_norm": 840.5177001953125,
+      "learning_rate": 9.766592591146353e-06,
+      "loss": 107.5938,
+      "step": 23250
+    },
+    {
+      "epoch": 0.19241427803284114,
+      "grad_norm": 1344.9217529296875,
+      "learning_rate": 9.766166517832188e-06,
+      "loss": 120.1126,
+      "step": 23260
+    },
+    {
+      "epoch": 0.19249700128221037,
+      "grad_norm": 1015.5147705078125,
+      "learning_rate": 9.765740065296025e-06,
+      "loss": 131.5013,
+      "step": 23270
+    },
+    {
+      "epoch": 0.1925797245315796,
+      "grad_norm": 574.64306640625,
+      "learning_rate": 9.765313233571798e-06,
+      "loss": 131.5927,
+      "step": 23280
+    },
+    {
+      "epoch": 0.19266244778094885,
+      "grad_norm": 1062.6119384765625,
+      "learning_rate": 9.76488602269347e-06,
+      "loss": 162.6192,
+      "step": 23290
+    },
+    {
+      "epoch": 0.19274517103031807,
+      "grad_norm": 1031.8822021484375,
+      "learning_rate": 9.764458432695026e-06,
+      "loss": 91.6339,
+      "step": 23300
+    },
+    {
+      "epoch": 0.1928278942796873,
+      "grad_norm": 715.5499877929688,
+      "learning_rate": 9.76403046361049e-06,
+      "loss": 98.5823,
+      "step": 23310
+    },
+    {
+      "epoch": 0.19291061752905656,
+      "grad_norm": 707.9616088867188,
+      "learning_rate": 9.763602115473914e-06,
+      "loss": 129.4643,
+      "step": 23320
+    },
+    {
+      "epoch": 0.19299334077842578,
+      "grad_norm": 1023.2615966796875,
+      "learning_rate": 9.763173388319381e-06,
+      "loss": 108.6447,
+      "step": 23330
+    },
+    {
+      "epoch": 0.193076064027795,
+      "grad_norm": 745.7240600585938,
+      "learning_rate": 9.762744282181e-06,
+      "loss": 160.1926,
+      "step": 23340
+    },
+    {
+      "epoch": 0.19315878727716423,
+      "grad_norm": 1169.117431640625,
+      "learning_rate": 9.762314797092916e-06,
+      "loss": 136.8101,
+      "step": 23350
+    },
+    {
+      "epoch": 0.1932415105265335,
+      "grad_norm": 2356.3876953125,
+      "learning_rate": 9.761884933089301e-06,
+      "loss": 149.6885,
+      "step": 23360
+    },
+    {
+      "epoch": 0.19332423377590272,
+      "grad_norm": 748.1530151367188,
+      "learning_rate": 9.761454690204352e-06,
+      "loss": 90.2527,
+      "step": 23370
+    },
+    {
+      "epoch": 0.19340695702527194,
+      "grad_norm": 773.795654296875,
+      "learning_rate": 9.76102406847231e-06,
+      "loss": 133.7801,
+      "step": 23380
+    },
+    {
+      "epoch": 0.1934896802746412,
+      "grad_norm": 785.1251831054688,
+      "learning_rate": 9.760593067927428e-06,
+      "loss": 92.1381,
+      "step": 23390
+    },
+    {
+      "epoch": 0.19357240352401042,
+      "grad_norm": 987.9070434570312,
+      "learning_rate": 9.760161688604008e-06,
+      "loss": 152.8994,
+      "step": 23400
+    },
+    {
+      "epoch": 0.19365512677337965,
+      "grad_norm": 1091.9166259765625,
+      "learning_rate": 9.759729930536367e-06,
+      "loss": 106.4156,
+      "step": 23410
+    },
+    {
+      "epoch": 0.1937378500227489,
+      "grad_norm": 828.226806640625,
+      "learning_rate": 9.75929779375886e-06,
+      "loss": 127.5234,
+      "step": 23420
+    },
+    {
+      "epoch": 0.19382057327211813,
+      "grad_norm": 1164.93359375,
+      "learning_rate": 9.75886527830587e-06,
+      "loss": 147.2189,
+      "step": 23430
+    },
+    {
+      "epoch": 0.19390329652148736,
+      "grad_norm": 1556.861572265625,
+      "learning_rate": 9.75843238421181e-06,
+      "loss": 132.9836,
+      "step": 23440
+    },
+    {
+      "epoch": 0.1939860197708566,
+      "grad_norm": 1154.7764892578125,
+      "learning_rate": 9.757999111511121e-06,
+      "loss": 131.7635,
+      "step": 23450
+    },
+    {
+      "epoch": 0.19406874302022584,
+      "grad_norm": 1170.080078125,
+      "learning_rate": 9.757565460238281e-06,
+      "loss": 120.619,
+      "step": 23460
+    },
+    {
+      "epoch": 0.19415146626959506,
+      "grad_norm": 1463.3924560546875,
+      "learning_rate": 9.757131430427791e-06,
+      "loss": 129.6818,
+      "step": 23470
+    },
+    {
+      "epoch": 0.19423418951896432,
+      "grad_norm": 1100.9061279296875,
+      "learning_rate": 9.756697022114185e-06,
+      "loss": 108.3849,
+      "step": 23480
+    },
+    {
+      "epoch": 0.19431691276833354,
+      "grad_norm": 1876.0504150390625,
+      "learning_rate": 9.756262235332029e-06,
+      "loss": 103.7323,
+      "step": 23490
+    },
+    {
+      "epoch": 0.19439963601770277,
+      "grad_norm": 974.3872680664062,
+      "learning_rate": 9.755827070115915e-06,
+      "loss": 105.9175,
+      "step": 23500
+    },
+    {
+      "epoch": 0.194482359267072,
+      "grad_norm": 888.02099609375,
+      "learning_rate": 9.755391526500466e-06,
+      "loss": 108.9482,
+      "step": 23510
+    },
+    {
+      "epoch": 0.19456508251644125,
+      "grad_norm": 863.0517578125,
+      "learning_rate": 9.75495560452034e-06,
+      "loss": 131.4029,
+      "step": 23520
+    },
+    {
+      "epoch": 0.19464780576581048,
+      "grad_norm": 1131.92431640625,
+      "learning_rate": 9.754519304210214e-06,
+      "loss": 122.4951,
+      "step": 23530
+    },
+    {
+      "epoch": 0.1947305290151797,
+      "grad_norm": 2271.8134765625,
+      "learning_rate": 9.754082625604812e-06,
+      "loss": 121.6546,
+      "step": 23540
+    },
+    {
+      "epoch": 0.19481325226454896,
+      "grad_norm": 919.2047119140625,
+      "learning_rate": 9.753645568738872e-06,
+      "loss": 138.8903,
+      "step": 23550
+    },
+    {
+      "epoch": 0.19489597551391818,
+      "grad_norm": 840.5191650390625,
+      "learning_rate": 9.75320813364717e-06,
+      "loss": 91.0436,
+      "step": 23560
+    },
+    {
+      "epoch": 0.1949786987632874,
+      "grad_norm": 621.6953125,
+      "learning_rate": 9.752770320364512e-06,
+      "loss": 128.6445,
+      "step": 23570
+    },
+    {
+      "epoch": 0.19506142201265667,
+      "grad_norm": 754.2584838867188,
+      "learning_rate": 9.752332128925732e-06,
+      "loss": 111.4495,
+      "step": 23580
+    },
+    {
+      "epoch": 0.1951441452620259,
+      "grad_norm": 1024.14501953125,
+      "learning_rate": 9.751893559365693e-06,
+      "loss": 170.7815,
+      "step": 23590
+    },
+    {
+      "epoch": 0.19522686851139512,
+      "grad_norm": 1205.4876708984375,
+      "learning_rate": 9.751454611719294e-06,
+      "loss": 121.6351,
+      "step": 23600
+    },
+    {
+      "epoch": 0.19530959176076437,
+      "grad_norm": 1503.17236328125,
+      "learning_rate": 9.751015286021455e-06,
+      "loss": 164.3172,
+      "step": 23610
+    },
+    {
+      "epoch": 0.1953923150101336,
+      "grad_norm": 937.5809936523438,
+      "learning_rate": 9.750575582307136e-06,
+      "loss": 160.1191,
+      "step": 23620
+    },
+    {
+      "epoch": 0.19547503825950283,
+      "grad_norm": 983.7006225585938,
+      "learning_rate": 9.75013550061132e-06,
+      "loss": 120.9869,
+      "step": 23630
+    },
+    {
+      "epoch": 0.19555776150887208,
+      "grad_norm": 1772.2957763671875,
+      "learning_rate": 9.749695040969022e-06,
+      "loss": 114.3774,
+      "step": 23640
+    },
+    {
+      "epoch": 0.1956404847582413,
+      "grad_norm": 924.7174072265625,
+      "learning_rate": 9.749254203415288e-06,
+      "loss": 124.1198,
+      "step": 23650
+    },
+    {
+      "epoch": 0.19572320800761053,
+      "grad_norm": 972.820068359375,
+      "learning_rate": 9.748812987985193e-06,
+      "loss": 123.3391,
+      "step": 23660
+    },
+    {
+      "epoch": 0.1958059312569798,
+      "grad_norm": 1330.291748046875,
+      "learning_rate": 9.748371394713842e-06,
+      "loss": 126.0948,
+      "step": 23670
+    },
+    {
+      "epoch": 0.195888654506349,
+      "grad_norm": 1056.2666015625,
+      "learning_rate": 9.747929423636372e-06,
+      "loss": 128.4867,
+      "step": 23680
+    },
+    {
+      "epoch": 0.19597137775571824,
+      "grad_norm": 670.9091186523438,
+      "learning_rate": 9.74748707478795e-06,
+      "loss": 140.6503,
+      "step": 23690
+    },
+    {
+      "epoch": 0.19605410100508747,
+      "grad_norm": 1226.66162109375,
+      "learning_rate": 9.747044348203766e-06,
+      "loss": 130.9208,
+      "step": 23700
+    },
+    {
+      "epoch": 0.19613682425445672,
+      "grad_norm": 907.8272705078125,
+      "learning_rate": 9.74660124391905e-06,
+      "loss": 127.5544,
+      "step": 23710
+    },
+    {
+      "epoch": 0.19621954750382595,
+      "grad_norm": 898.1268920898438,
+      "learning_rate": 9.746157761969058e-06,
+      "loss": 164.2529,
+      "step": 23720
+    },
+    {
+      "epoch": 0.19630227075319517,
+      "grad_norm": 1068.112060546875,
+      "learning_rate": 9.745713902389074e-06,
+      "loss": 143.5591,
+      "step": 23730
+    },
+    {
+      "epoch": 0.19638499400256443,
+      "grad_norm": 1102.2633056640625,
+      "learning_rate": 9.745269665214415e-06,
+      "loss": 122.1263,
+      "step": 23740
+    },
+    {
+      "epoch": 0.19646771725193365,
+      "grad_norm": 556.4323120117188,
+      "learning_rate": 9.744825050480425e-06,
+      "loss": 121.492,
+      "step": 23750
+    },
+    {
+      "epoch": 0.19655044050130288,
+      "grad_norm": 1045.6256103515625,
+      "learning_rate": 9.744380058222483e-06,
+      "loss": 131.9083,
+      "step": 23760
+    },
+    {
+      "epoch": 0.19663316375067214,
+      "grad_norm": 1136.8719482421875,
+      "learning_rate": 9.743934688475994e-06,
+      "loss": 115.8493,
+      "step": 23770
+    },
+    {
+      "epoch": 0.19671588700004136,
+      "grad_norm": 962.9896850585938,
+      "learning_rate": 9.743488941276394e-06,
+      "loss": 117.902,
+      "step": 23780
+    },
+    {
+      "epoch": 0.1967986102494106,
+      "grad_norm": 1086.8525390625,
+      "learning_rate": 9.743042816659147e-06,
+      "loss": 134.2704,
+      "step": 23790
+    },
+    {
+      "epoch": 0.19688133349877984,
+      "grad_norm": 1227.1202392578125,
+      "learning_rate": 9.742596314659751e-06,
+      "loss": 127.5619,
+      "step": 23800
+    },
+    {
+      "epoch": 0.19696405674814907,
+      "grad_norm": 1608.360595703125,
+      "learning_rate": 9.742149435313732e-06,
+      "loss": 120.5991,
+      "step": 23810
+    },
+    {
+      "epoch": 0.1970467799975183,
+      "grad_norm": 664.271728515625,
+      "learning_rate": 9.741702178656647e-06,
+      "loss": 151.2022,
+      "step": 23820
+    },
+    {
+      "epoch": 0.19712950324688755,
+      "grad_norm": 1345.0858154296875,
+      "learning_rate": 9.74125454472408e-06,
+      "loss": 136.8234,
+      "step": 23830
+    },
+    {
+      "epoch": 0.19721222649625678,
+      "grad_norm": 732.8429565429688,
+      "learning_rate": 9.740806533551647e-06,
+      "loss": 234.7962,
+      "step": 23840
+    },
+    {
+      "epoch": 0.197294949745626,
+      "grad_norm": 1542.8756103515625,
+      "learning_rate": 9.740358145174999e-06,
+      "loss": 130.3346,
+      "step": 23850
+    },
+    {
+      "epoch": 0.19737767299499526,
+      "grad_norm": 821.8853149414062,
+      "learning_rate": 9.739909379629805e-06,
+      "loss": 139.0029,
+      "step": 23860
+    },
+    {
+      "epoch": 0.19746039624436448,
+      "grad_norm": 812.029541015625,
+      "learning_rate": 9.739460236951778e-06,
+      "loss": 127.8701,
+      "step": 23870
+    },
+    {
+      "epoch": 0.1975431194937337,
+      "grad_norm": 969.9833984375,
+      "learning_rate": 9.739010717176649e-06,
+      "loss": 99.5856,
+      "step": 23880
+    },
+    {
+      "epoch": 0.19762584274310294,
+      "grad_norm": 1117.0540771484375,
+      "learning_rate": 9.738560820340189e-06,
+      "loss": 114.7225,
+      "step": 23890
+    },
+    {
+      "epoch": 0.1977085659924722,
+      "grad_norm": 601.694091796875,
+      "learning_rate": 9.738110546478188e-06,
+      "loss": 123.2165,
+      "step": 23900
+    },
+    {
+      "epoch": 0.19779128924184142,
+      "grad_norm": 1268.921630859375,
+      "learning_rate": 9.737659895626478e-06,
+      "loss": 130.8849,
+      "step": 23910
+    },
+    {
+      "epoch": 0.19787401249121064,
+      "grad_norm": 1429.4453125,
+      "learning_rate": 9.737208867820914e-06,
+      "loss": 123.7755,
+      "step": 23920
+    },
+    {
+      "epoch": 0.1979567357405799,
+      "grad_norm": 654.82080078125,
+      "learning_rate": 9.736757463097378e-06,
+      "loss": 152.5086,
+      "step": 23930
+    },
+    {
+      "epoch": 0.19803945898994912,
+      "grad_norm": 852.7525024414062,
+      "learning_rate": 9.736305681491792e-06,
+      "loss": 116.6475,
+      "step": 23940
+    },
+    {
+      "epoch": 0.19812218223931835,
+      "grad_norm": 1039.8662109375,
+      "learning_rate": 9.735853523040098e-06,
+      "loss": 138.9281,
+      "step": 23950
+    },
+    {
+      "epoch": 0.1982049054886876,
+      "grad_norm": 1178.4395751953125,
+      "learning_rate": 9.735400987778274e-06,
+      "loss": 143.172,
+      "step": 23960
+    },
+    {
+      "epoch": 0.19828762873805683,
+      "grad_norm": 1232.4547119140625,
+      "learning_rate": 9.734948075742328e-06,
+      "loss": 121.5333,
+      "step": 23970
+    },
+    {
+      "epoch": 0.19837035198742606,
+      "grad_norm": 865.7545166015625,
+      "learning_rate": 9.734494786968293e-06,
+      "loss": 154.3453,
+      "step": 23980
+    },
+    {
+      "epoch": 0.1984530752367953,
+      "grad_norm": 669.6997680664062,
+      "learning_rate": 9.734041121492235e-06,
+      "loss": 144.5734,
+      "step": 23990
+    },
+    {
+      "epoch": 0.19853579848616454,
+      "grad_norm": 1144.2825927734375,
+      "learning_rate": 9.733587079350254e-06,
+      "loss": 107.7752,
+      "step": 24000
+    },
+    {
+      "epoch": 0.19861852173553377,
+      "grad_norm": 2160.167236328125,
+      "learning_rate": 9.73313266057847e-06,
+      "loss": 129.0158,
+      "step": 24010
+    },
+    {
+      "epoch": 0.19870124498490302,
+      "grad_norm": 1064.593994140625,
+      "learning_rate": 9.732677865213044e-06,
+      "loss": 141.8949,
+      "step": 24020
+    },
+    {
+      "epoch": 0.19878396823427225,
+      "grad_norm": 1279.8077392578125,
+      "learning_rate": 9.73222269329016e-06,
+      "loss": 153.0952,
+      "step": 24030
+    },
+    {
+      "epoch": 0.19886669148364147,
+      "grad_norm": 950.4849853515625,
+      "learning_rate": 9.731767144846034e-06,
+      "loss": 146.848,
+      "step": 24040
+    },
+    {
+      "epoch": 0.1989494147330107,
+      "grad_norm": 1283.0186767578125,
+      "learning_rate": 9.731311219916912e-06,
+      "loss": 150.6841,
+      "step": 24050
+    },
+    {
+      "epoch": 0.19903213798237995,
+      "grad_norm": 704.3133544921875,
+      "learning_rate": 9.730854918539072e-06,
+      "loss": 149.4796,
+      "step": 24060
+    },
+    {
+      "epoch": 0.19911486123174918,
+      "grad_norm": 846.5486450195312,
+      "learning_rate": 9.730398240748816e-06,
+      "loss": 138.2696,
+      "step": 24070
+    },
+    {
+      "epoch": 0.1991975844811184,
+      "grad_norm": 779.280029296875,
+      "learning_rate": 9.729941186582482e-06,
+      "loss": 114.9246,
+      "step": 24080
+    },
+    {
+      "epoch": 0.19928030773048766,
+      "grad_norm": 1280.6629638671875,
+      "learning_rate": 9.729483756076436e-06,
+      "loss": 96.8995,
+      "step": 24090
+    },
+    {
+      "epoch": 0.1993630309798569,
+      "grad_norm": 626.4833374023438,
+      "learning_rate": 9.729025949267072e-06,
+      "loss": 117.7161,
+      "step": 24100
+    },
+    {
+      "epoch": 0.1994457542292261,
+      "grad_norm": 866.7782592773438,
+      "learning_rate": 9.728567766190817e-06,
+      "loss": 110.1413,
+      "step": 24110
+    },
+    {
+      "epoch": 0.19952847747859537,
+      "grad_norm": 1430.4652099609375,
+      "learning_rate": 9.728109206884125e-06,
+      "loss": 122.4254,
+      "step": 24120
+    },
+    {
+      "epoch": 0.1996112007279646,
+      "grad_norm": 1178.0247802734375,
+      "learning_rate": 9.727650271383485e-06,
+      "loss": 131.602,
+      "step": 24130
+    },
+    {
+      "epoch": 0.19969392397733382,
+      "grad_norm": 944.5925903320312,
+      "learning_rate": 9.727190959725407e-06,
+      "loss": 91.2385,
+      "step": 24140
+    },
+    {
+      "epoch": 0.19977664722670307,
+      "grad_norm": 1111.4093017578125,
+      "learning_rate": 9.72673127194644e-06,
+      "loss": 185.1948,
+      "step": 24150
+    },
+    {
+      "epoch": 0.1998593704760723,
+      "grad_norm": 1125.5455322265625,
+      "learning_rate": 9.72627120808316e-06,
+      "loss": 157.4787,
+      "step": 24160
+    },
+    {
+      "epoch": 0.19994209372544153,
+      "grad_norm": 1090.2935791015625,
+      "learning_rate": 9.725810768172169e-06,
+      "loss": 106.9974,
+      "step": 24170
+    },
+    {
+      "epoch": 0.20002481697481078,
+      "grad_norm": 620.5934448242188,
+      "learning_rate": 9.725349952250105e-06,
+      "loss": 115.3673,
+      "step": 24180
+    },
+    {
+      "epoch": 0.20010754022418,
+      "grad_norm": 1263.1217041015625,
+      "learning_rate": 9.724888760353631e-06,
+      "loss": 148.2358,
+      "step": 24190
+    },
+    {
+      "epoch": 0.20019026347354923,
+      "grad_norm": 1106.949951171875,
+      "learning_rate": 9.72442719251944e-06,
+      "loss": 170.5474,
+      "step": 24200
+    },
+    {
+      "epoch": 0.2002729867229185,
+      "grad_norm": 1334.36962890625,
+      "learning_rate": 9.723965248784264e-06,
+      "loss": 149.0543,
+      "step": 24210
+    },
+    {
+      "epoch": 0.20035570997228772,
+      "grad_norm": 1211.2431640625,
+      "learning_rate": 9.723502929184851e-06,
+      "loss": 126.5367,
+      "step": 24220
+    },
+    {
+      "epoch": 0.20043843322165694,
+      "grad_norm": 863.3626708984375,
+      "learning_rate": 9.723040233757987e-06,
+      "loss": 142.1387,
+      "step": 24230
+    },
+    {
+      "epoch": 0.20052115647102617,
+      "grad_norm": 1154.0657958984375,
+      "learning_rate": 9.722577162540489e-06,
+      "loss": 123.5952,
+      "step": 24240
+    },
+    {
+      "epoch": 0.20060387972039542,
+      "grad_norm": 1012.69140625,
+      "learning_rate": 9.7221137155692e-06,
+      "loss": 110.8877,
+      "step": 24250
+    },
+    {
+      "epoch": 0.20068660296976465,
+      "grad_norm": 1386.55859375,
+      "learning_rate": 9.721649892880995e-06,
+      "loss": 170.2592,
+      "step": 24260
+    },
+    {
+      "epoch": 0.20076932621913388,
+      "grad_norm": 1342.9095458984375,
+      "learning_rate": 9.721185694512776e-06,
+      "loss": 134.4591,
+      "step": 24270
+    },
+    {
+      "epoch": 0.20085204946850313,
+      "grad_norm": 392.2961120605469,
+      "learning_rate": 9.720721120501478e-06,
+      "loss": 134.1603,
+      "step": 24280
+    },
+    {
+      "epoch": 0.20093477271787236,
+      "grad_norm": 826.9261474609375,
+      "learning_rate": 9.720256170884066e-06,
+      "loss": 101.2419,
+      "step": 24290
+    },
+    {
+      "epoch": 0.20101749596724158,
+      "grad_norm": 791.0903930664062,
+      "learning_rate": 9.719790845697534e-06,
+      "loss": 129.4301,
+      "step": 24300
+    },
+    {
+      "epoch": 0.20110021921661084,
+      "grad_norm": 1188.9549560546875,
+      "learning_rate": 9.719325144978907e-06,
+      "loss": 145.1762,
+      "step": 24310
+    },
+    {
+      "epoch": 0.20118294246598006,
+      "grad_norm": 1141.0123291015625,
+      "learning_rate": 9.718859068765234e-06,
+      "loss": 107.9242,
+      "step": 24320
+    },
+    {
+      "epoch": 0.2012656657153493,
+      "grad_norm": 1065.128173828125,
+      "learning_rate": 9.718392617093602e-06,
+      "loss": 113.2393,
+      "step": 24330
+    },
+    {
+      "epoch": 0.20134838896471854,
+      "grad_norm": 1253.3646240234375,
+      "learning_rate": 9.717925790001125e-06,
+      "loss": 77.6727,
+      "step": 24340
+    },
+    {
+      "epoch": 0.20143111221408777,
+      "grad_norm": 1093.85498046875,
+      "learning_rate": 9.717458587524946e-06,
+      "loss": 112.4173,
+      "step": 24350
+    },
+    {
+      "epoch": 0.201513835463457,
+      "grad_norm": 787.4635009765625,
+      "learning_rate": 9.716991009702236e-06,
+      "loss": 114.5591,
+      "step": 24360
+    },
+    {
+      "epoch": 0.20159655871282625,
+      "grad_norm": 926.3908081054688,
+      "learning_rate": 9.7165230565702e-06,
+      "loss": 123.4194,
+      "step": 24370
+    },
+    {
+      "epoch": 0.20167928196219548,
+      "grad_norm": 792.05859375,
+      "learning_rate": 9.71605472816607e-06,
+      "loss": 128.6131,
+      "step": 24380
+    },
+    {
+      "epoch": 0.2017620052115647,
+      "grad_norm": 835.9586791992188,
+      "learning_rate": 9.71558602452711e-06,
+      "loss": 125.1441,
+      "step": 24390
+    },
+    {
+      "epoch": 0.20184472846093396,
+      "grad_norm": 1247.4454345703125,
+      "learning_rate": 9.71511694569061e-06,
+      "loss": 166.359,
+      "step": 24400
+    },
+    {
+      "epoch": 0.20192745171030319,
+      "grad_norm": 828.7860107421875,
+      "learning_rate": 9.714647491693897e-06,
+      "loss": 140.4719,
+      "step": 24410
+    },
+    {
+      "epoch": 0.2020101749596724,
+      "grad_norm": 843.8282470703125,
+      "learning_rate": 9.714177662574316e-06,
+      "loss": 101.7997,
+      "step": 24420
+    },
+    {
+      "epoch": 0.20209289820904164,
+      "grad_norm": 967.5045776367188,
+      "learning_rate": 9.713707458369258e-06,
+      "loss": 124.0601,
+      "step": 24430
+    },
+    {
+      "epoch": 0.2021756214584109,
+      "grad_norm": 692.8041381835938,
+      "learning_rate": 9.713236879116127e-06,
+      "loss": 120.8918,
+      "step": 24440
+    },
+    {
+      "epoch": 0.20225834470778012,
+      "grad_norm": 1022.600341796875,
+      "learning_rate": 9.71276592485237e-06,
+      "loss": 128.2274,
+      "step": 24450
+    },
+    {
+      "epoch": 0.20234106795714935,
+      "grad_norm": 1309.6890869140625,
+      "learning_rate": 9.712294595615458e-06,
+      "loss": 134.2852,
+      "step": 24460
+    },
+    {
+      "epoch": 0.2024237912065186,
+      "grad_norm": 596.5182495117188,
+      "learning_rate": 9.711822891442887e-06,
+      "loss": 141.3561,
+      "step": 24470
+    },
+    {
+      "epoch": 0.20250651445588783,
+      "grad_norm": 1456.3428955078125,
+      "learning_rate": 9.711350812372198e-06,
+      "loss": 166.3525,
+      "step": 24480
+    },
+    {
+      "epoch": 0.20258923770525705,
+      "grad_norm": 1364.7294921875,
+      "learning_rate": 9.710878358440945e-06,
+      "loss": 178.6096,
+      "step": 24490
+    },
+    {
+      "epoch": 0.2026719609546263,
+      "grad_norm": 941.8969116210938,
+      "learning_rate": 9.710405529686722e-06,
+      "loss": 154.1988,
+      "step": 24500
+    },
+    {
+      "epoch": 0.20275468420399553,
+      "grad_norm": 1231.3597412109375,
+      "learning_rate": 9.709932326147147e-06,
+      "loss": 109.8311,
+      "step": 24510
+    },
+    {
+      "epoch": 0.20283740745336476,
+      "grad_norm": 981.5293579101562,
+      "learning_rate": 9.709458747859874e-06,
+      "loss": 106.3588,
+      "step": 24520
+    },
+    {
+      "epoch": 0.20292013070273401,
+      "grad_norm": 882.2429809570312,
+      "learning_rate": 9.708984794862581e-06,
+      "loss": 106.4644,
+      "step": 24530
+    },
+    {
+      "epoch": 0.20300285395210324,
+      "grad_norm": 1536.478271484375,
+      "learning_rate": 9.708510467192981e-06,
+      "loss": 115.3478,
+      "step": 24540
+    },
+    {
+      "epoch": 0.20308557720147247,
+      "grad_norm": 480.1463928222656,
+      "learning_rate": 9.70803576488881e-06,
+      "loss": 131.8311,
+      "step": 24550
+    },
+    {
+      "epoch": 0.20316830045084172,
+      "grad_norm": 751.3448486328125,
+      "learning_rate": 9.707560687987843e-06,
+      "loss": 100.0164,
+      "step": 24560
+    },
+    {
+      "epoch": 0.20325102370021095,
+      "grad_norm": 983.8947143554688,
+      "learning_rate": 9.707085236527873e-06,
+      "loss": 124.7845,
+      "step": 24570
+    },
+    {
+      "epoch": 0.20333374694958017,
+      "grad_norm": 778.719970703125,
+      "learning_rate": 9.706609410546736e-06,
+      "loss": 70.2131,
+      "step": 24580
+    },
+    {
+      "epoch": 0.20341647019894943,
+      "grad_norm": 799.94580078125,
+      "learning_rate": 9.706133210082288e-06,
+      "loss": 107.5597,
+      "step": 24590
+    },
+    {
+      "epoch": 0.20349919344831865,
+      "grad_norm": 775.4839477539062,
+      "learning_rate": 9.705656635172418e-06,
+      "loss": 123.562,
+      "step": 24600
+    },
+    {
+      "epoch": 0.20358191669768788,
+      "grad_norm": 906.2435302734375,
+      "learning_rate": 9.705179685855048e-06,
+      "loss": 141.3586,
+      "step": 24610
+    },
+    {
+      "epoch": 0.2036646399470571,
+      "grad_norm": 572.1375732421875,
+      "learning_rate": 9.704702362168121e-06,
+      "loss": 174.9464,
+      "step": 24620
+    },
+    {
+      "epoch": 0.20374736319642636,
+      "grad_norm": 1068.9876708984375,
+      "learning_rate": 9.704224664149621e-06,
+      "loss": 122.0626,
+      "step": 24630
+    },
+    {
+      "epoch": 0.2038300864457956,
+      "grad_norm": 606.3220825195312,
+      "learning_rate": 9.703746591837552e-06,
+      "loss": 109.6907,
+      "step": 24640
+    },
+    {
+      "epoch": 0.20391280969516482,
+      "grad_norm": 1032.2161865234375,
+      "learning_rate": 9.703268145269957e-06,
+      "loss": 126.0382,
+      "step": 24650
+    },
+    {
+      "epoch": 0.20399553294453407,
+      "grad_norm": 1022.0555419921875,
+      "learning_rate": 9.702789324484898e-06,
+      "loss": 188.598,
+      "step": 24660
+    },
+    {
+      "epoch": 0.2040782561939033,
+      "grad_norm": 1047.5244140625,
+      "learning_rate": 9.702310129520476e-06,
+      "loss": 122.4435,
+      "step": 24670
+    },
+    {
+      "epoch": 0.20416097944327252,
+      "grad_norm": 1051.2161865234375,
+      "learning_rate": 9.701830560414817e-06,
+      "loss": 144.9207,
+      "step": 24680
+    },
+    {
+      "epoch": 0.20424370269264178,
+      "grad_norm": 1291.77099609375,
+      "learning_rate": 9.701350617206081e-06,
+      "loss": 141.4524,
+      "step": 24690
+    },
+    {
+      "epoch": 0.204326425942011,
+      "grad_norm": 1048.1988525390625,
+      "learning_rate": 9.700870299932453e-06,
+      "loss": 133.1601,
+      "step": 24700
+    },
+    {
+      "epoch": 0.20440914919138023,
+      "grad_norm": 1616.698486328125,
+      "learning_rate": 9.700389608632146e-06,
+      "loss": 159.5847,
+      "step": 24710
+    },
+    {
+      "epoch": 0.20449187244074948,
+      "grad_norm": 863.6107788085938,
+      "learning_rate": 9.699908543343413e-06,
+      "loss": 135.6566,
+      "step": 24720
+    },
+    {
+      "epoch": 0.2045745956901187,
+      "grad_norm": 1772.513916015625,
+      "learning_rate": 9.699427104104525e-06,
+      "loss": 202.6929,
+      "step": 24730
+    },
+    {
+      "epoch": 0.20465731893948794,
+      "grad_norm": 822.7219848632812,
+      "learning_rate": 9.698945290953789e-06,
+      "loss": 140.5493,
+      "step": 24740
+    },
+    {
+      "epoch": 0.2047400421888572,
+      "grad_norm": 811.2693481445312,
+      "learning_rate": 9.698463103929542e-06,
+      "loss": 81.8386,
+      "step": 24750
+    },
+    {
+      "epoch": 0.20482276543822642,
+      "grad_norm": 530.8605346679688,
+      "learning_rate": 9.69798054307015e-06,
+      "loss": 106.7644,
+      "step": 24760
+    },
+    {
+      "epoch": 0.20490548868759564,
+      "grad_norm": 2197.434326171875,
+      "learning_rate": 9.697497608414007e-06,
+      "loss": 189.2195,
+      "step": 24770
+    },
+    {
+      "epoch": 0.20498821193696487,
+      "grad_norm": 1628.6551513671875,
+      "learning_rate": 9.697014299999536e-06,
+      "loss": 110.0945,
+      "step": 24780
+    },
+    {
+      "epoch": 0.20507093518633412,
+      "grad_norm": 988.3732299804688,
+      "learning_rate": 9.696530617865197e-06,
+      "loss": 147.2918,
+      "step": 24790
+    },
+    {
+      "epoch": 0.20515365843570335,
+      "grad_norm": 934.0083618164062,
+      "learning_rate": 9.696046562049469e-06,
+      "loss": 124.3846,
+      "step": 24800
+    },
+    {
+      "epoch": 0.20523638168507258,
+      "grad_norm": 1064.4521484375,
+      "learning_rate": 9.695562132590865e-06,
+      "loss": 130.5005,
+      "step": 24810
+    },
+    {
+      "epoch": 0.20531910493444183,
+      "grad_norm": 936.8753662109375,
+      "learning_rate": 9.695077329527936e-06,
+      "loss": 79.4409,
+      "step": 24820
+    },
+    {
+      "epoch": 0.20540182818381106,
+      "grad_norm": 2023.431396484375,
+      "learning_rate": 9.694592152899249e-06,
+      "loss": 135.3671,
+      "step": 24830
+    },
+    {
+      "epoch": 0.20548455143318028,
+      "grad_norm": 1526.9307861328125,
+      "learning_rate": 9.694106602743411e-06,
+      "loss": 141.677,
+      "step": 24840
+    },
+    {
+      "epoch": 0.20556727468254954,
+      "grad_norm": 1296.6025390625,
+      "learning_rate": 9.693620679099055e-06,
+      "loss": 106.9513,
+      "step": 24850
+    },
+    {
+      "epoch": 0.20564999793191877,
+      "grad_norm": 675.1536865234375,
+      "learning_rate": 9.693134382004839e-06,
+      "loss": 122.0934,
+      "step": 24860
+    },
+    {
+      "epoch": 0.205732721181288,
+      "grad_norm": 1233.307373046875,
+      "learning_rate": 9.69264771149946e-06,
+      "loss": 152.0633,
+      "step": 24870
+    },
+    {
+      "epoch": 0.20581544443065725,
+      "grad_norm": 2084.801025390625,
+      "learning_rate": 9.692160667621639e-06,
+      "loss": 191.7889,
+      "step": 24880
+    },
+    {
+      "epoch": 0.20589816768002647,
+      "grad_norm": 1063.7586669921875,
+      "learning_rate": 9.69167325041013e-06,
+      "loss": 98.7458,
+      "step": 24890
+    },
+    {
+      "epoch": 0.2059808909293957,
+      "grad_norm": 1160.4176025390625,
+      "learning_rate": 9.69118545990371e-06,
+      "loss": 136.4698,
+      "step": 24900
+    },
+    {
+      "epoch": 0.20606361417876495,
+      "grad_norm": 1240.2447509765625,
+      "learning_rate": 9.690697296141194e-06,
+      "loss": 131.5934,
+      "step": 24910
+    },
+    {
+      "epoch": 0.20614633742813418,
+      "grad_norm": 1285.3697509765625,
+      "learning_rate": 9.690208759161418e-06,
+      "loss": 143.4328,
+      "step": 24920
+    },
+    {
+      "epoch": 0.2062290606775034,
+      "grad_norm": 899.1880493164062,
+      "learning_rate": 9.689719849003261e-06,
+      "loss": 150.501,
+      "step": 24930
+    },
+    {
+      "epoch": 0.20631178392687266,
+      "grad_norm": 1172.1856689453125,
+      "learning_rate": 9.689230565705617e-06,
+      "loss": 88.1189,
+      "step": 24940
+    },
+    {
+      "epoch": 0.2063945071762419,
+      "grad_norm": 940.1471557617188,
+      "learning_rate": 9.688740909307416e-06,
+      "loss": 126.9263,
+      "step": 24950
+    },
+    {
+      "epoch": 0.2064772304256111,
+      "grad_norm": 1031.7501220703125,
+      "learning_rate": 9.68825087984762e-06,
+      "loss": 145.9387,
+      "step": 24960
+    },
+    {
+      "epoch": 0.20655995367498034,
+      "grad_norm": 747.2893676757812,
+      "learning_rate": 9.687760477365217e-06,
+      "loss": 111.3116,
+      "step": 24970
+    },
+    {
+      "epoch": 0.2066426769243496,
+      "grad_norm": 1009.5612182617188,
+      "learning_rate": 9.687269701899228e-06,
+      "loss": 122.0988,
+      "step": 24980
+    },
+    {
+      "epoch": 0.20672540017371882,
+      "grad_norm": 1089.2601318359375,
+      "learning_rate": 9.6867785534887e-06,
+      "loss": 118.2881,
+      "step": 24990
+    },
+    {
+      "epoch": 0.20680812342308805,
+      "grad_norm": 1918.05908203125,
+      "learning_rate": 9.686287032172712e-06,
+      "loss": 119.0276,
+      "step": 25000
+    },
+    {
+      "epoch": 0.2068908466724573,
+      "grad_norm": 1754.23486328125,
+      "learning_rate": 9.685795137990372e-06,
+      "loss": 149.8688,
+      "step": 25010
+    },
+    {
+      "epoch": 0.20697356992182653,
+      "grad_norm": 701.3511352539062,
+      "learning_rate": 9.685302870980819e-06,
+      "loss": 111.5625,
+      "step": 25020
+    },
+    {
+      "epoch": 0.20705629317119575,
+      "grad_norm": 1214.149658203125,
+      "learning_rate": 9.684810231183218e-06,
+      "loss": 104.316,
+      "step": 25030
+    },
+    {
+      "epoch": 0.207139016420565,
+      "grad_norm": 1371.7374267578125,
+      "learning_rate": 9.684317218636767e-06,
+      "loss": 137.7989,
+      "step": 25040
+    },
+    {
+      "epoch": 0.20722173966993424,
+      "grad_norm": 976.9849243164062,
+      "learning_rate": 9.683823833380692e-06,
+      "loss": 107.9419,
+      "step": 25050
+    },
+    {
+      "epoch": 0.20730446291930346,
+      "grad_norm": 2080.160400390625,
+      "learning_rate": 9.683330075454252e-06,
+      "loss": 134.3795,
+      "step": 25060
+    },
+    {
+      "epoch": 0.20738718616867272,
+      "grad_norm": 1066.323974609375,
+      "learning_rate": 9.68283594489673e-06,
+      "loss": 113.7562,
+      "step": 25070
+    },
+    {
+      "epoch": 0.20746990941804194,
+      "grad_norm": 1623.5059814453125,
+      "learning_rate": 9.682341441747446e-06,
+      "loss": 124.4294,
+      "step": 25080
+    },
+    {
+      "epoch": 0.20755263266741117,
+      "grad_norm": 1231.952880859375,
+      "learning_rate": 9.68184656604574e-06,
+      "loss": 168.1638,
+      "step": 25090
+    },
+    {
+      "epoch": 0.20763535591678042,
+      "grad_norm": 604.9308471679688,
+      "learning_rate": 9.681351317830991e-06,
+      "loss": 150.3729,
+      "step": 25100
+    },
+    {
+      "epoch": 0.20771807916614965,
+      "grad_norm": 915.5381469726562,
+      "learning_rate": 9.680855697142601e-06,
+      "loss": 89.7146,
+      "step": 25110
+    },
+    {
+      "epoch": 0.20780080241551888,
+      "grad_norm": 517.771484375,
+      "learning_rate": 9.680359704020005e-06,
+      "loss": 110.3232,
+      "step": 25120
+    },
+    {
+      "epoch": 0.20788352566488813,
+      "grad_norm": 999.2185668945312,
+      "learning_rate": 9.67986333850267e-06,
+      "loss": 155.6853,
+      "step": 25130
+    },
+    {
+      "epoch": 0.20796624891425736,
+      "grad_norm": 837.3745727539062,
+      "learning_rate": 9.679366600630085e-06,
+      "loss": 123.4482,
+      "step": 25140
+    },
+    {
+      "epoch": 0.20804897216362658,
+      "grad_norm": 1082.25146484375,
+      "learning_rate": 9.678869490441775e-06,
+      "loss": 92.413,
+      "step": 25150
+    },
+    {
+      "epoch": 0.2081316954129958,
+      "grad_norm": 1187.2119140625,
+      "learning_rate": 9.678372007977292e-06,
+      "loss": 130.2228,
+      "step": 25160
+    },
+    {
+      "epoch": 0.20821441866236506,
+      "grad_norm": 1067.64306640625,
+      "learning_rate": 9.67787415327622e-06,
+      "loss": 106.1636,
+      "step": 25170
+    },
+    {
+      "epoch": 0.2082971419117343,
+      "grad_norm": 1119.31640625,
+      "learning_rate": 9.67737592637817e-06,
+      "loss": 125.416,
+      "step": 25180
+    },
+    {
+      "epoch": 0.20837986516110352,
+      "grad_norm": 1132.6695556640625,
+      "learning_rate": 9.676877327322785e-06,
+      "loss": 121.1855,
+      "step": 25190
+    },
+    {
+      "epoch": 0.20846258841047277,
+      "grad_norm": 921.9434204101562,
+      "learning_rate": 9.676378356149733e-06,
+      "loss": 119.0814,
+      "step": 25200
+    },
+    {
+      "epoch": 0.208545311659842,
+      "grad_norm": 1111.28857421875,
+      "learning_rate": 9.675879012898719e-06,
+      "loss": 112.7059,
+      "step": 25210
+    },
+    {
+      "epoch": 0.20862803490921122,
+      "grad_norm": 952.7010498046875,
+      "learning_rate": 9.67537929760947e-06,
+      "loss": 135.5189,
+      "step": 25220
+    },
+    {
+      "epoch": 0.20871075815858048,
+      "grad_norm": 668.4132080078125,
+      "learning_rate": 9.674879210321747e-06,
+      "loss": 126.6339,
+      "step": 25230
+    },
+    {
+      "epoch": 0.2087934814079497,
+      "grad_norm": 1562.3443603515625,
+      "learning_rate": 9.67437875107534e-06,
+      "loss": 157.3701,
+      "step": 25240
+    },
+    {
+      "epoch": 0.20887620465731893,
+      "grad_norm": 787.549072265625,
+      "learning_rate": 9.673877919910069e-06,
+      "loss": 144.3811,
+      "step": 25250
+    },
+    {
+      "epoch": 0.20895892790668819,
+      "grad_norm": 1168.210693359375,
+      "learning_rate": 9.673376716865781e-06,
+      "loss": 147.3342,
+      "step": 25260
+    },
+    {
+      "epoch": 0.2090416511560574,
+      "grad_norm": 922.6990356445312,
+      "learning_rate": 9.672875141982358e-06,
+      "loss": 134.1699,
+      "step": 25270
+    },
+    {
+      "epoch": 0.20912437440542664,
+      "grad_norm": 920.8440551757812,
+      "learning_rate": 9.672373195299704e-06,
+      "loss": 90.3388,
+      "step": 25280
+    },
+    {
+      "epoch": 0.2092070976547959,
+      "grad_norm": 742.2246704101562,
+      "learning_rate": 9.67187087685776e-06,
+      "loss": 127.1734,
+      "step": 25290
+    },
+    {
+      "epoch": 0.20928982090416512,
+      "grad_norm": 1413.235107421875,
+      "learning_rate": 9.671368186696488e-06,
+      "loss": 121.8572,
+      "step": 25300
+    },
+    {
+      "epoch": 0.20937254415353435,
+      "grad_norm": 1089.089111328125,
+      "learning_rate": 9.670865124855889e-06,
+      "loss": 155.726,
+      "step": 25310
+    },
+    {
+      "epoch": 0.2094552674029036,
+      "grad_norm": 1251.3277587890625,
+      "learning_rate": 9.67036169137599e-06,
+      "loss": 134.8554,
+      "step": 25320
+    },
+    {
+      "epoch": 0.20953799065227283,
+      "grad_norm": 975.4923706054688,
+      "learning_rate": 9.669857886296842e-06,
+      "loss": 137.952,
+      "step": 25330
+    },
+    {
+      "epoch": 0.20962071390164205,
+      "grad_norm": 1131.67822265625,
+      "learning_rate": 9.669353709658537e-06,
+      "loss": 119.2049,
+      "step": 25340
+    },
+    {
+      "epoch": 0.20970343715101128,
+      "grad_norm": 637.6780395507812,
+      "learning_rate": 9.668849161501186e-06,
+      "loss": 104.5548,
+      "step": 25350
+    },
+    {
+      "epoch": 0.20978616040038053,
+      "grad_norm": 1121.5142822265625,
+      "learning_rate": 9.668344241864934e-06,
+      "loss": 115.9487,
+      "step": 25360
+    },
+    {
+      "epoch": 0.20986888364974976,
+      "grad_norm": 937.0314331054688,
+      "learning_rate": 9.667838950789957e-06,
+      "loss": 112.8814,
+      "step": 25370
+    },
+    {
+      "epoch": 0.209951606899119,
+      "grad_norm": 1264.4849853515625,
+      "learning_rate": 9.667333288316454e-06,
+      "loss": 135.4153,
+      "step": 25380
+    },
+    {
+      "epoch": 0.21003433014848824,
+      "grad_norm": 495.7436828613281,
+      "learning_rate": 9.666827254484663e-06,
+      "loss": 111.5311,
+      "step": 25390
+    },
+    {
+      "epoch": 0.21011705339785747,
+      "grad_norm": 853.3590087890625,
+      "learning_rate": 9.666320849334846e-06,
+      "loss": 102.2723,
+      "step": 25400
+    },
+    {
+      "epoch": 0.2101997766472267,
+      "grad_norm": 1231.4791259765625,
+      "learning_rate": 9.665814072907293e-06,
+      "loss": 118.1443,
+      "step": 25410
+    },
+    {
+      "epoch": 0.21028249989659595,
+      "grad_norm": 688.8372192382812,
+      "learning_rate": 9.665306925242329e-06,
+      "loss": 120.3173,
+      "step": 25420
+    },
+    {
+      "epoch": 0.21036522314596517,
+      "grad_norm": 1288.0115966796875,
+      "learning_rate": 9.664799406380302e-06,
+      "loss": 129.2124,
+      "step": 25430
+    },
+    {
+      "epoch": 0.2104479463953344,
+      "grad_norm": 734.3385009765625,
+      "learning_rate": 9.664291516361597e-06,
+      "loss": 141.6067,
+      "step": 25440
+    },
+    {
+      "epoch": 0.21053066964470366,
+      "grad_norm": 808.9627685546875,
+      "learning_rate": 9.663783255226622e-06,
+      "loss": 128.5043,
+      "step": 25450
+    },
+    {
+      "epoch": 0.21061339289407288,
+      "grad_norm": 2131.609619140625,
+      "learning_rate": 9.663274623015816e-06,
+      "loss": 130.3591,
+      "step": 25460
+    },
+    {
+      "epoch": 0.2106961161434421,
+      "grad_norm": 1364.452880859375,
+      "learning_rate": 9.662765619769651e-06,
+      "loss": 186.9397,
+      "step": 25470
+    },
+    {
+      "epoch": 0.21077883939281136,
+      "grad_norm": 1547.8067626953125,
+      "learning_rate": 9.662256245528622e-06,
+      "loss": 130.2646,
+      "step": 25480
+    },
+    {
+      "epoch": 0.2108615626421806,
+      "grad_norm": 760.1724853515625,
+      "learning_rate": 9.661746500333265e-06,
+      "loss": 107.9425,
+      "step": 25490
+    },
+    {
+      "epoch": 0.21094428589154982,
+      "grad_norm": 691.5214233398438,
+      "learning_rate": 9.66123638422413e-06,
+      "loss": 131.0554,
+      "step": 25500
+    },
+    {
+      "epoch": 0.21102700914091904,
+      "grad_norm": 1053.4617919921875,
+      "learning_rate": 9.66072589724181e-06,
+      "loss": 139.2496,
+      "step": 25510
+    },
+    {
+      "epoch": 0.2111097323902883,
+      "grad_norm": 903.3119506835938,
+      "learning_rate": 9.66021503942692e-06,
+      "loss": 103.5156,
+      "step": 25520
+    },
+    {
+      "epoch": 0.21119245563965752,
+      "grad_norm": 1562.44482421875,
+      "learning_rate": 9.659703810820105e-06,
+      "loss": 154.1499,
+      "step": 25530
+    },
+    {
+      "epoch": 0.21127517888902675,
+      "grad_norm": 724.2481079101562,
+      "learning_rate": 9.659192211462043e-06,
+      "loss": 122.7316,
+      "step": 25540
+    },
+    {
+      "epoch": 0.211357902138396,
+      "grad_norm": 1171.6414794921875,
+      "learning_rate": 9.658680241393441e-06,
+      "loss": 112.3694,
+      "step": 25550
+    },
+    {
+      "epoch": 0.21144062538776523,
+      "grad_norm": 687.9328002929688,
+      "learning_rate": 9.658167900655032e-06,
+      "loss": 85.1699,
+      "step": 25560
+    },
+    {
+      "epoch": 0.21152334863713446,
+      "grad_norm": 1203.814208984375,
+      "learning_rate": 9.657655189287582e-06,
+      "loss": 111.8283,
+      "step": 25570
+    },
+    {
+      "epoch": 0.2116060718865037,
+      "grad_norm": 2333.49365234375,
+      "learning_rate": 9.657142107331883e-06,
+      "loss": 168.9521,
+      "step": 25580
+    },
+    {
+      "epoch": 0.21168879513587294,
+      "grad_norm": 919.4609375,
+      "learning_rate": 9.65662865482876e-06,
+      "loss": 110.3846,
+      "step": 25590
+    },
+    {
+      "epoch": 0.21177151838524216,
+      "grad_norm": 1061.5003662109375,
+      "learning_rate": 9.656114831819067e-06,
+      "loss": 133.6754,
+      "step": 25600
+    },
+    {
+      "epoch": 0.21185424163461142,
+      "grad_norm": 498.1741638183594,
+      "learning_rate": 9.655600638343685e-06,
+      "loss": 109.6762,
+      "step": 25610
+    },
+    {
+      "epoch": 0.21193696488398064,
+      "grad_norm": 946.8280639648438,
+      "learning_rate": 9.655086074443527e-06,
+      "loss": 183.0106,
+      "step": 25620
+    },
+    {
+      "epoch": 0.21201968813334987,
+      "grad_norm": 1837.9036865234375,
+      "learning_rate": 9.654571140159534e-06,
+      "loss": 122.9791,
+      "step": 25630
+    },
+    {
+      "epoch": 0.21210241138271912,
+      "grad_norm": 1461.47900390625,
+      "learning_rate": 9.654055835532676e-06,
+      "loss": 191.4103,
+      "step": 25640
+    },
+    {
+      "epoch": 0.21218513463208835,
+      "grad_norm": 820.8297119140625,
+      "learning_rate": 9.653540160603956e-06,
+      "loss": 115.1019,
+      "step": 25650
+    },
+    {
+      "epoch": 0.21226785788145758,
+      "grad_norm": 1157.8956298828125,
+      "learning_rate": 9.653024115414402e-06,
+      "loss": 142.3736,
+      "step": 25660
+    },
+    {
+      "epoch": 0.21235058113082683,
+      "grad_norm": 917.7392578125,
+      "learning_rate": 9.652507700005072e-06,
+      "loss": 130.4384,
+      "step": 25670
+    },
+    {
+      "epoch": 0.21243330438019606,
+      "grad_norm": 1544.7633056640625,
+      "learning_rate": 9.651990914417057e-06,
+      "loss": 135.7345,
+      "step": 25680
+    },
+    {
+      "epoch": 0.21251602762956529,
+      "grad_norm": 768.8818359375,
+      "learning_rate": 9.651473758691477e-06,
+      "loss": 99.1156,
+      "step": 25690
+    },
+    {
+      "epoch": 0.2125987508789345,
+      "grad_norm": 1055.4837646484375,
+      "learning_rate": 9.650956232869475e-06,
+      "loss": 113.1447,
+      "step": 25700
+    },
+    {
+      "epoch": 0.21268147412830377,
+      "grad_norm": 1085.100341796875,
+      "learning_rate": 9.650438336992231e-06,
+      "loss": 156.4204,
+      "step": 25710
+    },
+    {
+      "epoch": 0.212764197377673,
+      "grad_norm": 535.8530883789062,
+      "learning_rate": 9.64992007110095e-06,
+      "loss": 113.1042,
+      "step": 25720
+    },
+    {
+      "epoch": 0.21284692062704222,
+      "grad_norm": 1140.1822509765625,
+      "learning_rate": 9.64940143523687e-06,
+      "loss": 134.6167,
+      "step": 25730
+    },
+    {
+      "epoch": 0.21292964387641147,
+      "grad_norm": 1358.8885498046875,
+      "learning_rate": 9.648882429441258e-06,
+      "loss": 133.8381,
+      "step": 25740
+    },
+    {
+      "epoch": 0.2130123671257807,
+      "grad_norm": 1481.0440673828125,
+      "learning_rate": 9.648363053755406e-06,
+      "loss": 120.2028,
+      "step": 25750
+    },
+    {
+      "epoch": 0.21309509037514993,
+      "grad_norm": 1080.2623291015625,
+      "learning_rate": 9.647843308220636e-06,
+      "loss": 105.3537,
+      "step": 25760
+    },
+    {
+      "epoch": 0.21317781362451918,
+      "grad_norm": 712.189453125,
+      "learning_rate": 9.647323192878306e-06,
+      "loss": 101.1071,
+      "step": 25770
+    },
+    {
+      "epoch": 0.2132605368738884,
+      "grad_norm": 1282.9239501953125,
+      "learning_rate": 9.646802707769798e-06,
+      "loss": 121.2276,
+      "step": 25780
+    },
+    {
+      "epoch": 0.21334326012325763,
+      "grad_norm": 558.678955078125,
+      "learning_rate": 9.646281852936525e-06,
+      "loss": 92.6775,
+      "step": 25790
+    },
+    {
+      "epoch": 0.2134259833726269,
+      "grad_norm": 994.920654296875,
+      "learning_rate": 9.64576062841993e-06,
+      "loss": 124.3649,
+      "step": 25800
+    },
+    {
+      "epoch": 0.2135087066219961,
+      "grad_norm": 1420.953857421875,
+      "learning_rate": 9.64523903426148e-06,
+      "loss": 122.571,
+      "step": 25810
+    },
+    {
+      "epoch": 0.21359142987136534,
+      "grad_norm": 859.2244873046875,
+      "learning_rate": 9.64471707050268e-06,
+      "loss": 126.6968,
+      "step": 25820
+    },
+    {
+      "epoch": 0.2136741531207346,
+      "grad_norm": 1234.9310302734375,
+      "learning_rate": 9.644194737185058e-06,
+      "loss": 139.6012,
+      "step": 25830
+    },
+    {
+      "epoch": 0.21375687637010382,
+      "grad_norm": 1804.944091796875,
+      "learning_rate": 9.643672034350177e-06,
+      "loss": 150.6442,
+      "step": 25840
+    },
+    {
+      "epoch": 0.21383959961947305,
+      "grad_norm": 1387.7882080078125,
+      "learning_rate": 9.643148962039622e-06,
+      "loss": 93.8409,
+      "step": 25850
+    },
+    {
+      "epoch": 0.2139223228688423,
+      "grad_norm": 1455.9395751953125,
+      "learning_rate": 9.642625520295014e-06,
+      "loss": 174.9808,
+      "step": 25860
+    },
+    {
+      "epoch": 0.21400504611821153,
+      "grad_norm": 1184.424072265625,
+      "learning_rate": 9.642101709158001e-06,
+      "loss": 129.4004,
+      "step": 25870
+    },
+    {
+      "epoch": 0.21408776936758075,
+      "grad_norm": 1059.5274658203125,
+      "learning_rate": 9.641577528670257e-06,
+      "loss": 129.7015,
+      "step": 25880
+    },
+    {
+      "epoch": 0.21417049261694998,
+      "grad_norm": 753.4532470703125,
+      "learning_rate": 9.641052978873494e-06,
+      "loss": 109.8265,
+      "step": 25890
+    },
+    {
+      "epoch": 0.21425321586631924,
+      "grad_norm": 676.7515258789062,
+      "learning_rate": 9.640528059809442e-06,
+      "loss": 74.3417,
+      "step": 25900
+    },
+    {
+      "epoch": 0.21433593911568846,
+      "grad_norm": 2531.762939453125,
+      "learning_rate": 9.640002771519872e-06,
+      "loss": 133.8275,
+      "step": 25910
+    },
+    {
+      "epoch": 0.2144186623650577,
+      "grad_norm": 1403.41943359375,
+      "learning_rate": 9.639477114046575e-06,
+      "loss": 164.8685,
+      "step": 25920
+    },
+    {
+      "epoch": 0.21450138561442694,
+      "grad_norm": 922.8995971679688,
+      "learning_rate": 9.638951087431376e-06,
+      "loss": 114.9469,
+      "step": 25930
+    },
+    {
+      "epoch": 0.21458410886379617,
+      "grad_norm": 1005.201416015625,
+      "learning_rate": 9.638424691716129e-06,
+      "loss": 144.4065,
+      "step": 25940
+    },
+    {
+      "epoch": 0.2146668321131654,
+      "grad_norm": 917.942626953125,
+      "learning_rate": 9.637897926942716e-06,
+      "loss": 140.8008,
+      "step": 25950
+    },
+    {
+      "epoch": 0.21474955536253465,
+      "grad_norm": 1009.6735229492188,
+      "learning_rate": 9.637370793153051e-06,
+      "loss": 132.449,
+      "step": 25960
+    },
+    {
+      "epoch": 0.21483227861190388,
+      "grad_norm": 904.17529296875,
+      "learning_rate": 9.636843290389076e-06,
+      "loss": 114.3653,
+      "step": 25970
+    },
+    {
+      "epoch": 0.2149150018612731,
+      "grad_norm": 1091.709716796875,
+      "learning_rate": 9.636315418692759e-06,
+      "loss": 108.0948,
+      "step": 25980
+    },
+    {
+      "epoch": 0.21499772511064236,
+      "grad_norm": 2188.771240234375,
+      "learning_rate": 9.635787178106102e-06,
+      "loss": 107.5951,
+      "step": 25990
+    },
+    {
+      "epoch": 0.21508044836001158,
+      "grad_norm": 1075.4591064453125,
+      "learning_rate": 9.635258568671135e-06,
+      "loss": 137.4553,
+      "step": 26000
+    },
+    {
+      "epoch": 0.2151631716093808,
+      "grad_norm": 916.110107421875,
+      "learning_rate": 9.634729590429917e-06,
+      "loss": 123.9193,
+      "step": 26010
+    },
+    {
+      "epoch": 0.21524589485875006,
+      "grad_norm": 1345.30615234375,
+      "learning_rate": 9.634200243424535e-06,
+      "loss": 139.9196,
+      "step": 26020
+    },
+    {
+      "epoch": 0.2153286181081193,
+      "grad_norm": 1103.3697509765625,
+      "learning_rate": 9.633670527697108e-06,
+      "loss": 130.9413,
+      "step": 26030
+    },
+    {
+      "epoch": 0.21541134135748852,
+      "grad_norm": 1355.6485595703125,
+      "learning_rate": 9.633140443289784e-06,
+      "loss": 190.8187,
+      "step": 26040
+    },
+    {
+      "epoch": 0.21549406460685774,
+      "grad_norm": 575.96142578125,
+      "learning_rate": 9.632609990244737e-06,
+      "loss": 103.7051,
+      "step": 26050
+    },
+    {
+      "epoch": 0.215576787856227,
+      "grad_norm": 779.9686889648438,
+      "learning_rate": 9.632079168604175e-06,
+      "loss": 109.2936,
+      "step": 26060
+    },
+    {
+      "epoch": 0.21565951110559622,
+      "grad_norm": 1010.6686401367188,
+      "learning_rate": 9.63154797841033e-06,
+      "loss": 159.4392,
+      "step": 26070
+    },
+    {
+      "epoch": 0.21574223435496545,
+      "grad_norm": 826.314208984375,
+      "learning_rate": 9.63101641970547e-06,
+      "loss": 129.3246,
+      "step": 26080
+    },
+    {
+      "epoch": 0.2158249576043347,
+      "grad_norm": 898.4963989257812,
+      "learning_rate": 9.630484492531886e-06,
+      "loss": 129.947,
+      "step": 26090
+    },
+    {
+      "epoch": 0.21590768085370393,
+      "grad_norm": 843.03857421875,
+      "learning_rate": 9.629952196931902e-06,
+      "loss": 113.0155,
+      "step": 26100
+    },
+    {
+      "epoch": 0.21599040410307316,
+      "grad_norm": 660.1622314453125,
+      "learning_rate": 9.629419532947872e-06,
+      "loss": 156.4598,
+      "step": 26110
+    },
+    {
+      "epoch": 0.2160731273524424,
+      "grad_norm": 1190.5133056640625,
+      "learning_rate": 9.628886500622174e-06,
+      "loss": 128.8638,
+      "step": 26120
+    },
+    {
+      "epoch": 0.21615585060181164,
+      "grad_norm": 924.6058959960938,
+      "learning_rate": 9.62835309999722e-06,
+      "loss": 99.8876,
+      "step": 26130
+    },
+    {
+      "epoch": 0.21623857385118087,
+      "grad_norm": 1131.3111572265625,
+      "learning_rate": 9.627819331115453e-06,
+      "loss": 126.6344,
+      "step": 26140
+    },
+    {
+      "epoch": 0.21632129710055012,
+      "grad_norm": 784.7423706054688,
+      "learning_rate": 9.627285194019342e-06,
+      "loss": 102.5163,
+      "step": 26150
+    },
+    {
+      "epoch": 0.21640402034991935,
+      "grad_norm": 1271.62890625,
+      "learning_rate": 9.626750688751382e-06,
+      "loss": 115.6172,
+      "step": 26160
+    },
+    {
+      "epoch": 0.21648674359928857,
+      "grad_norm": 1133.4263916015625,
+      "learning_rate": 9.626215815354104e-06,
+      "loss": 98.0378,
+      "step": 26170
+    },
+    {
+      "epoch": 0.21656946684865783,
+      "grad_norm": 822.8490600585938,
+      "learning_rate": 9.625680573870067e-06,
+      "loss": 123.6515,
+      "step": 26180
+    },
+    {
+      "epoch": 0.21665219009802705,
+      "grad_norm": 796.753662109375,
+      "learning_rate": 9.625144964341853e-06,
+      "loss": 127.32,
+      "step": 26190
+    },
+    {
+      "epoch": 0.21673491334739628,
+      "grad_norm": 976.165771484375,
+      "learning_rate": 9.624608986812082e-06,
+      "loss": 113.2206,
+      "step": 26200
+    },
+    {
+      "epoch": 0.21681763659676553,
+      "grad_norm": 1549.4820556640625,
+      "learning_rate": 9.624072641323398e-06,
+      "loss": 121.0571,
+      "step": 26210
+    },
+    {
+      "epoch": 0.21690035984613476,
+      "grad_norm": 1164.861328125,
+      "learning_rate": 9.623535927918474e-06,
+      "loss": 168.508,
+      "step": 26220
+    },
+    {
+      "epoch": 0.216983083095504,
+      "grad_norm": 1254.7081298828125,
+      "learning_rate": 9.622998846640018e-06,
+      "loss": 114.7848,
+      "step": 26230
+    },
+    {
+      "epoch": 0.2170658063448732,
+      "grad_norm": 744.2569580078125,
+      "learning_rate": 9.62246139753076e-06,
+      "loss": 125.7201,
+      "step": 26240
+    },
+    {
+      "epoch": 0.21714852959424247,
+      "grad_norm": 1193.7801513671875,
+      "learning_rate": 9.621923580633462e-06,
+      "loss": 109.2309,
+      "step": 26250
+    },
+    {
+      "epoch": 0.2172312528436117,
+      "grad_norm": 933.6427612304688,
+      "learning_rate": 9.621385395990915e-06,
+      "loss": 137.9964,
+      "step": 26260
+    },
+    {
+      "epoch": 0.21731397609298092,
+      "grad_norm": 1021.1292724609375,
+      "learning_rate": 9.620846843645944e-06,
+      "loss": 105.2249,
+      "step": 26270
+    },
+    {
+      "epoch": 0.21739669934235017,
+      "grad_norm": 1435.4403076171875,
+      "learning_rate": 9.620307923641395e-06,
+      "loss": 120.3464,
+      "step": 26280
+    },
+    {
+      "epoch": 0.2174794225917194,
+      "grad_norm": 1647.5338134765625,
+      "learning_rate": 9.61976863602015e-06,
+      "loss": 115.3988,
+      "step": 26290
+    },
+    {
+      "epoch": 0.21756214584108863,
+      "grad_norm": 729.9489135742188,
+      "learning_rate": 9.619228980825114e-06,
+      "loss": 171.7545,
+      "step": 26300
+    },
+    {
+      "epoch": 0.21764486909045788,
+      "grad_norm": 1271.6209716796875,
+      "learning_rate": 9.61868895809923e-06,
+      "loss": 120.3744,
+      "step": 26310
+    },
+    {
+      "epoch": 0.2177275923398271,
+      "grad_norm": 844.6718139648438,
+      "learning_rate": 9.618148567885462e-06,
+      "loss": 142.1199,
+      "step": 26320
+    },
+    {
+      "epoch": 0.21781031558919633,
+      "grad_norm": 837.2586669921875,
+      "learning_rate": 9.617607810226806e-06,
+      "loss": 154.3331,
+      "step": 26330
+    },
+    {
+      "epoch": 0.2178930388385656,
+      "grad_norm": 1400.4326171875,
+      "learning_rate": 9.61706668516629e-06,
+      "loss": 124.3974,
+      "step": 26340
+    },
+    {
+      "epoch": 0.21797576208793482,
+      "grad_norm": 1115.8116455078125,
+      "learning_rate": 9.616525192746965e-06,
+      "loss": 105.8458,
+      "step": 26350
+    },
+    {
+      "epoch": 0.21805848533730404,
+      "grad_norm": 1421.916748046875,
+      "learning_rate": 9.61598333301192e-06,
+      "loss": 156.9738,
+      "step": 26360
+    },
+    {
+      "epoch": 0.2181412085866733,
+      "grad_norm": 798.7007446289062,
+      "learning_rate": 9.615441106004264e-06,
+      "loss": 111.3176,
+      "step": 26370
+    },
+    {
+      "epoch": 0.21822393183604252,
+      "grad_norm": 1311.8187255859375,
+      "learning_rate": 9.614898511767142e-06,
+      "loss": 112.8957,
+      "step": 26380
+    },
+    {
+      "epoch": 0.21830665508541175,
+      "grad_norm": 1265.518798828125,
+      "learning_rate": 9.614355550343724e-06,
+      "loss": 97.6749,
+      "step": 26390
+    },
+    {
+      "epoch": 0.218389378334781,
+      "grad_norm": 1181.8995361328125,
+      "learning_rate": 9.613812221777212e-06,
+      "loss": 126.788,
+      "step": 26400
+    },
+    {
+      "epoch": 0.21847210158415023,
+      "grad_norm": 921.5441284179688,
+      "learning_rate": 9.613268526110838e-06,
+      "loss": 126.4273,
+      "step": 26410
+    },
+    {
+      "epoch": 0.21855482483351946,
+      "grad_norm": 2307.983642578125,
+      "learning_rate": 9.612724463387857e-06,
+      "loss": 124.0576,
+      "step": 26420
+    },
+    {
+      "epoch": 0.21863754808288868,
+      "grad_norm": 1391.1617431640625,
+      "learning_rate": 9.612180033651561e-06,
+      "loss": 122.5418,
+      "step": 26430
+    },
+    {
+      "epoch": 0.21872027133225794,
+      "grad_norm": 883.1130981445312,
+      "learning_rate": 9.611635236945267e-06,
+      "loss": 105.9835,
+      "step": 26440
+    },
+    {
+      "epoch": 0.21880299458162716,
+      "grad_norm": 1083.864990234375,
+      "learning_rate": 9.61109007331232e-06,
+      "loss": 124.8244,
+      "step": 26450
+    },
+    {
+      "epoch": 0.2188857178309964,
+      "grad_norm": 1159.39697265625,
+      "learning_rate": 9.610544542796101e-06,
+      "loss": 120.3728,
+      "step": 26460
+    },
+    {
+      "epoch": 0.21896844108036564,
+      "grad_norm": 1433.8590087890625,
+      "learning_rate": 9.609998645440011e-06,
+      "loss": 132.6535,
+      "step": 26470
+    },
+    {
+      "epoch": 0.21905116432973487,
+      "grad_norm": 1183.7196044921875,
+      "learning_rate": 9.609452381287486e-06,
+      "loss": 133.8586,
+      "step": 26480
+    },
+    {
+      "epoch": 0.2191338875791041,
+      "grad_norm": 859.5003662109375,
+      "learning_rate": 9.608905750381988e-06,
+      "loss": 101.3727,
+      "step": 26490
+    },
+    {
+      "epoch": 0.21921661082847335,
+      "grad_norm": 800.9340209960938,
+      "learning_rate": 9.608358752767013e-06,
+      "loss": 142.6461,
+      "step": 26500
+    },
+    {
+      "epoch": 0.21929933407784258,
+      "grad_norm": 1471.0047607421875,
+      "learning_rate": 9.60781138848608e-06,
+      "loss": 110.0766,
+      "step": 26510
+    },
+    {
+      "epoch": 0.2193820573272118,
+      "grad_norm": 954.5092163085938,
+      "learning_rate": 9.607263657582744e-06,
+      "loss": 110.3453,
+      "step": 26520
+    },
+    {
+      "epoch": 0.21946478057658106,
+      "grad_norm": 847.3948974609375,
+      "learning_rate": 9.60671556010058e-06,
+      "loss": 119.9163,
+      "step": 26530
+    },
+    {
+      "epoch": 0.21954750382595029,
+      "grad_norm": 559.4891357421875,
+      "learning_rate": 9.606167096083205e-06,
+      "loss": 106.7365,
+      "step": 26540
+    },
+    {
+      "epoch": 0.2196302270753195,
+      "grad_norm": 1348.5836181640625,
+      "learning_rate": 9.60561826557425e-06,
+      "loss": 148.4306,
+      "step": 26550
+    },
+    {
+      "epoch": 0.21971295032468877,
+      "grad_norm": 912.4396362304688,
+      "learning_rate": 9.60506906861739e-06,
+      "loss": 116.1079,
+      "step": 26560
+    },
+    {
+      "epoch": 0.219795673574058,
+      "grad_norm": 939.4193115234375,
+      "learning_rate": 9.604519505256316e-06,
+      "loss": 127.2096,
+      "step": 26570
+    },
+    {
+      "epoch": 0.21987839682342722,
+      "grad_norm": 807.13623046875,
+      "learning_rate": 9.603969575534757e-06,
+      "loss": 102.2194,
+      "step": 26580
+    },
+    {
+      "epoch": 0.21996112007279647,
+      "grad_norm": 695.100830078125,
+      "learning_rate": 9.60341927949647e-06,
+      "loss": 113.9714,
+      "step": 26590
+    },
+    {
+      "epoch": 0.2200438433221657,
+      "grad_norm": 456.3763122558594,
+      "learning_rate": 9.602868617185238e-06,
+      "loss": 105.249,
+      "step": 26600
+    },
+    {
+      "epoch": 0.22012656657153493,
+      "grad_norm": 1128.632568359375,
+      "learning_rate": 9.602317588644872e-06,
+      "loss": 104.7491,
+      "step": 26610
+    },
+    {
+      "epoch": 0.22020928982090415,
+      "grad_norm": 948.9160766601562,
+      "learning_rate": 9.601766193919217e-06,
+      "loss": 104.4173,
+      "step": 26620
+    },
+    {
+      "epoch": 0.2202920130702734,
+      "grad_norm": 725.5731201171875,
+      "learning_rate": 9.601214433052147e-06,
+      "loss": 103.2853,
+      "step": 26630
+    },
+    {
+      "epoch": 0.22037473631964263,
+      "grad_norm": 979.1326293945312,
+      "learning_rate": 9.600662306087562e-06,
+      "loss": 122.0349,
+      "step": 26640
+    },
+    {
+      "epoch": 0.22045745956901186,
+      "grad_norm": 772.8959350585938,
+      "learning_rate": 9.600109813069389e-06,
+      "loss": 118.9232,
+      "step": 26650
+    },
+    {
+      "epoch": 0.22054018281838111,
+      "grad_norm": 879.559814453125,
+      "learning_rate": 9.599556954041591e-06,
+      "loss": 154.7716,
+      "step": 26660
+    },
+    {
+      "epoch": 0.22062290606775034,
+      "grad_norm": 852.1553344726562,
+      "learning_rate": 9.599003729048157e-06,
+      "loss": 115.1464,
+      "step": 26670
+    },
+    {
+      "epoch": 0.22070562931711957,
+      "grad_norm": 1091.1187744140625,
+      "learning_rate": 9.598450138133101e-06,
+      "loss": 124.5991,
+      "step": 26680
+    },
+    {
+      "epoch": 0.22078835256648882,
+      "grad_norm": 1003.1347045898438,
+      "learning_rate": 9.597896181340471e-06,
+      "loss": 133.6112,
+      "step": 26690
+    },
+    {
+      "epoch": 0.22087107581585805,
+      "grad_norm": 1310.1240234375,
+      "learning_rate": 9.597341858714344e-06,
+      "loss": 120.8151,
+      "step": 26700
+    },
+    {
+      "epoch": 0.22095379906522727,
+      "grad_norm": 973.50439453125,
+      "learning_rate": 9.596787170298824e-06,
+      "loss": 132.3573,
+      "step": 26710
+    },
+    {
+      "epoch": 0.22103652231459653,
+      "grad_norm": 838.1239013671875,
+      "learning_rate": 9.596232116138047e-06,
+      "loss": 135.2263,
+      "step": 26720
+    },
+    {
+      "epoch": 0.22111924556396576,
+      "grad_norm": 776.5439453125,
+      "learning_rate": 9.595676696276173e-06,
+      "loss": 140.2891,
+      "step": 26730
+    },
+    {
+      "epoch": 0.22120196881333498,
+      "grad_norm": 898.1226806640625,
+      "learning_rate": 9.595120910757396e-06,
+      "loss": 156.5662,
+      "step": 26740
+    },
+    {
+      "epoch": 0.22128469206270424,
+      "grad_norm": 988.5482788085938,
+      "learning_rate": 9.594564759625936e-06,
+      "loss": 119.9368,
+      "step": 26750
+    },
+    {
+      "epoch": 0.22136741531207346,
+      "grad_norm": 2739.469482421875,
+      "learning_rate": 9.594008242926046e-06,
+      "loss": 117.0178,
+      "step": 26760
+    },
+    {
+      "epoch": 0.2214501385614427,
+      "grad_norm": 1172.0687255859375,
+      "learning_rate": 9.593451360702003e-06,
+      "loss": 109.8631,
+      "step": 26770
+    },
+    {
+      "epoch": 0.22153286181081192,
+      "grad_norm": 1231.6884765625,
+      "learning_rate": 9.592894112998115e-06,
+      "loss": 123.6822,
+      "step": 26780
+    },
+    {
+      "epoch": 0.22161558506018117,
+      "grad_norm": 2576.349609375,
+      "learning_rate": 9.592336499858721e-06,
+      "loss": 134.2115,
+      "step": 26790
+    },
+    {
+      "epoch": 0.2216983083095504,
+      "grad_norm": 851.02197265625,
+      "learning_rate": 9.59177852132819e-06,
+      "loss": 130.6942,
+      "step": 26800
+    },
+    {
+      "epoch": 0.22178103155891962,
+      "grad_norm": 1379.3516845703125,
+      "learning_rate": 9.591220177450912e-06,
+      "loss": 148.0982,
+      "step": 26810
+    },
+    {
+      "epoch": 0.22186375480828888,
+      "grad_norm": 1033.928955078125,
+      "learning_rate": 9.590661468271319e-06,
+      "loss": 99.2162,
+      "step": 26820
+    },
+    {
+      "epoch": 0.2219464780576581,
+      "grad_norm": 987.3099365234375,
+      "learning_rate": 9.59010239383386e-06,
+      "loss": 138.9152,
+      "step": 26830
+    },
+    {
+      "epoch": 0.22202920130702733,
+      "grad_norm": 687.860595703125,
+      "learning_rate": 9.589542954183018e-06,
+      "loss": 112.7026,
+      "step": 26840
+    },
+    {
+      "epoch": 0.22211192455639658,
+      "grad_norm": 732.00537109375,
+      "learning_rate": 9.588983149363307e-06,
+      "loss": 123.2144,
+      "step": 26850
+    },
+    {
+      "epoch": 0.2221946478057658,
+      "grad_norm": 1079.510009765625,
+      "learning_rate": 9.588422979419267e-06,
+      "loss": 87.7841,
+      "step": 26860
+    },
+    {
+      "epoch": 0.22227737105513504,
+      "grad_norm": 1468.6038818359375,
+      "learning_rate": 9.587862444395471e-06,
+      "loss": 136.3903,
+      "step": 26870
+    },
+    {
+      "epoch": 0.2223600943045043,
+      "grad_norm": 889.5414428710938,
+      "learning_rate": 9.587301544336513e-06,
+      "loss": 115.5707,
+      "step": 26880
+    },
+    {
+      "epoch": 0.22244281755387352,
+      "grad_norm": 590.138916015625,
+      "learning_rate": 9.586740279287024e-06,
+      "loss": 117.9152,
+      "step": 26890
+    },
+    {
+      "epoch": 0.22252554080324274,
+      "grad_norm": 1332.5377197265625,
+      "learning_rate": 9.586178649291664e-06,
+      "loss": 131.7125,
+      "step": 26900
+    },
+    {
+      "epoch": 0.222608264052612,
+      "grad_norm": 1022.7705688476562,
+      "learning_rate": 9.585616654395113e-06,
+      "loss": 115.9927,
+      "step": 26910
+    },
+    {
+      "epoch": 0.22269098730198122,
+      "grad_norm": 847.87255859375,
+      "learning_rate": 9.585054294642093e-06,
+      "loss": 169.3321,
+      "step": 26920
+    },
+    {
+      "epoch": 0.22277371055135045,
+      "grad_norm": 1865.4990234375,
+      "learning_rate": 9.584491570077343e-06,
+      "loss": 128.1739,
+      "step": 26930
+    },
+    {
+      "epoch": 0.2228564338007197,
+      "grad_norm": 865.5852661132812,
+      "learning_rate": 9.58392848074564e-06,
+      "loss": 119.8757,
+      "step": 26940
+    },
+    {
+      "epoch": 0.22293915705008893,
+      "grad_norm": 1116.7584228515625,
+      "learning_rate": 9.583365026691785e-06,
+      "loss": 111.9066,
+      "step": 26950
+    },
+    {
+      "epoch": 0.22302188029945816,
+      "grad_norm": 1089.7393798828125,
+      "learning_rate": 9.58280120796061e-06,
+      "loss": 123.0073,
+      "step": 26960
+    },
+    {
+      "epoch": 0.22310460354882738,
+      "grad_norm": 854.9454956054688,
+      "learning_rate": 9.582237024596974e-06,
+      "loss": 131.7255,
+      "step": 26970
+    },
+    {
+      "epoch": 0.22318732679819664,
+      "grad_norm": 840.7254028320312,
+      "learning_rate": 9.581672476645768e-06,
+      "loss": 134.3853,
+      "step": 26980
+    },
+    {
+      "epoch": 0.22327005004756587,
+      "grad_norm": 1240.8424072265625,
+      "learning_rate": 9.58110756415191e-06,
+      "loss": 127.8951,
+      "step": 26990
+    },
+    {
+      "epoch": 0.2233527732969351,
+      "grad_norm": 1180.210205078125,
+      "learning_rate": 9.580542287160348e-06,
+      "loss": 127.6457,
+      "step": 27000
+    },
+    {
+      "epoch": 0.22343549654630435,
+      "grad_norm": 1015.6724853515625,
+      "learning_rate": 9.579976645716058e-06,
+      "loss": 130.7046,
+      "step": 27010
+    },
+    {
+      "epoch": 0.22351821979567357,
+      "grad_norm": 1194.459228515625,
+      "learning_rate": 9.579410639864046e-06,
+      "loss": 133.5198,
+      "step": 27020
+    },
+    {
+      "epoch": 0.2236009430450428,
+      "grad_norm": 1310.887451171875,
+      "learning_rate": 9.578844269649345e-06,
+      "loss": 123.0892,
+      "step": 27030
+    },
+    {
+      "epoch": 0.22368366629441205,
+      "grad_norm": 764.6407470703125,
+      "learning_rate": 9.578277535117022e-06,
+      "loss": 118.5598,
+      "step": 27040
+    },
+    {
+      "epoch": 0.22376638954378128,
+      "grad_norm": 1107.6348876953125,
+      "learning_rate": 9.577710436312164e-06,
+      "loss": 113.8774,
+      "step": 27050
+    },
+    {
+      "epoch": 0.2238491127931505,
+      "grad_norm": 1522.5843505859375,
+      "learning_rate": 9.577142973279896e-06,
+      "loss": 137.2552,
+      "step": 27060
+    },
+    {
+      "epoch": 0.22393183604251976,
+      "grad_norm": 984.351318359375,
+      "learning_rate": 9.576575146065369e-06,
+      "loss": 128.6748,
+      "step": 27070
+    },
+    {
+      "epoch": 0.224014559291889,
+      "grad_norm": 642.086181640625,
+      "learning_rate": 9.576006954713762e-06,
+      "loss": 117.2645,
+      "step": 27080
+    },
+    {
+      "epoch": 0.2240972825412582,
+      "grad_norm": 1214.22998046875,
+      "learning_rate": 9.57543839927028e-06,
+      "loss": 128.8348,
+      "step": 27090
+    },
+    {
+      "epoch": 0.22418000579062747,
+      "grad_norm": 1050.501953125,
+      "learning_rate": 9.574869479780165e-06,
+      "loss": 147.3516,
+      "step": 27100
+    },
+    {
+      "epoch": 0.2242627290399967,
+      "grad_norm": 1185.0849609375,
+      "learning_rate": 9.57430019628868e-06,
+      "loss": 147.9033,
+      "step": 27110
+    },
+    {
+      "epoch": 0.22434545228936592,
+      "grad_norm": 1115.831298828125,
+      "learning_rate": 9.573730548841122e-06,
+      "loss": 111.629,
+      "step": 27120
+    },
+    {
+      "epoch": 0.22442817553873518,
+      "grad_norm": 1058.506103515625,
+      "learning_rate": 9.573160537482816e-06,
+      "loss": 135.1445,
+      "step": 27130
+    },
+    {
+      "epoch": 0.2245108987881044,
+      "grad_norm": 1066.4493408203125,
+      "learning_rate": 9.572590162259112e-06,
+      "loss": 171.7336,
+      "step": 27140
+    },
+    {
+      "epoch": 0.22459362203747363,
+      "grad_norm": 735.5531005859375,
+      "learning_rate": 9.572019423215395e-06,
+      "loss": 110.0656,
+      "step": 27150
+    },
+    {
+      "epoch": 0.22467634528684285,
+      "grad_norm": 1075.277587890625,
+      "learning_rate": 9.571448320397076e-06,
+      "loss": 102.259,
+      "step": 27160
+    },
+    {
+      "epoch": 0.2247590685362121,
+      "grad_norm": 677.2155151367188,
+      "learning_rate": 9.570876853849593e-06,
+      "loss": 128.2644,
+      "step": 27170
+    },
+    {
+      "epoch": 0.22484179178558134,
+      "grad_norm": 993.1212768554688,
+      "learning_rate": 9.570305023618417e-06,
+      "loss": 185.3893,
+      "step": 27180
+    },
+    {
+      "epoch": 0.22492451503495056,
+      "grad_norm": 963.5702514648438,
+      "learning_rate": 9.569732829749045e-06,
+      "loss": 125.3146,
+      "step": 27190
+    },
+    {
+      "epoch": 0.22500723828431982,
+      "grad_norm": 950.5331420898438,
+      "learning_rate": 9.569160272287003e-06,
+      "loss": 126.4869,
+      "step": 27200
+    },
+    {
+      "epoch": 0.22508996153368904,
+      "grad_norm": 820.460205078125,
+      "learning_rate": 9.56858735127785e-06,
+      "loss": 76.8966,
+      "step": 27210
+    },
+    {
+      "epoch": 0.22517268478305827,
+      "grad_norm": 572.8544311523438,
+      "learning_rate": 9.568014066767166e-06,
+      "loss": 126.8829,
+      "step": 27220
+    },
+    {
+      "epoch": 0.22525540803242752,
+      "grad_norm": 998.837646484375,
+      "learning_rate": 9.567440418800569e-06,
+      "loss": 134.7057,
+      "step": 27230
+    },
+    {
+      "epoch": 0.22533813128179675,
+      "grad_norm": 956.578857421875,
+      "learning_rate": 9.566866407423698e-06,
+      "loss": 143.3908,
+      "step": 27240
+    },
+    {
+      "epoch": 0.22542085453116598,
+      "grad_norm": 1049.07861328125,
+      "learning_rate": 9.566292032682228e-06,
+      "loss": 137.0985,
+      "step": 27250
+    },
+    {
+      "epoch": 0.22550357778053523,
+      "grad_norm": 870.1298217773438,
+      "learning_rate": 9.565717294621856e-06,
+      "loss": 114.6946,
+      "step": 27260
+    },
+    {
+      "epoch": 0.22558630102990446,
+      "grad_norm": 784.1261596679688,
+      "learning_rate": 9.565142193288313e-06,
+      "loss": 150.3023,
+      "step": 27270
+    },
+    {
+      "epoch": 0.22566902427927368,
+      "grad_norm": 721.9214477539062,
+      "learning_rate": 9.564566728727358e-06,
+      "loss": 92.1085,
+      "step": 27280
+    },
+    {
+      "epoch": 0.22575174752864294,
+      "grad_norm": 819.4923706054688,
+      "learning_rate": 9.563990900984775e-06,
+      "loss": 103.506,
+      "step": 27290
+    },
+    {
+      "epoch": 0.22583447077801216,
+      "grad_norm": 2060.950927734375,
+      "learning_rate": 9.563414710106382e-06,
+      "loss": 212.3363,
+      "step": 27300
+    },
+    {
+      "epoch": 0.2259171940273814,
+      "grad_norm": 921.5272216796875,
+      "learning_rate": 9.562838156138025e-06,
+      "loss": 142.4072,
+      "step": 27310
+    },
+    {
+      "epoch": 0.22599991727675062,
+      "grad_norm": 935.8705444335938,
+      "learning_rate": 9.562261239125575e-06,
+      "loss": 122.9581,
+      "step": 27320
+    },
+    {
+      "epoch": 0.22608264052611987,
+      "grad_norm": 925.080810546875,
+      "learning_rate": 9.561683959114938e-06,
+      "loss": 127.3665,
+      "step": 27330
+    },
+    {
+      "epoch": 0.2261653637754891,
+      "grad_norm": 1195.8099365234375,
+      "learning_rate": 9.561106316152043e-06,
+      "loss": 133.1869,
+      "step": 27340
+    },
+    {
+      "epoch": 0.22624808702485832,
+      "grad_norm": 784.7255249023438,
+      "learning_rate": 9.56052831028285e-06,
+      "loss": 139.8995,
+      "step": 27350
+    },
+    {
+      "epoch": 0.22633081027422758,
+      "grad_norm": 1093.8494873046875,
+      "learning_rate": 9.559949941553351e-06,
+      "loss": 119.8402,
+      "step": 27360
+    },
+    {
+      "epoch": 0.2264135335235968,
+      "grad_norm": 973.4978637695312,
+      "learning_rate": 9.559371210009562e-06,
+      "loss": 124.6592,
+      "step": 27370
+    },
+    {
+      "epoch": 0.22649625677296603,
+      "grad_norm": 869.2335815429688,
+      "learning_rate": 9.55879211569753e-06,
+      "loss": 94.1615,
+      "step": 27380
+    },
+    {
+      "epoch": 0.22657898002233529,
+      "grad_norm": 1302.50244140625,
+      "learning_rate": 9.55821265866333e-06,
+      "loss": 123.9871,
+      "step": 27390
+    },
+    {
+      "epoch": 0.2266617032717045,
+      "grad_norm": 1908.8446044921875,
+      "learning_rate": 9.55763283895307e-06,
+      "loss": 140.2897,
+      "step": 27400
+    },
+    {
+      "epoch": 0.22674442652107374,
+      "grad_norm": 553.2281494140625,
+      "learning_rate": 9.557052656612882e-06,
+      "loss": 98.0197,
+      "step": 27410
+    },
+    {
+      "epoch": 0.226827149770443,
+      "grad_norm": 693.7083740234375,
+      "learning_rate": 9.556472111688928e-06,
+      "loss": 121.6574,
+      "step": 27420
+    },
+    {
+      "epoch": 0.22690987301981222,
+      "grad_norm": 1643.4599609375,
+      "learning_rate": 9.555891204227399e-06,
+      "loss": 107.1005,
+      "step": 27430
+    },
+    {
+      "epoch": 0.22699259626918145,
+      "grad_norm": 792.270263671875,
+      "learning_rate": 9.555309934274515e-06,
+      "loss": 177.7701,
+      "step": 27440
+    },
+    {
+      "epoch": 0.2270753195185507,
+      "grad_norm": 456.06939697265625,
+      "learning_rate": 9.554728301876525e-06,
+      "loss": 137.5959,
+      "step": 27450
+    },
+    {
+      "epoch": 0.22715804276791993,
+      "grad_norm": 966.9349365234375,
+      "learning_rate": 9.554146307079711e-06,
+      "loss": 99.2951,
+      "step": 27460
+    },
+    {
+      "epoch": 0.22724076601728915,
+      "grad_norm": 1171.660400390625,
+      "learning_rate": 9.553563949930374e-06,
+      "loss": 126.8152,
+      "step": 27470
+    },
+    {
+      "epoch": 0.2273234892666584,
+      "grad_norm": 631.3641967773438,
+      "learning_rate": 9.552981230474849e-06,
+      "loss": 101.0075,
+      "step": 27480
+    },
+    {
+      "epoch": 0.22740621251602763,
+      "grad_norm": 1808.03466796875,
+      "learning_rate": 9.552398148759506e-06,
+      "loss": 105.2634,
+      "step": 27490
+    },
+    {
+      "epoch": 0.22748893576539686,
+      "grad_norm": 1841.3206787109375,
+      "learning_rate": 9.551814704830734e-06,
+      "loss": 124.4455,
+      "step": 27500
+    },
+    {
+      "epoch": 0.2275716590147661,
+      "grad_norm": 2790.470947265625,
+      "learning_rate": 9.551230898734955e-06,
+      "loss": 143.8618,
+      "step": 27510
+    },
+    {
+      "epoch": 0.22765438226413534,
+      "grad_norm": 916.5533447265625,
+      "learning_rate": 9.550646730518623e-06,
+      "loss": 105.8261,
+      "step": 27520
+    },
+    {
+      "epoch": 0.22773710551350457,
+      "grad_norm": 1437.979248046875,
+      "learning_rate": 9.550062200228214e-06,
+      "loss": 101.92,
+      "step": 27530
+    },
+    {
+      "epoch": 0.2278198287628738,
+      "grad_norm": 1014.4559326171875,
+      "learning_rate": 9.549477307910238e-06,
+      "loss": 126.4266,
+      "step": 27540
+    },
+    {
+      "epoch": 0.22790255201224305,
+      "grad_norm": 1683.48779296875,
+      "learning_rate": 9.548892053611232e-06,
+      "loss": 133.186,
+      "step": 27550
+    },
+    {
+      "epoch": 0.22798527526161227,
+      "grad_norm": 864.87744140625,
+      "learning_rate": 9.54830643737776e-06,
+      "loss": 137.8783,
+      "step": 27560
+    },
+    {
+      "epoch": 0.2280679985109815,
+      "grad_norm": 1702.0152587890625,
+      "learning_rate": 9.54772045925642e-06,
+      "loss": 125.6688,
+      "step": 27570
+    },
+    {
+      "epoch": 0.22815072176035076,
+      "grad_norm": 1634.5037841796875,
+      "learning_rate": 9.547134119293835e-06,
+      "loss": 126.7895,
+      "step": 27580
+    },
+    {
+      "epoch": 0.22823344500971998,
+      "grad_norm": 1126.988525390625,
+      "learning_rate": 9.546547417536656e-06,
+      "loss": 117.5014,
+      "step": 27590
+    },
+    {
+      "epoch": 0.2283161682590892,
+      "grad_norm": 774.8274536132812,
+      "learning_rate": 9.545960354031564e-06,
+      "loss": 110.8326,
+      "step": 27600
+    },
+    {
+      "epoch": 0.22839889150845846,
+      "grad_norm": 920.1649780273438,
+      "learning_rate": 9.545372928825271e-06,
+      "loss": 85.081,
+      "step": 27610
+    },
+    {
+      "epoch": 0.2284816147578277,
+      "grad_norm": 550.8583374023438,
+      "learning_rate": 9.544785141964514e-06,
+      "loss": 82.1225,
+      "step": 27620
+    },
+    {
+      "epoch": 0.22856433800719692,
+      "grad_norm": 2031.975830078125,
+      "learning_rate": 9.544196993496062e-06,
+      "loss": 113.8713,
+      "step": 27630
+    },
+    {
+      "epoch": 0.22864706125656617,
+      "grad_norm": 660.8304443359375,
+      "learning_rate": 9.54360848346671e-06,
+      "loss": 111.5653,
+      "step": 27640
+    },
+    {
+      "epoch": 0.2287297845059354,
+      "grad_norm": 1158.63916015625,
+      "learning_rate": 9.543019611923283e-06,
+      "loss": 105.64,
+      "step": 27650
+    },
+    {
+      "epoch": 0.22881250775530462,
+      "grad_norm": 1066.19970703125,
+      "learning_rate": 9.542430378912634e-06,
+      "loss": 112.1879,
+      "step": 27660
+    },
+    {
+      "epoch": 0.22889523100467388,
+      "grad_norm": 1097.3199462890625,
+      "learning_rate": 9.541840784481648e-06,
+      "loss": 112.1237,
+      "step": 27670
+    },
+    {
+      "epoch": 0.2289779542540431,
+      "grad_norm": 1081.3756103515625,
+      "learning_rate": 9.541250828677235e-06,
+      "loss": 144.3419,
+      "step": 27680
+    },
+    {
+      "epoch": 0.22906067750341233,
+      "grad_norm": 899.7421264648438,
+      "learning_rate": 9.540660511546335e-06,
+      "loss": 129.8028,
+      "step": 27690
+    },
+    {
+      "epoch": 0.22914340075278156,
+      "grad_norm": 835.4717407226562,
+      "learning_rate": 9.540069833135917e-06,
+      "loss": 131.3196,
+      "step": 27700
+    },
+    {
+      "epoch": 0.2292261240021508,
+      "grad_norm": 603.4605712890625,
+      "learning_rate": 9.539478793492978e-06,
+      "loss": 94.5269,
+      "step": 27710
+    },
+    {
+      "epoch": 0.22930884725152004,
+      "grad_norm": 827.4793701171875,
+      "learning_rate": 9.538887392664544e-06,
+      "loss": 97.6406,
+      "step": 27720
+    },
+    {
+      "epoch": 0.22939157050088926,
+      "grad_norm": 947.7173461914062,
+      "learning_rate": 9.53829563069767e-06,
+      "loss": 134.4241,
+      "step": 27730
+    },
+    {
+      "epoch": 0.22947429375025852,
+      "grad_norm": 1094.2747802734375,
+      "learning_rate": 9.537703507639444e-06,
+      "loss": 102.8811,
+      "step": 27740
+    },
+    {
+      "epoch": 0.22955701699962774,
+      "grad_norm": 1054.0640869140625,
+      "learning_rate": 9.537111023536973e-06,
+      "loss": 110.9093,
+      "step": 27750
+    },
+    {
+      "epoch": 0.22963974024899697,
+      "grad_norm": 902.435302734375,
+      "learning_rate": 9.536518178437402e-06,
+      "loss": 120.4104,
+      "step": 27760
+    },
+    {
+      "epoch": 0.22972246349836623,
+      "grad_norm": 1348.26025390625,
+      "learning_rate": 9.535924972387898e-06,
+      "loss": 109.3034,
+      "step": 27770
+    },
+    {
+      "epoch": 0.22980518674773545,
+      "grad_norm": 857.2482299804688,
+      "learning_rate": 9.535331405435662e-06,
+      "loss": 125.6188,
+      "step": 27780
+    },
+    {
+      "epoch": 0.22988790999710468,
+      "grad_norm": 1263.4981689453125,
+      "learning_rate": 9.534737477627918e-06,
+      "loss": 152.1994,
+      "step": 27790
+    },
+    {
+      "epoch": 0.22997063324647393,
+      "grad_norm": 1043.9830322265625,
+      "learning_rate": 9.534143189011928e-06,
+      "loss": 139.1974,
+      "step": 27800
+    },
+    {
+      "epoch": 0.23005335649584316,
+      "grad_norm": 1181.6470947265625,
+      "learning_rate": 9.533548539634971e-06,
+      "loss": 124.807,
+      "step": 27810
+    },
+    {
+      "epoch": 0.23013607974521239,
+      "grad_norm": 1061.00244140625,
+      "learning_rate": 9.532953529544365e-06,
+      "loss": 127.3019,
+      "step": 27820
+    },
+    {
+      "epoch": 0.23021880299458164,
+      "grad_norm": 2072.89404296875,
+      "learning_rate": 9.532358158787446e-06,
+      "loss": 112.2069,
+      "step": 27830
+    },
+    {
+      "epoch": 0.23030152624395087,
+      "grad_norm": 726.18505859375,
+      "learning_rate": 9.531762427411592e-06,
+      "loss": 126.4197,
+      "step": 27840
+    },
+    {
+      "epoch": 0.2303842494933201,
+      "grad_norm": 1991.278564453125,
+      "learning_rate": 9.531166335464198e-06,
+      "loss": 257.9386,
+      "step": 27850
+    },
+    {
+      "epoch": 0.23046697274268935,
+      "grad_norm": 648.2035522460938,
+      "learning_rate": 9.530569882992698e-06,
+      "loss": 153.0686,
+      "step": 27860
+    },
+    {
+      "epoch": 0.23054969599205857,
+      "grad_norm": 913.1694946289062,
+      "learning_rate": 9.52997307004454e-06,
+      "loss": 122.5074,
+      "step": 27870
+    },
+    {
+      "epoch": 0.2306324192414278,
+      "grad_norm": 1065.9097900390625,
+      "learning_rate": 9.529375896667218e-06,
+      "loss": 95.8606,
+      "step": 27880
+    },
+    {
+      "epoch": 0.23071514249079703,
+      "grad_norm": 2741.90234375,
+      "learning_rate": 9.528778362908241e-06,
+      "loss": 141.1961,
+      "step": 27890
+    },
+    {
+      "epoch": 0.23079786574016628,
+      "grad_norm": 1098.0302734375,
+      "learning_rate": 9.528180468815155e-06,
+      "loss": 103.5171,
+      "step": 27900
+    },
+    {
+      "epoch": 0.2308805889895355,
+      "grad_norm": 899.1497192382812,
+      "learning_rate": 9.527582214435531e-06,
+      "loss": 143.412,
+      "step": 27910
+    },
+    {
+      "epoch": 0.23096331223890473,
+      "grad_norm": 311.1416320800781,
+      "learning_rate": 9.526983599816968e-06,
+      "loss": 91.3562,
+      "step": 27920
+    },
+    {
+      "epoch": 0.231046035488274,
+      "grad_norm": 1649.770263671875,
+      "learning_rate": 9.526384625007096e-06,
+      "loss": 124.463,
+      "step": 27930
+    },
+    {
+      "epoch": 0.23112875873764321,
+      "grad_norm": 804.8977661132812,
+      "learning_rate": 9.525785290053573e-06,
+      "loss": 138.7314,
+      "step": 27940
+    },
+    {
+      "epoch": 0.23121148198701244,
+      "grad_norm": 859.3971557617188,
+      "learning_rate": 9.525185595004085e-06,
+      "loss": 85.416,
+      "step": 27950
+    },
+    {
+      "epoch": 0.2312942052363817,
+      "grad_norm": 901.3175659179688,
+      "learning_rate": 9.524585539906345e-06,
+      "loss": 93.4797,
+      "step": 27960
+    },
+    {
+      "epoch": 0.23137692848575092,
+      "grad_norm": 1204.8818359375,
+      "learning_rate": 9.523985124808102e-06,
+      "loss": 140.1107,
+      "step": 27970
+    },
+    {
+      "epoch": 0.23145965173512015,
+      "grad_norm": 1186.2174072265625,
+      "learning_rate": 9.523384349757123e-06,
+      "loss": 111.8204,
+      "step": 27980
+    },
+    {
+      "epoch": 0.2315423749844894,
+      "grad_norm": 1178.4058837890625,
+      "learning_rate": 9.522783214801213e-06,
+      "loss": 164.7376,
+      "step": 27990
+    },
+    {
+      "epoch": 0.23162509823385863,
+      "grad_norm": 864.9526977539062,
+      "learning_rate": 9.522181719988196e-06,
+      "loss": 112.4256,
+      "step": 28000
+    },
+    {
+      "epoch": 0.23170782148322785,
+      "grad_norm": 1473.5499267578125,
+      "learning_rate": 9.521579865365935e-06,
+      "loss": 117.2831,
+      "step": 28010
+    },
+    {
+      "epoch": 0.2317905447325971,
+      "grad_norm": 525.5335083007812,
+      "learning_rate": 9.520977650982316e-06,
+      "loss": 150.6173,
+      "step": 28020
+    },
+    {
+      "epoch": 0.23187326798196634,
+      "grad_norm": 886.9190063476562,
+      "learning_rate": 9.520375076885253e-06,
+      "loss": 146.7385,
+      "step": 28030
+    },
+    {
+      "epoch": 0.23195599123133556,
+      "grad_norm": 937.6173095703125,
+      "learning_rate": 9.519772143122691e-06,
+      "loss": 104.6981,
+      "step": 28040
+    },
+    {
+      "epoch": 0.2320387144807048,
+      "grad_norm": 834.8984375,
+      "learning_rate": 9.519168849742603e-06,
+      "loss": 96.541,
+      "step": 28050
+    },
+    {
+      "epoch": 0.23212143773007404,
+      "grad_norm": 863.10986328125,
+      "learning_rate": 9.51856519679299e-06,
+      "loss": 93.61,
+      "step": 28060
+    },
+    {
+      "epoch": 0.23220416097944327,
+      "grad_norm": 683.4520263671875,
+      "learning_rate": 9.517961184321882e-06,
+      "loss": 131.8342,
+      "step": 28070
+    },
+    {
+      "epoch": 0.2322868842288125,
+      "grad_norm": 2610.283203125,
+      "learning_rate": 9.517356812377336e-06,
+      "loss": 189.3643,
+      "step": 28080
+    },
+    {
+      "epoch": 0.23236960747818175,
+      "grad_norm": 993.5272827148438,
+      "learning_rate": 9.516752081007441e-06,
+      "loss": 133.2566,
+      "step": 28090
+    },
+    {
+      "epoch": 0.23245233072755098,
+      "grad_norm": 1212.4417724609375,
+      "learning_rate": 9.51614699026031e-06,
+      "loss": 129.3865,
+      "step": 28100
+    },
+    {
+      "epoch": 0.2325350539769202,
+      "grad_norm": 812.443359375,
+      "learning_rate": 9.515541540184093e-06,
+      "loss": 95.0065,
+      "step": 28110
+    },
+    {
+      "epoch": 0.23261777722628946,
+      "grad_norm": 1204.5474853515625,
+      "learning_rate": 9.514935730826957e-06,
+      "loss": 145.4519,
+      "step": 28120
+    },
+    {
+      "epoch": 0.23270050047565868,
+      "grad_norm": 1555.590576171875,
+      "learning_rate": 9.514329562237107e-06,
+      "loss": 136.7933,
+      "step": 28130
+    },
+    {
+      "epoch": 0.2327832237250279,
+      "grad_norm": 870.7056274414062,
+      "learning_rate": 9.51372303446277e-06,
+      "loss": 122.1039,
+      "step": 28140
+    },
+    {
+      "epoch": 0.23286594697439716,
+      "grad_norm": 1071.2955322265625,
+      "learning_rate": 9.513116147552207e-06,
+      "loss": 102.6043,
+      "step": 28150
+    },
+    {
+      "epoch": 0.2329486702237664,
+      "grad_norm": 797.454833984375,
+      "learning_rate": 9.512508901553703e-06,
+      "loss": 140.6481,
+      "step": 28160
+    },
+    {
+      "epoch": 0.23303139347313562,
+      "grad_norm": 585.2443237304688,
+      "learning_rate": 9.511901296515578e-06,
+      "loss": 113.4713,
+      "step": 28170
+    },
+    {
+      "epoch": 0.23311411672250487,
+      "grad_norm": 870.2528076171875,
+      "learning_rate": 9.511293332486172e-06,
+      "loss": 130.058,
+      "step": 28180
+    },
+    {
+      "epoch": 0.2331968399718741,
+      "grad_norm": 1487.830322265625,
+      "learning_rate": 9.51068500951386e-06,
+      "loss": 132.6228,
+      "step": 28190
+    },
+    {
+      "epoch": 0.23327956322124332,
+      "grad_norm": 3511.711181640625,
+      "learning_rate": 9.510076327647043e-06,
+      "loss": 113.2403,
+      "step": 28200
+    },
+    {
+      "epoch": 0.23336228647061258,
+      "grad_norm": 1605.590087890625,
+      "learning_rate": 9.509467286934151e-06,
+      "loss": 150.1105,
+      "step": 28210
+    },
+    {
+      "epoch": 0.2334450097199818,
+      "grad_norm": 1058.44189453125,
+      "learning_rate": 9.508857887423644e-06,
+      "loss": 122.3643,
+      "step": 28220
+    },
+    {
+      "epoch": 0.23352773296935103,
+      "grad_norm": 820.7523803710938,
+      "learning_rate": 9.508248129164006e-06,
+      "loss": 80.4105,
+      "step": 28230
+    },
+    {
+      "epoch": 0.23361045621872026,
+      "grad_norm": 912.558837890625,
+      "learning_rate": 9.507638012203755e-06,
+      "loss": 152.501,
+      "step": 28240
+    },
+    {
+      "epoch": 0.2336931794680895,
+      "grad_norm": 1337.6898193359375,
+      "learning_rate": 9.507027536591436e-06,
+      "loss": 149.5806,
+      "step": 28250
+    },
+    {
+      "epoch": 0.23377590271745874,
+      "grad_norm": 1526.1043701171875,
+      "learning_rate": 9.506416702375618e-06,
+      "loss": 153.4466,
+      "step": 28260
+    },
+    {
+      "epoch": 0.23385862596682797,
+      "grad_norm": 933.4179077148438,
+      "learning_rate": 9.505805509604906e-06,
+      "loss": 106.969,
+      "step": 28270
+    },
+    {
+      "epoch": 0.23394134921619722,
+      "grad_norm": 1189.154296875,
+      "learning_rate": 9.505193958327927e-06,
+      "loss": 129.8097,
+      "step": 28280
+    },
+    {
+      "epoch": 0.23402407246556645,
+      "grad_norm": 691.6627197265625,
+      "learning_rate": 9.504582048593343e-06,
+      "loss": 99.3678,
+      "step": 28290
+    },
+    {
+      "epoch": 0.23410679571493567,
+      "grad_norm": 1236.778076171875,
+      "learning_rate": 9.503969780449838e-06,
+      "loss": 119.0243,
+      "step": 28300
+    },
+    {
+      "epoch": 0.23418951896430493,
+      "grad_norm": 1243.570068359375,
+      "learning_rate": 9.503357153946126e-06,
+      "loss": 104.1002,
+      "step": 28310
+    },
+    {
+      "epoch": 0.23427224221367415,
+      "grad_norm": 1092.5941162109375,
+      "learning_rate": 9.502744169130955e-06,
+      "loss": 97.2079,
+      "step": 28320
+    },
+    {
+      "epoch": 0.23435496546304338,
+      "grad_norm": 1051.38671875,
+      "learning_rate": 9.502130826053095e-06,
+      "loss": 132.0031,
+      "step": 28330
+    },
+    {
+      "epoch": 0.23443768871241263,
+      "grad_norm": 1343.4345703125,
+      "learning_rate": 9.501517124761347e-06,
+      "loss": 112.9695,
+      "step": 28340
+    },
+    {
+      "epoch": 0.23452041196178186,
+      "grad_norm": 637.9159545898438,
+      "learning_rate": 9.50090306530454e-06,
+      "loss": 85.6707,
+      "step": 28350
+    },
+    {
+      "epoch": 0.2346031352111511,
+      "grad_norm": 976.9581909179688,
+      "learning_rate": 9.500288647731533e-06,
+      "loss": 127.3839,
+      "step": 28360
+    },
+    {
+      "epoch": 0.23468585846052034,
+      "grad_norm": 1000.0506591796875,
+      "learning_rate": 9.49967387209121e-06,
+      "loss": 168.1098,
+      "step": 28370
+    },
+    {
+      "epoch": 0.23476858170988957,
+      "grad_norm": 838.3327026367188,
+      "learning_rate": 9.499058738432492e-06,
+      "loss": 112.4135,
+      "step": 28380
+    },
+    {
+      "epoch": 0.2348513049592588,
+      "grad_norm": 708.0794067382812,
+      "learning_rate": 9.498443246804314e-06,
+      "loss": 120.7116,
+      "step": 28390
+    },
+    {
+      "epoch": 0.23493402820862805,
+      "grad_norm": 1063.5574951171875,
+      "learning_rate": 9.497827397255655e-06,
+      "loss": 99.907,
+      "step": 28400
+    },
+    {
+      "epoch": 0.23501675145799727,
+      "grad_norm": 933.8026123046875,
+      "learning_rate": 9.49721118983551e-06,
+      "loss": 153.6268,
+      "step": 28410
+    },
+    {
+      "epoch": 0.2350994747073665,
+      "grad_norm": 836.0083618164062,
+      "learning_rate": 9.49659462459291e-06,
+      "loss": 123.5795,
+      "step": 28420
+    },
+    {
+      "epoch": 0.23518219795673573,
+      "grad_norm": 1091.86669921875,
+      "learning_rate": 9.495977701576913e-06,
+      "loss": 141.2293,
+      "step": 28430
+    },
+    {
+      "epoch": 0.23526492120610498,
+      "grad_norm": 960.7584838867188,
+      "learning_rate": 9.495360420836603e-06,
+      "loss": 113.6143,
+      "step": 28440
+    },
+    {
+      "epoch": 0.2353476444554742,
+      "grad_norm": 1572.40771484375,
+      "learning_rate": 9.494742782421099e-06,
+      "loss": 159.3734,
+      "step": 28450
+    },
+    {
+      "epoch": 0.23543036770484344,
+      "grad_norm": 1193.2608642578125,
+      "learning_rate": 9.494124786379535e-06,
+      "loss": 128.3347,
+      "step": 28460
+    },
+    {
+      "epoch": 0.2355130909542127,
+      "grad_norm": 1346.21630859375,
+      "learning_rate": 9.49350643276109e-06,
+      "loss": 105.2375,
+      "step": 28470
+    },
+    {
+      "epoch": 0.23559581420358192,
+      "grad_norm": 1731.52001953125,
+      "learning_rate": 9.49288772161496e-06,
+      "loss": 134.3214,
+      "step": 28480
+    },
+    {
+      "epoch": 0.23567853745295114,
+      "grad_norm": 982.36279296875,
+      "learning_rate": 9.492268652990374e-06,
+      "loss": 120.5295,
+      "step": 28490
+    },
+    {
+      "epoch": 0.2357612607023204,
+      "grad_norm": 1203.948486328125,
+      "learning_rate": 9.491649226936586e-06,
+      "loss": 149.1518,
+      "step": 28500
+    },
+    {
+      "epoch": 0.23584398395168962,
+      "grad_norm": 1445.0263671875,
+      "learning_rate": 9.491029443502884e-06,
+      "loss": 133.0336,
+      "step": 28510
+    },
+    {
+      "epoch": 0.23592670720105885,
+      "grad_norm": 1120.74462890625,
+      "learning_rate": 9.490409302738582e-06,
+      "loss": 104.1838,
+      "step": 28520
+    },
+    {
+      "epoch": 0.2360094304504281,
+      "grad_norm": 1055.50830078125,
+      "learning_rate": 9.489788804693017e-06,
+      "loss": 97.1542,
+      "step": 28530
+    },
+    {
+      "epoch": 0.23609215369979733,
+      "grad_norm": 645.5867309570312,
+      "learning_rate": 9.489167949415563e-06,
+      "loss": 124.6525,
+      "step": 28540
+    },
+    {
+      "epoch": 0.23617487694916656,
+      "grad_norm": 723.2225952148438,
+      "learning_rate": 9.48854673695562e-06,
+      "loss": 117.6365,
+      "step": 28550
+    },
+    {
+      "epoch": 0.2362576001985358,
+      "grad_norm": 1657.0540771484375,
+      "learning_rate": 9.48792516736261e-06,
+      "loss": 131.2895,
+      "step": 28560
+    },
+    {
+      "epoch": 0.23634032344790504,
+      "grad_norm": 967.8582763671875,
+      "learning_rate": 9.487303240685992e-06,
+      "loss": 100.9019,
+      "step": 28570
+    },
+    {
+      "epoch": 0.23642304669727426,
+      "grad_norm": 2011.5921630859375,
+      "learning_rate": 9.48668095697525e-06,
+      "loss": 167.2856,
+      "step": 28580
+    },
+    {
+      "epoch": 0.2365057699466435,
+      "grad_norm": 991.98291015625,
+      "learning_rate": 9.486058316279894e-06,
+      "loss": 158.8021,
+      "step": 28590
+    },
+    {
+      "epoch": 0.23658849319601274,
+      "grad_norm": 905.6279296875,
+      "learning_rate": 9.485435318649468e-06,
+      "loss": 124.5288,
+      "step": 28600
+    },
+    {
+      "epoch": 0.23667121644538197,
+      "grad_norm": 793.5623779296875,
+      "learning_rate": 9.484811964133537e-06,
+      "loss": 138.7439,
+      "step": 28610
+    },
+    {
+      "epoch": 0.2367539396947512,
+      "grad_norm": 1176.435791015625,
+      "learning_rate": 9.484188252781701e-06,
+      "loss": 109.5845,
+      "step": 28620
+    },
+    {
+      "epoch": 0.23683666294412045,
+      "grad_norm": 840.8705444335938,
+      "learning_rate": 9.483564184643586e-06,
+      "loss": 90.2001,
+      "step": 28630
+    },
+    {
+      "epoch": 0.23691938619348968,
+      "grad_norm": 705.6690673828125,
+      "learning_rate": 9.482939759768845e-06,
+      "loss": 145.6554,
+      "step": 28640
+    },
+    {
+      "epoch": 0.2370021094428589,
+      "grad_norm": 1252.253662109375,
+      "learning_rate": 9.48231497820716e-06,
+      "loss": 129.5605,
+      "step": 28650
+    },
+    {
+      "epoch": 0.23708483269222816,
+      "grad_norm": 1174.8831787109375,
+      "learning_rate": 9.481689840008246e-06,
+      "loss": 121.5843,
+      "step": 28660
+    },
+    {
+      "epoch": 0.23716755594159739,
+      "grad_norm": 1164.958740234375,
+      "learning_rate": 9.481064345221838e-06,
+      "loss": 130.8124,
+      "step": 28670
+    },
+    {
+      "epoch": 0.2372502791909666,
+      "grad_norm": 1460.279052734375,
+      "learning_rate": 9.480438493897707e-06,
+      "loss": 186.2501,
+      "step": 28680
+    },
+    {
+      "epoch": 0.23733300244033587,
+      "grad_norm": 1548.7734375,
+      "learning_rate": 9.479812286085645e-06,
+      "loss": 122.9342,
+      "step": 28690
+    },
+    {
+      "epoch": 0.2374157256897051,
+      "grad_norm": 485.3841857910156,
+      "learning_rate": 9.47918572183548e-06,
+      "loss": 116.2105,
+      "step": 28700
+    },
+    {
+      "epoch": 0.23749844893907432,
+      "grad_norm": 2227.711181640625,
+      "learning_rate": 9.478558801197065e-06,
+      "loss": 108.795,
+      "step": 28710
+    },
+    {
+      "epoch": 0.23758117218844357,
+      "grad_norm": 1480.4208984375,
+      "learning_rate": 9.47793152422028e-06,
+      "loss": 103.7922,
+      "step": 28720
+    },
+    {
+      "epoch": 0.2376638954378128,
+      "grad_norm": 845.8829956054688,
+      "learning_rate": 9.477303890955032e-06,
+      "loss": 112.599,
+      "step": 28730
+    },
+    {
+      "epoch": 0.23774661868718203,
+      "grad_norm": 593.84619140625,
+      "learning_rate": 9.476675901451264e-06,
+      "loss": 124.1586,
+      "step": 28740
+    },
+    {
+      "epoch": 0.23782934193655128,
+      "grad_norm": 636.6671752929688,
+      "learning_rate": 9.476047555758938e-06,
+      "loss": 172.5131,
+      "step": 28750
+    },
+    {
+      "epoch": 0.2379120651859205,
+      "grad_norm": 4360.0341796875,
+      "learning_rate": 9.475418853928051e-06,
+      "loss": 191.1747,
+      "step": 28760
+    },
+    {
+      "epoch": 0.23799478843528973,
+      "grad_norm": 483.2803649902344,
+      "learning_rate": 9.474789796008625e-06,
+      "loss": 138.2722,
+      "step": 28770
+    },
+    {
+      "epoch": 0.23807751168465896,
+      "grad_norm": 1149.00048828125,
+      "learning_rate": 9.474160382050711e-06,
+      "loss": 126.3032,
+      "step": 28780
+    },
+    {
+      "epoch": 0.23816023493402821,
+      "grad_norm": 1440.6688232421875,
+      "learning_rate": 9.47353061210439e-06,
+      "loss": 96.6904,
+      "step": 28790
+    },
+    {
+      "epoch": 0.23824295818339744,
+      "grad_norm": 2017.08837890625,
+      "learning_rate": 9.47290048621977e-06,
+      "loss": 116.2174,
+      "step": 28800
+    },
+    {
+      "epoch": 0.23832568143276667,
+      "grad_norm": 900.5809326171875,
+      "learning_rate": 9.472270004446984e-06,
+      "loss": 110.8572,
+      "step": 28810
+    },
+    {
+      "epoch": 0.23840840468213592,
+      "grad_norm": 853.006591796875,
+      "learning_rate": 9.4716391668362e-06,
+      "loss": 115.7237,
+      "step": 28820
+    },
+    {
+      "epoch": 0.23849112793150515,
+      "grad_norm": 1115.9259033203125,
+      "learning_rate": 9.471007973437607e-06,
+      "loss": 108.8435,
+      "step": 28830
+    },
+    {
+      "epoch": 0.23857385118087437,
+      "grad_norm": 1762.0460205078125,
+      "learning_rate": 9.470376424301432e-06,
+      "loss": 148.8191,
+      "step": 28840
+    },
+    {
+      "epoch": 0.23865657443024363,
+      "grad_norm": 1468.8228759765625,
+      "learning_rate": 9.46974451947792e-06,
+      "loss": 154.2129,
+      "step": 28850
+    },
+    {
+      "epoch": 0.23873929767961286,
+      "grad_norm": 1209.044921875,
+      "learning_rate": 9.469112259017349e-06,
+      "loss": 107.4766,
+      "step": 28860
+    },
+    {
+      "epoch": 0.23882202092898208,
+      "grad_norm": 1850.443603515625,
+      "learning_rate": 9.468479642970027e-06,
+      "loss": 117.6253,
+      "step": 28870
+    },
+    {
+      "epoch": 0.23890474417835134,
+      "grad_norm": 1324.356689453125,
+      "learning_rate": 9.467846671386287e-06,
+      "loss": 178.3749,
+      "step": 28880
+    },
+    {
+      "epoch": 0.23898746742772056,
+      "grad_norm": 1060.7711181640625,
+      "learning_rate": 9.467213344316493e-06,
+      "loss": 101.0151,
+      "step": 28890
+    },
+    {
+      "epoch": 0.2390701906770898,
+      "grad_norm": 2596.78076171875,
+      "learning_rate": 9.466579661811032e-06,
+      "loss": 149.5662,
+      "step": 28900
+    },
+    {
+      "epoch": 0.23915291392645904,
+      "grad_norm": 1160.0654296875,
+      "learning_rate": 9.46594562392033e-06,
+      "loss": 114.8278,
+      "step": 28910
+    },
+    {
+      "epoch": 0.23923563717582827,
+      "grad_norm": 1474.780029296875,
+      "learning_rate": 9.465311230694828e-06,
+      "loss": 94.9893,
+      "step": 28920
+    },
+    {
+      "epoch": 0.2393183604251975,
+      "grad_norm": 827.9722900390625,
+      "learning_rate": 9.464676482185005e-06,
+      "loss": 82.4494,
+      "step": 28930
+    },
+    {
+      "epoch": 0.23940108367456675,
+      "grad_norm": 490.88360595703125,
+      "learning_rate": 9.464041378441365e-06,
+      "loss": 160.122,
+      "step": 28940
+    },
+    {
+      "epoch": 0.23948380692393598,
+      "grad_norm": 2539.133544921875,
+      "learning_rate": 9.46340591951444e-06,
+      "loss": 135.2263,
+      "step": 28950
+    },
+    {
+      "epoch": 0.2395665301733052,
+      "grad_norm": 620.4109497070312,
+      "learning_rate": 9.462770105454789e-06,
+      "loss": 130.8018,
+      "step": 28960
+    },
+    {
+      "epoch": 0.23964925342267443,
+      "grad_norm": 575.00146484375,
+      "learning_rate": 9.462133936313002e-06,
+      "loss": 123.9707,
+      "step": 28970
+    },
+    {
+      "epoch": 0.23973197667204368,
+      "grad_norm": 764.5715942382812,
+      "learning_rate": 9.461497412139697e-06,
+      "loss": 103.3378,
+      "step": 28980
+    },
+    {
+      "epoch": 0.2398146999214129,
+      "grad_norm": 1179.3966064453125,
+      "learning_rate": 9.46086053298552e-06,
+      "loss": 139.9406,
+      "step": 28990
+    },
+    {
+      "epoch": 0.23989742317078214,
+      "grad_norm": 1064.7276611328125,
+      "learning_rate": 9.460223298901138e-06,
+      "loss": 100.375,
+      "step": 29000
+    },
+    {
+      "epoch": 0.2399801464201514,
+      "grad_norm": 833.2089233398438,
+      "learning_rate": 9.459585709937262e-06,
+      "loss": 120.6056,
+      "step": 29010
+    },
+    {
+      "epoch": 0.24006286966952062,
+      "grad_norm": 1361.6549072265625,
+      "learning_rate": 9.458947766144617e-06,
+      "loss": 129.4685,
+      "step": 29020
+    },
+    {
+      "epoch": 0.24014559291888984,
+      "grad_norm": 1131.559326171875,
+      "learning_rate": 9.458309467573963e-06,
+      "loss": 90.7656,
+      "step": 29030
+    },
+    {
+      "epoch": 0.2402283161682591,
+      "grad_norm": 1216.7392578125,
+      "learning_rate": 9.457670814276083e-06,
+      "loss": 105.4316,
+      "step": 29040
+    },
+    {
+      "epoch": 0.24031103941762832,
+      "grad_norm": 853.8714599609375,
+      "learning_rate": 9.457031806301795e-06,
+      "loss": 94.2898,
+      "step": 29050
+    },
+    {
+      "epoch": 0.24039376266699755,
+      "grad_norm": 826.33837890625,
+      "learning_rate": 9.456392443701943e-06,
+      "loss": 118.5048,
+      "step": 29060
+    },
+    {
+      "epoch": 0.2404764859163668,
+      "grad_norm": 969.9588012695312,
+      "learning_rate": 9.455752726527395e-06,
+      "loss": 158.1088,
+      "step": 29070
+    },
+    {
+      "epoch": 0.24055920916573603,
+      "grad_norm": 591.8674926757812,
+      "learning_rate": 9.45511265482905e-06,
+      "loss": 106.0139,
+      "step": 29080
+    },
+    {
+      "epoch": 0.24064193241510526,
+      "grad_norm": 1056.0914306640625,
+      "learning_rate": 9.454472228657841e-06,
+      "loss": 148.1635,
+      "step": 29090
+    },
+    {
+      "epoch": 0.2407246556644745,
+      "grad_norm": 1000.041259765625,
+      "learning_rate": 9.453831448064717e-06,
+      "loss": 119.1304,
+      "step": 29100
+    },
+    {
+      "epoch": 0.24080737891384374,
+      "grad_norm": 765.8133544921875,
+      "learning_rate": 9.453190313100666e-06,
+      "loss": 83.0749,
+      "step": 29110
+    },
+    {
+      "epoch": 0.24089010216321297,
+      "grad_norm": 963.5242919921875,
+      "learning_rate": 9.4525488238167e-06,
+      "loss": 153.7064,
+      "step": 29120
+    },
+    {
+      "epoch": 0.24097282541258222,
+      "grad_norm": 809.49267578125,
+      "learning_rate": 9.451906980263857e-06,
+      "loss": 122.1319,
+      "step": 29130
+    },
+    {
+      "epoch": 0.24105554866195145,
+      "grad_norm": 800.9232788085938,
+      "learning_rate": 9.451264782493208e-06,
+      "loss": 101.5012,
+      "step": 29140
+    },
+    {
+      "epoch": 0.24113827191132067,
+      "grad_norm": 1233.4398193359375,
+      "learning_rate": 9.450622230555849e-06,
+      "loss": 144.6246,
+      "step": 29150
+    },
+    {
+      "epoch": 0.2412209951606899,
+      "grad_norm": 692.1824951171875,
+      "learning_rate": 9.449979324502905e-06,
+      "loss": 160.0062,
+      "step": 29160
+    },
+    {
+      "epoch": 0.24130371841005915,
+      "grad_norm": 946.5017700195312,
+      "learning_rate": 9.449336064385529e-06,
+      "loss": 105.0953,
+      "step": 29170
+    },
+    {
+      "epoch": 0.24138644165942838,
+      "grad_norm": 1230.076416015625,
+      "learning_rate": 9.4486924502549e-06,
+      "loss": 120.0082,
+      "step": 29180
+    },
+    {
+      "epoch": 0.2414691649087976,
+      "grad_norm": 839.8562622070312,
+      "learning_rate": 9.448048482162231e-06,
+      "loss": 137.8697,
+      "step": 29190
+    },
+    {
+      "epoch": 0.24155188815816686,
+      "grad_norm": 882.9497680664062,
+      "learning_rate": 9.447404160158758e-06,
+      "loss": 119.5869,
+      "step": 29200
+    },
+    {
+      "epoch": 0.2416346114075361,
+      "grad_norm": 1333.1221923828125,
+      "learning_rate": 9.446759484295745e-06,
+      "loss": 116.6337,
+      "step": 29210
+    },
+    {
+      "epoch": 0.2417173346569053,
+      "grad_norm": 955.5294799804688,
+      "learning_rate": 9.44611445462449e-06,
+      "loss": 134.2271,
+      "step": 29220
+    },
+    {
+      "epoch": 0.24180005790627457,
+      "grad_norm": 1251.0631103515625,
+      "learning_rate": 9.445469071196312e-06,
+      "loss": 124.4641,
+      "step": 29230
+    },
+    {
+      "epoch": 0.2418827811556438,
+      "grad_norm": 834.0491943359375,
+      "learning_rate": 9.444823334062562e-06,
+      "loss": 116.2968,
+      "step": 29240
+    },
+    {
+      "epoch": 0.24196550440501302,
+      "grad_norm": 1349.356201171875,
+      "learning_rate": 9.444177243274619e-06,
+      "loss": 131.4607,
+      "step": 29250
+    },
+    {
+      "epoch": 0.24204822765438228,
+      "grad_norm": 1118.03173828125,
+      "learning_rate": 9.443530798883887e-06,
+      "loss": 107.2266,
+      "step": 29260
+    },
+    {
+      "epoch": 0.2421309509037515,
+      "grad_norm": 1354.8619384765625,
+      "learning_rate": 9.442884000941803e-06,
+      "loss": 129.6626,
+      "step": 29270
+    },
+    {
+      "epoch": 0.24221367415312073,
+      "grad_norm": 1301.5723876953125,
+      "learning_rate": 9.44223684949983e-06,
+      "loss": 126.9563,
+      "step": 29280
+    },
+    {
+      "epoch": 0.24229639740248998,
+      "grad_norm": 639.9192504882812,
+      "learning_rate": 9.441589344609457e-06,
+      "loss": 97.4439,
+      "step": 29290
+    },
+    {
+      "epoch": 0.2423791206518592,
+      "grad_norm": 1386.3795166015625,
+      "learning_rate": 9.440941486322205e-06,
+      "loss": 150.2773,
+      "step": 29300
+    },
+    {
+      "epoch": 0.24246184390122844,
+      "grad_norm": 1358.3714599609375,
+      "learning_rate": 9.44029327468962e-06,
+      "loss": 110.78,
+      "step": 29310
+    },
+    {
+      "epoch": 0.24254456715059766,
+      "grad_norm": 757.8045043945312,
+      "learning_rate": 9.439644709763276e-06,
+      "loss": 112.67,
+      "step": 29320
+    },
+    {
+      "epoch": 0.24262729039996692,
+      "grad_norm": 930.5925903320312,
+      "learning_rate": 9.43899579159478e-06,
+      "loss": 121.9743,
+      "step": 29330
+    },
+    {
+      "epoch": 0.24271001364933614,
+      "grad_norm": 1049.4073486328125,
+      "learning_rate": 9.438346520235759e-06,
+      "loss": 100.5406,
+      "step": 29340
+    },
+    {
+      "epoch": 0.24279273689870537,
+      "grad_norm": 1115.40966796875,
+      "learning_rate": 9.437696895737876e-06,
+      "loss": 121.6903,
+      "step": 29350
+    },
+    {
+      "epoch": 0.24287546014807462,
+      "grad_norm": 698.0397338867188,
+      "learning_rate": 9.437046918152817e-06,
+      "loss": 88.6896,
+      "step": 29360
+    },
+    {
+      "epoch": 0.24295818339744385,
+      "grad_norm": 824.7769165039062,
+      "learning_rate": 9.436396587532297e-06,
+      "loss": 126.8226,
+      "step": 29370
+    },
+    {
+      "epoch": 0.24304090664681308,
+      "grad_norm": 1229.65869140625,
+      "learning_rate": 9.435745903928062e-06,
+      "loss": 113.3302,
+      "step": 29380
+    },
+    {
+      "epoch": 0.24312362989618233,
+      "grad_norm": 845.088623046875,
+      "learning_rate": 9.435094867391881e-06,
+      "loss": 154.009,
+      "step": 29390
+    },
+    {
+      "epoch": 0.24320635314555156,
+      "grad_norm": 961.1011962890625,
+      "learning_rate": 9.434443477975557e-06,
+      "loss": 103.9956,
+      "step": 29400
+    },
+    {
+      "epoch": 0.24328907639492078,
+      "grad_norm": 757.7135009765625,
+      "learning_rate": 9.433791735730917e-06,
+      "loss": 98.1805,
+      "step": 29410
+    },
+    {
+      "epoch": 0.24337179964429004,
+      "grad_norm": 1421.3348388671875,
+      "learning_rate": 9.433139640709817e-06,
+      "loss": 132.9433,
+      "step": 29420
+    },
+    {
+      "epoch": 0.24345452289365926,
+      "grad_norm": 608.5304565429688,
+      "learning_rate": 9.432487192964142e-06,
+      "loss": 122.0067,
+      "step": 29430
+    },
+    {
+      "epoch": 0.2435372461430285,
+      "grad_norm": 1166.598876953125,
+      "learning_rate": 9.431834392545803e-06,
+      "loss": 127.8436,
+      "step": 29440
+    },
+    {
+      "epoch": 0.24361996939239774,
+      "grad_norm": 1254.7440185546875,
+      "learning_rate": 9.43118123950674e-06,
+      "loss": 124.5958,
+      "step": 29450
+    },
+    {
+      "epoch": 0.24370269264176697,
+      "grad_norm": 924.8063354492188,
+      "learning_rate": 9.430527733898922e-06,
+      "loss": 102.3073,
+      "step": 29460
+    },
+    {
+      "epoch": 0.2437854158911362,
+      "grad_norm": 957.7825927734375,
+      "learning_rate": 9.429873875774344e-06,
+      "loss": 112.2574,
+      "step": 29470
+    },
+    {
+      "epoch": 0.24386813914050545,
+      "grad_norm": 867.5020141601562,
+      "learning_rate": 9.429219665185034e-06,
+      "loss": 109.0799,
+      "step": 29480
+    },
+    {
+      "epoch": 0.24395086238987468,
+      "grad_norm": 1606.21435546875,
+      "learning_rate": 9.428565102183043e-06,
+      "loss": 114.3639,
+      "step": 29490
+    },
+    {
+      "epoch": 0.2440335856392439,
+      "grad_norm": 782.0588989257812,
+      "learning_rate": 9.42791018682045e-06,
+      "loss": 102.8368,
+      "step": 29500
+    },
+    {
+      "epoch": 0.24411630888861313,
+      "grad_norm": 1499.0159912109375,
+      "learning_rate": 9.427254919149367e-06,
+      "loss": 129.8493,
+      "step": 29510
+    },
+    {
+      "epoch": 0.24419903213798239,
+      "grad_norm": 1734.2025146484375,
+      "learning_rate": 9.426599299221925e-06,
+      "loss": 118.0028,
+      "step": 29520
+    },
+    {
+      "epoch": 0.2442817553873516,
+      "grad_norm": 1104.4652099609375,
+      "learning_rate": 9.425943327090295e-06,
+      "loss": 133.7769,
+      "step": 29530
+    },
+    {
+      "epoch": 0.24436447863672084,
+      "grad_norm": 809.5218505859375,
+      "learning_rate": 9.425287002806666e-06,
+      "loss": 101.5154,
+      "step": 29540
+    },
+    {
+      "epoch": 0.2444472018860901,
+      "grad_norm": 860.76123046875,
+      "learning_rate": 9.42463032642326e-06,
+      "loss": 126.5965,
+      "step": 29550
+    },
+    {
+      "epoch": 0.24452992513545932,
+      "grad_norm": 1226.3048095703125,
+      "learning_rate": 9.423973297992324e-06,
+      "loss": 133.2678,
+      "step": 29560
+    },
+    {
+      "epoch": 0.24461264838482855,
+      "grad_norm": 638.6930541992188,
+      "learning_rate": 9.423315917566137e-06,
+      "loss": 153.0996,
+      "step": 29570
+    },
+    {
+      "epoch": 0.2446953716341978,
+      "grad_norm": 598.7249755859375,
+      "learning_rate": 9.422658185197002e-06,
+      "loss": 122.7943,
+      "step": 29580
+    },
+    {
+      "epoch": 0.24477809488356703,
+      "grad_norm": 711.6947021484375,
+      "learning_rate": 9.422000100937253e-06,
+      "loss": 93.9475,
+      "step": 29590
+    },
+    {
+      "epoch": 0.24486081813293625,
+      "grad_norm": 1122.1689453125,
+      "learning_rate": 9.42134166483925e-06,
+      "loss": 97.7506,
+      "step": 29600
+    },
+    {
+      "epoch": 0.2449435413823055,
+      "grad_norm": 1915.1302490234375,
+      "learning_rate": 9.420682876955382e-06,
+      "loss": 115.1031,
+      "step": 29610
+    },
+    {
+      "epoch": 0.24502626463167473,
+      "grad_norm": 659.7858276367188,
+      "learning_rate": 9.420023737338065e-06,
+      "loss": 120.2869,
+      "step": 29620
+    },
+    {
+      "epoch": 0.24510898788104396,
+      "grad_norm": 834.023193359375,
+      "learning_rate": 9.419364246039745e-06,
+      "loss": 125.4224,
+      "step": 29630
+    },
+    {
+      "epoch": 0.24519171113041321,
+      "grad_norm": 1009.0372924804688,
+      "learning_rate": 9.418704403112894e-06,
+      "loss": 109.2442,
+      "step": 29640
+    },
+    {
+      "epoch": 0.24527443437978244,
+      "grad_norm": 2674.98193359375,
+      "learning_rate": 9.418044208610013e-06,
+      "loss": 156.5225,
+      "step": 29650
+    },
+    {
+      "epoch": 0.24535715762915167,
+      "grad_norm": 1011.0217895507812,
+      "learning_rate": 9.41738366258363e-06,
+      "loss": 126.1811,
+      "step": 29660
+    },
+    {
+      "epoch": 0.24543988087852092,
+      "grad_norm": 610.7017211914062,
+      "learning_rate": 9.416722765086304e-06,
+      "loss": 144.7449,
+      "step": 29670
+    },
+    {
+      "epoch": 0.24552260412789015,
+      "grad_norm": 1031.1539306640625,
+      "learning_rate": 9.416061516170615e-06,
+      "loss": 108.1692,
+      "step": 29680
+    },
+    {
+      "epoch": 0.24560532737725937,
+      "grad_norm": 1801.7032470703125,
+      "learning_rate": 9.415399915889179e-06,
+      "loss": 121.3443,
+      "step": 29690
+    },
+    {
+      "epoch": 0.2456880506266286,
+      "grad_norm": 1428.9144287109375,
+      "learning_rate": 9.414737964294636e-06,
+      "loss": 116.9526,
+      "step": 29700
+    },
+    {
+      "epoch": 0.24577077387599786,
+      "grad_norm": 937.070556640625,
+      "learning_rate": 9.414075661439653e-06,
+      "loss": 111.9231,
+      "step": 29710
+    },
+    {
+      "epoch": 0.24585349712536708,
+      "grad_norm": 1381.5968017578125,
+      "learning_rate": 9.413413007376928e-06,
+      "loss": 163.5947,
+      "step": 29720
+    },
+    {
+      "epoch": 0.2459362203747363,
+      "grad_norm": 1236.41552734375,
+      "learning_rate": 9.412750002159186e-06,
+      "loss": 110.7294,
+      "step": 29730
+    },
+    {
+      "epoch": 0.24601894362410556,
+      "grad_norm": 757.4229125976562,
+      "learning_rate": 9.412086645839177e-06,
+      "loss": 88.9742,
+      "step": 29740
+    },
+    {
+      "epoch": 0.2461016668734748,
+      "grad_norm": 1002.5419311523438,
+      "learning_rate": 9.411422938469683e-06,
+      "loss": 137.723,
+      "step": 29750
+    },
+    {
+      "epoch": 0.24618439012284402,
+      "grad_norm": 636.2584228515625,
+      "learning_rate": 9.41075888010351e-06,
+      "loss": 108.3714,
+      "step": 29760
+    },
+    {
+      "epoch": 0.24626711337221327,
+      "grad_norm": 737.2359008789062,
+      "learning_rate": 9.410094470793497e-06,
+      "loss": 135.6444,
+      "step": 29770
+    },
+    {
+      "epoch": 0.2463498366215825,
+      "grad_norm": 1887.1973876953125,
+      "learning_rate": 9.409429710592505e-06,
+      "loss": 126.0426,
+      "step": 29780
+    },
+    {
+      "epoch": 0.24643255987095172,
+      "grad_norm": 1311.354248046875,
+      "learning_rate": 9.408764599553429e-06,
+      "loss": 156.2838,
+      "step": 29790
+    },
+    {
+      "epoch": 0.24651528312032098,
+      "grad_norm": 1149.238525390625,
+      "learning_rate": 9.408099137729188e-06,
+      "loss": 130.7976,
+      "step": 29800
+    },
+    {
+      "epoch": 0.2465980063696902,
+      "grad_norm": 1126.7828369140625,
+      "learning_rate": 9.407433325172727e-06,
+      "loss": 153.4184,
+      "step": 29810
+    },
+    {
+      "epoch": 0.24668072961905943,
+      "grad_norm": 1250.0152587890625,
+      "learning_rate": 9.406767161937025e-06,
+      "loss": 142.3581,
+      "step": 29820
+    },
+    {
+      "epoch": 0.24676345286842868,
+      "grad_norm": 1395.9898681640625,
+      "learning_rate": 9.406100648075084e-06,
+      "loss": 122.4098,
+      "step": 29830
+    },
+    {
+      "epoch": 0.2468461761177979,
+      "grad_norm": 1577.0543212890625,
+      "learning_rate": 9.405433783639936e-06,
+      "loss": 112.6034,
+      "step": 29840
+    },
+    {
+      "epoch": 0.24692889936716714,
+      "grad_norm": 1184.2935791015625,
+      "learning_rate": 9.40476656868464e-06,
+      "loss": 148.0747,
+      "step": 29850
+    },
+    {
+      "epoch": 0.24701162261653636,
+      "grad_norm": 805.5813598632812,
+      "learning_rate": 9.404099003262282e-06,
+      "loss": 155.5525,
+      "step": 29860
+    },
+    {
+      "epoch": 0.24709434586590562,
+      "grad_norm": 1295.2099609375,
+      "learning_rate": 9.40343108742598e-06,
+      "loss": 149.4768,
+      "step": 29870
+    },
+    {
+      "epoch": 0.24717706911527484,
+      "grad_norm": 995.8720092773438,
+      "learning_rate": 9.402762821228875e-06,
+      "loss": 140.2816,
+      "step": 29880
+    },
+    {
+      "epoch": 0.24725979236464407,
+      "grad_norm": 866.3977661132812,
+      "learning_rate": 9.402094204724138e-06,
+      "loss": 129.4959,
+      "step": 29890
+    },
+    {
+      "epoch": 0.24734251561401333,
+      "grad_norm": 1852.4481201171875,
+      "learning_rate": 9.401425237964966e-06,
+      "loss": 102.9619,
+      "step": 29900
+    },
+    {
+      "epoch": 0.24742523886338255,
+      "grad_norm": 743.556640625,
+      "learning_rate": 9.400755921004592e-06,
+      "loss": 85.2109,
+      "step": 29910
+    },
+    {
+      "epoch": 0.24750796211275178,
+      "grad_norm": 809.3905639648438,
+      "learning_rate": 9.400086253896264e-06,
+      "loss": 106.4736,
+      "step": 29920
+    },
+    {
+      "epoch": 0.24759068536212103,
+      "grad_norm": 783.1993408203125,
+      "learning_rate": 9.399416236693264e-06,
+      "loss": 125.8943,
+      "step": 29930
+    },
+    {
+      "epoch": 0.24767340861149026,
+      "grad_norm": 772.568115234375,
+      "learning_rate": 9.398745869448909e-06,
+      "loss": 123.8559,
+      "step": 29940
+    },
+    {
+      "epoch": 0.24775613186085949,
+      "grad_norm": 1002.6859130859375,
+      "learning_rate": 9.39807515221653e-06,
+      "loss": 101.8871,
+      "step": 29950
+    },
+    {
+      "epoch": 0.24783885511022874,
+      "grad_norm": 1258.36572265625,
+      "learning_rate": 9.397404085049496e-06,
+      "loss": 98.4138,
+      "step": 29960
+    },
+    {
+      "epoch": 0.24792157835959797,
+      "grad_norm": 1145.0703125,
+      "learning_rate": 9.3967326680012e-06,
+      "loss": 118.3927,
+      "step": 29970
+    },
+    {
+      "epoch": 0.2480043016089672,
+      "grad_norm": 1031.3804931640625,
+      "learning_rate": 9.396060901125064e-06,
+      "loss": 105.3649,
+      "step": 29980
+    },
+    {
+      "epoch": 0.24808702485833645,
+      "grad_norm": 2366.289794921875,
+      "learning_rate": 9.395388784474538e-06,
+      "loss": 168.2479,
+      "step": 29990
+    },
+    {
+      "epoch": 0.24816974810770567,
+      "grad_norm": 369.7084045410156,
+      "learning_rate": 9.394716318103098e-06,
+      "loss": 121.3149,
+      "step": 30000
+    },
+    {
+      "epoch": 0.2482524713570749,
+      "grad_norm": 1409.655029296875,
+      "learning_rate": 9.394043502064249e-06,
+      "loss": 105.2097,
+      "step": 30010
+    },
+    {
+      "epoch": 0.24833519460644415,
+      "grad_norm": 1403.9825439453125,
+      "learning_rate": 9.393370336411527e-06,
+      "loss": 147.3934,
+      "step": 30020
+    },
+    {
+      "epoch": 0.24841791785581338,
+      "grad_norm": 929.0065307617188,
+      "learning_rate": 9.392696821198488e-06,
+      "loss": 124.7842,
+      "step": 30030
+    },
+    {
+      "epoch": 0.2485006411051826,
+      "grad_norm": 828.625244140625,
+      "learning_rate": 9.392022956478724e-06,
+      "loss": 112.9368,
+      "step": 30040
+    },
+    {
+      "epoch": 0.24858336435455183,
+      "grad_norm": 3000.644287109375,
+      "learning_rate": 9.391348742305849e-06,
+      "loss": 148.1125,
+      "step": 30050
+    },
+    {
+      "epoch": 0.2486660876039211,
+      "grad_norm": 908.979248046875,
+      "learning_rate": 9.390674178733508e-06,
+      "loss": 109.6535,
+      "step": 30060
+    },
+    {
+      "epoch": 0.24874881085329031,
+      "grad_norm": 1394.7421875,
+      "learning_rate": 9.389999265815373e-06,
+      "loss": 112.9092,
+      "step": 30070
+    },
+    {
+      "epoch": 0.24883153410265954,
+      "grad_norm": 1016.4574584960938,
+      "learning_rate": 9.389324003605144e-06,
+      "loss": 168.127,
+      "step": 30080
+    },
+    {
+      "epoch": 0.2489142573520288,
+      "grad_norm": 1174.79443359375,
+      "learning_rate": 9.388648392156547e-06,
+      "loss": 112.0588,
+      "step": 30090
+    },
+    {
+      "epoch": 0.24899698060139802,
+      "grad_norm": 2049.481689453125,
+      "learning_rate": 9.387972431523341e-06,
+      "loss": 127.4066,
+      "step": 30100
+    },
+    {
+      "epoch": 0.24907970385076725,
+      "grad_norm": 712.5939331054688,
+      "learning_rate": 9.387296121759305e-06,
+      "loss": 98.8517,
+      "step": 30110
+    },
+    {
+      "epoch": 0.2491624271001365,
+      "grad_norm": 872.814208984375,
+      "learning_rate": 9.386619462918254e-06,
+      "loss": 100.3602,
+      "step": 30120
+    },
+    {
+      "epoch": 0.24924515034950573,
+      "grad_norm": 649.6997680664062,
+      "learning_rate": 9.385942455054022e-06,
+      "loss": 119.4873,
+      "step": 30130
+    },
+    {
+      "epoch": 0.24932787359887496,
+      "grad_norm": 655.9243774414062,
+      "learning_rate": 9.385265098220478e-06,
+      "loss": 124.5341,
+      "step": 30140
+    },
+    {
+      "epoch": 0.2494105968482442,
+      "grad_norm": 822.2200927734375,
+      "learning_rate": 9.384587392471516e-06,
+      "loss": 162.9077,
+      "step": 30150
+    },
+    {
+      "epoch": 0.24949332009761344,
+      "grad_norm": 940.898193359375,
+      "learning_rate": 9.383909337861058e-06,
+      "loss": 118.583,
+      "step": 30160
+    },
+    {
+      "epoch": 0.24957604334698266,
+      "grad_norm": 1128.41943359375,
+      "learning_rate": 9.383230934443053e-06,
+      "loss": 136.6669,
+      "step": 30170
+    },
+    {
+      "epoch": 0.24965876659635192,
+      "grad_norm": 631.8690795898438,
+      "learning_rate": 9.382552182271478e-06,
+      "loss": 97.5566,
+      "step": 30180
+    },
+    {
+      "epoch": 0.24974148984572114,
+      "grad_norm": 1021.1989135742188,
+      "learning_rate": 9.38187308140034e-06,
+      "loss": 146.495,
+      "step": 30190
+    },
+    {
+      "epoch": 0.24982421309509037,
+      "grad_norm": 1181.3828125,
+      "learning_rate": 9.381193631883672e-06,
+      "loss": 150.6252,
+      "step": 30200
+    },
+    {
+      "epoch": 0.24990693634445962,
+      "grad_norm": 814.0835571289062,
+      "learning_rate": 9.380513833775531e-06,
+      "loss": 114.7124,
+      "step": 30210
+    },
+    {
+      "epoch": 0.24998965959382885,
+      "grad_norm": 1297.4193115234375,
+      "learning_rate": 9.37983368713001e-06,
+      "loss": 97.1973,
+      "step": 30220
+    },
+    {
+      "epoch": 0.2500723828431981,
+      "grad_norm": 800.78564453125,
+      "learning_rate": 9.379153192001223e-06,
+      "loss": 98.411,
+      "step": 30230
+    },
+    {
+      "epoch": 0.2501551060925673,
+      "grad_norm": 1123.505859375,
+      "learning_rate": 9.378472348443315e-06,
+      "loss": 119.3296,
+      "step": 30240
+    },
+    {
+      "epoch": 0.25023782934193656,
+      "grad_norm": 888.609375,
+      "learning_rate": 9.377791156510456e-06,
+      "loss": 74.0182,
+      "step": 30250
+    },
+    {
+      "epoch": 0.2503205525913058,
+      "grad_norm": 713.3021240234375,
+      "learning_rate": 9.377109616256846e-06,
+      "loss": 147.7178,
+      "step": 30260
+    },
+    {
+      "epoch": 0.250403275840675,
+      "grad_norm": 1878.27880859375,
+      "learning_rate": 9.37642772773671e-06,
+      "loss": 154.91,
+      "step": 30270
+    },
+    {
+      "epoch": 0.25048599909004426,
+      "grad_norm": 625.2662353515625,
+      "learning_rate": 9.375745491004307e-06,
+      "loss": 90.2972,
+      "step": 30280
+    },
+    {
+      "epoch": 0.2505687223394135,
+      "grad_norm": 621.6907958984375,
+      "learning_rate": 9.375062906113916e-06,
+      "loss": 126.9956,
+      "step": 30290
+    },
+    {
+      "epoch": 0.2506514455887827,
+      "grad_norm": 1250.3077392578125,
+      "learning_rate": 9.37437997311985e-06,
+      "loss": 107.0872,
+      "step": 30300
+    },
+    {
+      "epoch": 0.25073416883815197,
+      "grad_norm": 720.9534912109375,
+      "learning_rate": 9.373696692076446e-06,
+      "loss": 105.0815,
+      "step": 30310
+    },
+    {
+      "epoch": 0.25081689208752117,
+      "grad_norm": 684.220703125,
+      "learning_rate": 9.373013063038066e-06,
+      "loss": 129.8487,
+      "step": 30320
+    },
+    {
+      "epoch": 0.2508996153368904,
+      "grad_norm": 1063.6759033203125,
+      "learning_rate": 9.372329086059108e-06,
+      "loss": 135.9542,
+      "step": 30330
+    },
+    {
+      "epoch": 0.2509823385862597,
+      "grad_norm": 1607.3919677734375,
+      "learning_rate": 9.37164476119399e-06,
+      "loss": 142.6617,
+      "step": 30340
+    },
+    {
+      "epoch": 0.2510650618356289,
+      "grad_norm": 542.721435546875,
+      "learning_rate": 9.370960088497162e-06,
+      "loss": 106.839,
+      "step": 30350
+    },
+    {
+      "epoch": 0.25114778508499813,
+      "grad_norm": 549.24560546875,
+      "learning_rate": 9.370275068023097e-06,
+      "loss": 129.0447,
+      "step": 30360
+    },
+    {
+      "epoch": 0.2512305083343674,
+      "grad_norm": 960.9791259765625,
+      "learning_rate": 9.369589699826306e-06,
+      "loss": 140.4398,
+      "step": 30370
+    },
+    {
+      "epoch": 0.2513132315837366,
+      "grad_norm": 1057.4302978515625,
+      "learning_rate": 9.368903983961315e-06,
+      "loss": 138.126,
+      "step": 30380
+    },
+    {
+      "epoch": 0.25139595483310584,
+      "grad_norm": 1365.5726318359375,
+      "learning_rate": 9.368217920482684e-06,
+      "loss": 139.174,
+      "step": 30390
+    },
+    {
+      "epoch": 0.2514786780824751,
+      "grad_norm": 887.7735595703125,
+      "learning_rate": 9.367531509445001e-06,
+      "loss": 129.102,
+      "step": 30400
+    },
+    {
+      "epoch": 0.2515614013318443,
+      "grad_norm": 1251.561767578125,
+      "learning_rate": 9.366844750902878e-06,
+      "loss": 121.9665,
+      "step": 30410
+    },
+    {
+      "epoch": 0.25164412458121355,
+      "grad_norm": 881.740234375,
+      "learning_rate": 9.36615764491096e-06,
+      "loss": 82.5869,
+      "step": 30420
+    },
+    {
+      "epoch": 0.2517268478305828,
+      "grad_norm": 821.780029296875,
+      "learning_rate": 9.365470191523917e-06,
+      "loss": 146.2663,
+      "step": 30430
+    },
+    {
+      "epoch": 0.251809571079952,
+      "grad_norm": 626.3407592773438,
+      "learning_rate": 9.364782390796446e-06,
+      "loss": 86.4238,
+      "step": 30440
+    },
+    {
+      "epoch": 0.25189229432932125,
+      "grad_norm": 1124.6002197265625,
+      "learning_rate": 9.364094242783272e-06,
+      "loss": 146.8187,
+      "step": 30450
+    },
+    {
+      "epoch": 0.2519750175786905,
+      "grad_norm": 631.0712890625,
+      "learning_rate": 9.363405747539147e-06,
+      "loss": 98.5037,
+      "step": 30460
+    },
+    {
+      "epoch": 0.2520577408280597,
+      "grad_norm": 949.3443603515625,
+      "learning_rate": 9.362716905118851e-06,
+      "loss": 139.6968,
+      "step": 30470
+    },
+    {
+      "epoch": 0.25214046407742896,
+      "grad_norm": 513.1497802734375,
+      "learning_rate": 9.362027715577195e-06,
+      "loss": 118.3806,
+      "step": 30480
+    },
+    {
+      "epoch": 0.2522231873267982,
+      "grad_norm": 1057.8067626953125,
+      "learning_rate": 9.361338178969012e-06,
+      "loss": 108.9348,
+      "step": 30490
+    },
+    {
+      "epoch": 0.2523059105761674,
+      "grad_norm": 903.6969604492188,
+      "learning_rate": 9.360648295349165e-06,
+      "loss": 105.4085,
+      "step": 30500
+    },
+    {
+      "epoch": 0.25238863382553667,
+      "grad_norm": 2535.44189453125,
+      "learning_rate": 9.359958064772547e-06,
+      "loss": 161.6714,
+      "step": 30510
+    },
+    {
+      "epoch": 0.2524713570749059,
+      "grad_norm": 1677.7100830078125,
+      "learning_rate": 9.359267487294075e-06,
+      "loss": 128.2102,
+      "step": 30520
+    },
+    {
+      "epoch": 0.2525540803242751,
+      "grad_norm": 1912.9716796875,
+      "learning_rate": 9.358576562968695e-06,
+      "loss": 118.4899,
+      "step": 30530
+    },
+    {
+      "epoch": 0.2526368035736444,
+      "grad_norm": 763.476318359375,
+      "learning_rate": 9.357885291851382e-06,
+      "loss": 124.9722,
+      "step": 30540
+    },
+    {
+      "epoch": 0.25271952682301363,
+      "grad_norm": 1949.4473876953125,
+      "learning_rate": 9.357193673997133e-06,
+      "loss": 104.1943,
+      "step": 30550
+    },
+    {
+      "epoch": 0.25280225007238283,
+      "grad_norm": 1901.4537353515625,
+      "learning_rate": 9.356501709460984e-06,
+      "loss": 108.5047,
+      "step": 30560
+    },
+    {
+      "epoch": 0.2528849733217521,
+      "grad_norm": 773.7173461914062,
+      "learning_rate": 9.355809398297986e-06,
+      "loss": 95.4959,
+      "step": 30570
+    },
+    {
+      "epoch": 0.25296769657112134,
+      "grad_norm": 1136.826171875,
+      "learning_rate": 9.355116740563225e-06,
+      "loss": 136.653,
+      "step": 30580
+    },
+    {
+      "epoch": 0.25305041982049054,
+      "grad_norm": 826.9710693359375,
+      "learning_rate": 9.354423736311813e-06,
+      "loss": 119.1377,
+      "step": 30590
+    },
+    {
+      "epoch": 0.2531331430698598,
+      "grad_norm": 1087.69677734375,
+      "learning_rate": 9.353730385598887e-06,
+      "loss": 101.276,
+      "step": 30600
+    },
+    {
+      "epoch": 0.25321586631922904,
+      "grad_norm": 567.2242431640625,
+      "learning_rate": 9.353036688479615e-06,
+      "loss": 116.7849,
+      "step": 30610
+    },
+    {
+      "epoch": 0.25329858956859824,
+      "grad_norm": 1647.7808837890625,
+      "learning_rate": 9.352342645009193e-06,
+      "loss": 142.3532,
+      "step": 30620
+    },
+    {
+      "epoch": 0.2533813128179675,
+      "grad_norm": 1223.5712890625,
+      "learning_rate": 9.35164825524284e-06,
+      "loss": 106.8768,
+      "step": 30630
+    },
+    {
+      "epoch": 0.25346403606733675,
+      "grad_norm": 1215.446044921875,
+      "learning_rate": 9.350953519235807e-06,
+      "loss": 142.7279,
+      "step": 30640
+    },
+    {
+      "epoch": 0.25354675931670595,
+      "grad_norm": 1093.0865478515625,
+      "learning_rate": 9.35025843704337e-06,
+      "loss": 133.1846,
+      "step": 30650
+    },
+    {
+      "epoch": 0.2536294825660752,
+      "grad_norm": 603.0365600585938,
+      "learning_rate": 9.349563008720836e-06,
+      "loss": 143.9578,
+      "step": 30660
+    },
+    {
+      "epoch": 0.25371220581544446,
+      "grad_norm": 926.9697265625,
+      "learning_rate": 9.348867234323534e-06,
+      "loss": 115.379,
+      "step": 30670
+    },
+    {
+      "epoch": 0.25379492906481366,
+      "grad_norm": 1196.4434814453125,
+      "learning_rate": 9.348171113906826e-06,
+      "loss": 128.1764,
+      "step": 30680
+    },
+    {
+      "epoch": 0.2538776523141829,
+      "grad_norm": 750.9150390625,
+      "learning_rate": 9.347474647526095e-06,
+      "loss": 194.924,
+      "step": 30690
+    },
+    {
+      "epoch": 0.2539603755635521,
+      "grad_norm": 1341.235595703125,
+      "learning_rate": 9.34677783523676e-06,
+      "loss": 137.8295,
+      "step": 30700
+    },
+    {
+      "epoch": 0.25404309881292136,
+      "grad_norm": 1126.3736572265625,
+      "learning_rate": 9.346080677094262e-06,
+      "loss": 132.7227,
+      "step": 30710
+    },
+    {
+      "epoch": 0.2541258220622906,
+      "grad_norm": 824.2003784179688,
+      "learning_rate": 9.345383173154072e-06,
+      "loss": 133.4808,
+      "step": 30720
+    },
+    {
+      "epoch": 0.2542085453116598,
+      "grad_norm": 929.5851440429688,
+      "learning_rate": 9.344685323471682e-06,
+      "loss": 109.8865,
+      "step": 30730
+    },
+    {
+      "epoch": 0.25429126856102907,
+      "grad_norm": 763.9591674804688,
+      "learning_rate": 9.343987128102624e-06,
+      "loss": 114.478,
+      "step": 30740
+    },
+    {
+      "epoch": 0.2543739918103983,
+      "grad_norm": 896.4277954101562,
+      "learning_rate": 9.343288587102444e-06,
+      "loss": 139.7716,
+      "step": 30750
+    },
+    {
+      "epoch": 0.2544567150597675,
+      "grad_norm": 890.5599975585938,
+      "learning_rate": 9.342589700526725e-06,
+      "loss": 119.8424,
+      "step": 30760
+    },
+    {
+      "epoch": 0.2545394383091368,
+      "grad_norm": 954.55322265625,
+      "learning_rate": 9.341890468431072e-06,
+      "loss": 197.9463,
+      "step": 30770
+    },
+    {
+      "epoch": 0.25462216155850603,
+      "grad_norm": 962.0372314453125,
+      "learning_rate": 9.341190890871123e-06,
+      "loss": 173.233,
+      "step": 30780
+    },
+    {
+      "epoch": 0.25470488480787523,
+      "grad_norm": 1072.42724609375,
+      "learning_rate": 9.340490967902535e-06,
+      "loss": 114.3112,
+      "step": 30790
+    },
+    {
+      "epoch": 0.2547876080572445,
+      "grad_norm": 927.2454223632812,
+      "learning_rate": 9.339790699581004e-06,
+      "loss": 98.9923,
+      "step": 30800
+    },
+    {
+      "epoch": 0.25487033130661374,
+      "grad_norm": 485.0035400390625,
+      "learning_rate": 9.339090085962244e-06,
+      "loss": 109.2545,
+      "step": 30810
+    },
+    {
+      "epoch": 0.25495305455598294,
+      "grad_norm": 1289.7406005859375,
+      "learning_rate": 9.338389127101998e-06,
+      "loss": 137.8362,
+      "step": 30820
+    },
+    {
+      "epoch": 0.2550357778053522,
+      "grad_norm": 655.8922119140625,
+      "learning_rate": 9.337687823056041e-06,
+      "loss": 101.9889,
+      "step": 30830
+    },
+    {
+      "epoch": 0.25511850105472145,
+      "grad_norm": 1242.1337890625,
+      "learning_rate": 9.336986173880169e-06,
+      "loss": 106.3836,
+      "step": 30840
+    },
+    {
+      "epoch": 0.25520122430409065,
+      "grad_norm": 580.3970947265625,
+      "learning_rate": 9.336284179630215e-06,
+      "loss": 94.8493,
+      "step": 30850
+    },
+    {
+      "epoch": 0.2552839475534599,
+      "grad_norm": 789.4066162109375,
+      "learning_rate": 9.335581840362026e-06,
+      "loss": 74.5354,
+      "step": 30860
+    },
+    {
+      "epoch": 0.25536667080282915,
+      "grad_norm": 1426.3404541015625,
+      "learning_rate": 9.33487915613149e-06,
+      "loss": 108.0914,
+      "step": 30870
+    },
+    {
+      "epoch": 0.25544939405219835,
+      "grad_norm": 826.1908569335938,
+      "learning_rate": 9.334176126994512e-06,
+      "loss": 109.946,
+      "step": 30880
+    },
+    {
+      "epoch": 0.2555321173015676,
+      "grad_norm": 755.2938232421875,
+      "learning_rate": 9.333472753007031e-06,
+      "loss": 111.354,
+      "step": 30890
+    },
+    {
+      "epoch": 0.25561484055093686,
+      "grad_norm": 780.1597290039062,
+      "learning_rate": 9.332769034225012e-06,
+      "loss": 142.2512,
+      "step": 30900
+    },
+    {
+      "epoch": 0.25569756380030606,
+      "grad_norm": 731.4649047851562,
+      "learning_rate": 9.332064970704445e-06,
+      "loss": 156.2841,
+      "step": 30910
+    },
+    {
+      "epoch": 0.2557802870496753,
+      "grad_norm": 1020.5748291015625,
+      "learning_rate": 9.33136056250135e-06,
+      "loss": 127.7295,
+      "step": 30920
+    },
+    {
+      "epoch": 0.25586301029904457,
+      "grad_norm": 698.0821533203125,
+      "learning_rate": 9.330655809671773e-06,
+      "loss": 92.2535,
+      "step": 30930
+    },
+    {
+      "epoch": 0.25594573354841377,
+      "grad_norm": 698.9208374023438,
+      "learning_rate": 9.32995071227179e-06,
+      "loss": 124.9385,
+      "step": 30940
+    },
+    {
+      "epoch": 0.256028456797783,
+      "grad_norm": 1320.0196533203125,
+      "learning_rate": 9.3292452703575e-06,
+      "loss": 128.1827,
+      "step": 30950
+    },
+    {
+      "epoch": 0.2561111800471523,
+      "grad_norm": 2012.84033203125,
+      "learning_rate": 9.328539483985031e-06,
+      "loss": 155.1285,
+      "step": 30960
+    },
+    {
+      "epoch": 0.2561939032965215,
+      "grad_norm": 1151.90625,
+      "learning_rate": 9.327833353210541e-06,
+      "loss": 111.3364,
+      "step": 30970
+    },
+    {
+      "epoch": 0.25627662654589073,
+      "grad_norm": 1791.881103515625,
+      "learning_rate": 9.327126878090214e-06,
+      "loss": 118.88,
+      "step": 30980
+    },
+    {
+      "epoch": 0.25635934979526,
+      "grad_norm": 741.9896850585938,
+      "learning_rate": 9.32642005868026e-06,
+      "loss": 122.3429,
+      "step": 30990
+    },
+    {
+      "epoch": 0.2564420730446292,
+      "grad_norm": 712.4248657226562,
+      "learning_rate": 9.325712895036916e-06,
+      "loss": 125.5105,
+      "step": 31000
+    },
+    {
+      "epoch": 0.25652479629399844,
+      "grad_norm": 1130.374267578125,
+      "learning_rate": 9.32500538721645e-06,
+      "loss": 86.682,
+      "step": 31010
+    },
+    {
+      "epoch": 0.2566075195433677,
+      "grad_norm": 1284.844970703125,
+      "learning_rate": 9.324297535275156e-06,
+      "loss": 114.092,
+      "step": 31020
+    },
+    {
+      "epoch": 0.2566902427927369,
+      "grad_norm": 911.753173828125,
+      "learning_rate": 9.323589339269352e-06,
+      "loss": 106.8176,
+      "step": 31030
+    },
+    {
+      "epoch": 0.25677296604210614,
+      "grad_norm": 3918.962646484375,
+      "learning_rate": 9.322880799255385e-06,
+      "loss": 160.4931,
+      "step": 31040
+    },
+    {
+      "epoch": 0.25685568929147534,
+      "grad_norm": 835.4883422851562,
+      "learning_rate": 9.322171915289635e-06,
+      "loss": 108.218,
+      "step": 31050
+    },
+    {
+      "epoch": 0.2569384125408446,
+      "grad_norm": 745.3687133789062,
+      "learning_rate": 9.321462687428499e-06,
+      "loss": 103.9572,
+      "step": 31060
+    },
+    {
+      "epoch": 0.25702113579021385,
+      "grad_norm": 2105.262939453125,
+      "learning_rate": 9.320753115728413e-06,
+      "loss": 137.9624,
+      "step": 31070
+    },
+    {
+      "epoch": 0.25710385903958305,
+      "grad_norm": 1638.893798828125,
+      "learning_rate": 9.320043200245829e-06,
+      "loss": 76.6734,
+      "step": 31080
+    },
+    {
+      "epoch": 0.2571865822889523,
+      "grad_norm": 1251.9464111328125,
+      "learning_rate": 9.319332941037235e-06,
+      "loss": 128.9104,
+      "step": 31090
+    },
+    {
+      "epoch": 0.25726930553832156,
+      "grad_norm": 946.0052490234375,
+      "learning_rate": 9.31862233815914e-06,
+      "loss": 83.5885,
+      "step": 31100
+    },
+    {
+      "epoch": 0.25735202878769076,
+      "grad_norm": 613.86181640625,
+      "learning_rate": 9.317911391668087e-06,
+      "loss": 88.8766,
+      "step": 31110
+    },
+    {
+      "epoch": 0.25743475203706,
+      "grad_norm": 980.6512451171875,
+      "learning_rate": 9.317200101620641e-06,
+      "loss": 111.9333,
+      "step": 31120
+    },
+    {
+      "epoch": 0.25751747528642926,
+      "grad_norm": 959.6453857421875,
+      "learning_rate": 9.316488468073397e-06,
+      "loss": 87.5497,
+      "step": 31130
+    },
+    {
+      "epoch": 0.25760019853579846,
+      "grad_norm": 784.59033203125,
+      "learning_rate": 9.315776491082973e-06,
+      "loss": 140.7631,
+      "step": 31140
+    },
+    {
+      "epoch": 0.2576829217851677,
+      "grad_norm": 1294.2275390625,
+      "learning_rate": 9.315064170706023e-06,
+      "loss": 114.7354,
+      "step": 31150
+    },
+    {
+      "epoch": 0.25776564503453697,
+      "grad_norm": 1613.03857421875,
+      "learning_rate": 9.31435150699922e-06,
+      "loss": 123.3567,
+      "step": 31160
+    },
+    {
+      "epoch": 0.25784836828390617,
+      "grad_norm": 1174.7305908203125,
+      "learning_rate": 9.313638500019267e-06,
+      "loss": 133.3073,
+      "step": 31170
+    },
+    {
+      "epoch": 0.2579310915332754,
+      "grad_norm": 587.86572265625,
+      "learning_rate": 9.312925149822895e-06,
+      "loss": 90.9177,
+      "step": 31180
+    },
+    {
+      "epoch": 0.2580138147826447,
+      "grad_norm": 934.0054931640625,
+      "learning_rate": 9.312211456466862e-06,
+      "loss": 127.4864,
+      "step": 31190
+    },
+    {
+      "epoch": 0.2580965380320139,
+      "grad_norm": 712.6873779296875,
+      "learning_rate": 9.311497420007955e-06,
+      "loss": 111.8241,
+      "step": 31200
+    },
+    {
+      "epoch": 0.25817926128138313,
+      "grad_norm": 1259.027587890625,
+      "learning_rate": 9.310783040502987e-06,
+      "loss": 120.1594,
+      "step": 31210
+    },
+    {
+      "epoch": 0.2582619845307524,
+      "grad_norm": 1388.41162109375,
+      "learning_rate": 9.310068318008794e-06,
+      "loss": 121.605,
+      "step": 31220
+    },
+    {
+      "epoch": 0.2583447077801216,
+      "grad_norm": 1037.8282470703125,
+      "learning_rate": 9.309353252582246e-06,
+      "loss": 138.8729,
+      "step": 31230
+    },
+    {
+      "epoch": 0.25842743102949084,
+      "grad_norm": 943.52490234375,
+      "learning_rate": 9.308637844280236e-06,
+      "loss": 132.2363,
+      "step": 31240
+    },
+    {
+      "epoch": 0.2585101542788601,
+      "grad_norm": 1297.0338134765625,
+      "learning_rate": 9.307922093159688e-06,
+      "loss": 113.9879,
+      "step": 31250
+    },
+    {
+      "epoch": 0.2585928775282293,
+      "grad_norm": 739.4756469726562,
+      "learning_rate": 9.30720599927755e-06,
+      "loss": 79.2995,
+      "step": 31260
+    },
+    {
+      "epoch": 0.25867560077759855,
+      "grad_norm": 1136.6614990234375,
+      "learning_rate": 9.306489562690797e-06,
+      "loss": 148.8123,
+      "step": 31270
+    },
+    {
+      "epoch": 0.2587583240269678,
+      "grad_norm": 1102.057861328125,
+      "learning_rate": 9.305772783456435e-06,
+      "loss": 126.115,
+      "step": 31280
+    },
+    {
+      "epoch": 0.258841047276337,
+      "grad_norm": 1000.919677734375,
+      "learning_rate": 9.305055661631493e-06,
+      "loss": 128.0628,
+      "step": 31290
+    },
+    {
+      "epoch": 0.25892377052570625,
+      "grad_norm": 1486.086669921875,
+      "learning_rate": 9.304338197273029e-06,
+      "loss": 141.9742,
+      "step": 31300
+    },
+    {
+      "epoch": 0.2590064937750755,
+      "grad_norm": 1208.7861328125,
+      "learning_rate": 9.303620390438128e-06,
+      "loss": 119.3574,
+      "step": 31310
+    },
+    {
+      "epoch": 0.2590892170244447,
+      "grad_norm": 1793.4461669921875,
+      "learning_rate": 9.302902241183905e-06,
+      "loss": 115.7504,
+      "step": 31320
+    },
+    {
+      "epoch": 0.25917194027381396,
+      "grad_norm": 1034.2620849609375,
+      "learning_rate": 9.302183749567498e-06,
+      "loss": 104.3807,
+      "step": 31330
+    },
+    {
+      "epoch": 0.2592546635231832,
+      "grad_norm": 1191.4996337890625,
+      "learning_rate": 9.301464915646074e-06,
+      "loss": 95.0326,
+      "step": 31340
+    },
+    {
+      "epoch": 0.2593373867725524,
+      "grad_norm": 663.1774291992188,
+      "learning_rate": 9.30074573947683e-06,
+      "loss": 105.2758,
+      "step": 31350
+    },
+    {
+      "epoch": 0.25942011002192167,
+      "grad_norm": 1111.489501953125,
+      "learning_rate": 9.30002622111698e-06,
+      "loss": 120.7122,
+      "step": 31360
+    },
+    {
+      "epoch": 0.2595028332712909,
+      "grad_norm": 1140.2496337890625,
+      "learning_rate": 9.299306360623782e-06,
+      "loss": 111.4868,
+      "step": 31370
+    },
+    {
+      "epoch": 0.2595855565206601,
+      "grad_norm": 1038.2596435546875,
+      "learning_rate": 9.298586158054508e-06,
+      "loss": 119.8149,
+      "step": 31380
+    },
+    {
+      "epoch": 0.2596682797700294,
+      "grad_norm": 1008.992431640625,
+      "learning_rate": 9.297865613466459e-06,
+      "loss": 145.3494,
+      "step": 31390
+    },
+    {
+      "epoch": 0.25975100301939863,
+      "grad_norm": 1030.713623046875,
+      "learning_rate": 9.29714472691697e-06,
+      "loss": 116.2307,
+      "step": 31400
+    },
+    {
+      "epoch": 0.25983372626876783,
+      "grad_norm": 1515.673095703125,
+      "learning_rate": 9.296423498463396e-06,
+      "loss": 125.331,
+      "step": 31410
+    },
+    {
+      "epoch": 0.2599164495181371,
+      "grad_norm": 1273.73486328125,
+      "learning_rate": 9.29570192816312e-06,
+      "loss": 140.6214,
+      "step": 31420
+    },
+    {
+      "epoch": 0.2599991727675063,
+      "grad_norm": 710.17236328125,
+      "learning_rate": 9.29498001607356e-06,
+      "loss": 108.083,
+      "step": 31430
+    },
+    {
+      "epoch": 0.26008189601687554,
+      "grad_norm": 900.10107421875,
+      "learning_rate": 9.294257762252148e-06,
+      "loss": 98.9134,
+      "step": 31440
+    },
+    {
+      "epoch": 0.2601646192662448,
+      "grad_norm": 875.4248657226562,
+      "learning_rate": 9.293535166756356e-06,
+      "loss": 174.0914,
+      "step": 31450
+    },
+    {
+      "epoch": 0.260247342515614,
+      "grad_norm": 1023.4577026367188,
+      "learning_rate": 9.292812229643674e-06,
+      "loss": 96.2018,
+      "step": 31460
+    },
+    {
+      "epoch": 0.26033006576498324,
+      "grad_norm": 1871.7161865234375,
+      "learning_rate": 9.292088950971624e-06,
+      "loss": 135.2347,
+      "step": 31470
+    },
+    {
+      "epoch": 0.2604127890143525,
+      "grad_norm": 1158.376953125,
+      "learning_rate": 9.291365330797755e-06,
+      "loss": 131.9809,
+      "step": 31480
+    },
+    {
+      "epoch": 0.2604955122637217,
+      "grad_norm": 962.1968383789062,
+      "learning_rate": 9.290641369179643e-06,
+      "loss": 109.7965,
+      "step": 31490
+    },
+    {
+      "epoch": 0.26057823551309095,
+      "grad_norm": 4627.09521484375,
+      "learning_rate": 9.289917066174887e-06,
+      "loss": 133.032,
+      "step": 31500
+    },
+    {
+      "epoch": 0.2606609587624602,
+      "grad_norm": 1069.57177734375,
+      "learning_rate": 9.289192421841116e-06,
+      "loss": 114.866,
+      "step": 31510
+    },
+    {
+      "epoch": 0.2607436820118294,
+      "grad_norm": 1533.0517578125,
+      "learning_rate": 9.288467436235992e-06,
+      "loss": 135.5069,
+      "step": 31520
+    },
+    {
+      "epoch": 0.26082640526119866,
+      "grad_norm": 876.2843627929688,
+      "learning_rate": 9.287742109417194e-06,
+      "loss": 165.9743,
+      "step": 31530
+    },
+    {
+      "epoch": 0.2609091285105679,
+      "grad_norm": 849.820556640625,
+      "learning_rate": 9.287016441442435e-06,
+      "loss": 113.8865,
+      "step": 31540
+    },
+    {
+      "epoch": 0.2609918517599371,
+      "grad_norm": 1214.044189453125,
+      "learning_rate": 9.28629043236945e-06,
+      "loss": 129.5201,
+      "step": 31550
+    },
+    {
+      "epoch": 0.26107457500930636,
+      "grad_norm": 974.0680541992188,
+      "learning_rate": 9.285564082256011e-06,
+      "loss": 106.2931,
+      "step": 31560
+    },
+    {
+      "epoch": 0.2611572982586756,
+      "grad_norm": 591.6702270507812,
+      "learning_rate": 9.284837391159904e-06,
+      "loss": 77.5611,
+      "step": 31570
+    },
+    {
+      "epoch": 0.2612400215080448,
+      "grad_norm": 995.6359252929688,
+      "learning_rate": 9.284110359138951e-06,
+      "loss": 169.7267,
+      "step": 31580
+    },
+    {
+      "epoch": 0.26132274475741407,
+      "grad_norm": 993.3253784179688,
+      "learning_rate": 9.283382986250997e-06,
+      "loss": 117.8098,
+      "step": 31590
+    },
+    {
+      "epoch": 0.2614054680067833,
+      "grad_norm": 720.0477905273438,
+      "learning_rate": 9.282655272553917e-06,
+      "loss": 164.0745,
+      "step": 31600
+    },
+    {
+      "epoch": 0.2614881912561525,
+      "grad_norm": 1000.5869750976562,
+      "learning_rate": 9.281927218105613e-06,
+      "loss": 103.8817,
+      "step": 31610
+    },
+    {
+      "epoch": 0.2615709145055218,
+      "grad_norm": 701.7263793945312,
+      "learning_rate": 9.281198822964011e-06,
+      "loss": 115.0276,
+      "step": 31620
+    },
+    {
+      "epoch": 0.26165363775489103,
+      "grad_norm": 1523.617919921875,
+      "learning_rate": 9.280470087187066e-06,
+      "loss": 150.8629,
+      "step": 31630
+    },
+    {
+      "epoch": 0.26173636100426023,
+      "grad_norm": 1352.073486328125,
+      "learning_rate": 9.279741010832761e-06,
+      "loss": 111.8819,
+      "step": 31640
+    },
+    {
+      "epoch": 0.2618190842536295,
+      "grad_norm": 784.3294677734375,
+      "learning_rate": 9.279011593959107e-06,
+      "loss": 134.8354,
+      "step": 31650
+    },
+    {
+      "epoch": 0.26190180750299874,
+      "grad_norm": 518.8139038085938,
+      "learning_rate": 9.278281836624137e-06,
+      "loss": 109.1452,
+      "step": 31660
+    },
+    {
+      "epoch": 0.26198453075236794,
+      "grad_norm": 738.8187255859375,
+      "learning_rate": 9.277551738885915e-06,
+      "loss": 137.3162,
+      "step": 31670
+    },
+    {
+      "epoch": 0.2620672540017372,
+      "grad_norm": 1180.13037109375,
+      "learning_rate": 9.276821300802535e-06,
+      "loss": 101.0768,
+      "step": 31680
+    },
+    {
+      "epoch": 0.26214997725110645,
+      "grad_norm": 1431.0516357421875,
+      "learning_rate": 9.276090522432109e-06,
+      "loss": 99.368,
+      "step": 31690
+    },
+    {
+      "epoch": 0.26223270050047565,
+      "grad_norm": 688.6452026367188,
+      "learning_rate": 9.275359403832787e-06,
+      "loss": 123.8677,
+      "step": 31700
+    },
+    {
+      "epoch": 0.2623154237498449,
+      "grad_norm": 854.5988159179688,
+      "learning_rate": 9.274627945062738e-06,
+      "loss": 111.6149,
+      "step": 31710
+    },
+    {
+      "epoch": 0.26239814699921415,
+      "grad_norm": 1075.212158203125,
+      "learning_rate": 9.27389614618016e-06,
+      "loss": 138.4687,
+      "step": 31720
+    },
+    {
+      "epoch": 0.26248087024858335,
+      "grad_norm": 573.2298583984375,
+      "learning_rate": 9.273164007243281e-06,
+      "loss": 108.3331,
+      "step": 31730
+    },
+    {
+      "epoch": 0.2625635934979526,
+      "grad_norm": 990.3570556640625,
+      "learning_rate": 9.272431528310354e-06,
+      "loss": 112.5479,
+      "step": 31740
+    },
+    {
+      "epoch": 0.26264631674732186,
+      "grad_norm": 1721.805419921875,
+      "learning_rate": 9.271698709439658e-06,
+      "loss": 140.3488,
+      "step": 31750
+    },
+    {
+      "epoch": 0.26272903999669106,
+      "grad_norm": 1286.5728759765625,
+      "learning_rate": 9.2709655506895e-06,
+      "loss": 168.1867,
+      "step": 31760
+    },
+    {
+      "epoch": 0.2628117632460603,
+      "grad_norm": 582.025146484375,
+      "learning_rate": 9.270232052118214e-06,
+      "loss": 119.0196,
+      "step": 31770
+    },
+    {
+      "epoch": 0.2628944864954295,
+      "grad_norm": 1799.970703125,
+      "learning_rate": 9.26949821378416e-06,
+      "loss": 130.6446,
+      "step": 31780
+    },
+    {
+      "epoch": 0.26297720974479877,
+      "grad_norm": 810.4905395507812,
+      "learning_rate": 9.268764035745727e-06,
+      "loss": 123.3437,
+      "step": 31790
+    },
+    {
+      "epoch": 0.263059932994168,
+      "grad_norm": 1139.095947265625,
+      "learning_rate": 9.268029518061335e-06,
+      "loss": 138.0163,
+      "step": 31800
+    },
+    {
+      "epoch": 0.2631426562435372,
+      "grad_norm": 728.7420654296875,
+      "learning_rate": 9.267294660789417e-06,
+      "loss": 118.9001,
+      "step": 31810
+    },
+    {
+      "epoch": 0.2632253794929065,
+      "grad_norm": 937.4639282226562,
+      "learning_rate": 9.26655946398845e-06,
+      "loss": 101.5111,
+      "step": 31820
+    },
+    {
+      "epoch": 0.26330810274227573,
+      "grad_norm": 951.8093872070312,
+      "learning_rate": 9.265823927716927e-06,
+      "loss": 114.6193,
+      "step": 31830
+    },
+    {
+      "epoch": 0.26339082599164493,
+      "grad_norm": 1131.0379638671875,
+      "learning_rate": 9.26508805203337e-06,
+      "loss": 164.4967,
+      "step": 31840
+    },
+    {
+      "epoch": 0.2634735492410142,
+      "grad_norm": 916.825439453125,
+      "learning_rate": 9.264351836996332e-06,
+      "loss": 99.9893,
+      "step": 31850
+    },
+    {
+      "epoch": 0.26355627249038344,
+      "grad_norm": 781.8618774414062,
+      "learning_rate": 9.26361528266439e-06,
+      "loss": 147.4806,
+      "step": 31860
+    },
+    {
+      "epoch": 0.26363899573975264,
+      "grad_norm": 1081.302001953125,
+      "learning_rate": 9.262878389096147e-06,
+      "loss": 107.9612,
+      "step": 31870
+    },
+    {
+      "epoch": 0.2637217189891219,
+      "grad_norm": 1674.4130859375,
+      "learning_rate": 9.262141156350233e-06,
+      "loss": 120.8496,
+      "step": 31880
+    },
+    {
+      "epoch": 0.26380444223849114,
+      "grad_norm": 599.2373046875,
+      "learning_rate": 9.261403584485308e-06,
+      "loss": 130.4039,
+      "step": 31890
+    },
+    {
+      "epoch": 0.26388716548786034,
+      "grad_norm": 931.727783203125,
+      "learning_rate": 9.260665673560058e-06,
+      "loss": 94.2291,
+      "step": 31900
+    },
+    {
+      "epoch": 0.2639698887372296,
+      "grad_norm": 695.3705444335938,
+      "learning_rate": 9.259927423633193e-06,
+      "loss": 173.953,
+      "step": 31910
+    },
+    {
+      "epoch": 0.26405261198659885,
+      "grad_norm": 643.5379638671875,
+      "learning_rate": 9.259188834763455e-06,
+      "loss": 91.6798,
+      "step": 31920
+    },
+    {
+      "epoch": 0.26413533523596805,
+      "grad_norm": 1540.7181396484375,
+      "learning_rate": 9.258449907009607e-06,
+      "loss": 126.6724,
+      "step": 31930
+    },
+    {
+      "epoch": 0.2642180584853373,
+      "grad_norm": 1498.8092041015625,
+      "learning_rate": 9.257710640430444e-06,
+      "loss": 110.8607,
+      "step": 31940
+    },
+    {
+      "epoch": 0.26430078173470656,
+      "grad_norm": 626.6985473632812,
+      "learning_rate": 9.256971035084786e-06,
+      "loss": 85.7513,
+      "step": 31950
+    },
+    {
+      "epoch": 0.26438350498407576,
+      "grad_norm": 1363.011962890625,
+      "learning_rate": 9.256231091031477e-06,
+      "loss": 101.1794,
+      "step": 31960
+    },
+    {
+      "epoch": 0.264466228233445,
+      "grad_norm": 1134.742919921875,
+      "learning_rate": 9.255490808329397e-06,
+      "loss": 212.7933,
+      "step": 31970
+    },
+    {
+      "epoch": 0.26454895148281427,
+      "grad_norm": 1176.7347412109375,
+      "learning_rate": 9.254750187037443e-06,
+      "loss": 122.8415,
+      "step": 31980
+    },
+    {
+      "epoch": 0.26463167473218346,
+      "grad_norm": 843.20458984375,
+      "learning_rate": 9.254009227214543e-06,
+      "loss": 115.774,
+      "step": 31990
+    },
+    {
+      "epoch": 0.2647143979815527,
+      "grad_norm": 990.8587646484375,
+      "learning_rate": 9.253267928919652e-06,
+      "loss": 141.8495,
+      "step": 32000
+    },
+    {
+      "epoch": 0.264797121230922,
+      "grad_norm": 1063.32763671875,
+      "learning_rate": 9.25252629221175e-06,
+      "loss": 118.555,
+      "step": 32010
+    },
+    {
+      "epoch": 0.26487984448029117,
+      "grad_norm": 1893.1072998046875,
+      "learning_rate": 9.251784317149848e-06,
+      "loss": 122.1342,
+      "step": 32020
+    },
+    {
+      "epoch": 0.2649625677296604,
+      "grad_norm": 1173.83349609375,
+      "learning_rate": 9.251042003792983e-06,
+      "loss": 161.0942,
+      "step": 32030
+    },
+    {
+      "epoch": 0.2650452909790297,
+      "grad_norm": 927.9155883789062,
+      "learning_rate": 9.250299352200214e-06,
+      "loss": 78.0564,
+      "step": 32040
+    },
+    {
+      "epoch": 0.2651280142283989,
+      "grad_norm": 1478.896484375,
+      "learning_rate": 9.249556362430631e-06,
+      "loss": 129.6906,
+      "step": 32050
+    },
+    {
+      "epoch": 0.26521073747776813,
+      "grad_norm": 772.536865234375,
+      "learning_rate": 9.248813034543353e-06,
+      "loss": 102.5596,
+      "step": 32060
+    },
+    {
+      "epoch": 0.2652934607271374,
+      "grad_norm": 728.9833984375,
+      "learning_rate": 9.24806936859752e-06,
+      "loss": 111.6626,
+      "step": 32070
+    },
+    {
+      "epoch": 0.2653761839765066,
+      "grad_norm": 1042.043701171875,
+      "learning_rate": 9.247325364652304e-06,
+      "loss": 132.3886,
+      "step": 32080
+    },
+    {
+      "epoch": 0.26545890722587584,
+      "grad_norm": 490.30419921875,
+      "learning_rate": 9.2465810227669e-06,
+      "loss": 128.4408,
+      "step": 32090
+    },
+    {
+      "epoch": 0.2655416304752451,
+      "grad_norm": 1035.492919921875,
+      "learning_rate": 9.245836343000534e-06,
+      "loss": 102.5217,
+      "step": 32100
+    },
+    {
+      "epoch": 0.2656243537246143,
+      "grad_norm": 1005.1189575195312,
+      "learning_rate": 9.245091325412456e-06,
+      "loss": 112.0046,
+      "step": 32110
+    },
+    {
+      "epoch": 0.26570707697398355,
+      "grad_norm": 864.4721069335938,
+      "learning_rate": 9.244345970061944e-06,
+      "loss": 109.3595,
+      "step": 32120
+    },
+    {
+      "epoch": 0.26578980022335275,
+      "grad_norm": 803.18896484375,
+      "learning_rate": 9.243600277008301e-06,
+      "loss": 123.4932,
+      "step": 32130
+    },
+    {
+      "epoch": 0.265872523472722,
+      "grad_norm": 1339.6492919921875,
+      "learning_rate": 9.24285424631086e-06,
+      "loss": 131.0302,
+      "step": 32140
+    },
+    {
+      "epoch": 0.26595524672209125,
+      "grad_norm": 331.6441650390625,
+      "learning_rate": 9.242107878028978e-06,
+      "loss": 84.0776,
+      "step": 32150
+    },
+    {
+      "epoch": 0.26603796997146045,
+      "grad_norm": 1174.8720703125,
+      "learning_rate": 9.241361172222043e-06,
+      "loss": 144.0584,
+      "step": 32160
+    },
+    {
+      "epoch": 0.2661206932208297,
+      "grad_norm": 955.9444580078125,
+      "learning_rate": 9.240614128949463e-06,
+      "loss": 79.2642,
+      "step": 32170
+    },
+    {
+      "epoch": 0.26620341647019896,
+      "grad_norm": 408.5090637207031,
+      "learning_rate": 9.239866748270679e-06,
+      "loss": 80.5909,
+      "step": 32180
+    },
+    {
+      "epoch": 0.26628613971956816,
+      "grad_norm": 1054.4498291015625,
+      "learning_rate": 9.239119030245156e-06,
+      "loss": 105.1165,
+      "step": 32190
+    },
+    {
+      "epoch": 0.2663688629689374,
+      "grad_norm": 1652.845703125,
+      "learning_rate": 9.238370974932387e-06,
+      "loss": 144.1475,
+      "step": 32200
+    },
+    {
+      "epoch": 0.26645158621830667,
+      "grad_norm": 897.6002807617188,
+      "learning_rate": 9.23762258239189e-06,
+      "loss": 85.8102,
+      "step": 32210
+    },
+    {
+      "epoch": 0.26653430946767587,
+      "grad_norm": 848.9139404296875,
+      "learning_rate": 9.236873852683213e-06,
+      "loss": 107.4219,
+      "step": 32220
+    },
+    {
+      "epoch": 0.2666170327170451,
+      "grad_norm": 900.8207397460938,
+      "learning_rate": 9.23612478586593e-06,
+      "loss": 77.9687,
+      "step": 32230
+    },
+    {
+      "epoch": 0.2666997559664144,
+      "grad_norm": 1326.0458984375,
+      "learning_rate": 9.235375381999636e-06,
+      "loss": 106.4991,
+      "step": 32240
+    },
+    {
+      "epoch": 0.2667824792157836,
+      "grad_norm": 2422.508056640625,
+      "learning_rate": 9.234625641143962e-06,
+      "loss": 138.341,
+      "step": 32250
+    },
+    {
+      "epoch": 0.26686520246515283,
+      "grad_norm": 964.7281494140625,
+      "learning_rate": 9.233875563358559e-06,
+      "loss": 122.4212,
+      "step": 32260
+    },
+    {
+      "epoch": 0.2669479257145221,
+      "grad_norm": 1445.0108642578125,
+      "learning_rate": 9.23312514870311e-06,
+      "loss": 107.632,
+      "step": 32270
+    },
+    {
+      "epoch": 0.2670306489638913,
+      "grad_norm": 628.1731567382812,
+      "learning_rate": 9.232374397237318e-06,
+      "loss": 118.4748,
+      "step": 32280
+    },
+    {
+      "epoch": 0.26711337221326054,
+      "grad_norm": 1020.5704345703125,
+      "learning_rate": 9.231623309020922e-06,
+      "loss": 132.8099,
+      "step": 32290
+    },
+    {
+      "epoch": 0.2671960954626298,
+      "grad_norm": 949.1611938476562,
+      "learning_rate": 9.230871884113679e-06,
+      "loss": 128.9596,
+      "step": 32300
+    },
+    {
+      "epoch": 0.267278818711999,
+      "grad_norm": 853.702880859375,
+      "learning_rate": 9.230120122575376e-06,
+      "loss": 117.7804,
+      "step": 32310
+    },
+    {
+      "epoch": 0.26736154196136824,
+      "grad_norm": 856.49462890625,
+      "learning_rate": 9.22936802446583e-06,
+      "loss": 179.9062,
+      "step": 32320
+    },
+    {
+      "epoch": 0.2674442652107375,
+      "grad_norm": 847.5012817382812,
+      "learning_rate": 9.228615589844879e-06,
+      "loss": 84.1749,
+      "step": 32330
+    },
+    {
+      "epoch": 0.2675269884601067,
+      "grad_norm": 1034.2694091796875,
+      "learning_rate": 9.227862818772392e-06,
+      "loss": 131.7186,
+      "step": 32340
+    },
+    {
+      "epoch": 0.26760971170947595,
+      "grad_norm": 1057.1470947265625,
+      "learning_rate": 9.227109711308265e-06,
+      "loss": 94.973,
+      "step": 32350
+    },
+    {
+      "epoch": 0.2676924349588452,
+      "grad_norm": 374.79473876953125,
+      "learning_rate": 9.226356267512417e-06,
+      "loss": 107.7693,
+      "step": 32360
+    },
+    {
+      "epoch": 0.2677751582082144,
+      "grad_norm": 817.5911865234375,
+      "learning_rate": 9.225602487444799e-06,
+      "loss": 107.2883,
+      "step": 32370
+    },
+    {
+      "epoch": 0.26785788145758366,
+      "grad_norm": 1157.15234375,
+      "learning_rate": 9.224848371165382e-06,
+      "loss": 170.9429,
+      "step": 32380
+    },
+    {
+      "epoch": 0.2679406047069529,
+      "grad_norm": 1658.9010009765625,
+      "learning_rate": 9.224093918734172e-06,
+      "loss": 202.5666,
+      "step": 32390
+    },
+    {
+      "epoch": 0.2680233279563221,
+      "grad_norm": 1403.2574462890625,
+      "learning_rate": 9.223339130211194e-06,
+      "loss": 113.1494,
+      "step": 32400
+    },
+    {
+      "epoch": 0.26810605120569136,
+      "grad_norm": 939.0480346679688,
+      "learning_rate": 9.222584005656501e-06,
+      "loss": 114.9759,
+      "step": 32410
+    },
+    {
+      "epoch": 0.2681887744550606,
+      "grad_norm": 578.2918701171875,
+      "learning_rate": 9.22182854513018e-06,
+      "loss": 94.9374,
+      "step": 32420
+    },
+    {
+      "epoch": 0.2682714977044298,
+      "grad_norm": 2604.21484375,
+      "learning_rate": 9.221072748692336e-06,
+      "loss": 109.7514,
+      "step": 32430
+    },
+    {
+      "epoch": 0.26835422095379907,
+      "grad_norm": 962.8878173828125,
+      "learning_rate": 9.220316616403109e-06,
+      "loss": 104.4484,
+      "step": 32440
+    },
+    {
+      "epoch": 0.2684369442031683,
+      "grad_norm": 759.3588256835938,
+      "learning_rate": 9.219560148322655e-06,
+      "loss": 101.3538,
+      "step": 32450
+    },
+    {
+      "epoch": 0.2685196674525375,
+      "grad_norm": 1004.7863159179688,
+      "learning_rate": 9.218803344511165e-06,
+      "loss": 134.46,
+      "step": 32460
+    },
+    {
+      "epoch": 0.2686023907019068,
+      "grad_norm": 1546.0279541015625,
+      "learning_rate": 9.218046205028854e-06,
+      "loss": 102.795,
+      "step": 32470
+    },
+    {
+      "epoch": 0.26868511395127603,
+      "grad_norm": 422.1183776855469,
+      "learning_rate": 9.217288729935966e-06,
+      "loss": 100.5324,
+      "step": 32480
+    },
+    {
+      "epoch": 0.26876783720064523,
+      "grad_norm": 953.4854736328125,
+      "learning_rate": 9.216530919292768e-06,
+      "loss": 147.8428,
+      "step": 32490
+    },
+    {
+      "epoch": 0.2688505604500145,
+      "grad_norm": 910.6680297851562,
+      "learning_rate": 9.215772773159556e-06,
+      "loss": 138.0076,
+      "step": 32500
+    },
+    {
+      "epoch": 0.2689332836993837,
+      "grad_norm": 1041.43505859375,
+      "learning_rate": 9.215014291596653e-06,
+      "loss": 120.8348,
+      "step": 32510
+    },
+    {
+      "epoch": 0.26901600694875294,
+      "grad_norm": 1151.9285888671875,
+      "learning_rate": 9.214255474664405e-06,
+      "loss": 150.9121,
+      "step": 32520
+    },
+    {
+      "epoch": 0.2690987301981222,
+      "grad_norm": 921.2622680664062,
+      "learning_rate": 9.213496322423193e-06,
+      "loss": 83.8476,
+      "step": 32530
+    },
+    {
+      "epoch": 0.2691814534474914,
+      "grad_norm": 411.46826171875,
+      "learning_rate": 9.212736834933413e-06,
+      "loss": 129.1243,
+      "step": 32540
+    },
+    {
+      "epoch": 0.26926417669686065,
+      "grad_norm": 3256.450927734375,
+      "learning_rate": 9.211977012255497e-06,
+      "loss": 87.7072,
+      "step": 32550
+    },
+    {
+      "epoch": 0.2693468999462299,
+      "grad_norm": 1524.965087890625,
+      "learning_rate": 9.211216854449903e-06,
+      "loss": 128.8632,
+      "step": 32560
+    },
+    {
+      "epoch": 0.2694296231955991,
+      "grad_norm": 953.8908081054688,
+      "learning_rate": 9.210456361577109e-06,
+      "loss": 113.7588,
+      "step": 32570
+    },
+    {
+      "epoch": 0.26951234644496835,
+      "grad_norm": 1008.9974975585938,
+      "learning_rate": 9.209695533697624e-06,
+      "loss": 122.1501,
+      "step": 32580
+    },
+    {
+      "epoch": 0.2695950696943376,
+      "grad_norm": 805.004150390625,
+      "learning_rate": 9.208934370871989e-06,
+      "loss": 113.7434,
+      "step": 32590
+    },
+    {
+      "epoch": 0.2696777929437068,
+      "grad_norm": 636.7761840820312,
+      "learning_rate": 9.20817287316076e-06,
+      "loss": 113.1953,
+      "step": 32600
+    },
+    {
+      "epoch": 0.26976051619307606,
+      "grad_norm": 763.083984375,
+      "learning_rate": 9.20741104062453e-06,
+      "loss": 101.9306,
+      "step": 32610
+    },
+    {
+      "epoch": 0.2698432394424453,
+      "grad_norm": 1067.39208984375,
+      "learning_rate": 9.206648873323912e-06,
+      "loss": 104.6595,
+      "step": 32620
+    },
+    {
+      "epoch": 0.2699259626918145,
+      "grad_norm": 1282.15576171875,
+      "learning_rate": 9.205886371319548e-06,
+      "loss": 112.2921,
+      "step": 32630
+    },
+    {
+      "epoch": 0.27000868594118377,
+      "grad_norm": 1269.3165283203125,
+      "learning_rate": 9.20512353467211e-06,
+      "loss": 101.4175,
+      "step": 32640
+    },
+    {
+      "epoch": 0.270091409190553,
+      "grad_norm": 1416.7222900390625,
+      "learning_rate": 9.204360363442288e-06,
+      "loss": 94.3014,
+      "step": 32650
+    },
+    {
+      "epoch": 0.2701741324399222,
+      "grad_norm": 577.2730712890625,
+      "learning_rate": 9.20359685769081e-06,
+      "loss": 178.5912,
+      "step": 32660
+    },
+    {
+      "epoch": 0.2702568556892915,
+      "grad_norm": 976.02880859375,
+      "learning_rate": 9.202833017478421e-06,
+      "loss": 142.0586,
+      "step": 32670
+    },
+    {
+      "epoch": 0.27033957893866073,
+      "grad_norm": 1126.535888671875,
+      "learning_rate": 9.2020688428659e-06,
+      "loss": 100.6238,
+      "step": 32680
+    },
+    {
+      "epoch": 0.27042230218802993,
+      "grad_norm": 945.1123657226562,
+      "learning_rate": 9.201304333914042e-06,
+      "loss": 124.4394,
+      "step": 32690
+    },
+    {
+      "epoch": 0.2705050254373992,
+      "grad_norm": 600.9089965820312,
+      "learning_rate": 9.200539490683682e-06,
+      "loss": 93.1028,
+      "step": 32700
+    },
+    {
+      "epoch": 0.27058774868676844,
+      "grad_norm": 704.4324951171875,
+      "learning_rate": 9.19977431323567e-06,
+      "loss": 124.6708,
+      "step": 32710
+    },
+    {
+      "epoch": 0.27067047193613764,
+      "grad_norm": 938.8729858398438,
+      "learning_rate": 9.199008801630893e-06,
+      "loss": 119.1146,
+      "step": 32720
+    },
+    {
+      "epoch": 0.2707531951855069,
+      "grad_norm": 1016.2319946289062,
+      "learning_rate": 9.198242955930257e-06,
+      "loss": 126.7218,
+      "step": 32730
+    },
+    {
+      "epoch": 0.27083591843487614,
+      "grad_norm": 832.0853881835938,
+      "learning_rate": 9.197476776194693e-06,
+      "loss": 102.6724,
+      "step": 32740
+    },
+    {
+      "epoch": 0.27091864168424534,
+      "grad_norm": 1260.37548828125,
+      "learning_rate": 9.196710262485168e-06,
+      "loss": 107.5099,
+      "step": 32750
+    },
+    {
+      "epoch": 0.2710013649336146,
+      "grad_norm": 840.14990234375,
+      "learning_rate": 9.195943414862667e-06,
+      "loss": 124.9764,
+      "step": 32760
+    },
+    {
+      "epoch": 0.27108408818298385,
+      "grad_norm": 486.6063537597656,
+      "learning_rate": 9.195176233388206e-06,
+      "loss": 92.0499,
+      "step": 32770
+    },
+    {
+      "epoch": 0.27116681143235305,
+      "grad_norm": 907.8724975585938,
+      "learning_rate": 9.194408718122825e-06,
+      "loss": 120.6719,
+      "step": 32780
+    },
+    {
+      "epoch": 0.2712495346817223,
+      "grad_norm": 572.4683837890625,
+      "learning_rate": 9.193640869127592e-06,
+      "loss": 124.2721,
+      "step": 32790
+    },
+    {
+      "epoch": 0.27133225793109156,
+      "grad_norm": 957.0466918945312,
+      "learning_rate": 9.192872686463601e-06,
+      "loss": 131.9941,
+      "step": 32800
+    },
+    {
+      "epoch": 0.27141498118046076,
+      "grad_norm": 691.4140014648438,
+      "learning_rate": 9.192104170191973e-06,
+      "loss": 101.0082,
+      "step": 32810
+    },
+    {
+      "epoch": 0.27149770442983,
+      "grad_norm": 791.294677734375,
+      "learning_rate": 9.191335320373856e-06,
+      "loss": 138.9451,
+      "step": 32820
+    },
+    {
+      "epoch": 0.27158042767919927,
+      "grad_norm": 1051.594482421875,
+      "learning_rate": 9.190566137070422e-06,
+      "loss": 107.2597,
+      "step": 32830
+    },
+    {
+      "epoch": 0.27166315092856846,
+      "grad_norm": 1549.462158203125,
+      "learning_rate": 9.189796620342875e-06,
+      "loss": 153.74,
+      "step": 32840
+    },
+    {
+      "epoch": 0.2717458741779377,
+      "grad_norm": 751.6979370117188,
+      "learning_rate": 9.189026770252437e-06,
+      "loss": 108.3263,
+      "step": 32850
+    },
+    {
+      "epoch": 0.2718285974273069,
+      "grad_norm": 4171.58203125,
+      "learning_rate": 9.188256586860365e-06,
+      "loss": 177.8506,
+      "step": 32860
+    },
+    {
+      "epoch": 0.27191132067667617,
+      "grad_norm": 880.1036987304688,
+      "learning_rate": 9.187486070227938e-06,
+      "loss": 124.1478,
+      "step": 32870
+    },
+    {
+      "epoch": 0.2719940439260454,
+      "grad_norm": 875.8194580078125,
+      "learning_rate": 9.186715220416463e-06,
+      "loss": 80.9601,
+      "step": 32880
+    },
+    {
+      "epoch": 0.2720767671754146,
+      "grad_norm": 517.209716796875,
+      "learning_rate": 9.185944037487271e-06,
+      "loss": 105.6458,
+      "step": 32890
+    },
+    {
+      "epoch": 0.2721594904247839,
+      "grad_norm": 783.9718627929688,
+      "learning_rate": 9.185172521501723e-06,
+      "loss": 103.928,
+      "step": 32900
+    },
+    {
+      "epoch": 0.27224221367415313,
+      "grad_norm": 809.2305297851562,
+      "learning_rate": 9.184400672521204e-06,
+      "loss": 119.6438,
+      "step": 32910
+    },
+    {
+      "epoch": 0.27232493692352233,
+      "grad_norm": 847.0447998046875,
+      "learning_rate": 9.183628490607129e-06,
+      "loss": 118.6409,
+      "step": 32920
+    },
+    {
+      "epoch": 0.2724076601728916,
+      "grad_norm": 803.6466674804688,
+      "learning_rate": 9.182855975820934e-06,
+      "loss": 86.8706,
+      "step": 32930
+    },
+    {
+      "epoch": 0.27249038342226084,
+      "grad_norm": 1112.55322265625,
+      "learning_rate": 9.182083128224086e-06,
+      "loss": 108.3938,
+      "step": 32940
+    },
+    {
+      "epoch": 0.27257310667163004,
+      "grad_norm": 758.2708740234375,
+      "learning_rate": 9.181309947878077e-06,
+      "loss": 136.1542,
+      "step": 32950
+    },
+    {
+      "epoch": 0.2726558299209993,
+      "grad_norm": 655.5614013671875,
+      "learning_rate": 9.180536434844426e-06,
+      "loss": 93.3358,
+      "step": 32960
+    },
+    {
+      "epoch": 0.27273855317036855,
+      "grad_norm": 746.388671875,
+      "learning_rate": 9.179762589184676e-06,
+      "loss": 137.545,
+      "step": 32970
+    },
+    {
+      "epoch": 0.27282127641973775,
+      "grad_norm": 919.367431640625,
+      "learning_rate": 9.1789884109604e-06,
+      "loss": 105.9232,
+      "step": 32980
+    },
+    {
+      "epoch": 0.272903999669107,
+      "grad_norm": 1165.487548828125,
+      "learning_rate": 9.178213900233193e-06,
+      "loss": 119.8975,
+      "step": 32990
+    },
+    {
+      "epoch": 0.27298672291847625,
+      "grad_norm": 1274.4398193359375,
+      "learning_rate": 9.177439057064684e-06,
+      "loss": 108.2796,
+      "step": 33000
+    },
+    {
+      "epoch": 0.27306944616784545,
+      "grad_norm": 2099.3134765625,
+      "learning_rate": 9.17666388151652e-06,
+      "loss": 88.9919,
+      "step": 33010
+    },
+    {
+      "epoch": 0.2731521694172147,
+      "grad_norm": 1373.3428955078125,
+      "learning_rate": 9.175888373650377e-06,
+      "loss": 109.9396,
+      "step": 33020
+    },
+    {
+      "epoch": 0.27323489266658396,
+      "grad_norm": 1429.390869140625,
+      "learning_rate": 9.175112533527963e-06,
+      "loss": 94.565,
+      "step": 33030
+    },
+    {
+      "epoch": 0.27331761591595316,
+      "grad_norm": 1118.853271484375,
+      "learning_rate": 9.174336361211007e-06,
+      "loss": 101.014,
+      "step": 33040
+    },
+    {
+      "epoch": 0.2734003391653224,
+      "grad_norm": 2152.85107421875,
+      "learning_rate": 9.173559856761262e-06,
+      "loss": 153.9467,
+      "step": 33050
+    },
+    {
+      "epoch": 0.27348306241469167,
+      "grad_norm": 909.071533203125,
+      "learning_rate": 9.172783020240514e-06,
+      "loss": 102.0454,
+      "step": 33060
+    },
+    {
+      "epoch": 0.27356578566406087,
+      "grad_norm": 890.9006958007812,
+      "learning_rate": 9.172005851710573e-06,
+      "loss": 130.9717,
+      "step": 33070
+    },
+    {
+      "epoch": 0.2736485089134301,
+      "grad_norm": 1000.61279296875,
+      "learning_rate": 9.171228351233272e-06,
+      "loss": 150.3027,
+      "step": 33080
+    },
+    {
+      "epoch": 0.2737312321627994,
+      "grad_norm": 1430.8470458984375,
+      "learning_rate": 9.170450518870475e-06,
+      "loss": 149.3742,
+      "step": 33090
+    },
+    {
+      "epoch": 0.2738139554121686,
+      "grad_norm": 1026.9654541015625,
+      "learning_rate": 9.169672354684069e-06,
+      "loss": 123.5882,
+      "step": 33100
+    },
+    {
+      "epoch": 0.27389667866153783,
+      "grad_norm": 841.4974975585938,
+      "learning_rate": 9.168893858735972e-06,
+      "loss": 92.7002,
+      "step": 33110
+    },
+    {
+      "epoch": 0.2739794019109071,
+      "grad_norm": 1173.48681640625,
+      "learning_rate": 9.168115031088122e-06,
+      "loss": 89.6682,
+      "step": 33120
+    },
+    {
+      "epoch": 0.2740621251602763,
+      "grad_norm": 733.5807495117188,
+      "learning_rate": 9.167335871802488e-06,
+      "loss": 86.3547,
+      "step": 33130
+    },
+    {
+      "epoch": 0.27414484840964554,
+      "grad_norm": 1434.056640625,
+      "learning_rate": 9.166556380941063e-06,
+      "loss": 125.5328,
+      "step": 33140
+    },
+    {
+      "epoch": 0.2742275716590148,
+      "grad_norm": 820.1240844726562,
+      "learning_rate": 9.16577655856587e-06,
+      "loss": 130.4958,
+      "step": 33150
+    },
+    {
+      "epoch": 0.274310294908384,
+      "grad_norm": 840.7511596679688,
+      "learning_rate": 9.164996404738955e-06,
+      "loss": 160.6511,
+      "step": 33160
+    },
+    {
+      "epoch": 0.27439301815775324,
+      "grad_norm": 915.5698852539062,
+      "learning_rate": 9.16421591952239e-06,
+      "loss": 127.9161,
+      "step": 33170
+    },
+    {
+      "epoch": 0.2744757414071225,
+      "grad_norm": 1336.1126708984375,
+      "learning_rate": 9.163435102978276e-06,
+      "loss": 122.7304,
+      "step": 33180
+    },
+    {
+      "epoch": 0.2745584646564917,
+      "grad_norm": 982.8937377929688,
+      "learning_rate": 9.162653955168739e-06,
+      "loss": 118.9783,
+      "step": 33190
+    },
+    {
+      "epoch": 0.27464118790586095,
+      "grad_norm": 756.2153930664062,
+      "learning_rate": 9.161872476155929e-06,
+      "loss": 101.5269,
+      "step": 33200
+    },
+    {
+      "epoch": 0.2747239111552302,
+      "grad_norm": 732.1403198242188,
+      "learning_rate": 9.161090666002029e-06,
+      "loss": 124.4535,
+      "step": 33210
+    },
+    {
+      "epoch": 0.2748066344045994,
+      "grad_norm": 753.6777954101562,
+      "learning_rate": 9.16030852476924e-06,
+      "loss": 109.3473,
+      "step": 33220
+    },
+    {
+      "epoch": 0.27488935765396866,
+      "grad_norm": 822.023681640625,
+      "learning_rate": 9.159526052519794e-06,
+      "loss": 120.7444,
+      "step": 33230
+    },
+    {
+      "epoch": 0.27497208090333786,
+      "grad_norm": 925.0752563476562,
+      "learning_rate": 9.15874324931595e-06,
+      "loss": 100.8344,
+      "step": 33240
+    },
+    {
+      "epoch": 0.2750548041527071,
+      "grad_norm": 686.9509887695312,
+      "learning_rate": 9.157960115219993e-06,
+      "loss": 113.704,
+      "step": 33250
+    },
+    {
+      "epoch": 0.27513752740207637,
+      "grad_norm": 996.56787109375,
+      "learning_rate": 9.157176650294231e-06,
+      "loss": 133.5279,
+      "step": 33260
+    },
+    {
+      "epoch": 0.27522025065144556,
+      "grad_norm": 728.7750244140625,
+      "learning_rate": 9.156392854601001e-06,
+      "loss": 133.3526,
+      "step": 33270
+    },
+    {
+      "epoch": 0.2753029739008148,
+      "grad_norm": 1302.504638671875,
+      "learning_rate": 9.155608728202669e-06,
+      "loss": 113.5402,
+      "step": 33280
+    },
+    {
+      "epoch": 0.2753856971501841,
+      "grad_norm": 857.5109252929688,
+      "learning_rate": 9.154824271161621e-06,
+      "loss": 83.4826,
+      "step": 33290
+    },
+    {
+      "epoch": 0.27546842039955327,
+      "grad_norm": 910.7625122070312,
+      "learning_rate": 9.154039483540273e-06,
+      "loss": 110.0397,
+      "step": 33300
+    },
+    {
+      "epoch": 0.2755511436489225,
+      "grad_norm": 529.6849365234375,
+      "learning_rate": 9.153254365401069e-06,
+      "loss": 125.4888,
+      "step": 33310
+    },
+    {
+      "epoch": 0.2756338668982918,
+      "grad_norm": 879.7425537109375,
+      "learning_rate": 9.152468916806477e-06,
+      "loss": 103.2796,
+      "step": 33320
+    },
+    {
+      "epoch": 0.275716590147661,
+      "grad_norm": 1299.931884765625,
+      "learning_rate": 9.151683137818989e-06,
+      "loss": 119.4664,
+      "step": 33330
+    },
+    {
+      "epoch": 0.27579931339703023,
+      "grad_norm": 698.8023681640625,
+      "learning_rate": 9.150897028501126e-06,
+      "loss": 100.8363,
+      "step": 33340
+    },
+    {
+      "epoch": 0.2758820366463995,
+      "grad_norm": 1195.03466796875,
+      "learning_rate": 9.15011058891544e-06,
+      "loss": 119.996,
+      "step": 33350
+    },
+    {
+      "epoch": 0.2759647598957687,
+      "grad_norm": 1072.2655029296875,
+      "learning_rate": 9.149323819124498e-06,
+      "loss": 113.2403,
+      "step": 33360
+    },
+    {
+      "epoch": 0.27604748314513794,
+      "grad_norm": 897.378662109375,
+      "learning_rate": 9.148536719190904e-06,
+      "loss": 131.1827,
+      "step": 33370
+    },
+    {
+      "epoch": 0.2761302063945072,
+      "grad_norm": 900.53955078125,
+      "learning_rate": 9.147749289177282e-06,
+      "loss": 141.6734,
+      "step": 33380
+    },
+    {
+      "epoch": 0.2762129296438764,
+      "grad_norm": 809.6741333007812,
+      "learning_rate": 9.146961529146285e-06,
+      "loss": 99.829,
+      "step": 33390
+    },
+    {
+      "epoch": 0.27629565289324565,
+      "grad_norm": 757.1586303710938,
+      "learning_rate": 9.146173439160591e-06,
+      "loss": 117.2545,
+      "step": 33400
+    },
+    {
+      "epoch": 0.2763783761426149,
+      "grad_norm": 763.6544799804688,
+      "learning_rate": 9.145385019282904e-06,
+      "loss": 135.9243,
+      "step": 33410
+    },
+    {
+      "epoch": 0.2764610993919841,
+      "grad_norm": 1427.904296875,
+      "learning_rate": 9.144596269575957e-06,
+      "loss": 125.229,
+      "step": 33420
+    },
+    {
+      "epoch": 0.27654382264135335,
+      "grad_norm": 1012.0225219726562,
+      "learning_rate": 9.143807190102504e-06,
+      "loss": 126.2279,
+      "step": 33430
+    },
+    {
+      "epoch": 0.2766265458907226,
+      "grad_norm": 2056.858154296875,
+      "learning_rate": 9.143017780925331e-06,
+      "loss": 153.6504,
+      "step": 33440
+    },
+    {
+      "epoch": 0.2767092691400918,
+      "grad_norm": 1273.417236328125,
+      "learning_rate": 9.142228042107248e-06,
+      "loss": 109.7093,
+      "step": 33450
+    },
+    {
+      "epoch": 0.27679199238946106,
+      "grad_norm": 1077.05126953125,
+      "learning_rate": 9.141437973711092e-06,
+      "loss": 106.108,
+      "step": 33460
+    },
+    {
+      "epoch": 0.2768747156388303,
+      "grad_norm": 1240.419189453125,
+      "learning_rate": 9.14064757579972e-06,
+      "loss": 95.5216,
+      "step": 33470
+    },
+    {
+      "epoch": 0.2769574388881995,
+      "grad_norm": 1002.42724609375,
+      "learning_rate": 9.139856848436023e-06,
+      "loss": 117.1653,
+      "step": 33480
+    },
+    {
+      "epoch": 0.27704016213756877,
+      "grad_norm": 1525.692626953125,
+      "learning_rate": 9.139065791682916e-06,
+      "loss": 161.3095,
+      "step": 33490
+    },
+    {
+      "epoch": 0.277122885386938,
+      "grad_norm": 773.1032104492188,
+      "learning_rate": 9.138274405603342e-06,
+      "loss": 168.1776,
+      "step": 33500
+    },
+    {
+      "epoch": 0.2772056086363072,
+      "grad_norm": 768.6693115234375,
+      "learning_rate": 9.137482690260265e-06,
+      "loss": 156.8843,
+      "step": 33510
+    },
+    {
+      "epoch": 0.2772883318856765,
+      "grad_norm": 307.3286437988281,
+      "learning_rate": 9.13669064571668e-06,
+      "loss": 109.4442,
+      "step": 33520
+    },
+    {
+      "epoch": 0.27737105513504573,
+      "grad_norm": 3717.694580078125,
+      "learning_rate": 9.135898272035601e-06,
+      "loss": 167.7826,
+      "step": 33530
+    },
+    {
+      "epoch": 0.27745377838441493,
+      "grad_norm": 986.0308227539062,
+      "learning_rate": 9.13510556928008e-06,
+      "loss": 87.5478,
+      "step": 33540
+    },
+    {
+      "epoch": 0.2775365016337842,
+      "grad_norm": 652.1424560546875,
+      "learning_rate": 9.134312537513188e-06,
+      "loss": 106.2764,
+      "step": 33550
+    },
+    {
+      "epoch": 0.27761922488315344,
+      "grad_norm": 720.5858154296875,
+      "learning_rate": 9.133519176798021e-06,
+      "loss": 152.1906,
+      "step": 33560
+    },
+    {
+      "epoch": 0.27770194813252264,
+      "grad_norm": 1833.9097900390625,
+      "learning_rate": 9.132725487197701e-06,
+      "loss": 116.5092,
+      "step": 33570
+    },
+    {
+      "epoch": 0.2777846713818919,
+      "grad_norm": 1374.428955078125,
+      "learning_rate": 9.131931468775382e-06,
+      "loss": 132.6865,
+      "step": 33580
+    },
+    {
+      "epoch": 0.2778673946312611,
+      "grad_norm": 724.4324340820312,
+      "learning_rate": 9.131137121594239e-06,
+      "loss": 127.3931,
+      "step": 33590
+    },
+    {
+      "epoch": 0.27795011788063034,
+      "grad_norm": 594.5848999023438,
+      "learning_rate": 9.130342445717474e-06,
+      "loss": 117.9793,
+      "step": 33600
+    },
+    {
+      "epoch": 0.2780328411299996,
+      "grad_norm": 1347.4599609375,
+      "learning_rate": 9.129547441208317e-06,
+      "loss": 123.5553,
+      "step": 33610
+    },
+    {
+      "epoch": 0.2781155643793688,
+      "grad_norm": 825.2183837890625,
+      "learning_rate": 9.128752108130022e-06,
+      "loss": 109.409,
+      "step": 33620
+    },
+    {
+      "epoch": 0.27819828762873805,
+      "grad_norm": 952.2738037109375,
+      "learning_rate": 9.12795644654587e-06,
+      "loss": 149.2746,
+      "step": 33630
+    },
+    {
+      "epoch": 0.2782810108781073,
+      "grad_norm": 6804.1904296875,
+      "learning_rate": 9.127160456519168e-06,
+      "loss": 132.9842,
+      "step": 33640
+    },
+    {
+      "epoch": 0.2783637341274765,
+      "grad_norm": 953.0177612304688,
+      "learning_rate": 9.126364138113251e-06,
+      "loss": 119.1077,
+      "step": 33650
+    },
+    {
+      "epoch": 0.27844645737684576,
+      "grad_norm": 771.2464599609375,
+      "learning_rate": 9.125567491391476e-06,
+      "loss": 117.123,
+      "step": 33660
+    },
+    {
+      "epoch": 0.278529180626215,
+      "grad_norm": 1822.6439208984375,
+      "learning_rate": 9.12477051641723e-06,
+      "loss": 151.1186,
+      "step": 33670
+    },
+    {
+      "epoch": 0.2786119038755842,
+      "grad_norm": 872.4586791992188,
+      "learning_rate": 9.123973213253923e-06,
+      "loss": 112.0873,
+      "step": 33680
+    },
+    {
+      "epoch": 0.27869462712495346,
+      "grad_norm": 494.63134765625,
+      "learning_rate": 9.123175581964995e-06,
+      "loss": 100.9059,
+      "step": 33690
+    },
+    {
+      "epoch": 0.2787773503743227,
+      "grad_norm": 959.332275390625,
+      "learning_rate": 9.122377622613909e-06,
+      "loss": 106.5335,
+      "step": 33700
+    },
+    {
+      "epoch": 0.2788600736236919,
+      "grad_norm": 886.007568359375,
+      "learning_rate": 9.121579335264155e-06,
+      "loss": 128.5588,
+      "step": 33710
+    },
+    {
+      "epoch": 0.27894279687306117,
+      "grad_norm": 1174.72265625,
+      "learning_rate": 9.120780719979248e-06,
+      "loss": 92.5889,
+      "step": 33720
+    },
+    {
+      "epoch": 0.2790255201224304,
+      "grad_norm": 2207.4111328125,
+      "learning_rate": 9.11998177682273e-06,
+      "loss": 132.5646,
+      "step": 33730
+    },
+    {
+      "epoch": 0.2791082433717996,
+      "grad_norm": 730.7549438476562,
+      "learning_rate": 9.11918250585817e-06,
+      "loss": 82.5129,
+      "step": 33740
+    },
+    {
+      "epoch": 0.2791909666211689,
+      "grad_norm": 876.688232421875,
+      "learning_rate": 9.118382907149164e-06,
+      "loss": 109.9412,
+      "step": 33750
+    },
+    {
+      "epoch": 0.27927368987053813,
+      "grad_norm": 831.474609375,
+      "learning_rate": 9.117582980759332e-06,
+      "loss": 124.5468,
+      "step": 33760
+    },
+    {
+      "epoch": 0.27935641311990733,
+      "grad_norm": 820.8474731445312,
+      "learning_rate": 9.116782726752317e-06,
+      "loss": 126.4644,
+      "step": 33770
+    },
+    {
+      "epoch": 0.2794391363692766,
+      "grad_norm": 921.0737915039062,
+      "learning_rate": 9.115982145191796e-06,
+      "loss": 116.8273,
+      "step": 33780
+    },
+    {
+      "epoch": 0.27952185961864584,
+      "grad_norm": 1426.6785888671875,
+      "learning_rate": 9.115181236141463e-06,
+      "loss": 127.0457,
+      "step": 33790
+    },
+    {
+      "epoch": 0.27960458286801504,
+      "grad_norm": 1093.164794921875,
+      "learning_rate": 9.114379999665047e-06,
+      "loss": 85.9147,
+      "step": 33800
+    },
+    {
+      "epoch": 0.2796873061173843,
+      "grad_norm": 1163.352783203125,
+      "learning_rate": 9.113578435826295e-06,
+      "loss": 140.9147,
+      "step": 33810
+    },
+    {
+      "epoch": 0.27977002936675355,
+      "grad_norm": 412.13787841796875,
+      "learning_rate": 9.112776544688988e-06,
+      "loss": 127.3203,
+      "step": 33820
+    },
+    {
+      "epoch": 0.27985275261612275,
+      "grad_norm": 785.0323486328125,
+      "learning_rate": 9.111974326316926e-06,
+      "loss": 103.5417,
+      "step": 33830
+    },
+    {
+      "epoch": 0.279935475865492,
+      "grad_norm": 1045.812255859375,
+      "learning_rate": 9.111171780773938e-06,
+      "loss": 120.1187,
+      "step": 33840
+    },
+    {
+      "epoch": 0.28001819911486125,
+      "grad_norm": 916.0906372070312,
+      "learning_rate": 9.110368908123878e-06,
+      "loss": 139.0241,
+      "step": 33850
+    },
+    {
+      "epoch": 0.28010092236423045,
+      "grad_norm": 1398.31982421875,
+      "learning_rate": 9.10956570843063e-06,
+      "loss": 132.5252,
+      "step": 33860
+    },
+    {
+      "epoch": 0.2801836456135997,
+      "grad_norm": 1527.2587890625,
+      "learning_rate": 9.108762181758096e-06,
+      "loss": 130.4855,
+      "step": 33870
+    },
+    {
+      "epoch": 0.28026636886296896,
+      "grad_norm": 751.3243408203125,
+      "learning_rate": 9.107958328170215e-06,
+      "loss": 106.5942,
+      "step": 33880
+    },
+    {
+      "epoch": 0.28034909211233816,
+      "grad_norm": 1024.8853759765625,
+      "learning_rate": 9.10715414773094e-06,
+      "loss": 109.6669,
+      "step": 33890
+    },
+    {
+      "epoch": 0.2804318153617074,
+      "grad_norm": 1099.5294189453125,
+      "learning_rate": 9.10634964050426e-06,
+      "loss": 119.2499,
+      "step": 33900
+    },
+    {
+      "epoch": 0.28051453861107667,
+      "grad_norm": 775.4506225585938,
+      "learning_rate": 9.105544806554184e-06,
+      "loss": 157.9587,
+      "step": 33910
+    },
+    {
+      "epoch": 0.28059726186044587,
+      "grad_norm": 1175.566162109375,
+      "learning_rate": 9.104739645944752e-06,
+      "loss": 179.6702,
+      "step": 33920
+    },
+    {
+      "epoch": 0.2806799851098151,
+      "grad_norm": 1368.4671630859375,
+      "learning_rate": 9.103934158740023e-06,
+      "loss": 129.1513,
+      "step": 33930
+    },
+    {
+      "epoch": 0.2807627083591844,
+      "grad_norm": 1385.238037109375,
+      "learning_rate": 9.10312834500409e-06,
+      "loss": 150.3397,
+      "step": 33940
+    },
+    {
+      "epoch": 0.2808454316085536,
+      "grad_norm": 1050.946044921875,
+      "learning_rate": 9.102322204801062e-06,
+      "loss": 118.2614,
+      "step": 33950
+    },
+    {
+      "epoch": 0.28092815485792283,
+      "grad_norm": 2881.64013671875,
+      "learning_rate": 9.101515738195084e-06,
+      "loss": 100.6495,
+      "step": 33960
+    },
+    {
+      "epoch": 0.28101087810729203,
+      "grad_norm": 394.7975769042969,
+      "learning_rate": 9.100708945250322e-06,
+      "loss": 81.3734,
+      "step": 33970
+    },
+    {
+      "epoch": 0.2810936013566613,
+      "grad_norm": 1190.456298828125,
+      "learning_rate": 9.099901826030969e-06,
+      "loss": 130.64,
+      "step": 33980
+    },
+    {
+      "epoch": 0.28117632460603054,
+      "grad_norm": 1374.1256103515625,
+      "learning_rate": 9.099094380601244e-06,
+      "loss": 119.4305,
+      "step": 33990
+    },
+    {
+      "epoch": 0.28125904785539974,
+      "grad_norm": 1412.7530517578125,
+      "learning_rate": 9.098286609025392e-06,
+      "loss": 106.0938,
+      "step": 34000
+    },
+    {
+      "epoch": 0.281341771104769,
+      "grad_norm": 617.6422119140625,
+      "learning_rate": 9.097478511367682e-06,
+      "loss": 119.852,
+      "step": 34010
+    },
+    {
+      "epoch": 0.28142449435413824,
+      "grad_norm": 857.2781982421875,
+      "learning_rate": 9.096670087692413e-06,
+      "loss": 168.8287,
+      "step": 34020
+    },
+    {
+      "epoch": 0.28150721760350744,
+      "grad_norm": 0.0,
+      "learning_rate": 9.095861338063906e-06,
+      "loss": 99.5146,
+      "step": 34030
+    },
+    {
+      "epoch": 0.2815899408528767,
+      "grad_norm": 1164.4888916015625,
+      "learning_rate": 9.09505226254651e-06,
+      "loss": 139.5455,
+      "step": 34040
+    },
+    {
+      "epoch": 0.28167266410224595,
+      "grad_norm": 1813.4522705078125,
+      "learning_rate": 9.094242861204598e-06,
+      "loss": 153.4502,
+      "step": 34050
+    },
+    {
+      "epoch": 0.28175538735161515,
+      "grad_norm": 918.6724243164062,
+      "learning_rate": 9.093433134102572e-06,
+      "loss": 95.0513,
+      "step": 34060
+    },
+    {
+      "epoch": 0.2818381106009844,
+      "grad_norm": 758.8914794921875,
+      "learning_rate": 9.09262308130486e-06,
+      "loss": 127.6234,
+      "step": 34070
+    },
+    {
+      "epoch": 0.28192083385035366,
+      "grad_norm": 1214.897705078125,
+      "learning_rate": 9.091812702875908e-06,
+      "loss": 135.7131,
+      "step": 34080
+    },
+    {
+      "epoch": 0.28200355709972286,
+      "grad_norm": 1762.4208984375,
+      "learning_rate": 9.0910019988802e-06,
+      "loss": 116.575,
+      "step": 34090
+    },
+    {
+      "epoch": 0.2820862803490921,
+      "grad_norm": 1027.57958984375,
+      "learning_rate": 9.09019096938224e-06,
+      "loss": 115.0406,
+      "step": 34100
+    },
+    {
+      "epoch": 0.28216900359846137,
+      "grad_norm": 549.1436157226562,
+      "learning_rate": 9.089379614446554e-06,
+      "loss": 101.1465,
+      "step": 34110
+    },
+    {
+      "epoch": 0.28225172684783056,
+      "grad_norm": 1149.4486083984375,
+      "learning_rate": 9.0885679341377e-06,
+      "loss": 111.7543,
+      "step": 34120
+    },
+    {
+      "epoch": 0.2823344500971998,
+      "grad_norm": 827.5548095703125,
+      "learning_rate": 9.08775592852026e-06,
+      "loss": 84.8386,
+      "step": 34130
+    },
+    {
+      "epoch": 0.2824171733465691,
+      "grad_norm": 2282.6884765625,
+      "learning_rate": 9.08694359765884e-06,
+      "loss": 122.2059,
+      "step": 34140
+    },
+    {
+      "epoch": 0.28249989659593827,
+      "grad_norm": 1563.5760498046875,
+      "learning_rate": 9.086130941618075e-06,
+      "loss": 127.1692,
+      "step": 34150
+    },
+    {
+      "epoch": 0.2825826198453075,
+      "grad_norm": 546.21337890625,
+      "learning_rate": 9.085317960462625e-06,
+      "loss": 91.0043,
+      "step": 34160
+    },
+    {
+      "epoch": 0.2826653430946768,
+      "grad_norm": 1092.4716796875,
+      "learning_rate": 9.084504654257173e-06,
+      "loss": 126.0462,
+      "step": 34170
+    },
+    {
+      "epoch": 0.282748066344046,
+      "grad_norm": 1006.257568359375,
+      "learning_rate": 9.08369102306643e-06,
+      "loss": 104.315,
+      "step": 34180
+    },
+    {
+      "epoch": 0.28283078959341523,
+      "grad_norm": 1649.3040771484375,
+      "learning_rate": 9.082877066955135e-06,
+      "loss": 101.4608,
+      "step": 34190
+    },
+    {
+      "epoch": 0.2829135128427845,
+      "grad_norm": 1023.9491577148438,
+      "learning_rate": 9.08206278598805e-06,
+      "loss": 133.0885,
+      "step": 34200
+    },
+    {
+      "epoch": 0.2829962360921537,
+      "grad_norm": 1013.92041015625,
+      "learning_rate": 9.081248180229963e-06,
+      "loss": 93.8945,
+      "step": 34210
+    },
+    {
+      "epoch": 0.28307895934152294,
+      "grad_norm": 655.7772827148438,
+      "learning_rate": 9.080433249745688e-06,
+      "loss": 104.3141,
+      "step": 34220
+    },
+    {
+      "epoch": 0.2831616825908922,
+      "grad_norm": 1191.8131103515625,
+      "learning_rate": 9.079617994600066e-06,
+      "loss": 147.7726,
+      "step": 34230
+    },
+    {
+      "epoch": 0.2832444058402614,
+      "grad_norm": 628.782470703125,
+      "learning_rate": 9.078802414857963e-06,
+      "loss": 96.9786,
+      "step": 34240
+    },
+    {
+      "epoch": 0.28332712908963065,
+      "grad_norm": 695.0119018554688,
+      "learning_rate": 9.077986510584273e-06,
+      "loss": 111.2695,
+      "step": 34250
+    },
+    {
+      "epoch": 0.2834098523389999,
+      "grad_norm": 1148.8243408203125,
+      "learning_rate": 9.07717028184391e-06,
+      "loss": 104.8876,
+      "step": 34260
+    },
+    {
+      "epoch": 0.2834925755883691,
+      "grad_norm": 766.49072265625,
+      "learning_rate": 9.07635372870182e-06,
+      "loss": 84.2609,
+      "step": 34270
+    },
+    {
+      "epoch": 0.28357529883773835,
+      "grad_norm": 1150.672119140625,
+      "learning_rate": 9.07553685122297e-06,
+      "loss": 146.0944,
+      "step": 34280
+    },
+    {
+      "epoch": 0.2836580220871076,
+      "grad_norm": 1328.8944091796875,
+      "learning_rate": 9.074719649472358e-06,
+      "loss": 115.0404,
+      "step": 34290
+    },
+    {
+      "epoch": 0.2837407453364768,
+      "grad_norm": 1080.066650390625,
+      "learning_rate": 9.073902123515005e-06,
+      "loss": 144.9194,
+      "step": 34300
+    },
+    {
+      "epoch": 0.28382346858584606,
+      "grad_norm": 676.450439453125,
+      "learning_rate": 9.073084273415956e-06,
+      "loss": 79.431,
+      "step": 34310
+    },
+    {
+      "epoch": 0.28390619183521526,
+      "grad_norm": 1445.7745361328125,
+      "learning_rate": 9.072266099240286e-06,
+      "loss": 99.8113,
+      "step": 34320
+    },
+    {
+      "epoch": 0.2839889150845845,
+      "grad_norm": 739.0885009765625,
+      "learning_rate": 9.07144760105309e-06,
+      "loss": 142.6431,
+      "step": 34330
+    },
+    {
+      "epoch": 0.28407163833395377,
+      "grad_norm": 920.0739135742188,
+      "learning_rate": 9.070628778919493e-06,
+      "loss": 93.1577,
+      "step": 34340
+    },
+    {
+      "epoch": 0.28415436158332297,
+      "grad_norm": 1188.843017578125,
+      "learning_rate": 9.069809632904647e-06,
+      "loss": 111.0898,
+      "step": 34350
+    },
+    {
+      "epoch": 0.2842370848326922,
+      "grad_norm": 991.0109252929688,
+      "learning_rate": 9.068990163073726e-06,
+      "loss": 87.4462,
+      "step": 34360
+    },
+    {
+      "epoch": 0.2843198080820615,
+      "grad_norm": 845.0187377929688,
+      "learning_rate": 9.068170369491932e-06,
+      "loss": 95.4965,
+      "step": 34370
+    },
+    {
+      "epoch": 0.2844025313314307,
+      "grad_norm": 861.9247436523438,
+      "learning_rate": 9.067350252224491e-06,
+      "loss": 76.6258,
+      "step": 34380
+    },
+    {
+      "epoch": 0.28448525458079993,
+      "grad_norm": 1371.8582763671875,
+      "learning_rate": 9.066529811336658e-06,
+      "loss": 117.6796,
+      "step": 34390
+    },
+    {
+      "epoch": 0.2845679778301692,
+      "grad_norm": 909.2398681640625,
+      "learning_rate": 9.06570904689371e-06,
+      "loss": 96.8476,
+      "step": 34400
+    },
+    {
+      "epoch": 0.2846507010795384,
+      "grad_norm": 941.552978515625,
+      "learning_rate": 9.064887958960953e-06,
+      "loss": 89.4982,
+      "step": 34410
+    },
+    {
+      "epoch": 0.28473342432890764,
+      "grad_norm": 2689.89404296875,
+      "learning_rate": 9.064066547603716e-06,
+      "loss": 116.0555,
+      "step": 34420
+    },
+    {
+      "epoch": 0.2848161475782769,
+      "grad_norm": 1277.172607421875,
+      "learning_rate": 9.063244812887357e-06,
+      "loss": 111.9663,
+      "step": 34430
+    },
+    {
+      "epoch": 0.2848988708276461,
+      "grad_norm": 872.6428833007812,
+      "learning_rate": 9.062422754877253e-06,
+      "loss": 124.443,
+      "step": 34440
+    },
+    {
+      "epoch": 0.28498159407701534,
+      "grad_norm": 698.310546875,
+      "learning_rate": 9.061600373638816e-06,
+      "loss": 127.376,
+      "step": 34450
+    },
+    {
+      "epoch": 0.2850643173263846,
+      "grad_norm": 914.8101806640625,
+      "learning_rate": 9.06077766923748e-06,
+      "loss": 113.9157,
+      "step": 34460
+    },
+    {
+      "epoch": 0.2851470405757538,
+      "grad_norm": 1122.4351806640625,
+      "learning_rate": 9.059954641738697e-06,
+      "loss": 126.3374,
+      "step": 34470
+    },
+    {
+      "epoch": 0.28522976382512305,
+      "grad_norm": 1123.8736572265625,
+      "learning_rate": 9.059131291207958e-06,
+      "loss": 105.3611,
+      "step": 34480
+    },
+    {
+      "epoch": 0.2853124870744923,
+      "grad_norm": 656.4722290039062,
+      "learning_rate": 9.058307617710771e-06,
+      "loss": 142.5772,
+      "step": 34490
+    },
+    {
+      "epoch": 0.2853952103238615,
+      "grad_norm": 608.4116821289062,
+      "learning_rate": 9.057483621312671e-06,
+      "loss": 115.3732,
+      "step": 34500
+    },
+    {
+      "epoch": 0.28547793357323076,
+      "grad_norm": 581.153564453125,
+      "learning_rate": 9.056659302079222e-06,
+      "loss": 118.1142,
+      "step": 34510
+    },
+    {
+      "epoch": 0.2855606568226,
+      "grad_norm": 1226.16552734375,
+      "learning_rate": 9.055834660076008e-06,
+      "loss": 110.4029,
+      "step": 34520
+    },
+    {
+      "epoch": 0.2856433800719692,
+      "grad_norm": 1013.7431640625,
+      "learning_rate": 9.055009695368646e-06,
+      "loss": 135.0033,
+      "step": 34530
+    },
+    {
+      "epoch": 0.28572610332133846,
+      "grad_norm": 749.7244262695312,
+      "learning_rate": 9.054184408022772e-06,
+      "loss": 157.7157,
+      "step": 34540
+    },
+    {
+      "epoch": 0.2858088265707077,
+      "grad_norm": 456.33026123046875,
+      "learning_rate": 9.05335879810405e-06,
+      "loss": 145.5047,
+      "step": 34550
+    },
+    {
+      "epoch": 0.2858915498200769,
+      "grad_norm": 622.3046264648438,
+      "learning_rate": 9.052532865678171e-06,
+      "loss": 123.2549,
+      "step": 34560
+    },
+    {
+      "epoch": 0.28597427306944617,
+      "grad_norm": 1478.7266845703125,
+      "learning_rate": 9.05170661081085e-06,
+      "loss": 127.2245,
+      "step": 34570
+    },
+    {
+      "epoch": 0.2860569963188154,
+      "grad_norm": 619.5127563476562,
+      "learning_rate": 9.050880033567831e-06,
+      "loss": 132.0401,
+      "step": 34580
+    },
+    {
+      "epoch": 0.2861397195681846,
+      "grad_norm": 1351.7064208984375,
+      "learning_rate": 9.050053134014878e-06,
+      "loss": 100.6227,
+      "step": 34590
+    },
+    {
+      "epoch": 0.2862224428175539,
+      "grad_norm": 1285.1185302734375,
+      "learning_rate": 9.049225912217782e-06,
+      "loss": 102.2231,
+      "step": 34600
+    },
+    {
+      "epoch": 0.28630516606692313,
+      "grad_norm": 1479.285888671875,
+      "learning_rate": 9.048398368242365e-06,
+      "loss": 93.523,
+      "step": 34610
+    },
+    {
+      "epoch": 0.28638788931629233,
+      "grad_norm": 845.21826171875,
+      "learning_rate": 9.047570502154471e-06,
+      "loss": 97.2673,
+      "step": 34620
+    },
+    {
+      "epoch": 0.2864706125656616,
+      "grad_norm": 1318.1756591796875,
+      "learning_rate": 9.046742314019968e-06,
+      "loss": 136.1264,
+      "step": 34630
+    },
+    {
+      "epoch": 0.28655333581503084,
+      "grad_norm": 1689.4859619140625,
+      "learning_rate": 9.045913803904748e-06,
+      "loss": 127.8525,
+      "step": 34640
+    },
+    {
+      "epoch": 0.28663605906440004,
+      "grad_norm": 759.2603149414062,
+      "learning_rate": 9.045084971874738e-06,
+      "loss": 119.6263,
+      "step": 34650
+    },
+    {
+      "epoch": 0.2867187823137693,
+      "grad_norm": 1472.703857421875,
+      "learning_rate": 9.04425581799588e-06,
+      "loss": 122.0518,
+      "step": 34660
+    },
+    {
+      "epoch": 0.2868015055631385,
+      "grad_norm": 903.3499755859375,
+      "learning_rate": 9.043426342334147e-06,
+      "loss": 104.4996,
+      "step": 34670
+    },
+    {
+      "epoch": 0.28688422881250775,
+      "grad_norm": 726.6737670898438,
+      "learning_rate": 9.042596544955538e-06,
+      "loss": 93.7296,
+      "step": 34680
+    },
+    {
+      "epoch": 0.286966952061877,
+      "grad_norm": 1467.8472900390625,
+      "learning_rate": 9.041766425926073e-06,
+      "loss": 109.8673,
+      "step": 34690
+    },
+    {
+      "epoch": 0.2870496753112462,
+      "grad_norm": 1312.5286865234375,
+      "learning_rate": 9.040935985311804e-06,
+      "loss": 83.5798,
+      "step": 34700
+    },
+    {
+      "epoch": 0.28713239856061545,
+      "grad_norm": 1040.613525390625,
+      "learning_rate": 9.040105223178803e-06,
+      "loss": 105.7686,
+      "step": 34710
+    },
+    {
+      "epoch": 0.2872151218099847,
+      "grad_norm": 737.3902587890625,
+      "learning_rate": 9.039274139593173e-06,
+      "loss": 100.6217,
+      "step": 34720
+    },
+    {
+      "epoch": 0.2872978450593539,
+      "grad_norm": 1900.810302734375,
+      "learning_rate": 9.038442734621034e-06,
+      "loss": 108.4988,
+      "step": 34730
+    },
+    {
+      "epoch": 0.28738056830872316,
+      "grad_norm": 823.1213989257812,
+      "learning_rate": 9.037611008328544e-06,
+      "loss": 87.0847,
+      "step": 34740
+    },
+    {
+      "epoch": 0.2874632915580924,
+      "grad_norm": 1027.710693359375,
+      "learning_rate": 9.036778960781874e-06,
+      "loss": 96.1623,
+      "step": 34750
+    },
+    {
+      "epoch": 0.2875460148074616,
+      "grad_norm": 1012.5924682617188,
+      "learning_rate": 9.03594659204723e-06,
+      "loss": 99.4886,
+      "step": 34760
+    },
+    {
+      "epoch": 0.28762873805683087,
+      "grad_norm": 1981.74609375,
+      "learning_rate": 9.035113902190838e-06,
+      "loss": 131.8146,
+      "step": 34770
+    },
+    {
+      "epoch": 0.2877114613062001,
+      "grad_norm": 838.7272338867188,
+      "learning_rate": 9.03428089127895e-06,
+      "loss": 138.8772,
+      "step": 34780
+    },
+    {
+      "epoch": 0.2877941845555693,
+      "grad_norm": 614.944091796875,
+      "learning_rate": 9.033447559377847e-06,
+      "loss": 102.8104,
+      "step": 34790
+    },
+    {
+      "epoch": 0.2878769078049386,
+      "grad_norm": 1577.7645263671875,
+      "learning_rate": 9.032613906553833e-06,
+      "loss": 128.1048,
+      "step": 34800
+    },
+    {
+      "epoch": 0.28795963105430783,
+      "grad_norm": 1021.2303466796875,
+      "learning_rate": 9.031779932873238e-06,
+      "loss": 107.6142,
+      "step": 34810
+    },
+    {
+      "epoch": 0.28804235430367703,
+      "grad_norm": 644.7843017578125,
+      "learning_rate": 9.030945638402415e-06,
+      "loss": 103.4213,
+      "step": 34820
+    },
+    {
+      "epoch": 0.2881250775530463,
+      "grad_norm": 873.067138671875,
+      "learning_rate": 9.030111023207751e-06,
+      "loss": 93.0126,
+      "step": 34830
+    },
+    {
+      "epoch": 0.28820780080241554,
+      "grad_norm": 1150.903564453125,
+      "learning_rate": 9.029276087355646e-06,
+      "loss": 121.5422,
+      "step": 34840
+    },
+    {
+      "epoch": 0.28829052405178474,
+      "grad_norm": 573.4194946289062,
+      "learning_rate": 9.028440830912536e-06,
+      "loss": 120.833,
+      "step": 34850
+    },
+    {
+      "epoch": 0.288373247301154,
+      "grad_norm": 951.9305419921875,
+      "learning_rate": 9.027605253944874e-06,
+      "loss": 146.5991,
+      "step": 34860
+    },
+    {
+      "epoch": 0.28845597055052324,
+      "grad_norm": 646.8169555664062,
+      "learning_rate": 9.026769356519149e-06,
+      "loss": 73.8264,
+      "step": 34870
+    },
+    {
+      "epoch": 0.28853869379989244,
+      "grad_norm": 2599.574462890625,
+      "learning_rate": 9.025933138701865e-06,
+      "loss": 121.7481,
+      "step": 34880
+    },
+    {
+      "epoch": 0.2886214170492617,
+      "grad_norm": 964.8515625,
+      "learning_rate": 9.02509660055956e-06,
+      "loss": 98.844,
+      "step": 34890
+    },
+    {
+      "epoch": 0.28870414029863095,
+      "grad_norm": 665.4266357421875,
+      "learning_rate": 9.02425974215879e-06,
+      "loss": 126.3558,
+      "step": 34900
+    },
+    {
+      "epoch": 0.28878686354800015,
+      "grad_norm": 747.7141723632812,
+      "learning_rate": 9.02342256356614e-06,
+      "loss": 94.1504,
+      "step": 34910
+    },
+    {
+      "epoch": 0.2888695867973694,
+      "grad_norm": 1255.60302734375,
+      "learning_rate": 9.022585064848222e-06,
+      "loss": 101.1276,
+      "step": 34920
+    },
+    {
+      "epoch": 0.28895231004673866,
+      "grad_norm": 1393.7789306640625,
+      "learning_rate": 9.021747246071673e-06,
+      "loss": 106.2699,
+      "step": 34930
+    },
+    {
+      "epoch": 0.28903503329610786,
+      "grad_norm": 860.56689453125,
+      "learning_rate": 9.020909107303152e-06,
+      "loss": 124.7073,
+      "step": 34940
+    },
+    {
+      "epoch": 0.2891177565454771,
+      "grad_norm": 982.3703002929688,
+      "learning_rate": 9.020070648609347e-06,
+      "loss": 112.9233,
+      "step": 34950
+    },
+    {
+      "epoch": 0.28920047979484637,
+      "grad_norm": 1855.6322021484375,
+      "learning_rate": 9.01923187005697e-06,
+      "loss": 108.5488,
+      "step": 34960
+    },
+    {
+      "epoch": 0.28928320304421556,
+      "grad_norm": 904.826416015625,
+      "learning_rate": 9.018392771712758e-06,
+      "loss": 125.191,
+      "step": 34970
+    },
+    {
+      "epoch": 0.2893659262935848,
+      "grad_norm": 704.2764892578125,
+      "learning_rate": 9.017553353643479e-06,
+      "loss": 126.2956,
+      "step": 34980
+    },
+    {
+      "epoch": 0.2894486495429541,
+      "grad_norm": 956.4945678710938,
+      "learning_rate": 9.016713615915913e-06,
+      "loss": 81.5747,
+      "step": 34990
+    },
+    {
+      "epoch": 0.28953137279232327,
+      "grad_norm": 783.7576293945312,
+      "learning_rate": 9.01587355859688e-06,
+      "loss": 92.5522,
+      "step": 35000
+    },
+    {
+      "epoch": 0.2896140960416925,
+      "grad_norm": 1208.607421875,
+      "learning_rate": 9.015033181753219e-06,
+      "loss": 106.08,
+      "step": 35010
+    },
+    {
+      "epoch": 0.2896968192910618,
+      "grad_norm": 852.6314697265625,
+      "learning_rate": 9.014192485451794e-06,
+      "loss": 116.7134,
+      "step": 35020
+    },
+    {
+      "epoch": 0.289779542540431,
+      "grad_norm": 1022.0242309570312,
+      "learning_rate": 9.013351469759497e-06,
+      "loss": 144.6814,
+      "step": 35030
+    },
+    {
+      "epoch": 0.28986226578980023,
+      "grad_norm": 1243.4373779296875,
+      "learning_rate": 9.01251013474324e-06,
+      "loss": 104.1561,
+      "step": 35040
+    },
+    {
+      "epoch": 0.28994498903916943,
+      "grad_norm": 960.918701171875,
+      "learning_rate": 9.011668480469969e-06,
+      "loss": 105.9182,
+      "step": 35050
+    },
+    {
+      "epoch": 0.2900277122885387,
+      "grad_norm": 601.8880004882812,
+      "learning_rate": 9.010826507006644e-06,
+      "loss": 94.2774,
+      "step": 35060
+    },
+    {
+      "epoch": 0.29011043553790794,
+      "grad_norm": 642.360107421875,
+      "learning_rate": 9.009984214420265e-06,
+      "loss": 142.8863,
+      "step": 35070
+    },
+    {
+      "epoch": 0.29019315878727714,
+      "grad_norm": 1095.9666748046875,
+      "learning_rate": 9.009141602777845e-06,
+      "loss": 128.7609,
+      "step": 35080
+    },
+    {
+      "epoch": 0.2902758820366464,
+      "grad_norm": 1230.556396484375,
+      "learning_rate": 9.008298672146425e-06,
+      "loss": 111.0792,
+      "step": 35090
+    },
+    {
+      "epoch": 0.29035860528601565,
+      "grad_norm": 1034.6533203125,
+      "learning_rate": 9.007455422593077e-06,
+      "loss": 111.8609,
+      "step": 35100
+    },
+    {
+      "epoch": 0.29044132853538485,
+      "grad_norm": 509.1571960449219,
+      "learning_rate": 9.006611854184893e-06,
+      "loss": 88.8575,
+      "step": 35110
+    },
+    {
+      "epoch": 0.2905240517847541,
+      "grad_norm": 832.6845703125,
+      "learning_rate": 9.00576796698899e-06,
+      "loss": 113.1765,
+      "step": 35120
+    },
+    {
+      "epoch": 0.29060677503412335,
+      "grad_norm": 1436.23388671875,
+      "learning_rate": 9.004923761072515e-06,
+      "loss": 103.5024,
+      "step": 35130
+    },
+    {
+      "epoch": 0.29068949828349255,
+      "grad_norm": 1610.103759765625,
+      "learning_rate": 9.004079236502636e-06,
+      "loss": 113.4215,
+      "step": 35140
+    },
+    {
+      "epoch": 0.2907722215328618,
+      "grad_norm": 881.2527465820312,
+      "learning_rate": 9.00323439334655e-06,
+      "loss": 123.1393,
+      "step": 35150
+    },
+    {
+      "epoch": 0.29085494478223106,
+      "grad_norm": 734.708740234375,
+      "learning_rate": 9.002389231671474e-06,
+      "loss": 121.4382,
+      "step": 35160
+    },
+    {
+      "epoch": 0.29093766803160026,
+      "grad_norm": 787.5501708984375,
+      "learning_rate": 9.001543751544654e-06,
+      "loss": 107.295,
+      "step": 35170
+    },
+    {
+      "epoch": 0.2910203912809695,
+      "grad_norm": 1131.2176513671875,
+      "learning_rate": 9.000697953033364e-06,
+      "loss": 107.077,
+      "step": 35180
+    },
+    {
+      "epoch": 0.29110311453033877,
+      "grad_norm": 1214.8603515625,
+      "learning_rate": 8.999851836204901e-06,
+      "loss": 103.6586,
+      "step": 35190
+    },
+    {
+      "epoch": 0.29118583777970797,
+      "grad_norm": 895.4462280273438,
+      "learning_rate": 8.99900540112658e-06,
+      "loss": 83.9514,
+      "step": 35200
+    },
+    {
+      "epoch": 0.2912685610290772,
+      "grad_norm": 543.6385498046875,
+      "learning_rate": 8.998158647865753e-06,
+      "loss": 107.0998,
+      "step": 35210
+    },
+    {
+      "epoch": 0.2913512842784465,
+      "grad_norm": 1422.5147705078125,
+      "learning_rate": 8.997311576489793e-06,
+      "loss": 127.3802,
+      "step": 35220
+    },
+    {
+      "epoch": 0.2914340075278157,
+      "grad_norm": 1031.476806640625,
+      "learning_rate": 8.996464187066096e-06,
+      "loss": 132.6045,
+      "step": 35230
+    },
+    {
+      "epoch": 0.29151673077718493,
+      "grad_norm": 962.57568359375,
+      "learning_rate": 8.995616479662084e-06,
+      "loss": 76.9806,
+      "step": 35240
+    },
+    {
+      "epoch": 0.2915994540265542,
+      "grad_norm": 888.2576293945312,
+      "learning_rate": 8.994768454345207e-06,
+      "loss": 88.927,
+      "step": 35250
+    },
+    {
+      "epoch": 0.2916821772759234,
+      "grad_norm": 991.271484375,
+      "learning_rate": 8.993920111182937e-06,
+      "loss": 116.7842,
+      "step": 35260
+    },
+    {
+      "epoch": 0.29176490052529264,
+      "grad_norm": 2302.41064453125,
+      "learning_rate": 8.993071450242775e-06,
+      "loss": 99.9801,
+      "step": 35270
+    },
+    {
+      "epoch": 0.2918476237746619,
+      "grad_norm": 729.2933349609375,
+      "learning_rate": 8.99222247159224e-06,
+      "loss": 108.2754,
+      "step": 35280
+    },
+    {
+      "epoch": 0.2919303470240311,
+      "grad_norm": 940.2220458984375,
+      "learning_rate": 8.991373175298887e-06,
+      "loss": 104.497,
+      "step": 35290
+    },
+    {
+      "epoch": 0.29201307027340034,
+      "grad_norm": 893.2850952148438,
+      "learning_rate": 8.99052356143029e-06,
+      "loss": 104.8119,
+      "step": 35300
+    },
+    {
+      "epoch": 0.2920957935227696,
+      "grad_norm": 1792.8408203125,
+      "learning_rate": 8.989673630054044e-06,
+      "loss": 134.8155,
+      "step": 35310
+    },
+    {
+      "epoch": 0.2921785167721388,
+      "grad_norm": 823.2114868164062,
+      "learning_rate": 8.988823381237778e-06,
+      "loss": 91.5063,
+      "step": 35320
+    },
+    {
+      "epoch": 0.29226124002150805,
+      "grad_norm": 1062.8973388671875,
+      "learning_rate": 8.987972815049144e-06,
+      "loss": 130.8569,
+      "step": 35330
+    },
+    {
+      "epoch": 0.2923439632708773,
+      "grad_norm": 676.7745361328125,
+      "learning_rate": 8.987121931555814e-06,
+      "loss": 124.0575,
+      "step": 35340
+    },
+    {
+      "epoch": 0.2924266865202465,
+      "grad_norm": 1002.80029296875,
+      "learning_rate": 8.986270730825489e-06,
+      "loss": 101.8287,
+      "step": 35350
+    },
+    {
+      "epoch": 0.29250940976961576,
+      "grad_norm": 1145.8223876953125,
+      "learning_rate": 8.985419212925898e-06,
+      "loss": 110.8668,
+      "step": 35360
+    },
+    {
+      "epoch": 0.292592133018985,
+      "grad_norm": 992.4735717773438,
+      "learning_rate": 8.98456737792479e-06,
+      "loss": 80.9995,
+      "step": 35370
+    },
+    {
+      "epoch": 0.2926748562683542,
+      "grad_norm": 488.52618408203125,
+      "learning_rate": 8.983715225889942e-06,
+      "loss": 111.5234,
+      "step": 35380
+    },
+    {
+      "epoch": 0.29275757951772347,
+      "grad_norm": 1136.89794921875,
+      "learning_rate": 8.982862756889158e-06,
+      "loss": 162.2252,
+      "step": 35390
+    },
+    {
+      "epoch": 0.29284030276709266,
+      "grad_norm": 1113.3115234375,
+      "learning_rate": 8.982009970990262e-06,
+      "loss": 89.9033,
+      "step": 35400
+    },
+    {
+      "epoch": 0.2929230260164619,
+      "grad_norm": 1392.5189208984375,
+      "learning_rate": 8.98115686826111e-06,
+      "loss": 131.7283,
+      "step": 35410
+    },
+    {
+      "epoch": 0.2930057492658312,
+      "grad_norm": 617.0336303710938,
+      "learning_rate": 8.980303448769574e-06,
+      "loss": 99.9001,
+      "step": 35420
+    },
+    {
+      "epoch": 0.29308847251520037,
+      "grad_norm": 2442.5810546875,
+      "learning_rate": 8.979449712583562e-06,
+      "loss": 112.8064,
+      "step": 35430
+    },
+    {
+      "epoch": 0.2931711957645696,
+      "grad_norm": 544.6372680664062,
+      "learning_rate": 8.978595659770997e-06,
+      "loss": 109.6494,
+      "step": 35440
+    },
+    {
+      "epoch": 0.2932539190139389,
+      "grad_norm": 860.6021118164062,
+      "learning_rate": 8.977741290399836e-06,
+      "loss": 106.3515,
+      "step": 35450
+    },
+    {
+      "epoch": 0.2933366422633081,
+      "grad_norm": 1385.670654296875,
+      "learning_rate": 8.976886604538055e-06,
+      "loss": 117.2203,
+      "step": 35460
+    },
+    {
+      "epoch": 0.29341936551267733,
+      "grad_norm": 653.8483276367188,
+      "learning_rate": 8.976031602253661e-06,
+      "loss": 91.5749,
+      "step": 35470
+    },
+    {
+      "epoch": 0.2935020887620466,
+      "grad_norm": 707.4151000976562,
+      "learning_rate": 8.975176283614677e-06,
+      "loss": 137.304,
+      "step": 35480
+    },
+    {
+      "epoch": 0.2935848120114158,
+      "grad_norm": 724.526123046875,
+      "learning_rate": 8.97432064868916e-06,
+      "loss": 117.3225,
+      "step": 35490
+    },
+    {
+      "epoch": 0.29366753526078504,
+      "grad_norm": 684.9421997070312,
+      "learning_rate": 8.973464697545191e-06,
+      "loss": 118.9565,
+      "step": 35500
+    },
+    {
+      "epoch": 0.2937502585101543,
+      "grad_norm": 1034.4615478515625,
+      "learning_rate": 8.97260843025087e-06,
+      "loss": 117.1274,
+      "step": 35510
+    },
+    {
+      "epoch": 0.2938329817595235,
+      "grad_norm": 808.4039916992188,
+      "learning_rate": 8.971751846874329e-06,
+      "loss": 109.1277,
+      "step": 35520
+    },
+    {
+      "epoch": 0.29391570500889275,
+      "grad_norm": 793.2233276367188,
+      "learning_rate": 8.97089494748372e-06,
+      "loss": 150.5428,
+      "step": 35530
+    },
+    {
+      "epoch": 0.293998428258262,
+      "grad_norm": 1373.2613525390625,
+      "learning_rate": 8.970037732147226e-06,
+      "loss": 144.7835,
+      "step": 35540
+    },
+    {
+      "epoch": 0.2940811515076312,
+      "grad_norm": 876.5359497070312,
+      "learning_rate": 8.969180200933048e-06,
+      "loss": 175.0752,
+      "step": 35550
+    },
+    {
+      "epoch": 0.29416387475700045,
+      "grad_norm": 1215.04443359375,
+      "learning_rate": 8.968322353909417e-06,
+      "loss": 105.4971,
+      "step": 35560
+    },
+    {
+      "epoch": 0.2942465980063697,
+      "grad_norm": 1001.4893188476562,
+      "learning_rate": 8.96746419114459e-06,
+      "loss": 109.2281,
+      "step": 35570
+    },
+    {
+      "epoch": 0.2943293212557389,
+      "grad_norm": 1082.677001953125,
+      "learning_rate": 8.966605712706844e-06,
+      "loss": 90.3954,
+      "step": 35580
+    },
+    {
+      "epoch": 0.29441204450510816,
+      "grad_norm": 962.15380859375,
+      "learning_rate": 8.965746918664486e-06,
+      "loss": 96.3644,
+      "step": 35590
+    },
+    {
+      "epoch": 0.2944947677544774,
+      "grad_norm": 1004.3434448242188,
+      "learning_rate": 8.964887809085846e-06,
+      "loss": 128.4367,
+      "step": 35600
+    },
+    {
+      "epoch": 0.2945774910038466,
+      "grad_norm": 535.0782470703125,
+      "learning_rate": 8.96402838403928e-06,
+      "loss": 101.0775,
+      "step": 35610
+    },
+    {
+      "epoch": 0.29466021425321587,
+      "grad_norm": 451.95941162109375,
+      "learning_rate": 8.96316864359317e-06,
+      "loss": 103.9834,
+      "step": 35620
+    },
+    {
+      "epoch": 0.2947429375025851,
+      "grad_norm": 861.1315307617188,
+      "learning_rate": 8.962308587815916e-06,
+      "loss": 95.3887,
+      "step": 35630
+    },
+    {
+      "epoch": 0.2948256607519543,
+      "grad_norm": 873.6254272460938,
+      "learning_rate": 8.961448216775955e-06,
+      "loss": 106.6663,
+      "step": 35640
+    },
+    {
+      "epoch": 0.2949083840013236,
+      "grad_norm": 647.5992431640625,
+      "learning_rate": 8.960587530541737e-06,
+      "loss": 113.9285,
+      "step": 35650
+    },
+    {
+      "epoch": 0.29499110725069283,
+      "grad_norm": 818.12744140625,
+      "learning_rate": 8.959726529181748e-06,
+      "loss": 93.2586,
+      "step": 35660
+    },
+    {
+      "epoch": 0.29507383050006203,
+      "grad_norm": 2233.892578125,
+      "learning_rate": 8.95886521276449e-06,
+      "loss": 99.4201,
+      "step": 35670
+    },
+    {
+      "epoch": 0.2951565537494313,
+      "grad_norm": 1236.3218994140625,
+      "learning_rate": 8.958003581358498e-06,
+      "loss": 122.5037,
+      "step": 35680
+    },
+    {
+      "epoch": 0.29523927699880054,
+      "grad_norm": 1387.99462890625,
+      "learning_rate": 8.957141635032325e-06,
+      "loss": 103.0061,
+      "step": 35690
+    },
+    {
+      "epoch": 0.29532200024816974,
+      "grad_norm": 363.5679626464844,
+      "learning_rate": 8.956279373854553e-06,
+      "loss": 297.6621,
+      "step": 35700
+    },
+    {
+      "epoch": 0.295404723497539,
+      "grad_norm": 848.06884765625,
+      "learning_rate": 8.955416797893787e-06,
+      "loss": 133.0075,
+      "step": 35710
+    },
+    {
+      "epoch": 0.29548744674690824,
+      "grad_norm": 767.8993530273438,
+      "learning_rate": 8.95455390721866e-06,
+      "loss": 86.9913,
+      "step": 35720
+    },
+    {
+      "epoch": 0.29557016999627744,
+      "grad_norm": 947.8824462890625,
+      "learning_rate": 8.953690701897827e-06,
+      "loss": 126.0984,
+      "step": 35730
+    },
+    {
+      "epoch": 0.2956528932456467,
+      "grad_norm": 866.94970703125,
+      "learning_rate": 8.952827181999973e-06,
+      "loss": 100.0804,
+      "step": 35740
+    },
+    {
+      "epoch": 0.29573561649501595,
+      "grad_norm": 997.9763793945312,
+      "learning_rate": 8.951963347593797e-06,
+      "loss": 158.5358,
+      "step": 35750
+    },
+    {
+      "epoch": 0.29581833974438515,
+      "grad_norm": 682.8516235351562,
+      "learning_rate": 8.951099198748036e-06,
+      "loss": 101.2486,
+      "step": 35760
+    },
+    {
+      "epoch": 0.2959010629937544,
+      "grad_norm": 804.0764770507812,
+      "learning_rate": 8.950234735531445e-06,
+      "loss": 98.5626,
+      "step": 35770
+    },
+    {
+      "epoch": 0.2959837862431236,
+      "grad_norm": 10388.48046875,
+      "learning_rate": 8.949369958012806e-06,
+      "loss": 155.4089,
+      "step": 35780
+    },
+    {
+      "epoch": 0.29606650949249286,
+      "grad_norm": 886.81298828125,
+      "learning_rate": 8.948504866260924e-06,
+      "loss": 113.0202,
+      "step": 35790
+    },
+    {
+      "epoch": 0.2961492327418621,
+      "grad_norm": 897.130859375,
+      "learning_rate": 8.94763946034463e-06,
+      "loss": 142.2092,
+      "step": 35800
+    },
+    {
+      "epoch": 0.2962319559912313,
+      "grad_norm": 966.3784790039062,
+      "learning_rate": 8.946773740332781e-06,
+      "loss": 157.32,
+      "step": 35810
+    },
+    {
+      "epoch": 0.29631467924060056,
+      "grad_norm": 1265.8228759765625,
+      "learning_rate": 8.945907706294262e-06,
+      "loss": 87.6832,
+      "step": 35820
+    },
+    {
+      "epoch": 0.2963974024899698,
+      "grad_norm": 434.8976135253906,
+      "learning_rate": 8.945041358297973e-06,
+      "loss": 115.3741,
+      "step": 35830
+    },
+    {
+      "epoch": 0.296480125739339,
+      "grad_norm": 1037.9794921875,
+      "learning_rate": 8.94417469641285e-06,
+      "loss": 101.735,
+      "step": 35840
+    },
+    {
+      "epoch": 0.29656284898870827,
+      "grad_norm": 1543.5999755859375,
+      "learning_rate": 8.943307720707846e-06,
+      "loss": 149.339,
+      "step": 35850
+    },
+    {
+      "epoch": 0.2966455722380775,
+      "grad_norm": 851.0203247070312,
+      "learning_rate": 8.942440431251947e-06,
+      "loss": 126.8035,
+      "step": 35860
+    },
+    {
+      "epoch": 0.2967282954874467,
+      "grad_norm": 1087.814697265625,
+      "learning_rate": 8.941572828114154e-06,
+      "loss": 154.3589,
+      "step": 35870
+    },
+    {
+      "epoch": 0.296811018736816,
+      "grad_norm": 1085.237548828125,
+      "learning_rate": 8.9407049113635e-06,
+      "loss": 133.6983,
+      "step": 35880
+    },
+    {
+      "epoch": 0.29689374198618523,
+      "grad_norm": 657.667724609375,
+      "learning_rate": 8.939836681069042e-06,
+      "loss": 77.88,
+      "step": 35890
+    },
+    {
+      "epoch": 0.29697646523555443,
+      "grad_norm": 937.2635498046875,
+      "learning_rate": 8.938968137299861e-06,
+      "loss": 121.5767,
+      "step": 35900
+    },
+    {
+      "epoch": 0.2970591884849237,
+      "grad_norm": 692.5191650390625,
+      "learning_rate": 8.938099280125064e-06,
+      "loss": 110.6443,
+      "step": 35910
+    },
+    {
+      "epoch": 0.29714191173429294,
+      "grad_norm": 1191.0335693359375,
+      "learning_rate": 8.937230109613778e-06,
+      "loss": 125.6926,
+      "step": 35920
+    },
+    {
+      "epoch": 0.29722463498366214,
+      "grad_norm": 600.857177734375,
+      "learning_rate": 8.936360625835164e-06,
+      "loss": 114.9589,
+      "step": 35930
+    },
+    {
+      "epoch": 0.2973073582330314,
+      "grad_norm": 786.0075073242188,
+      "learning_rate": 8.935490828858399e-06,
+      "loss": 156.1116,
+      "step": 35940
+    },
+    {
+      "epoch": 0.29739008148240065,
+      "grad_norm": 923.1871948242188,
+      "learning_rate": 8.934620718752691e-06,
+      "loss": 102.9856,
+      "step": 35950
+    },
+    {
+      "epoch": 0.29747280473176985,
+      "grad_norm": 785.8792114257812,
+      "learning_rate": 8.933750295587269e-06,
+      "loss": 121.2862,
+      "step": 35960
+    },
+    {
+      "epoch": 0.2975555279811391,
+      "grad_norm": 1262.131591796875,
+      "learning_rate": 8.932879559431392e-06,
+      "loss": 144.9996,
+      "step": 35970
+    },
+    {
+      "epoch": 0.29763825123050835,
+      "grad_norm": 1094.29296875,
+      "learning_rate": 8.932008510354336e-06,
+      "loss": 99.4907,
+      "step": 35980
+    },
+    {
+      "epoch": 0.29772097447987755,
+      "grad_norm": 1515.66748046875,
+      "learning_rate": 8.931137148425407e-06,
+      "loss": 117.0325,
+      "step": 35990
+    },
+    {
+      "epoch": 0.2978036977292468,
+      "grad_norm": 702.0181274414062,
+      "learning_rate": 8.930265473713939e-06,
+      "loss": 89.8551,
+      "step": 36000
+    },
+    {
+      "epoch": 0.29788642097861606,
+      "grad_norm": 924.1445922851562,
+      "learning_rate": 8.929393486289283e-06,
+      "loss": 91.7574,
+      "step": 36010
+    },
+    {
+      "epoch": 0.29796914422798526,
+      "grad_norm": 886.2630615234375,
+      "learning_rate": 8.928521186220822e-06,
+      "loss": 134.8864,
+      "step": 36020
+    },
+    {
+      "epoch": 0.2980518674773545,
+      "grad_norm": 571.9080200195312,
+      "learning_rate": 8.92764857357796e-06,
+      "loss": 166.6288,
+      "step": 36030
+    },
+    {
+      "epoch": 0.29813459072672377,
+      "grad_norm": 1017.0745239257812,
+      "learning_rate": 8.926775648430124e-06,
+      "loss": 97.3446,
+      "step": 36040
+    },
+    {
+      "epoch": 0.29821731397609297,
+      "grad_norm": 1384.15869140625,
+      "learning_rate": 8.925902410846774e-06,
+      "loss": 102.3454,
+      "step": 36050
+    },
+    {
+      "epoch": 0.2983000372254622,
+      "grad_norm": 627.1011962890625,
+      "learning_rate": 8.925028860897384e-06,
+      "loss": 99.8053,
+      "step": 36060
+    },
+    {
+      "epoch": 0.2983827604748315,
+      "grad_norm": 1164.1707763671875,
+      "learning_rate": 8.924154998651461e-06,
+      "loss": 150.2465,
+      "step": 36070
+    },
+    {
+      "epoch": 0.2984654837242007,
+      "grad_norm": 806.7046508789062,
+      "learning_rate": 8.923280824178538e-06,
+      "loss": 127.2189,
+      "step": 36080
+    },
+    {
+      "epoch": 0.29854820697356993,
+      "grad_norm": 936.9692993164062,
+      "learning_rate": 8.922406337548162e-06,
+      "loss": 106.9401,
+      "step": 36090
+    },
+    {
+      "epoch": 0.2986309302229392,
+      "grad_norm": 861.8311157226562,
+      "learning_rate": 8.921531538829917e-06,
+      "loss": 92.577,
+      "step": 36100
+    },
+    {
+      "epoch": 0.2987136534723084,
+      "grad_norm": 901.859375,
+      "learning_rate": 8.920656428093403e-06,
+      "loss": 83.8378,
+      "step": 36110
+    },
+    {
+      "epoch": 0.29879637672167764,
+      "grad_norm": 1293.734375,
+      "learning_rate": 8.919781005408251e-06,
+      "loss": 114.6592,
+      "step": 36120
+    },
+    {
+      "epoch": 0.29887909997104684,
+      "grad_norm": 697.5694580078125,
+      "learning_rate": 8.918905270844113e-06,
+      "loss": 141.8754,
+      "step": 36130
+    },
+    {
+      "epoch": 0.2989618232204161,
+      "grad_norm": 822.4385986328125,
+      "learning_rate": 8.918029224470671e-06,
+      "loss": 101.8231,
+      "step": 36140
+    },
+    {
+      "epoch": 0.29904454646978534,
+      "grad_norm": 886.5821533203125,
+      "learning_rate": 8.917152866357621e-06,
+      "loss": 97.08,
+      "step": 36150
+    },
+    {
+      "epoch": 0.29912726971915454,
+      "grad_norm": 1727.11279296875,
+      "learning_rate": 8.916276196574698e-06,
+      "loss": 124.1994,
+      "step": 36160
+    },
+    {
+      "epoch": 0.2992099929685238,
+      "grad_norm": 649.4053955078125,
+      "learning_rate": 8.91539921519165e-06,
+      "loss": 113.0979,
+      "step": 36170
+    },
+    {
+      "epoch": 0.29929271621789305,
+      "grad_norm": 762.2130737304688,
+      "learning_rate": 8.914521922278255e-06,
+      "loss": 118.0666,
+      "step": 36180
+    },
+    {
+      "epoch": 0.29937543946726225,
+      "grad_norm": 995.956787109375,
+      "learning_rate": 8.913644317904317e-06,
+      "loss": 125.7407,
+      "step": 36190
+    },
+    {
+      "epoch": 0.2994581627166315,
+      "grad_norm": 1194.2001953125,
+      "learning_rate": 8.912766402139662e-06,
+      "loss": 124.1992,
+      "step": 36200
+    },
+    {
+      "epoch": 0.29954088596600076,
+      "grad_norm": 1277.1484375,
+      "learning_rate": 8.91188817505414e-06,
+      "loss": 146.9492,
+      "step": 36210
+    },
+    {
+      "epoch": 0.29962360921536996,
+      "grad_norm": 992.379638671875,
+      "learning_rate": 8.91100963671763e-06,
+      "loss": 107.3992,
+      "step": 36220
+    },
+    {
+      "epoch": 0.2997063324647392,
+      "grad_norm": 820.5222778320312,
+      "learning_rate": 8.910130787200032e-06,
+      "loss": 93.3464,
+      "step": 36230
+    },
+    {
+      "epoch": 0.29978905571410847,
+      "grad_norm": 2073.3251953125,
+      "learning_rate": 8.909251626571273e-06,
+      "loss": 101.3619,
+      "step": 36240
+    },
+    {
+      "epoch": 0.29987177896347766,
+      "grad_norm": 767.80615234375,
+      "learning_rate": 8.908372154901302e-06,
+      "loss": 89.5982,
+      "step": 36250
+    },
+    {
+      "epoch": 0.2999545022128469,
+      "grad_norm": 1339.1114501953125,
+      "learning_rate": 8.907492372260096e-06,
+      "loss": 118.8273,
+      "step": 36260
+    },
+    {
+      "epoch": 0.3000372254622162,
+      "grad_norm": 1395.037353515625,
+      "learning_rate": 8.906612278717657e-06,
+      "loss": 114.5038,
+      "step": 36270
+    },
+    {
+      "epoch": 0.30011994871158537,
+      "grad_norm": 1279.881591796875,
+      "learning_rate": 8.905731874344005e-06,
+      "loss": 110.8277,
+      "step": 36280
+    },
+    {
+      "epoch": 0.3002026719609546,
+      "grad_norm": 1153.6810302734375,
+      "learning_rate": 8.904851159209193e-06,
+      "loss": 111.3379,
+      "step": 36290
+    },
+    {
+      "epoch": 0.3002853952103239,
+      "grad_norm": 910.88623046875,
+      "learning_rate": 8.903970133383297e-06,
+      "loss": 83.3806,
+      "step": 36300
+    },
+    {
+      "epoch": 0.3003681184596931,
+      "grad_norm": 993.2054443359375,
+      "learning_rate": 8.903088796936414e-06,
+      "loss": 117.134,
+      "step": 36310
+    },
+    {
+      "epoch": 0.30045084170906233,
+      "grad_norm": 1444.7095947265625,
+      "learning_rate": 8.902207149938667e-06,
+      "loss": 118.84,
+      "step": 36320
+    },
+    {
+      "epoch": 0.3005335649584316,
+      "grad_norm": 722.3483276367188,
+      "learning_rate": 8.901325192460206e-06,
+      "loss": 100.5878,
+      "step": 36330
+    },
+    {
+      "epoch": 0.3006162882078008,
+      "grad_norm": 2218.97119140625,
+      "learning_rate": 8.900442924571204e-06,
+      "loss": 125.5503,
+      "step": 36340
+    },
+    {
+      "epoch": 0.30069901145717004,
+      "grad_norm": 933.83984375,
+      "learning_rate": 8.89956034634186e-06,
+      "loss": 100.4451,
+      "step": 36350
+    },
+    {
+      "epoch": 0.3007817347065393,
+      "grad_norm": 1003.083251953125,
+      "learning_rate": 8.898677457842394e-06,
+      "loss": 97.8874,
+      "step": 36360
+    },
+    {
+      "epoch": 0.3008644579559085,
+      "grad_norm": 1543.9967041015625,
+      "learning_rate": 8.897794259143057e-06,
+      "loss": 144.2935,
+      "step": 36370
+    },
+    {
+      "epoch": 0.30094718120527775,
+      "grad_norm": 1334.9381103515625,
+      "learning_rate": 8.896910750314118e-06,
+      "loss": 91.8307,
+      "step": 36380
+    },
+    {
+      "epoch": 0.301029904454647,
+      "grad_norm": 1186.411865234375,
+      "learning_rate": 8.896026931425876e-06,
+      "loss": 131.1232,
+      "step": 36390
+    },
+    {
+      "epoch": 0.3011126277040162,
+      "grad_norm": 1868.9034423828125,
+      "learning_rate": 8.895142802548653e-06,
+      "loss": 124.4849,
+      "step": 36400
+    },
+    {
+      "epoch": 0.30119535095338545,
+      "grad_norm": 840.4161987304688,
+      "learning_rate": 8.89425836375279e-06,
+      "loss": 81.0873,
+      "step": 36410
+    },
+    {
+      "epoch": 0.3012780742027547,
+      "grad_norm": 1163.7984619140625,
+      "learning_rate": 8.893373615108663e-06,
+      "loss": 136.477,
+      "step": 36420
+    },
+    {
+      "epoch": 0.3013607974521239,
+      "grad_norm": 1086.72216796875,
+      "learning_rate": 8.892488556686665e-06,
+      "loss": 126.1113,
+      "step": 36430
+    },
+    {
+      "epoch": 0.30144352070149316,
+      "grad_norm": 879.4309692382812,
+      "learning_rate": 8.891603188557218e-06,
+      "loss": 96.8926,
+      "step": 36440
+    },
+    {
+      "epoch": 0.3015262439508624,
+      "grad_norm": 1197.5047607421875,
+      "learning_rate": 8.890717510790763e-06,
+      "loss": 110.1916,
+      "step": 36450
+    },
+    {
+      "epoch": 0.3016089672002316,
+      "grad_norm": 349.7181701660156,
+      "learning_rate": 8.889831523457773e-06,
+      "loss": 154.8354,
+      "step": 36460
+    },
+    {
+      "epoch": 0.30169169044960087,
+      "grad_norm": 1460.5762939453125,
+      "learning_rate": 8.888945226628742e-06,
+      "loss": 140.0803,
+      "step": 36470
+    },
+    {
+      "epoch": 0.3017744136989701,
+      "grad_norm": 665.8645629882812,
+      "learning_rate": 8.888058620374185e-06,
+      "loss": 92.3529,
+      "step": 36480
+    },
+    {
+      "epoch": 0.3018571369483393,
+      "grad_norm": 879.9156494140625,
+      "learning_rate": 8.887171704764647e-06,
+      "loss": 129.1003,
+      "step": 36490
+    },
+    {
+      "epoch": 0.3019398601977086,
+      "grad_norm": 815.3402709960938,
+      "learning_rate": 8.8862844798707e-06,
+      "loss": 118.2584,
+      "step": 36500
+    },
+    {
+      "epoch": 0.3020225834470778,
+      "grad_norm": 1314.0321044921875,
+      "learning_rate": 8.885396945762928e-06,
+      "loss": 126.3761,
+      "step": 36510
+    },
+    {
+      "epoch": 0.30210530669644703,
+      "grad_norm": 858.8338012695312,
+      "learning_rate": 8.884509102511956e-06,
+      "loss": 98.5212,
+      "step": 36520
+    },
+    {
+      "epoch": 0.3021880299458163,
+      "grad_norm": 1700.220458984375,
+      "learning_rate": 8.883620950188422e-06,
+      "loss": 106.6714,
+      "step": 36530
+    },
+    {
+      "epoch": 0.3022707531951855,
+      "grad_norm": 696.0690307617188,
+      "learning_rate": 8.882732488862988e-06,
+      "loss": 118.2453,
+      "step": 36540
+    },
+    {
+      "epoch": 0.30235347644455474,
+      "grad_norm": 793.1904907226562,
+      "learning_rate": 8.881843718606353e-06,
+      "loss": 125.0236,
+      "step": 36550
+    },
+    {
+      "epoch": 0.302436199693924,
+      "grad_norm": 970.5441284179688,
+      "learning_rate": 8.880954639489227e-06,
+      "loss": 108.3671,
+      "step": 36560
+    },
+    {
+      "epoch": 0.3025189229432932,
+      "grad_norm": 1777.4910888671875,
+      "learning_rate": 8.880065251582354e-06,
+      "loss": 160.0988,
+      "step": 36570
+    },
+    {
+      "epoch": 0.30260164619266244,
+      "grad_norm": 1037.1177978515625,
+      "learning_rate": 8.879175554956495e-06,
+      "loss": 104.0029,
+      "step": 36580
+    },
+    {
+      "epoch": 0.3026843694420317,
+      "grad_norm": 806.9476928710938,
+      "learning_rate": 8.87828554968244e-06,
+      "loss": 88.7504,
+      "step": 36590
+    },
+    {
+      "epoch": 0.3027670926914009,
+      "grad_norm": 1243.7362060546875,
+      "learning_rate": 8.877395235831002e-06,
+      "loss": 111.0507,
+      "step": 36600
+    },
+    {
+      "epoch": 0.30284981594077015,
+      "grad_norm": 1789.92529296875,
+      "learning_rate": 8.876504613473019e-06,
+      "loss": 154.0786,
+      "step": 36610
+    },
+    {
+      "epoch": 0.3029325391901394,
+      "grad_norm": 718.7664184570312,
+      "learning_rate": 8.875613682679356e-06,
+      "loss": 108.9329,
+      "step": 36620
+    },
+    {
+      "epoch": 0.3030152624395086,
+      "grad_norm": 748.9815063476562,
+      "learning_rate": 8.874722443520898e-06,
+      "loss": 123.7988,
+      "step": 36630
+    },
+    {
+      "epoch": 0.30309798568887786,
+      "grad_norm": 606.9013671875,
+      "learning_rate": 8.873830896068559e-06,
+      "loss": 107.3505,
+      "step": 36640
+    },
+    {
+      "epoch": 0.3031807089382471,
+      "grad_norm": 878.5547485351562,
+      "learning_rate": 8.872939040393274e-06,
+      "loss": 113.0779,
+      "step": 36650
+    },
+    {
+      "epoch": 0.3032634321876163,
+      "grad_norm": 1300.362060546875,
+      "learning_rate": 8.872046876566003e-06,
+      "loss": 130.3682,
+      "step": 36660
+    },
+    {
+      "epoch": 0.30334615543698557,
+      "grad_norm": 886.26025390625,
+      "learning_rate": 8.871154404657734e-06,
+      "loss": 106.1408,
+      "step": 36670
+    },
+    {
+      "epoch": 0.3034288786863548,
+      "grad_norm": 1735.70947265625,
+      "learning_rate": 8.870261624739474e-06,
+      "loss": 120.6958,
+      "step": 36680
+    },
+    {
+      "epoch": 0.303511601935724,
+      "grad_norm": 886.688232421875,
+      "learning_rate": 8.869368536882258e-06,
+      "loss": 102.5698,
+      "step": 36690
+    },
+    {
+      "epoch": 0.3035943251850933,
+      "grad_norm": 632.3603515625,
+      "learning_rate": 8.868475141157146e-06,
+      "loss": 88.5606,
+      "step": 36700
+    },
+    {
+      "epoch": 0.3036770484344625,
+      "grad_norm": 764.1319580078125,
+      "learning_rate": 8.867581437635221e-06,
+      "loss": 107.0108,
+      "step": 36710
+    },
+    {
+      "epoch": 0.3037597716838317,
+      "grad_norm": 1273.7620849609375,
+      "learning_rate": 8.866687426387592e-06,
+      "loss": 159.5809,
+      "step": 36720
+    },
+    {
+      "epoch": 0.303842494933201,
+      "grad_norm": 685.6810302734375,
+      "learning_rate": 8.86579310748539e-06,
+      "loss": 99.7657,
+      "step": 36730
+    },
+    {
+      "epoch": 0.30392521818257023,
+      "grad_norm": 777.0698852539062,
+      "learning_rate": 8.86489848099977e-06,
+      "loss": 120.2746,
+      "step": 36740
+    },
+    {
+      "epoch": 0.30400794143193943,
+      "grad_norm": 746.861572265625,
+      "learning_rate": 8.864003547001916e-06,
+      "loss": 106.8211,
+      "step": 36750
+    },
+    {
+      "epoch": 0.3040906646813087,
+      "grad_norm": 853.1625366210938,
+      "learning_rate": 8.863108305563035e-06,
+      "loss": 91.7284,
+      "step": 36760
+    },
+    {
+      "epoch": 0.30417338793067794,
+      "grad_norm": 1231.1922607421875,
+      "learning_rate": 8.862212756754354e-06,
+      "loss": 155.4766,
+      "step": 36770
+    },
+    {
+      "epoch": 0.30425611118004714,
+      "grad_norm": 1215.257568359375,
+      "learning_rate": 8.861316900647129e-06,
+      "loss": 149.8827,
+      "step": 36780
+    },
+    {
+      "epoch": 0.3043388344294164,
+      "grad_norm": 3567.228515625,
+      "learning_rate": 8.860420737312638e-06,
+      "loss": 121.5637,
+      "step": 36790
+    },
+    {
+      "epoch": 0.30442155767878565,
+      "grad_norm": 694.6704711914062,
+      "learning_rate": 8.859524266822188e-06,
+      "loss": 101.825,
+      "step": 36800
+    },
+    {
+      "epoch": 0.30450428092815485,
+      "grad_norm": 1621.055908203125,
+      "learning_rate": 8.858627489247105e-06,
+      "loss": 128.4847,
+      "step": 36810
+    },
+    {
+      "epoch": 0.3045870041775241,
+      "grad_norm": 974.5038452148438,
+      "learning_rate": 8.85773040465874e-06,
+      "loss": 157.9853,
+      "step": 36820
+    },
+    {
+      "epoch": 0.30466972742689336,
+      "grad_norm": 1332.6234130859375,
+      "learning_rate": 8.856833013128472e-06,
+      "loss": 107.964,
+      "step": 36830
+    },
+    {
+      "epoch": 0.30475245067626255,
+      "grad_norm": 1167.7923583984375,
+      "learning_rate": 8.855935314727702e-06,
+      "loss": 109.8553,
+      "step": 36840
+    },
+    {
+      "epoch": 0.3048351739256318,
+      "grad_norm": 704.8397216796875,
+      "learning_rate": 8.855037309527854e-06,
+      "loss": 115.8193,
+      "step": 36850
+    },
+    {
+      "epoch": 0.304917897175001,
+      "grad_norm": 883.3186645507812,
+      "learning_rate": 8.854138997600382e-06,
+      "loss": 91.2245,
+      "step": 36860
+    },
+    {
+      "epoch": 0.30500062042437026,
+      "grad_norm": 660.802001953125,
+      "learning_rate": 8.853240379016757e-06,
+      "loss": 88.3629,
+      "step": 36870
+    },
+    {
+      "epoch": 0.3050833436737395,
+      "grad_norm": 1194.5704345703125,
+      "learning_rate": 8.852341453848477e-06,
+      "loss": 113.7338,
+      "step": 36880
+    },
+    {
+      "epoch": 0.3051660669231087,
+      "grad_norm": 1196.6671142578125,
+      "learning_rate": 8.851442222167068e-06,
+      "loss": 96.392,
+      "step": 36890
+    },
+    {
+      "epoch": 0.30524879017247797,
+      "grad_norm": 1096.315673828125,
+      "learning_rate": 8.850542684044078e-06,
+      "loss": 132.9981,
+      "step": 36900
+    },
+    {
+      "epoch": 0.3053315134218472,
+      "grad_norm": 843.752685546875,
+      "learning_rate": 8.849642839551079e-06,
+      "loss": 120.9463,
+      "step": 36910
+    },
+    {
+      "epoch": 0.3054142366712164,
+      "grad_norm": 1700.408203125,
+      "learning_rate": 8.848742688759666e-06,
+      "loss": 106.4173,
+      "step": 36920
+    },
+    {
+      "epoch": 0.3054969599205857,
+      "grad_norm": 1714.5902099609375,
+      "learning_rate": 8.847842231741462e-06,
+      "loss": 119.7494,
+      "step": 36930
+    },
+    {
+      "epoch": 0.30557968316995493,
+      "grad_norm": 998.3058471679688,
+      "learning_rate": 8.846941468568108e-06,
+      "loss": 104.204,
+      "step": 36940
+    },
+    {
+      "epoch": 0.30566240641932413,
+      "grad_norm": 1132.1826171875,
+      "learning_rate": 8.846040399311278e-06,
+      "loss": 112.0986,
+      "step": 36950
+    },
+    {
+      "epoch": 0.3057451296686934,
+      "grad_norm": 661.6875,
+      "learning_rate": 8.845139024042664e-06,
+      "loss": 81.744,
+      "step": 36960
+    },
+    {
+      "epoch": 0.30582785291806264,
+      "grad_norm": 932.6510009765625,
+      "learning_rate": 8.844237342833985e-06,
+      "loss": 127.1373,
+      "step": 36970
+    },
+    {
+      "epoch": 0.30591057616743184,
+      "grad_norm": 780.3368530273438,
+      "learning_rate": 8.843335355756983e-06,
+      "loss": 75.523,
+      "step": 36980
+    },
+    {
+      "epoch": 0.3059932994168011,
+      "grad_norm": 754.1832275390625,
+      "learning_rate": 8.842433062883427e-06,
+      "loss": 89.4337,
+      "step": 36990
+    },
+    {
+      "epoch": 0.30607602266617034,
+      "grad_norm": 764.4425659179688,
+      "learning_rate": 8.841530464285105e-06,
+      "loss": 129.0274,
+      "step": 37000
+    },
+    {
+      "epoch": 0.30615874591553954,
+      "grad_norm": 861.7986450195312,
+      "learning_rate": 8.840627560033833e-06,
+      "loss": 82.3673,
+      "step": 37010
+    },
+    {
+      "epoch": 0.3062414691649088,
+      "grad_norm": 988.38916015625,
+      "learning_rate": 8.839724350201452e-06,
+      "loss": 114.8146,
+      "step": 37020
+    },
+    {
+      "epoch": 0.30632419241427805,
+      "grad_norm": 530.607421875,
+      "learning_rate": 8.838820834859829e-06,
+      "loss": 103.93,
+      "step": 37030
+    },
+    {
+      "epoch": 0.30640691566364725,
+      "grad_norm": 941.2070922851562,
+      "learning_rate": 8.837917014080849e-06,
+      "loss": 89.9074,
+      "step": 37040
+    },
+    {
+      "epoch": 0.3064896389130165,
+      "grad_norm": 688.3717651367188,
+      "learning_rate": 8.837012887936426e-06,
+      "loss": 111.2304,
+      "step": 37050
+    },
+    {
+      "epoch": 0.30657236216238576,
+      "grad_norm": 1092.7115478515625,
+      "learning_rate": 8.836108456498497e-06,
+      "loss": 93.2177,
+      "step": 37060
+    },
+    {
+      "epoch": 0.30665508541175496,
+      "grad_norm": 970.1156616210938,
+      "learning_rate": 8.835203719839024e-06,
+      "loss": 113.6382,
+      "step": 37070
+    },
+    {
+      "epoch": 0.3067378086611242,
+      "grad_norm": 1119.5540771484375,
+      "learning_rate": 8.834298678029988e-06,
+      "loss": 120.2044,
+      "step": 37080
+    },
+    {
+      "epoch": 0.30682053191049347,
+      "grad_norm": 840.0438842773438,
+      "learning_rate": 8.833393331143409e-06,
+      "loss": 103.9367,
+      "step": 37090
+    },
+    {
+      "epoch": 0.30690325515986266,
+      "grad_norm": 1232.6182861328125,
+      "learning_rate": 8.832487679251311e-06,
+      "loss": 107.9942,
+      "step": 37100
+    },
+    {
+      "epoch": 0.3069859784092319,
+      "grad_norm": 702.6771850585938,
+      "learning_rate": 8.831581722425761e-06,
+      "loss": 107.5534,
+      "step": 37110
+    },
+    {
+      "epoch": 0.3070687016586012,
+      "grad_norm": 933.55615234375,
+      "learning_rate": 8.830675460738835e-06,
+      "loss": 92.3703,
+      "step": 37120
+    },
+    {
+      "epoch": 0.30715142490797037,
+      "grad_norm": 1156.6966552734375,
+      "learning_rate": 8.829768894262644e-06,
+      "loss": 118.0975,
+      "step": 37130
+    },
+    {
+      "epoch": 0.3072341481573396,
+      "grad_norm": 766.1238403320312,
+      "learning_rate": 8.82886202306932e-06,
+      "loss": 91.8914,
+      "step": 37140
+    },
+    {
+      "epoch": 0.3073168714067089,
+      "grad_norm": 1076.5308837890625,
+      "learning_rate": 8.827954847231016e-06,
+      "loss": 115.0902,
+      "step": 37150
+    },
+    {
+      "epoch": 0.3073995946560781,
+      "grad_norm": 867.0706176757812,
+      "learning_rate": 8.82704736681991e-06,
+      "loss": 128.4827,
+      "step": 37160
+    },
+    {
+      "epoch": 0.30748231790544733,
+      "grad_norm": 963.2921752929688,
+      "learning_rate": 8.826139581908211e-06,
+      "loss": 112.9323,
+      "step": 37170
+    },
+    {
+      "epoch": 0.3075650411548166,
+      "grad_norm": 735.8433227539062,
+      "learning_rate": 8.825231492568146e-06,
+      "loss": 114.1932,
+      "step": 37180
+    },
+    {
+      "epoch": 0.3076477644041858,
+      "grad_norm": 1258.63818359375,
+      "learning_rate": 8.824323098871966e-06,
+      "loss": 136.7632,
+      "step": 37190
+    },
+    {
+      "epoch": 0.30773048765355504,
+      "grad_norm": 1213.376220703125,
+      "learning_rate": 8.823414400891948e-06,
+      "loss": 140.9363,
+      "step": 37200
+    },
+    {
+      "epoch": 0.30781321090292424,
+      "grad_norm": 816.0035400390625,
+      "learning_rate": 8.822505398700395e-06,
+      "loss": 97.0494,
+      "step": 37210
+    },
+    {
+      "epoch": 0.3078959341522935,
+      "grad_norm": 908.9609375,
+      "learning_rate": 8.821596092369627e-06,
+      "loss": 112.1852,
+      "step": 37220
+    },
+    {
+      "epoch": 0.30797865740166275,
+      "grad_norm": 1214.3897705078125,
+      "learning_rate": 8.820686481971998e-06,
+      "loss": 133.1782,
+      "step": 37230
+    },
+    {
+      "epoch": 0.30806138065103195,
+      "grad_norm": 1155.935546875,
+      "learning_rate": 8.81977656757988e-06,
+      "loss": 91.5354,
+      "step": 37240
+    },
+    {
+      "epoch": 0.3081441039004012,
+      "grad_norm": 1258.32275390625,
+      "learning_rate": 8.81886634926567e-06,
+      "loss": 108.436,
+      "step": 37250
+    },
+    {
+      "epoch": 0.30822682714977045,
+      "grad_norm": 1130.4833984375,
+      "learning_rate": 8.817955827101794e-06,
+      "loss": 133.1508,
+      "step": 37260
+    },
+    {
+      "epoch": 0.30830955039913965,
+      "grad_norm": 738.8419799804688,
+      "learning_rate": 8.817045001160693e-06,
+      "loss": 122.5803,
+      "step": 37270
+    },
+    {
+      "epoch": 0.3083922736485089,
+      "grad_norm": 913.4530639648438,
+      "learning_rate": 8.816133871514838e-06,
+      "loss": 108.7282,
+      "step": 37280
+    },
+    {
+      "epoch": 0.30847499689787816,
+      "grad_norm": 600.1754150390625,
+      "learning_rate": 8.815222438236726e-06,
+      "loss": 91.7117,
+      "step": 37290
+    },
+    {
+      "epoch": 0.30855772014724736,
+      "grad_norm": 987.5612182617188,
+      "learning_rate": 8.814310701398873e-06,
+      "loss": 111.6003,
+      "step": 37300
+    },
+    {
+      "epoch": 0.3086404433966166,
+      "grad_norm": 718.8853759765625,
+      "learning_rate": 8.813398661073823e-06,
+      "loss": 120.6641,
+      "step": 37310
+    },
+    {
+      "epoch": 0.30872316664598587,
+      "grad_norm": 969.3789672851562,
+      "learning_rate": 8.812486317334145e-06,
+      "loss": 112.9521,
+      "step": 37320
+    },
+    {
+      "epoch": 0.30880588989535507,
+      "grad_norm": 1069.70849609375,
+      "learning_rate": 8.811573670252426e-06,
+      "loss": 102.9678,
+      "step": 37330
+    },
+    {
+      "epoch": 0.3088886131447243,
+      "grad_norm": 1160.0357666015625,
+      "learning_rate": 8.810660719901283e-06,
+      "loss": 142.8662,
+      "step": 37340
+    },
+    {
+      "epoch": 0.3089713363940936,
+      "grad_norm": 923.1897583007812,
+      "learning_rate": 8.809747466353356e-06,
+      "loss": 134.5727,
+      "step": 37350
+    },
+    {
+      "epoch": 0.3090540596434628,
+      "grad_norm": 1227.94580078125,
+      "learning_rate": 8.808833909681305e-06,
+      "loss": 144.0661,
+      "step": 37360
+    },
+    {
+      "epoch": 0.30913678289283203,
+      "grad_norm": 636.2891235351562,
+      "learning_rate": 8.80792004995782e-06,
+      "loss": 120.0885,
+      "step": 37370
+    },
+    {
+      "epoch": 0.3092195061422013,
+      "grad_norm": 1097.7366943359375,
+      "learning_rate": 8.807005887255615e-06,
+      "loss": 138.4748,
+      "step": 37380
+    },
+    {
+      "epoch": 0.3093022293915705,
+      "grad_norm": 814.5217895507812,
+      "learning_rate": 8.806091421647423e-06,
+      "loss": 113.6995,
+      "step": 37390
+    },
+    {
+      "epoch": 0.30938495264093974,
+      "grad_norm": 737.350341796875,
+      "learning_rate": 8.805176653206004e-06,
+      "loss": 116.3498,
+      "step": 37400
+    },
+    {
+      "epoch": 0.309467675890309,
+      "grad_norm": 2945.936279296875,
+      "learning_rate": 8.80426158200414e-06,
+      "loss": 109.3581,
+      "step": 37410
+    },
+    {
+      "epoch": 0.3095503991396782,
+      "grad_norm": 1349.6761474609375,
+      "learning_rate": 8.803346208114643e-06,
+      "loss": 117.0218,
+      "step": 37420
+    },
+    {
+      "epoch": 0.30963312238904744,
+      "grad_norm": 744.024658203125,
+      "learning_rate": 8.802430531610344e-06,
+      "loss": 101.151,
+      "step": 37430
+    },
+    {
+      "epoch": 0.3097158456384167,
+      "grad_norm": 1006.9130859375,
+      "learning_rate": 8.801514552564097e-06,
+      "loss": 87.0184,
+      "step": 37440
+    },
+    {
+      "epoch": 0.3097985688877859,
+      "grad_norm": 1106.60693359375,
+      "learning_rate": 8.800598271048784e-06,
+      "loss": 159.6884,
+      "step": 37450
+    },
+    {
+      "epoch": 0.30988129213715515,
+      "grad_norm": 1054.00244140625,
+      "learning_rate": 8.799681687137309e-06,
+      "loss": 97.845,
+      "step": 37460
+    },
+    {
+      "epoch": 0.3099640153865244,
+      "grad_norm": 551.2701416015625,
+      "learning_rate": 8.7987648009026e-06,
+      "loss": 87.5821,
+      "step": 37470
+    },
+    {
+      "epoch": 0.3100467386358936,
+      "grad_norm": 2352.765869140625,
+      "learning_rate": 8.79784761241761e-06,
+      "loss": 144.6846,
+      "step": 37480
+    },
+    {
+      "epoch": 0.31012946188526286,
+      "grad_norm": 899.6213989257812,
+      "learning_rate": 8.796930121755315e-06,
+      "loss": 110.2097,
+      "step": 37490
+    },
+    {
+      "epoch": 0.3102121851346321,
+      "grad_norm": 527.784912109375,
+      "learning_rate": 8.796012328988716e-06,
+      "loss": 98.5166,
+      "step": 37500
+    },
+    {
+      "epoch": 0.3102949083840013,
+      "grad_norm": 1046.642822265625,
+      "learning_rate": 8.795094234190837e-06,
+      "loss": 94.5682,
+      "step": 37510
+    },
+    {
+      "epoch": 0.31037763163337057,
+      "grad_norm": 1084.1600341796875,
+      "learning_rate": 8.794175837434729e-06,
+      "loss": 141.1946,
+      "step": 37520
+    },
+    {
+      "epoch": 0.3104603548827398,
+      "grad_norm": 761.6773071289062,
+      "learning_rate": 8.79325713879346e-06,
+      "loss": 87.328,
+      "step": 37530
+    },
+    {
+      "epoch": 0.310543078132109,
+      "grad_norm": 1374.055908203125,
+      "learning_rate": 8.792338138340131e-06,
+      "loss": 102.905,
+      "step": 37540
+    },
+    {
+      "epoch": 0.3106258013814783,
+      "grad_norm": 935.9447631835938,
+      "learning_rate": 8.791418836147858e-06,
+      "loss": 146.4921,
+      "step": 37550
+    },
+    {
+      "epoch": 0.3107085246308475,
+      "grad_norm": 1152.4400634765625,
+      "learning_rate": 8.790499232289793e-06,
+      "loss": 128.4351,
+      "step": 37560
+    },
+    {
+      "epoch": 0.3107912478802167,
+      "grad_norm": 658.3630981445312,
+      "learning_rate": 8.789579326839097e-06,
+      "loss": 121.7294,
+      "step": 37570
+    },
+    {
+      "epoch": 0.310873971129586,
+      "grad_norm": 830.0960083007812,
+      "learning_rate": 8.788659119868966e-06,
+      "loss": 133.9257,
+      "step": 37580
+    },
+    {
+      "epoch": 0.3109566943789552,
+      "grad_norm": 866.0687255859375,
+      "learning_rate": 8.787738611452616e-06,
+      "loss": 116.4662,
+      "step": 37590
+    },
+    {
+      "epoch": 0.31103941762832443,
+      "grad_norm": 1266.32177734375,
+      "learning_rate": 8.78681780166329e-06,
+      "loss": 120.4589,
+      "step": 37600
+    },
+    {
+      "epoch": 0.3111221408776937,
+      "grad_norm": 876.0301513671875,
+      "learning_rate": 8.785896690574248e-06,
+      "loss": 103.7038,
+      "step": 37610
+    },
+    {
+      "epoch": 0.3112048641270629,
+      "grad_norm": 822.327880859375,
+      "learning_rate": 8.784975278258783e-06,
+      "loss": 146.5088,
+      "step": 37620
+    },
+    {
+      "epoch": 0.31128758737643214,
+      "grad_norm": 976.1931762695312,
+      "learning_rate": 8.784053564790205e-06,
+      "loss": 110.9248,
+      "step": 37630
+    },
+    {
+      "epoch": 0.3113703106258014,
+      "grad_norm": 1754.4827880859375,
+      "learning_rate": 8.783131550241853e-06,
+      "loss": 132.1888,
+      "step": 37640
+    },
+    {
+      "epoch": 0.3114530338751706,
+      "grad_norm": 889.1416015625,
+      "learning_rate": 8.782209234687083e-06,
+      "loss": 98.5607,
+      "step": 37650
+    },
+    {
+      "epoch": 0.31153575712453985,
+      "grad_norm": 578.06689453125,
+      "learning_rate": 8.781286618199285e-06,
+      "loss": 93.0681,
+      "step": 37660
+    },
+    {
+      "epoch": 0.3116184803739091,
+      "grad_norm": 625.0341796875,
+      "learning_rate": 8.780363700851863e-06,
+      "loss": 84.4234,
+      "step": 37670
+    },
+    {
+      "epoch": 0.3117012036232783,
+      "grad_norm": 1180.2471923828125,
+      "learning_rate": 8.779440482718251e-06,
+      "loss": 126.0896,
+      "step": 37680
+    },
+    {
+      "epoch": 0.31178392687264755,
+      "grad_norm": 1036.988525390625,
+      "learning_rate": 8.778516963871904e-06,
+      "loss": 109.445,
+      "step": 37690
+    },
+    {
+      "epoch": 0.3118666501220168,
+      "grad_norm": 771.7061157226562,
+      "learning_rate": 8.777593144386305e-06,
+      "loss": 106.4233,
+      "step": 37700
+    },
+    {
+      "epoch": 0.311949373371386,
+      "grad_norm": 853.5861206054688,
+      "learning_rate": 8.776669024334955e-06,
+      "loss": 149.2146,
+      "step": 37710
+    },
+    {
+      "epoch": 0.31203209662075526,
+      "grad_norm": 1038.280029296875,
+      "learning_rate": 8.775744603791385e-06,
+      "loss": 87.6942,
+      "step": 37720
+    },
+    {
+      "epoch": 0.3121148198701245,
+      "grad_norm": 785.39892578125,
+      "learning_rate": 8.774819882829144e-06,
+      "loss": 101.3138,
+      "step": 37730
+    },
+    {
+      "epoch": 0.3121975431194937,
+      "grad_norm": 810.301513671875,
+      "learning_rate": 8.77389486152181e-06,
+      "loss": 104.7605,
+      "step": 37740
+    },
+    {
+      "epoch": 0.31228026636886297,
+      "grad_norm": 907.918212890625,
+      "learning_rate": 8.772969539942981e-06,
+      "loss": 93.2778,
+      "step": 37750
+    },
+    {
+      "epoch": 0.3123629896182322,
+      "grad_norm": 1630.78955078125,
+      "learning_rate": 8.772043918166282e-06,
+      "loss": 118.6197,
+      "step": 37760
+    },
+    {
+      "epoch": 0.3124457128676014,
+      "grad_norm": 792.9736328125,
+      "learning_rate": 8.771117996265358e-06,
+      "loss": 110.336,
+      "step": 37770
+    },
+    {
+      "epoch": 0.3125284361169707,
+      "grad_norm": 515.8519287109375,
+      "learning_rate": 8.770191774313883e-06,
+      "loss": 98.6496,
+      "step": 37780
+    },
+    {
+      "epoch": 0.31261115936633993,
+      "grad_norm": 889.776611328125,
+      "learning_rate": 8.769265252385552e-06,
+      "loss": 165.6621,
+      "step": 37790
+    },
+    {
+      "epoch": 0.31269388261570913,
+      "grad_norm": 780.7835693359375,
+      "learning_rate": 8.768338430554083e-06,
+      "loss": 91.5842,
+      "step": 37800
+    },
+    {
+      "epoch": 0.3127766058650784,
+      "grad_norm": 1138.0804443359375,
+      "learning_rate": 8.76741130889322e-06,
+      "loss": 100.7134,
+      "step": 37810
+    },
+    {
+      "epoch": 0.31285932911444764,
+      "grad_norm": 942.5914916992188,
+      "learning_rate": 8.766483887476727e-06,
+      "loss": 105.6115,
+      "step": 37820
+    },
+    {
+      "epoch": 0.31294205236381684,
+      "grad_norm": 618.5468139648438,
+      "learning_rate": 8.7655561663784e-06,
+      "loss": 75.4478,
+      "step": 37830
+    },
+    {
+      "epoch": 0.3130247756131861,
+      "grad_norm": 1693.069580078125,
+      "learning_rate": 8.764628145672048e-06,
+      "loss": 127.7167,
+      "step": 37840
+    },
+    {
+      "epoch": 0.31310749886255534,
+      "grad_norm": 1026.7965087890625,
+      "learning_rate": 8.763699825431513e-06,
+      "loss": 138.911,
+      "step": 37850
+    },
+    {
+      "epoch": 0.31319022211192454,
+      "grad_norm": 1648.09765625,
+      "learning_rate": 8.762771205730656e-06,
+      "loss": 117.1393,
+      "step": 37860
+    },
+    {
+      "epoch": 0.3132729453612938,
+      "grad_norm": 933.2096557617188,
+      "learning_rate": 8.761842286643362e-06,
+      "loss": 96.2922,
+      "step": 37870
+    },
+    {
+      "epoch": 0.31335566861066305,
+      "grad_norm": 802.694580078125,
+      "learning_rate": 8.760913068243542e-06,
+      "loss": 144.4842,
+      "step": 37880
+    },
+    {
+      "epoch": 0.31343839186003225,
+      "grad_norm": 1243.0850830078125,
+      "learning_rate": 8.759983550605132e-06,
+      "loss": 128.3055,
+      "step": 37890
+    },
+    {
+      "epoch": 0.3135211151094015,
+      "grad_norm": 1257.254638671875,
+      "learning_rate": 8.759053733802083e-06,
+      "loss": 100.2831,
+      "step": 37900
+    },
+    {
+      "epoch": 0.31360383835877076,
+      "grad_norm": 934.783935546875,
+      "learning_rate": 8.758123617908383e-06,
+      "loss": 100.1143,
+      "step": 37910
+    },
+    {
+      "epoch": 0.31368656160813996,
+      "grad_norm": 989.8101196289062,
+      "learning_rate": 8.757193202998033e-06,
+      "loss": 127.9963,
+      "step": 37920
+    },
+    {
+      "epoch": 0.3137692848575092,
+      "grad_norm": 1011.3324584960938,
+      "learning_rate": 8.756262489145061e-06,
+      "loss": 112.5696,
+      "step": 37930
+    },
+    {
+      "epoch": 0.3138520081068784,
+      "grad_norm": 925.4760131835938,
+      "learning_rate": 8.755331476423526e-06,
+      "loss": 79.9976,
+      "step": 37940
+    },
+    {
+      "epoch": 0.31393473135624766,
+      "grad_norm": 932.2153930664062,
+      "learning_rate": 8.754400164907496e-06,
+      "loss": 118.1142,
+      "step": 37950
+    },
+    {
+      "epoch": 0.3140174546056169,
+      "grad_norm": 869.6834106445312,
+      "learning_rate": 8.753468554671078e-06,
+      "loss": 122.9429,
+      "step": 37960
+    },
+    {
+      "epoch": 0.3141001778549861,
+      "grad_norm": 1097.6431884765625,
+      "learning_rate": 8.752536645788391e-06,
+      "loss": 116.1235,
+      "step": 37970
+    },
+    {
+      "epoch": 0.31418290110435537,
+      "grad_norm": 943.837158203125,
+      "learning_rate": 8.751604438333587e-06,
+      "loss": 120.1827,
+      "step": 37980
+    },
+    {
+      "epoch": 0.3142656243537246,
+      "grad_norm": 965.2254028320312,
+      "learning_rate": 8.750671932380834e-06,
+      "loss": 111.8385,
+      "step": 37990
+    },
+    {
+      "epoch": 0.3143483476030938,
+      "grad_norm": 2054.935546875,
+      "learning_rate": 8.749739128004329e-06,
+      "loss": 105.5353,
+      "step": 38000
+    },
+    {
+      "epoch": 0.3144310708524631,
+      "grad_norm": 885.7263793945312,
+      "learning_rate": 8.748806025278292e-06,
+      "loss": 113.4429,
+      "step": 38010
+    },
+    {
+      "epoch": 0.31451379410183233,
+      "grad_norm": 1407.1341552734375,
+      "learning_rate": 8.747872624276963e-06,
+      "loss": 99.107,
+      "step": 38020
+    },
+    {
+      "epoch": 0.31459651735120153,
+      "grad_norm": 1092.1849365234375,
+      "learning_rate": 8.746938925074609e-06,
+      "loss": 130.1728,
+      "step": 38030
+    },
+    {
+      "epoch": 0.3146792406005708,
+      "grad_norm": 977.42578125,
+      "learning_rate": 8.746004927745522e-06,
+      "loss": 116.9955,
+      "step": 38040
+    },
+    {
+      "epoch": 0.31476196384994004,
+      "grad_norm": 813.2716064453125,
+      "learning_rate": 8.745070632364014e-06,
+      "loss": 103.2874,
+      "step": 38050
+    },
+    {
+      "epoch": 0.31484468709930924,
+      "grad_norm": 823.4459838867188,
+      "learning_rate": 8.744136039004422e-06,
+      "loss": 122.7185,
+      "step": 38060
+    },
+    {
+      "epoch": 0.3149274103486785,
+      "grad_norm": 675.5857543945312,
+      "learning_rate": 8.743201147741112e-06,
+      "loss": 117.528,
+      "step": 38070
+    },
+    {
+      "epoch": 0.31501013359804775,
+      "grad_norm": 1719.105224609375,
+      "learning_rate": 8.742265958648464e-06,
+      "loss": 110.0581,
+      "step": 38080
+    },
+    {
+      "epoch": 0.31509285684741695,
+      "grad_norm": 1066.7659912109375,
+      "learning_rate": 8.741330471800888e-06,
+      "loss": 89.2473,
+      "step": 38090
+    },
+    {
+      "epoch": 0.3151755800967862,
+      "grad_norm": 776.9163818359375,
+      "learning_rate": 8.740394687272817e-06,
+      "loss": 124.1178,
+      "step": 38100
+    },
+    {
+      "epoch": 0.31525830334615546,
+      "grad_norm": 760.9815063476562,
+      "learning_rate": 8.739458605138706e-06,
+      "loss": 119.1256,
+      "step": 38110
+    },
+    {
+      "epoch": 0.31534102659552465,
+      "grad_norm": 974.4657592773438,
+      "learning_rate": 8.738522225473036e-06,
+      "loss": 105.6252,
+      "step": 38120
+    },
+    {
+      "epoch": 0.3154237498448939,
+      "grad_norm": 634.0014038085938,
+      "learning_rate": 8.737585548350312e-06,
+      "loss": 119.6853,
+      "step": 38130
+    },
+    {
+      "epoch": 0.31550647309426316,
+      "grad_norm": 765.5140380859375,
+      "learning_rate": 8.736648573845057e-06,
+      "loss": 99.3297,
+      "step": 38140
+    },
+    {
+      "epoch": 0.31558919634363236,
+      "grad_norm": 1316.4825439453125,
+      "learning_rate": 8.735711302031824e-06,
+      "loss": 123.088,
+      "step": 38150
+    },
+    {
+      "epoch": 0.3156719195930016,
+      "grad_norm": 1433.5438232421875,
+      "learning_rate": 8.734773732985186e-06,
+      "loss": 116.7357,
+      "step": 38160
+    },
+    {
+      "epoch": 0.31575464284237087,
+      "grad_norm": 826.6362915039062,
+      "learning_rate": 8.733835866779745e-06,
+      "loss": 93.5203,
+      "step": 38170
+    },
+    {
+      "epoch": 0.31583736609174007,
+      "grad_norm": 1368.983642578125,
+      "learning_rate": 8.73289770349012e-06,
+      "loss": 90.895,
+      "step": 38180
+    },
+    {
+      "epoch": 0.3159200893411093,
+      "grad_norm": 764.3739013671875,
+      "learning_rate": 8.731959243190955e-06,
+      "loss": 115.1358,
+      "step": 38190
+    },
+    {
+      "epoch": 0.3160028125904786,
+      "grad_norm": 1166.0986328125,
+      "learning_rate": 8.73102048595692e-06,
+      "loss": 110.7284,
+      "step": 38200
+    },
+    {
+      "epoch": 0.3160855358398478,
+      "grad_norm": 900.78759765625,
+      "learning_rate": 8.730081431862709e-06,
+      "loss": 114.7286,
+      "step": 38210
+    },
+    {
+      "epoch": 0.31616825908921703,
+      "grad_norm": 694.7711791992188,
+      "learning_rate": 8.729142080983037e-06,
+      "loss": 99.7621,
+      "step": 38220
+    },
+    {
+      "epoch": 0.3162509823385863,
+      "grad_norm": 1478.10546875,
+      "learning_rate": 8.728202433392645e-06,
+      "loss": 103.1368,
+      "step": 38230
+    },
+    {
+      "epoch": 0.3163337055879555,
+      "grad_norm": 739.2174682617188,
+      "learning_rate": 8.727262489166295e-06,
+      "loss": 91.7107,
+      "step": 38240
+    },
+    {
+      "epoch": 0.31641642883732474,
+      "grad_norm": 1041.38525390625,
+      "learning_rate": 8.726322248378775e-06,
+      "loss": 133.2948,
+      "step": 38250
+    },
+    {
+      "epoch": 0.316499152086694,
+      "grad_norm": 709.4502563476562,
+      "learning_rate": 8.725381711104894e-06,
+      "loss": 134.3007,
+      "step": 38260
+    },
+    {
+      "epoch": 0.3165818753360632,
+      "grad_norm": 1081.0482177734375,
+      "learning_rate": 8.724440877419487e-06,
+      "loss": 107.7189,
+      "step": 38270
+    },
+    {
+      "epoch": 0.31666459858543244,
+      "grad_norm": 1248.0484619140625,
+      "learning_rate": 8.723499747397415e-06,
+      "loss": 105.8039,
+      "step": 38280
+    },
+    {
+      "epoch": 0.3167473218348017,
+      "grad_norm": 1431.8619384765625,
+      "learning_rate": 8.722558321113555e-06,
+      "loss": 108.0174,
+      "step": 38290
+    },
+    {
+      "epoch": 0.3168300450841709,
+      "grad_norm": 972.2454223632812,
+      "learning_rate": 8.721616598642812e-06,
+      "loss": 124.3465,
+      "step": 38300
+    },
+    {
+      "epoch": 0.31691276833354015,
+      "grad_norm": 1106.1824951171875,
+      "learning_rate": 8.720674580060117e-06,
+      "loss": 92.7966,
+      "step": 38310
+    },
+    {
+      "epoch": 0.31699549158290935,
+      "grad_norm": 1259.550048828125,
+      "learning_rate": 8.719732265440423e-06,
+      "loss": 91.808,
+      "step": 38320
+    },
+    {
+      "epoch": 0.3170782148322786,
+      "grad_norm": 1003.4591064453125,
+      "learning_rate": 8.718789654858702e-06,
+      "loss": 97.9086,
+      "step": 38330
+    },
+    {
+      "epoch": 0.31716093808164786,
+      "grad_norm": 1087.510498046875,
+      "learning_rate": 8.717846748389956e-06,
+      "loss": 190.7288,
+      "step": 38340
+    },
+    {
+      "epoch": 0.31724366133101706,
+      "grad_norm": 1307.7303466796875,
+      "learning_rate": 8.716903546109208e-06,
+      "loss": 100.8898,
+      "step": 38350
+    },
+    {
+      "epoch": 0.3173263845803863,
+      "grad_norm": 389.17926025390625,
+      "learning_rate": 8.715960048091502e-06,
+      "loss": 105.3628,
+      "step": 38360
+    },
+    {
+      "epoch": 0.31740910782975557,
+      "grad_norm": 634.9244995117188,
+      "learning_rate": 8.715016254411908e-06,
+      "loss": 93.6207,
+      "step": 38370
+    },
+    {
+      "epoch": 0.31749183107912476,
+      "grad_norm": 1184.2578125,
+      "learning_rate": 8.714072165145521e-06,
+      "loss": 135.4022,
+      "step": 38380
+    },
+    {
+      "epoch": 0.317574554328494,
+      "grad_norm": 619.5433349609375,
+      "learning_rate": 8.713127780367458e-06,
+      "loss": 82.4095,
+      "step": 38390
+    },
+    {
+      "epoch": 0.3176572775778633,
+      "grad_norm": 1421.2760009765625,
+      "learning_rate": 8.712183100152858e-06,
+      "loss": 104.6683,
+      "step": 38400
+    },
+    {
+      "epoch": 0.31774000082723247,
+      "grad_norm": 569.2383422851562,
+      "learning_rate": 8.711238124576884e-06,
+      "loss": 100.9128,
+      "step": 38410
+    },
+    {
+      "epoch": 0.3178227240766017,
+      "grad_norm": 486.7202453613281,
+      "learning_rate": 8.710292853714726e-06,
+      "loss": 102.4042,
+      "step": 38420
+    },
+    {
+      "epoch": 0.317905447325971,
+      "grad_norm": 1279.0054931640625,
+      "learning_rate": 8.709347287641593e-06,
+      "loss": 107.4948,
+      "step": 38430
+    },
+    {
+      "epoch": 0.3179881705753402,
+      "grad_norm": 777.2363891601562,
+      "learning_rate": 8.70840142643272e-06,
+      "loss": 126.2411,
+      "step": 38440
+    },
+    {
+      "epoch": 0.31807089382470943,
+      "grad_norm": 680.6844482421875,
+      "learning_rate": 8.707455270163365e-06,
+      "loss": 98.6142,
+      "step": 38450
+    },
+    {
+      "epoch": 0.3181536170740787,
+      "grad_norm": 1288.6944580078125,
+      "learning_rate": 8.70650881890881e-06,
+      "loss": 125.0557,
+      "step": 38460
+    },
+    {
+      "epoch": 0.3182363403234479,
+      "grad_norm": 760.6031494140625,
+      "learning_rate": 8.705562072744358e-06,
+      "loss": 138.251,
+      "step": 38470
+    },
+    {
+      "epoch": 0.31831906357281714,
+      "grad_norm": 883.0706176757812,
+      "learning_rate": 8.704615031745337e-06,
+      "loss": 111.4153,
+      "step": 38480
+    },
+    {
+      "epoch": 0.3184017868221864,
+      "grad_norm": 1131.5177001953125,
+      "learning_rate": 8.703667695987102e-06,
+      "loss": 113.8998,
+      "step": 38490
+    },
+    {
+      "epoch": 0.3184845100715556,
+      "grad_norm": 973.4132690429688,
+      "learning_rate": 8.702720065545024e-06,
+      "loss": 106.239,
+      "step": 38500
+    },
+    {
+      "epoch": 0.31856723332092485,
+      "grad_norm": 1401.8857421875,
+      "learning_rate": 8.701772140494504e-06,
+      "loss": 131.2176,
+      "step": 38510
+    },
+    {
+      "epoch": 0.3186499565702941,
+      "grad_norm": 735.4816284179688,
+      "learning_rate": 8.700823920910964e-06,
+      "loss": 124.5568,
+      "step": 38520
+    },
+    {
+      "epoch": 0.3187326798196633,
+      "grad_norm": 1129.6258544921875,
+      "learning_rate": 8.699875406869848e-06,
+      "loss": 103.7197,
+      "step": 38530
+    },
+    {
+      "epoch": 0.31881540306903255,
+      "grad_norm": 802.9367065429688,
+      "learning_rate": 8.69892659844663e-06,
+      "loss": 129.3652,
+      "step": 38540
+    },
+    {
+      "epoch": 0.3188981263184018,
+      "grad_norm": 1019.4291381835938,
+      "learning_rate": 8.697977495716794e-06,
+      "loss": 113.1963,
+      "step": 38550
+    },
+    {
+      "epoch": 0.318980849567771,
+      "grad_norm": 1141.768798828125,
+      "learning_rate": 8.697028098755863e-06,
+      "loss": 75.0446,
+      "step": 38560
+    },
+    {
+      "epoch": 0.31906357281714026,
+      "grad_norm": 561.0872192382812,
+      "learning_rate": 8.69607840763937e-06,
+      "loss": 107.2292,
+      "step": 38570
+    },
+    {
+      "epoch": 0.3191462960665095,
+      "grad_norm": 567.6842651367188,
+      "learning_rate": 8.695128422442882e-06,
+      "loss": 105.8062,
+      "step": 38580
+    },
+    {
+      "epoch": 0.3192290193158787,
+      "grad_norm": 1812.7880859375,
+      "learning_rate": 8.694178143241984e-06,
+      "loss": 116.8599,
+      "step": 38590
+    },
+    {
+      "epoch": 0.31931174256524797,
+      "grad_norm": 821.3114013671875,
+      "learning_rate": 8.693227570112285e-06,
+      "loss": 113.9192,
+      "step": 38600
+    },
+    {
+      "epoch": 0.3193944658146172,
+      "grad_norm": 1383.909423828125,
+      "learning_rate": 8.692276703129421e-06,
+      "loss": 123.0948,
+      "step": 38610
+    },
+    {
+      "epoch": 0.3194771890639864,
+      "grad_norm": 559.3286743164062,
+      "learning_rate": 8.691325542369041e-06,
+      "loss": 81.4486,
+      "step": 38620
+    },
+    {
+      "epoch": 0.3195599123133557,
+      "grad_norm": 1106.412109375,
+      "learning_rate": 8.69037408790683e-06,
+      "loss": 119.6945,
+      "step": 38630
+    },
+    {
+      "epoch": 0.31964263556272493,
+      "grad_norm": 1364.677490234375,
+      "learning_rate": 8.689422339818489e-06,
+      "loss": 140.9282,
+      "step": 38640
+    },
+    {
+      "epoch": 0.31972535881209413,
+      "grad_norm": 0.0,
+      "learning_rate": 8.688470298179746e-06,
+      "loss": 140.5661,
+      "step": 38650
+    },
+    {
+      "epoch": 0.3198080820614634,
+      "grad_norm": 859.6204833984375,
+      "learning_rate": 8.687517963066347e-06,
+      "loss": 110.4718,
+      "step": 38660
+    },
+    {
+      "epoch": 0.3198908053108326,
+      "grad_norm": 1258.269775390625,
+      "learning_rate": 8.686565334554069e-06,
+      "loss": 126.0004,
+      "step": 38670
+    },
+    {
+      "epoch": 0.31997352856020184,
+      "grad_norm": 1039.6004638671875,
+      "learning_rate": 8.685612412718704e-06,
+      "loss": 119.8658,
+      "step": 38680
+    },
+    {
+      "epoch": 0.3200562518095711,
+      "grad_norm": 923.8244018554688,
+      "learning_rate": 8.684659197636076e-06,
+      "loss": 124.2017,
+      "step": 38690
+    },
+    {
+      "epoch": 0.3201389750589403,
+      "grad_norm": 1117.4451904296875,
+      "learning_rate": 8.683705689382025e-06,
+      "loss": 107.7295,
+      "step": 38700
+    },
+    {
+      "epoch": 0.32022169830830954,
+      "grad_norm": 1363.1929931640625,
+      "learning_rate": 8.682751888032419e-06,
+      "loss": 99.3945,
+      "step": 38710
+    },
+    {
+      "epoch": 0.3203044215576788,
+      "grad_norm": 701.418212890625,
+      "learning_rate": 8.681797793663147e-06,
+      "loss": 120.6914,
+      "step": 38720
+    },
+    {
+      "epoch": 0.320387144807048,
+      "grad_norm": 1962.092041015625,
+      "learning_rate": 8.680843406350122e-06,
+      "loss": 105.0907,
+      "step": 38730
+    },
+    {
+      "epoch": 0.32046986805641725,
+      "grad_norm": 1574.8798828125,
+      "learning_rate": 8.679888726169277e-06,
+      "loss": 123.2075,
+      "step": 38740
+    },
+    {
+      "epoch": 0.3205525913057865,
+      "grad_norm": 979.2647094726562,
+      "learning_rate": 8.678933753196577e-06,
+      "loss": 117.9523,
+      "step": 38750
+    },
+    {
+      "epoch": 0.3206353145551557,
+      "grad_norm": 961.4496459960938,
+      "learning_rate": 8.677978487508002e-06,
+      "loss": 130.495,
+      "step": 38760
+    },
+    {
+      "epoch": 0.32071803780452496,
+      "grad_norm": 1239.4903564453125,
+      "learning_rate": 8.677022929179558e-06,
+      "loss": 116.15,
+      "step": 38770
+    },
+    {
+      "epoch": 0.3208007610538942,
+      "grad_norm": 936.8592529296875,
+      "learning_rate": 8.676067078287276e-06,
+      "loss": 102.5058,
+      "step": 38780
+    },
+    {
+      "epoch": 0.3208834843032634,
+      "grad_norm": 1102.019775390625,
+      "learning_rate": 8.675110934907206e-06,
+      "loss": 105.1739,
+      "step": 38790
+    },
+    {
+      "epoch": 0.32096620755263267,
+      "grad_norm": 1137.979736328125,
+      "learning_rate": 8.674154499115426e-06,
+      "loss": 103.8995,
+      "step": 38800
+    },
+    {
+      "epoch": 0.3210489308020019,
+      "grad_norm": 983.4441528320312,
+      "learning_rate": 8.673197770988034e-06,
+      "loss": 100.5983,
+      "step": 38810
+    },
+    {
+      "epoch": 0.3211316540513711,
+      "grad_norm": 601.077392578125,
+      "learning_rate": 8.672240750601152e-06,
+      "loss": 100.5274,
+      "step": 38820
+    },
+    {
+      "epoch": 0.3212143773007404,
+      "grad_norm": 897.4487915039062,
+      "learning_rate": 8.67128343803093e-06,
+      "loss": 99.9841,
+      "step": 38830
+    },
+    {
+      "epoch": 0.3212971005501096,
+      "grad_norm": 1025.739013671875,
+      "learning_rate": 8.670325833353532e-06,
+      "loss": 89.5816,
+      "step": 38840
+    },
+    {
+      "epoch": 0.3213798237994788,
+      "grad_norm": 677.2406005859375,
+      "learning_rate": 8.669367936645152e-06,
+      "loss": 105.6764,
+      "step": 38850
+    },
+    {
+      "epoch": 0.3214625470488481,
+      "grad_norm": 912.070068359375,
+      "learning_rate": 8.668409747982005e-06,
+      "loss": 129.2276,
+      "step": 38860
+    },
+    {
+      "epoch": 0.32154527029821733,
+      "grad_norm": 909.7315673828125,
+      "learning_rate": 8.667451267440332e-06,
+      "loss": 98.6507,
+      "step": 38870
+    },
+    {
+      "epoch": 0.32162799354758653,
+      "grad_norm": 934.35400390625,
+      "learning_rate": 8.666492495096391e-06,
+      "loss": 121.8479,
+      "step": 38880
+    },
+    {
+      "epoch": 0.3217107167969558,
+      "grad_norm": 1550.7830810546875,
+      "learning_rate": 8.66553343102647e-06,
+      "loss": 124.7869,
+      "step": 38890
+    },
+    {
+      "epoch": 0.32179344004632504,
+      "grad_norm": 1913.998291015625,
+      "learning_rate": 8.664574075306876e-06,
+      "loss": 109.4713,
+      "step": 38900
+    },
+    {
+      "epoch": 0.32187616329569424,
+      "grad_norm": 665.229736328125,
+      "learning_rate": 8.66361442801394e-06,
+      "loss": 94.5389,
+      "step": 38910
+    },
+    {
+      "epoch": 0.3219588865450635,
+      "grad_norm": 762.2039794921875,
+      "learning_rate": 8.662654489224018e-06,
+      "loss": 94.5307,
+      "step": 38920
+    },
+    {
+      "epoch": 0.32204160979443275,
+      "grad_norm": 1010.405029296875,
+      "learning_rate": 8.661694259013489e-06,
+      "loss": 143.722,
+      "step": 38930
+    },
+    {
+      "epoch": 0.32212433304380195,
+      "grad_norm": 808.5379638671875,
+      "learning_rate": 8.660733737458751e-06,
+      "loss": 134.8724,
+      "step": 38940
+    },
+    {
+      "epoch": 0.3222070562931712,
+      "grad_norm": 964.7113037109375,
+      "learning_rate": 8.659772924636232e-06,
+      "loss": 122.8288,
+      "step": 38950
+    },
+    {
+      "epoch": 0.32228977954254046,
+      "grad_norm": 908.0009765625,
+      "learning_rate": 8.658811820622376e-06,
+      "loss": 90.802,
+      "step": 38960
+    },
+    {
+      "epoch": 0.32237250279190965,
+      "grad_norm": 1564.2470703125,
+      "learning_rate": 8.657850425493656e-06,
+      "loss": 129.7668,
+      "step": 38970
+    },
+    {
+      "epoch": 0.3224552260412789,
+      "grad_norm": 1255.0084228515625,
+      "learning_rate": 8.656888739326564e-06,
+      "loss": 96.6529,
+      "step": 38980
+    },
+    {
+      "epoch": 0.32253794929064816,
+      "grad_norm": 960.5479125976562,
+      "learning_rate": 8.65592676219762e-06,
+      "loss": 103.2335,
+      "step": 38990
+    },
+    {
+      "epoch": 0.32262067254001736,
+      "grad_norm": 815.9246826171875,
+      "learning_rate": 8.65496449418336e-06,
+      "loss": 94.21,
+      "step": 39000
+    },
+    {
+      "epoch": 0.3227033957893866,
+      "grad_norm": 942.2803344726562,
+      "learning_rate": 8.654001935360349e-06,
+      "loss": 108.4447,
+      "step": 39010
+    },
+    {
+      "epoch": 0.32278611903875587,
+      "grad_norm": 1212.7076416015625,
+      "learning_rate": 8.653039085805174e-06,
+      "loss": 97.4576,
+      "step": 39020
+    },
+    {
+      "epoch": 0.32286884228812507,
+      "grad_norm": 1259.3065185546875,
+      "learning_rate": 8.652075945594444e-06,
+      "loss": 96.3901,
+      "step": 39030
+    },
+    {
+      "epoch": 0.3229515655374943,
+      "grad_norm": 673.8052368164062,
+      "learning_rate": 8.651112514804793e-06,
+      "loss": 94.6694,
+      "step": 39040
+    },
+    {
+      "epoch": 0.3230342887868635,
+      "grad_norm": 1086.0980224609375,
+      "learning_rate": 8.650148793512874e-06,
+      "loss": 161.4135,
+      "step": 39050
+    },
+    {
+      "epoch": 0.3231170120362328,
+      "grad_norm": 1190.9241943359375,
+      "learning_rate": 8.649184781795367e-06,
+      "loss": 122.2091,
+      "step": 39060
+    },
+    {
+      "epoch": 0.32319973528560203,
+      "grad_norm": 750.9871215820312,
+      "learning_rate": 8.648220479728976e-06,
+      "loss": 129.1647,
+      "step": 39070
+    },
+    {
+      "epoch": 0.32328245853497123,
+      "grad_norm": 675.1224365234375,
+      "learning_rate": 8.647255887390425e-06,
+      "loss": 87.7561,
+      "step": 39080
+    },
+    {
+      "epoch": 0.3233651817843405,
+      "grad_norm": 960.1796875,
+      "learning_rate": 8.64629100485646e-06,
+      "loss": 107.4958,
+      "step": 39090
+    },
+    {
+      "epoch": 0.32344790503370974,
+      "grad_norm": 1386.735595703125,
+      "learning_rate": 8.645325832203855e-06,
+      "loss": 135.6421,
+      "step": 39100
+    },
+    {
+      "epoch": 0.32353062828307894,
+      "grad_norm": 589.93896484375,
+      "learning_rate": 8.644360369509403e-06,
+      "loss": 102.7022,
+      "step": 39110
+    },
+    {
+      "epoch": 0.3236133515324482,
+      "grad_norm": 1394.318603515625,
+      "learning_rate": 8.64339461684992e-06,
+      "loss": 109.0257,
+      "step": 39120
+    },
+    {
+      "epoch": 0.32369607478181744,
+      "grad_norm": 1033.558349609375,
+      "learning_rate": 8.64242857430225e-06,
+      "loss": 120.0752,
+      "step": 39130
+    },
+    {
+      "epoch": 0.32377879803118664,
+      "grad_norm": 877.220947265625,
+      "learning_rate": 8.641462241943255e-06,
+      "loss": 149.2554,
+      "step": 39140
+    },
+    {
+      "epoch": 0.3238615212805559,
+      "grad_norm": 1799.2706298828125,
+      "learning_rate": 8.640495619849821e-06,
+      "loss": 106.8699,
+      "step": 39150
+    },
+    {
+      "epoch": 0.32394424452992515,
+      "grad_norm": 354.8163146972656,
+      "learning_rate": 8.639528708098858e-06,
+      "loss": 138.8774,
+      "step": 39160
+    },
+    {
+      "epoch": 0.32402696777929435,
+      "grad_norm": 734.3604125976562,
+      "learning_rate": 8.6385615067673e-06,
+      "loss": 113.2349,
+      "step": 39170
+    },
+    {
+      "epoch": 0.3241096910286636,
+      "grad_norm": 1038.3038330078125,
+      "learning_rate": 8.6375940159321e-06,
+      "loss": 80.891,
+      "step": 39180
+    },
+    {
+      "epoch": 0.32419241427803286,
+      "grad_norm": 1265.29638671875,
+      "learning_rate": 8.63662623567024e-06,
+      "loss": 100.5664,
+      "step": 39190
+    },
+    {
+      "epoch": 0.32427513752740206,
+      "grad_norm": 975.9793701171875,
+      "learning_rate": 8.63565816605872e-06,
+      "loss": 98.3975,
+      "step": 39200
+    },
+    {
+      "epoch": 0.3243578607767713,
+      "grad_norm": 1260.0736083984375,
+      "learning_rate": 8.634689807174564e-06,
+      "loss": 122.2016,
+      "step": 39210
+    },
+    {
+      "epoch": 0.32444058402614057,
+      "grad_norm": 745.4339599609375,
+      "learning_rate": 8.633721159094823e-06,
+      "loss": 118.936,
+      "step": 39220
+    },
+    {
+      "epoch": 0.32452330727550976,
+      "grad_norm": 803.7348022460938,
+      "learning_rate": 8.632752221896562e-06,
+      "loss": 76.6836,
+      "step": 39230
+    },
+    {
+      "epoch": 0.324606030524879,
+      "grad_norm": 901.3794555664062,
+      "learning_rate": 8.631782995656884e-06,
+      "loss": 114.3698,
+      "step": 39240
+    },
+    {
+      "epoch": 0.3246887537742483,
+      "grad_norm": 2110.507080078125,
+      "learning_rate": 8.630813480452898e-06,
+      "loss": 109.549,
+      "step": 39250
+    },
+    {
+      "epoch": 0.32477147702361747,
+      "grad_norm": 868.3207397460938,
+      "learning_rate": 8.629843676361747e-06,
+      "loss": 147.7418,
+      "step": 39260
+    },
+    {
+      "epoch": 0.3248542002729867,
+      "grad_norm": 776.3412475585938,
+      "learning_rate": 8.628873583460593e-06,
+      "loss": 126.7695,
+      "step": 39270
+    },
+    {
+      "epoch": 0.324936923522356,
+      "grad_norm": 1359.6387939453125,
+      "learning_rate": 8.627903201826622e-06,
+      "loss": 120.8187,
+      "step": 39280
+    },
+    {
+      "epoch": 0.3250196467717252,
+      "grad_norm": 1282.950439453125,
+      "learning_rate": 8.626932531537042e-06,
+      "loss": 123.0786,
+      "step": 39290
+    },
+    {
+      "epoch": 0.32510237002109443,
+      "grad_norm": 530.9898071289062,
+      "learning_rate": 8.625961572669087e-06,
+      "loss": 120.1885,
+      "step": 39300
+    },
+    {
+      "epoch": 0.3251850932704637,
+      "grad_norm": 997.0206298828125,
+      "learning_rate": 8.62499032530001e-06,
+      "loss": 120.9154,
+      "step": 39310
+    },
+    {
+      "epoch": 0.3252678165198329,
+      "grad_norm": 676.1962890625,
+      "learning_rate": 8.624018789507091e-06,
+      "loss": 97.3104,
+      "step": 39320
+    },
+    {
+      "epoch": 0.32535053976920214,
+      "grad_norm": 692.87255859375,
+      "learning_rate": 8.62304696536763e-06,
+      "loss": 118.6817,
+      "step": 39330
+    },
+    {
+      "epoch": 0.3254332630185714,
+      "grad_norm": 980.8681030273438,
+      "learning_rate": 8.622074852958946e-06,
+      "loss": 112.1015,
+      "step": 39340
+    },
+    {
+      "epoch": 0.3255159862679406,
+      "grad_norm": 1021.6939697265625,
+      "learning_rate": 8.621102452358393e-06,
+      "loss": 190.7402,
+      "step": 39350
+    },
+    {
+      "epoch": 0.32559870951730985,
+      "grad_norm": 1223.068115234375,
+      "learning_rate": 8.620129763643333e-06,
+      "loss": 128.2917,
+      "step": 39360
+    },
+    {
+      "epoch": 0.3256814327666791,
+      "grad_norm": 1219.437744140625,
+      "learning_rate": 8.619156786891162e-06,
+      "loss": 136.7339,
+      "step": 39370
+    },
+    {
+      "epoch": 0.3257641560160483,
+      "grad_norm": 1412.034912109375,
+      "learning_rate": 8.618183522179295e-06,
+      "loss": 131.6702,
+      "step": 39380
+    },
+    {
+      "epoch": 0.32584687926541755,
+      "grad_norm": 1203.2330322265625,
+      "learning_rate": 8.617209969585171e-06,
+      "loss": 88.7958,
+      "step": 39390
+    },
+    {
+      "epoch": 0.32592960251478675,
+      "grad_norm": 809.6597290039062,
+      "learning_rate": 8.616236129186252e-06,
+      "loss": 102.6644,
+      "step": 39400
+    },
+    {
+      "epoch": 0.326012325764156,
+      "grad_norm": 3100.604736328125,
+      "learning_rate": 8.615262001060019e-06,
+      "loss": 176.8819,
+      "step": 39410
+    },
+    {
+      "epoch": 0.32609504901352526,
+      "grad_norm": 1474.5286865234375,
+      "learning_rate": 8.61428758528398e-06,
+      "loss": 123.6517,
+      "step": 39420
+    },
+    {
+      "epoch": 0.32617777226289446,
+      "grad_norm": 893.0943603515625,
+      "learning_rate": 8.613312881935667e-06,
+      "loss": 118.6461,
+      "step": 39430
+    },
+    {
+      "epoch": 0.3262604955122637,
+      "grad_norm": 1108.7327880859375,
+      "learning_rate": 8.61233789109263e-06,
+      "loss": 129.676,
+      "step": 39440
+    },
+    {
+      "epoch": 0.32634321876163297,
+      "grad_norm": 799.1575317382812,
+      "learning_rate": 8.611362612832445e-06,
+      "loss": 109.5865,
+      "step": 39450
+    },
+    {
+      "epoch": 0.32642594201100217,
+      "grad_norm": 1402.5484619140625,
+      "learning_rate": 8.610387047232711e-06,
+      "loss": 103.1031,
+      "step": 39460
+    },
+    {
+      "epoch": 0.3265086652603714,
+      "grad_norm": 1590.6834716796875,
+      "learning_rate": 8.609411194371049e-06,
+      "loss": 114.6393,
+      "step": 39470
+    },
+    {
+      "epoch": 0.3265913885097407,
+      "grad_norm": 1389.5260009765625,
+      "learning_rate": 8.608435054325103e-06,
+      "loss": 100.0405,
+      "step": 39480
+    },
+    {
+      "epoch": 0.3266741117591099,
+      "grad_norm": 1510.9293212890625,
+      "learning_rate": 8.60745862717254e-06,
+      "loss": 78.9952,
+      "step": 39490
+    },
+    {
+      "epoch": 0.32675683500847913,
+      "grad_norm": 2515.405029296875,
+      "learning_rate": 8.606481912991052e-06,
+      "loss": 125.8343,
+      "step": 39500
+    },
+    {
+      "epoch": 0.3268395582578484,
+      "grad_norm": 1044.6246337890625,
+      "learning_rate": 8.605504911858347e-06,
+      "loss": 95.9947,
+      "step": 39510
+    },
+    {
+      "epoch": 0.3269222815072176,
+      "grad_norm": 742.9393920898438,
+      "learning_rate": 8.604527623852165e-06,
+      "loss": 129.3403,
+      "step": 39520
+    },
+    {
+      "epoch": 0.32700500475658684,
+      "grad_norm": 843.3123779296875,
+      "learning_rate": 8.603550049050262e-06,
+      "loss": 124.6452,
+      "step": 39530
+    },
+    {
+      "epoch": 0.3270877280059561,
+      "grad_norm": 629.9082641601562,
+      "learning_rate": 8.602572187530421e-06,
+      "loss": 103.7542,
+      "step": 39540
+    },
+    {
+      "epoch": 0.3271704512553253,
+      "grad_norm": 1079.95556640625,
+      "learning_rate": 8.601594039370441e-06,
+      "loss": 117.5058,
+      "step": 39550
+    },
+    {
+      "epoch": 0.32725317450469454,
+      "grad_norm": 785.5013427734375,
+      "learning_rate": 8.600615604648155e-06,
+      "loss": 67.8067,
+      "step": 39560
+    },
+    {
+      "epoch": 0.3273358977540638,
+      "grad_norm": 831.0355224609375,
+      "learning_rate": 8.599636883441408e-06,
+      "loss": 123.9131,
+      "step": 39570
+    },
+    {
+      "epoch": 0.327418621003433,
+      "grad_norm": 2647.583251953125,
+      "learning_rate": 8.598657875828078e-06,
+      "loss": 107.162,
+      "step": 39580
+    },
+    {
+      "epoch": 0.32750134425280225,
+      "grad_norm": 2026.9219970703125,
+      "learning_rate": 8.597678581886055e-06,
+      "loss": 130.6936,
+      "step": 39590
+    },
+    {
+      "epoch": 0.3275840675021715,
+      "grad_norm": 991.1618041992188,
+      "learning_rate": 8.596699001693257e-06,
+      "loss": 107.1374,
+      "step": 39600
+    },
+    {
+      "epoch": 0.3276667907515407,
+      "grad_norm": 1521.5604248046875,
+      "learning_rate": 8.595719135327627e-06,
+      "loss": 98.5977,
+      "step": 39610
+    },
+    {
+      "epoch": 0.32774951400090996,
+      "grad_norm": 613.502685546875,
+      "learning_rate": 8.594738982867126e-06,
+      "loss": 100.9653,
+      "step": 39620
+    },
+    {
+      "epoch": 0.3278322372502792,
+      "grad_norm": 1424.407470703125,
+      "learning_rate": 8.593758544389743e-06,
+      "loss": 119.1369,
+      "step": 39630
+    },
+    {
+      "epoch": 0.3279149604996484,
+      "grad_norm": 519.1179809570312,
+      "learning_rate": 8.592777819973486e-06,
+      "loss": 121.2218,
+      "step": 39640
+    },
+    {
+      "epoch": 0.32799768374901767,
+      "grad_norm": 1449.661865234375,
+      "learning_rate": 8.591796809696386e-06,
+      "loss": 114.1455,
+      "step": 39650
+    },
+    {
+      "epoch": 0.3280804069983869,
+      "grad_norm": 986.8948364257812,
+      "learning_rate": 8.590815513636498e-06,
+      "loss": 111.6402,
+      "step": 39660
+    },
+    {
+      "epoch": 0.3281631302477561,
+      "grad_norm": 839.0859375,
+      "learning_rate": 8.5898339318719e-06,
+      "loss": 85.4794,
+      "step": 39670
+    },
+    {
+      "epoch": 0.3282458534971254,
+      "grad_norm": 1216.8238525390625,
+      "learning_rate": 8.58885206448069e-06,
+      "loss": 126.5229,
+      "step": 39680
+    },
+    {
+      "epoch": 0.3283285767464946,
+      "grad_norm": 1210.4658203125,
+      "learning_rate": 8.587869911540993e-06,
+      "loss": 131.2425,
+      "step": 39690
+    },
+    {
+      "epoch": 0.3284112999958638,
+      "grad_norm": 638.7323608398438,
+      "learning_rate": 8.586887473130951e-06,
+      "loss": 117.0074,
+      "step": 39700
+    },
+    {
+      "epoch": 0.3284940232452331,
+      "grad_norm": 1674.9326171875,
+      "learning_rate": 8.585904749328736e-06,
+      "loss": 101.3178,
+      "step": 39710
+    },
+    {
+      "epoch": 0.32857674649460233,
+      "grad_norm": 730.9718627929688,
+      "learning_rate": 8.584921740212537e-06,
+      "loss": 79.7682,
+      "step": 39720
+    },
+    {
+      "epoch": 0.32865946974397153,
+      "grad_norm": 849.2908325195312,
+      "learning_rate": 8.583938445860569e-06,
+      "loss": 134.0528,
+      "step": 39730
+    },
+    {
+      "epoch": 0.3287421929933408,
+      "grad_norm": 1976.713134765625,
+      "learning_rate": 8.582954866351065e-06,
+      "loss": 109.2086,
+      "step": 39740
+    },
+    {
+      "epoch": 0.32882491624271004,
+      "grad_norm": 1140.69140625,
+      "learning_rate": 8.581971001762287e-06,
+      "loss": 115.7576,
+      "step": 39750
+    },
+    {
+      "epoch": 0.32890763949207924,
+      "grad_norm": 1253.4771728515625,
+      "learning_rate": 8.580986852172514e-06,
+      "loss": 99.3701,
+      "step": 39760
+    },
+    {
+      "epoch": 0.3289903627414485,
+      "grad_norm": 1577.8370361328125,
+      "learning_rate": 8.580002417660054e-06,
+      "loss": 137.5488,
+      "step": 39770
+    },
+    {
+      "epoch": 0.3290730859908177,
+      "grad_norm": 759.9320068359375,
+      "learning_rate": 8.579017698303228e-06,
+      "loss": 110.6118,
+      "step": 39780
+    },
+    {
+      "epoch": 0.32915580924018695,
+      "grad_norm": 807.0444946289062,
+      "learning_rate": 8.578032694180394e-06,
+      "loss": 108.5404,
+      "step": 39790
+    },
+    {
+      "epoch": 0.3292385324895562,
+      "grad_norm": 901.5609741210938,
+      "learning_rate": 8.577047405369916e-06,
+      "loss": 92.3528,
+      "step": 39800
+    },
+    {
+      "epoch": 0.3293212557389254,
+      "grad_norm": 815.9768676757812,
+      "learning_rate": 8.576061831950193e-06,
+      "loss": 116.8808,
+      "step": 39810
+    },
+    {
+      "epoch": 0.32940397898829465,
+      "grad_norm": 676.6227416992188,
+      "learning_rate": 8.575075973999642e-06,
+      "loss": 104.0332,
+      "step": 39820
+    },
+    {
+      "epoch": 0.3294867022376639,
+      "grad_norm": 655.098876953125,
+      "learning_rate": 8.574089831596703e-06,
+      "loss": 114.4098,
+      "step": 39830
+    },
+    {
+      "epoch": 0.3295694254870331,
+      "grad_norm": 1169.68359375,
+      "learning_rate": 8.57310340481984e-06,
+      "loss": 88.758,
+      "step": 39840
+    },
+    {
+      "epoch": 0.32965214873640236,
+      "grad_norm": 635.2750244140625,
+      "learning_rate": 8.572116693747537e-06,
+      "loss": 98.1875,
+      "step": 39850
+    },
+    {
+      "epoch": 0.3297348719857716,
+      "grad_norm": 797.3588256835938,
+      "learning_rate": 8.571129698458302e-06,
+      "loss": 101.033,
+      "step": 39860
+    },
+    {
+      "epoch": 0.3298175952351408,
+      "grad_norm": 1276.1683349609375,
+      "learning_rate": 8.570142419030668e-06,
+      "loss": 111.7359,
+      "step": 39870
+    },
+    {
+      "epoch": 0.32990031848451007,
+      "grad_norm": 700.5169677734375,
+      "learning_rate": 8.569154855543184e-06,
+      "loss": 101.829,
+      "step": 39880
+    },
+    {
+      "epoch": 0.3299830417338793,
+      "grad_norm": 1665.83984375,
+      "learning_rate": 8.56816700807443e-06,
+      "loss": 120.2223,
+      "step": 39890
+    },
+    {
+      "epoch": 0.3300657649832485,
+      "grad_norm": 619.46435546875,
+      "learning_rate": 8.567178876703002e-06,
+      "loss": 101.4117,
+      "step": 39900
+    },
+    {
+      "epoch": 0.3301484882326178,
+      "grad_norm": 1413.1790771484375,
+      "learning_rate": 8.566190461507521e-06,
+      "loss": 108.3938,
+      "step": 39910
+    },
+    {
+      "epoch": 0.33023121148198703,
+      "grad_norm": 917.2999877929688,
+      "learning_rate": 8.565201762566632e-06,
+      "loss": 80.6623,
+      "step": 39920
+    },
+    {
+      "epoch": 0.33031393473135623,
+      "grad_norm": 787.8756713867188,
+      "learning_rate": 8.564212779959003e-06,
+      "loss": 130.3724,
+      "step": 39930
+    },
+    {
+      "epoch": 0.3303966579807255,
+      "grad_norm": 649.1900634765625,
+      "learning_rate": 8.563223513763319e-06,
+      "loss": 107.5673,
+      "step": 39940
+    },
+    {
+      "epoch": 0.33047938123009474,
+      "grad_norm": 656.0354614257812,
+      "learning_rate": 8.562233964058294e-06,
+      "loss": 138.9998,
+      "step": 39950
+    },
+    {
+      "epoch": 0.33056210447946394,
+      "grad_norm": 1630.0479736328125,
+      "learning_rate": 8.561244130922658e-06,
+      "loss": 79.6873,
+      "step": 39960
+    },
+    {
+      "epoch": 0.3306448277288332,
+      "grad_norm": 908.9981689453125,
+      "learning_rate": 8.560254014435172e-06,
+      "loss": 124.0382,
+      "step": 39970
+    },
+    {
+      "epoch": 0.33072755097820244,
+      "grad_norm": 1071.929931640625,
+      "learning_rate": 8.559263614674615e-06,
+      "loss": 102.3747,
+      "step": 39980
+    },
+    {
+      "epoch": 0.33081027422757164,
+      "grad_norm": 827.8428344726562,
+      "learning_rate": 8.558272931719785e-06,
+      "loss": 100.2324,
+      "step": 39990
+    },
+    {
+      "epoch": 0.3308929974769409,
+      "grad_norm": 2593.23779296875,
+      "learning_rate": 8.557281965649508e-06,
+      "loss": 107.4415,
+      "step": 40000
+    },
+    {
+      "epoch": 0.33097572072631015,
+      "grad_norm": 1010.6577758789062,
+      "learning_rate": 8.556290716542632e-06,
+      "loss": 84.2611,
+      "step": 40010
+    },
+    {
+      "epoch": 0.33105844397567935,
+      "grad_norm": 787.0457153320312,
+      "learning_rate": 8.555299184478026e-06,
+      "loss": 129.8781,
+      "step": 40020
+    },
+    {
+      "epoch": 0.3311411672250486,
+      "grad_norm": 953.2777099609375,
+      "learning_rate": 8.554307369534577e-06,
+      "loss": 103.5916,
+      "step": 40030
+    },
+    {
+      "epoch": 0.33122389047441786,
+      "grad_norm": 1413.6817626953125,
+      "learning_rate": 8.553315271791207e-06,
+      "loss": 92.5186,
+      "step": 40040
+    },
+    {
+      "epoch": 0.33130661372378706,
+      "grad_norm": 1188.9342041015625,
+      "learning_rate": 8.552322891326846e-06,
+      "loss": 98.2379,
+      "step": 40050
+    },
+    {
+      "epoch": 0.3313893369731563,
+      "grad_norm": 1236.0517578125,
+      "learning_rate": 8.551330228220454e-06,
+      "loss": 107.0516,
+      "step": 40060
+    },
+    {
+      "epoch": 0.33147206022252557,
+      "grad_norm": 919.5791015625,
+      "learning_rate": 8.550337282551016e-06,
+      "loss": 101.3186,
+      "step": 40070
+    },
+    {
+      "epoch": 0.33155478347189477,
+      "grad_norm": 1503.684326171875,
+      "learning_rate": 8.549344054397533e-06,
+      "loss": 108.6517,
+      "step": 40080
+    },
+    {
+      "epoch": 0.331637506721264,
+      "grad_norm": 419.35028076171875,
+      "learning_rate": 8.548350543839034e-06,
+      "loss": 85.9801,
+      "step": 40090
+    },
+    {
+      "epoch": 0.3317202299706333,
+      "grad_norm": 1511.6007080078125,
+      "learning_rate": 8.547356750954568e-06,
+      "loss": 109.9655,
+      "step": 40100
+    },
+    {
+      "epoch": 0.3318029532200025,
+      "grad_norm": 811.8706665039062,
+      "learning_rate": 8.546362675823204e-06,
+      "loss": 186.2005,
+      "step": 40110
+    },
+    {
+      "epoch": 0.3318856764693717,
+      "grad_norm": 910.4181518554688,
+      "learning_rate": 8.545368318524036e-06,
+      "loss": 113.1569,
+      "step": 40120
+    },
+    {
+      "epoch": 0.3319683997187409,
+      "grad_norm": 841.266357421875,
+      "learning_rate": 8.544373679136184e-06,
+      "loss": 102.7385,
+      "step": 40130
+    },
+    {
+      "epoch": 0.3320511229681102,
+      "grad_norm": 1087.2705078125,
+      "learning_rate": 8.543378757738785e-06,
+      "loss": 83.0132,
+      "step": 40140
+    },
+    {
+      "epoch": 0.33213384621747943,
+      "grad_norm": 844.1511840820312,
+      "learning_rate": 8.542383554411e-06,
+      "loss": 97.4727,
+      "step": 40150
+    },
+    {
+      "epoch": 0.33221656946684863,
+      "grad_norm": 1138.1463623046875,
+      "learning_rate": 8.541388069232012e-06,
+      "loss": 96.1207,
+      "step": 40160
+    },
+    {
+      "epoch": 0.3322992927162179,
+      "grad_norm": 1386.313232421875,
+      "learning_rate": 8.54039230228103e-06,
+      "loss": 121.9133,
+      "step": 40170
+    },
+    {
+      "epoch": 0.33238201596558714,
+      "grad_norm": 1644.196044921875,
+      "learning_rate": 8.53939625363728e-06,
+      "loss": 110.2846,
+      "step": 40180
+    },
+    {
+      "epoch": 0.33246473921495634,
+      "grad_norm": 663.9628295898438,
+      "learning_rate": 8.538399923380011e-06,
+      "loss": 147.1378,
+      "step": 40190
+    },
+    {
+      "epoch": 0.3325474624643256,
+      "grad_norm": 921.0543212890625,
+      "learning_rate": 8.537403311588502e-06,
+      "loss": 94.7127,
+      "step": 40200
+    },
+    {
+      "epoch": 0.33263018571369485,
+      "grad_norm": 1512.567626953125,
+      "learning_rate": 8.536406418342044e-06,
+      "loss": 87.8837,
+      "step": 40210
+    },
+    {
+      "epoch": 0.33271290896306405,
+      "grad_norm": 1098.78369140625,
+      "learning_rate": 8.53540924371996e-06,
+      "loss": 108.6445,
+      "step": 40220
+    },
+    {
+      "epoch": 0.3327956322124333,
+      "grad_norm": 1846.2921142578125,
+      "learning_rate": 8.534411787801586e-06,
+      "loss": 95.6519,
+      "step": 40230
+    },
+    {
+      "epoch": 0.33287835546180256,
+      "grad_norm": 1103.0107421875,
+      "learning_rate": 8.533414050666287e-06,
+      "loss": 109.1561,
+      "step": 40240
+    },
+    {
+      "epoch": 0.33296107871117175,
+      "grad_norm": 903.0642700195312,
+      "learning_rate": 8.532416032393447e-06,
+      "loss": 99.2833,
+      "step": 40250
+    },
+    {
+      "epoch": 0.333043801960541,
+      "grad_norm": 1340.9583740234375,
+      "learning_rate": 8.531417733062476e-06,
+      "loss": 116.9413,
+      "step": 40260
+    },
+    {
+      "epoch": 0.33312652520991026,
+      "grad_norm": 1119.4525146484375,
+      "learning_rate": 8.530419152752804e-06,
+      "loss": 124.4811,
+      "step": 40270
+    },
+    {
+      "epoch": 0.33320924845927946,
+      "grad_norm": 1566.3560791015625,
+      "learning_rate": 8.529420291543882e-06,
+      "loss": 127.6215,
+      "step": 40280
+    },
+    {
+      "epoch": 0.3332919717086487,
+      "grad_norm": 1316.3895263671875,
+      "learning_rate": 8.528421149515185e-06,
+      "loss": 107.4906,
+      "step": 40290
+    },
+    {
+      "epoch": 0.33337469495801797,
+      "grad_norm": 825.5098266601562,
+      "learning_rate": 8.52742172674621e-06,
+      "loss": 129.04,
+      "step": 40300
+    },
+    {
+      "epoch": 0.33345741820738717,
+      "grad_norm": 1096.2099609375,
+      "learning_rate": 8.526422023316478e-06,
+      "loss": 91.2496,
+      "step": 40310
+    },
+    {
+      "epoch": 0.3335401414567564,
+      "grad_norm": 781.985107421875,
+      "learning_rate": 8.525422039305529e-06,
+      "loss": 124.4936,
+      "step": 40320
+    },
+    {
+      "epoch": 0.3336228647061257,
+      "grad_norm": 1158.1458740234375,
+      "learning_rate": 8.524421774792926e-06,
+      "loss": 84.0795,
+      "step": 40330
+    },
+    {
+      "epoch": 0.3337055879554949,
+      "grad_norm": 1207.7098388671875,
+      "learning_rate": 8.52342122985826e-06,
+      "loss": 91.0112,
+      "step": 40340
+    },
+    {
+      "epoch": 0.33378831120486413,
+      "grad_norm": 847.1415405273438,
+      "learning_rate": 8.522420404581135e-06,
+      "loss": 101.6437,
+      "step": 40350
+    },
+    {
+      "epoch": 0.3338710344542334,
+      "grad_norm": 995.0086669921875,
+      "learning_rate": 8.521419299041185e-06,
+      "loss": 93.0817,
+      "step": 40360
+    },
+    {
+      "epoch": 0.3339537577036026,
+      "grad_norm": 892.8756103515625,
+      "learning_rate": 8.520417913318065e-06,
+      "loss": 110.2036,
+      "step": 40370
+    },
+    {
+      "epoch": 0.33403648095297184,
+      "grad_norm": 1183.2650146484375,
+      "learning_rate": 8.519416247491445e-06,
+      "loss": 126.0844,
+      "step": 40380
+    },
+    {
+      "epoch": 0.3341192042023411,
+      "grad_norm": 605.9612426757812,
+      "learning_rate": 8.518414301641027e-06,
+      "loss": 69.7784,
+      "step": 40390
+    },
+    {
+      "epoch": 0.3342019274517103,
+      "grad_norm": 1190.9295654296875,
+      "learning_rate": 8.517412075846529e-06,
+      "loss": 138.7514,
+      "step": 40400
+    },
+    {
+      "epoch": 0.33428465070107954,
+      "grad_norm": 792.2731323242188,
+      "learning_rate": 8.516409570187698e-06,
+      "loss": 98.212,
+      "step": 40410
+    },
+    {
+      "epoch": 0.3343673739504488,
+      "grad_norm": 820.3512573242188,
+      "learning_rate": 8.515406784744294e-06,
+      "loss": 87.1192,
+      "step": 40420
+    },
+    {
+      "epoch": 0.334450097199818,
+      "grad_norm": 1030.4779052734375,
+      "learning_rate": 8.514403719596104e-06,
+      "loss": 112.4568,
+      "step": 40430
+    },
+    {
+      "epoch": 0.33453282044918725,
+      "grad_norm": 649.7733764648438,
+      "learning_rate": 8.513400374822942e-06,
+      "loss": 129.4392,
+      "step": 40440
+    },
+    {
+      "epoch": 0.3346155436985565,
+      "grad_norm": 1169.7542724609375,
+      "learning_rate": 8.512396750504635e-06,
+      "loss": 96.2116,
+      "step": 40450
+    },
+    {
+      "epoch": 0.3346982669479257,
+      "grad_norm": 1340.5599365234375,
+      "learning_rate": 8.511392846721037e-06,
+      "loss": 130.6511,
+      "step": 40460
+    },
+    {
+      "epoch": 0.33478099019729496,
+      "grad_norm": 1084.3248291015625,
+      "learning_rate": 8.510388663552027e-06,
+      "loss": 96.2522,
+      "step": 40470
+    },
+    {
+      "epoch": 0.33486371344666416,
+      "grad_norm": 940.4993896484375,
+      "learning_rate": 8.509384201077502e-06,
+      "loss": 182.6661,
+      "step": 40480
+    },
+    {
+      "epoch": 0.3349464366960334,
+      "grad_norm": 473.4619445800781,
+      "learning_rate": 8.508379459377381e-06,
+      "loss": 98.2326,
+      "step": 40490
+    },
+    {
+      "epoch": 0.33502915994540267,
+      "grad_norm": 915.9678344726562,
+      "learning_rate": 8.507374438531606e-06,
+      "loss": 90.029,
+      "step": 40500
+    },
+    {
+      "epoch": 0.33511188319477186,
+      "grad_norm": 644.4666748046875,
+      "learning_rate": 8.506369138620148e-06,
+      "loss": 160.186,
+      "step": 40510
+    },
+    {
+      "epoch": 0.3351946064441411,
+      "grad_norm": 771.171875,
+      "learning_rate": 8.505363559722985e-06,
+      "loss": 96.5032,
+      "step": 40520
+    },
+    {
+      "epoch": 0.3352773296935104,
+      "grad_norm": 575.1885375976562,
+      "learning_rate": 8.504357701920134e-06,
+      "loss": 78.7146,
+      "step": 40530
+    },
+    {
+      "epoch": 0.33536005294287957,
+      "grad_norm": 728.8525390625,
+      "learning_rate": 8.503351565291622e-06,
+      "loss": 130.1776,
+      "step": 40540
+    },
+    {
+      "epoch": 0.3354427761922488,
+      "grad_norm": 1049.326416015625,
+      "learning_rate": 8.502345149917506e-06,
+      "loss": 91.4142,
+      "step": 40550
+    },
+    {
+      "epoch": 0.3355254994416181,
+      "grad_norm": 1567.1068115234375,
+      "learning_rate": 8.501338455877859e-06,
+      "loss": 128.1109,
+      "step": 40560
+    },
+    {
+      "epoch": 0.3356082226909873,
+      "grad_norm": 639.6390380859375,
+      "learning_rate": 8.50033148325278e-06,
+      "loss": 89.2162,
+      "step": 40570
+    },
+    {
+      "epoch": 0.33569094594035653,
+      "grad_norm": 1309.4464111328125,
+      "learning_rate": 8.499324232122389e-06,
+      "loss": 119.5868,
+      "step": 40580
+    },
+    {
+      "epoch": 0.3357736691897258,
+      "grad_norm": 961.0704345703125,
+      "learning_rate": 8.498316702566828e-06,
+      "loss": 108.5671,
+      "step": 40590
+    },
+    {
+      "epoch": 0.335856392439095,
+      "grad_norm": 1294.3653564453125,
+      "learning_rate": 8.497308894666263e-06,
+      "loss": 114.0025,
+      "step": 40600
+    },
+    {
+      "epoch": 0.33593911568846424,
+      "grad_norm": 1210.564697265625,
+      "learning_rate": 8.496300808500878e-06,
+      "loss": 122.3642,
+      "step": 40610
+    },
+    {
+      "epoch": 0.3360218389378335,
+      "grad_norm": 1512.3568115234375,
+      "learning_rate": 8.495292444150887e-06,
+      "loss": 146.3031,
+      "step": 40620
+    },
+    {
+      "epoch": 0.3361045621872027,
+      "grad_norm": 956.16162109375,
+      "learning_rate": 8.494283801696514e-06,
+      "loss": 140.3855,
+      "step": 40630
+    },
+    {
+      "epoch": 0.33618728543657195,
+      "grad_norm": 862.5502319335938,
+      "learning_rate": 8.493274881218017e-06,
+      "loss": 92.3681,
+      "step": 40640
+    },
+    {
+      "epoch": 0.3362700086859412,
+      "grad_norm": 715.0096435546875,
+      "learning_rate": 8.49226568279567e-06,
+      "loss": 93.8356,
+      "step": 40650
+    },
+    {
+      "epoch": 0.3363527319353104,
+      "grad_norm": 858.4218139648438,
+      "learning_rate": 8.49125620650977e-06,
+      "loss": 89.4689,
+      "step": 40660
+    },
+    {
+      "epoch": 0.33643545518467965,
+      "grad_norm": 797.36328125,
+      "learning_rate": 8.490246452440636e-06,
+      "loss": 131.3191,
+      "step": 40670
+    },
+    {
+      "epoch": 0.3365181784340489,
+      "grad_norm": 887.1189575195312,
+      "learning_rate": 8.48923642066861e-06,
+      "loss": 143.5478,
+      "step": 40680
+    },
+    {
+      "epoch": 0.3366009016834181,
+      "grad_norm": 688.75732421875,
+      "learning_rate": 8.488226111274055e-06,
+      "loss": 129.7013,
+      "step": 40690
+    },
+    {
+      "epoch": 0.33668362493278736,
+      "grad_norm": 770.3273315429688,
+      "learning_rate": 8.487215524337357e-06,
+      "loss": 77.0376,
+      "step": 40700
+    },
+    {
+      "epoch": 0.3367663481821566,
+      "grad_norm": 960.6439819335938,
+      "learning_rate": 8.486204659938924e-06,
+      "loss": 127.0703,
+      "step": 40710
+    },
+    {
+      "epoch": 0.3368490714315258,
+      "grad_norm": 764.961669921875,
+      "learning_rate": 8.485193518159186e-06,
+      "loss": 111.9176,
+      "step": 40720
+    },
+    {
+      "epoch": 0.33693179468089507,
+      "grad_norm": 900.38134765625,
+      "learning_rate": 8.484182099078596e-06,
+      "loss": 112.1536,
+      "step": 40730
+    },
+    {
+      "epoch": 0.3370145179302643,
+      "grad_norm": 960.113525390625,
+      "learning_rate": 8.483170402777624e-06,
+      "loss": 127.3206,
+      "step": 40740
+    },
+    {
+      "epoch": 0.3370972411796335,
+      "grad_norm": 1646.762939453125,
+      "learning_rate": 8.482158429336769e-06,
+      "loss": 143.5467,
+      "step": 40750
+    },
+    {
+      "epoch": 0.3371799644290028,
+      "grad_norm": 985.1450805664062,
+      "learning_rate": 8.48114617883655e-06,
+      "loss": 114.6298,
+      "step": 40760
+    },
+    {
+      "epoch": 0.33726268767837203,
+      "grad_norm": 759.6780395507812,
+      "learning_rate": 8.480133651357507e-06,
+      "loss": 116.8154,
+      "step": 40770
+    },
+    {
+      "epoch": 0.33734541092774123,
+      "grad_norm": 885.9656982421875,
+      "learning_rate": 8.479120846980197e-06,
+      "loss": 108.7685,
+      "step": 40780
+    },
+    {
+      "epoch": 0.3374281341771105,
+      "grad_norm": 575.7235717773438,
+      "learning_rate": 8.478107765785212e-06,
+      "loss": 88.1911,
+      "step": 40790
+    },
+    {
+      "epoch": 0.33751085742647974,
+      "grad_norm": 666.4598388671875,
+      "learning_rate": 8.477094407853153e-06,
+      "loss": 69.0146,
+      "step": 40800
+    },
+    {
+      "epoch": 0.33759358067584894,
+      "grad_norm": 645.81201171875,
+      "learning_rate": 8.47608077326465e-06,
+      "loss": 94.4932,
+      "step": 40810
+    },
+    {
+      "epoch": 0.3376763039252182,
+      "grad_norm": 994.3779907226562,
+      "learning_rate": 8.475066862100352e-06,
+      "loss": 88.9415,
+      "step": 40820
+    },
+    {
+      "epoch": 0.33775902717458745,
+      "grad_norm": 759.6505737304688,
+      "learning_rate": 8.474052674440934e-06,
+      "loss": 117.1598,
+      "step": 40830
+    },
+    {
+      "epoch": 0.33784175042395664,
+      "grad_norm": 869.6111450195312,
+      "learning_rate": 8.473038210367086e-06,
+      "loss": 78.2449,
+      "step": 40840
+    },
+    {
+      "epoch": 0.3379244736733259,
+      "grad_norm": 947.123046875,
+      "learning_rate": 8.47202346995953e-06,
+      "loss": 93.2359,
+      "step": 40850
+    },
+    {
+      "epoch": 0.3380071969226951,
+      "grad_norm": 1361.4940185546875,
+      "learning_rate": 8.471008453298998e-06,
+      "loss": 134.8301,
+      "step": 40860
+    },
+    {
+      "epoch": 0.33808992017206435,
+      "grad_norm": 729.060546875,
+      "learning_rate": 8.469993160466254e-06,
+      "loss": 94.2659,
+      "step": 40870
+    },
+    {
+      "epoch": 0.3381726434214336,
+      "grad_norm": 643.9381713867188,
+      "learning_rate": 8.46897759154208e-06,
+      "loss": 88.6729,
+      "step": 40880
+    },
+    {
+      "epoch": 0.3382553666708028,
+      "grad_norm": 927.8340454101562,
+      "learning_rate": 8.467961746607279e-06,
+      "loss": 127.7991,
+      "step": 40890
+    },
+    {
+      "epoch": 0.33833808992017206,
+      "grad_norm": 604.8875122070312,
+      "learning_rate": 8.466945625742678e-06,
+      "loss": 76.509,
+      "step": 40900
+    },
+    {
+      "epoch": 0.3384208131695413,
+      "grad_norm": 672.4412841796875,
+      "learning_rate": 8.465929229029124e-06,
+      "loss": 165.3088,
+      "step": 40910
+    },
+    {
+      "epoch": 0.3385035364189105,
+      "grad_norm": 649.953369140625,
+      "learning_rate": 8.464912556547486e-06,
+      "loss": 112.9105,
+      "step": 40920
+    },
+    {
+      "epoch": 0.33858625966827977,
+      "grad_norm": 1930.8785400390625,
+      "learning_rate": 8.46389560837866e-06,
+      "loss": 108.5322,
+      "step": 40930
+    },
+    {
+      "epoch": 0.338668982917649,
+      "grad_norm": 622.740966796875,
+      "learning_rate": 8.462878384603558e-06,
+      "loss": 117.5824,
+      "step": 40940
+    },
+    {
+      "epoch": 0.3387517061670182,
+      "grad_norm": 474.0765686035156,
+      "learning_rate": 8.461860885303116e-06,
+      "loss": 99.5456,
+      "step": 40950
+    },
+    {
+      "epoch": 0.3388344294163875,
+      "grad_norm": 952.2894287109375,
+      "learning_rate": 8.460843110558287e-06,
+      "loss": 124.9169,
+      "step": 40960
+    },
+    {
+      "epoch": 0.3389171526657567,
+      "grad_norm": 654.9668579101562,
+      "learning_rate": 8.459825060450058e-06,
+      "loss": 90.4174,
+      "step": 40970
+    },
+    {
+      "epoch": 0.3389998759151259,
+      "grad_norm": 1069.80029296875,
+      "learning_rate": 8.458806735059428e-06,
+      "loss": 134.0334,
+      "step": 40980
+    },
+    {
+      "epoch": 0.3390825991644952,
+      "grad_norm": 657.5958862304688,
+      "learning_rate": 8.45778813446742e-06,
+      "loss": 97.927,
+      "step": 40990
+    },
+    {
+      "epoch": 0.33916532241386443,
+      "grad_norm": 1148.0079345703125,
+      "learning_rate": 8.456769258755078e-06,
+      "loss": 111.504,
+      "step": 41000
+    },
+    {
+      "epoch": 0.33924804566323363,
+      "grad_norm": 1073.7718505859375,
+      "learning_rate": 8.455750108003468e-06,
+      "loss": 78.7796,
+      "step": 41010
+    },
+    {
+      "epoch": 0.3393307689126029,
+      "grad_norm": 530.935302734375,
+      "learning_rate": 8.454730682293686e-06,
+      "loss": 76.4729,
+      "step": 41020
+    },
+    {
+      "epoch": 0.33941349216197214,
+      "grad_norm": 584.3358764648438,
+      "learning_rate": 8.453710981706838e-06,
+      "loss": 100.3047,
+      "step": 41030
+    },
+    {
+      "epoch": 0.33949621541134134,
+      "grad_norm": 1019.848388671875,
+      "learning_rate": 8.452691006324055e-06,
+      "loss": 101.324,
+      "step": 41040
+    },
+    {
+      "epoch": 0.3395789386607106,
+      "grad_norm": 991.2681274414062,
+      "learning_rate": 8.451670756226496e-06,
+      "loss": 75.0817,
+      "step": 41050
+    },
+    {
+      "epoch": 0.33966166191007985,
+      "grad_norm": 730.0400390625,
+      "learning_rate": 8.450650231495336e-06,
+      "loss": 85.305,
+      "step": 41060
+    },
+    {
+      "epoch": 0.33974438515944905,
+      "grad_norm": 929.1596069335938,
+      "learning_rate": 8.449629432211774e-06,
+      "loss": 92.8536,
+      "step": 41070
+    },
+    {
+      "epoch": 0.3398271084088183,
+      "grad_norm": 1377.75146484375,
+      "learning_rate": 8.44860835845703e-06,
+      "loss": 103.3764,
+      "step": 41080
+    },
+    {
+      "epoch": 0.33990983165818756,
+      "grad_norm": 522.6357421875,
+      "learning_rate": 8.447587010312343e-06,
+      "loss": 107.57,
+      "step": 41090
+    },
+    {
+      "epoch": 0.33999255490755675,
+      "grad_norm": 2292.6142578125,
+      "learning_rate": 8.44656538785898e-06,
+      "loss": 135.6061,
+      "step": 41100
+    },
+    {
+      "epoch": 0.340075278156926,
+      "grad_norm": 419.8533630371094,
+      "learning_rate": 8.44554349117823e-06,
+      "loss": 80.6014,
+      "step": 41110
+    },
+    {
+      "epoch": 0.34015800140629526,
+      "grad_norm": 1163.7928466796875,
+      "learning_rate": 8.444521320351397e-06,
+      "loss": 110.4075,
+      "step": 41120
+    },
+    {
+      "epoch": 0.34024072465566446,
+      "grad_norm": 942.8880615234375,
+      "learning_rate": 8.44349887545981e-06,
+      "loss": 118.3985,
+      "step": 41130
+    },
+    {
+      "epoch": 0.3403234479050337,
+      "grad_norm": 509.7681884765625,
+      "learning_rate": 8.442476156584818e-06,
+      "loss": 133.2833,
+      "step": 41140
+    },
+    {
+      "epoch": 0.34040617115440297,
+      "grad_norm": 1343.72705078125,
+      "learning_rate": 8.4414531638078e-06,
+      "loss": 134.3201,
+      "step": 41150
+    },
+    {
+      "epoch": 0.34048889440377217,
+      "grad_norm": 1334.9111328125,
+      "learning_rate": 8.440429897210148e-06,
+      "loss": 94.3114,
+      "step": 41160
+    },
+    {
+      "epoch": 0.3405716176531414,
+      "grad_norm": 918.8824462890625,
+      "learning_rate": 8.439406356873279e-06,
+      "loss": 105.6756,
+      "step": 41170
+    },
+    {
+      "epoch": 0.3406543409025107,
+      "grad_norm": 435.05010986328125,
+      "learning_rate": 8.43838254287863e-06,
+      "loss": 86.3829,
+      "step": 41180
+    },
+    {
+      "epoch": 0.3407370641518799,
+      "grad_norm": 934.824462890625,
+      "learning_rate": 8.43735845530766e-06,
+      "loss": 122.4491,
+      "step": 41190
+    },
+    {
+      "epoch": 0.34081978740124913,
+      "grad_norm": 1095.854248046875,
+      "learning_rate": 8.436334094241855e-06,
+      "loss": 110.3371,
+      "step": 41200
+    },
+    {
+      "epoch": 0.34090251065061833,
+      "grad_norm": 1048.58154296875,
+      "learning_rate": 8.435309459762718e-06,
+      "loss": 135.8438,
+      "step": 41210
+    },
+    {
+      "epoch": 0.3409852338999876,
+      "grad_norm": 722.0227661132812,
+      "learning_rate": 8.434284551951772e-06,
+      "loss": 86.2307,
+      "step": 41220
+    },
+    {
+      "epoch": 0.34106795714935684,
+      "grad_norm": 835.3677368164062,
+      "learning_rate": 8.433259370890565e-06,
+      "loss": 79.5151,
+      "step": 41230
+    },
+    {
+      "epoch": 0.34115068039872604,
+      "grad_norm": 1341.2940673828125,
+      "learning_rate": 8.432233916660669e-06,
+      "loss": 102.6455,
+      "step": 41240
+    },
+    {
+      "epoch": 0.3412334036480953,
+      "grad_norm": 972.4404907226562,
+      "learning_rate": 8.43120818934367e-06,
+      "loss": 105.9913,
+      "step": 41250
+    },
+    {
+      "epoch": 0.34131612689746454,
+      "grad_norm": 752.3978881835938,
+      "learning_rate": 8.43018218902118e-06,
+      "loss": 97.6527,
+      "step": 41260
+    },
+    {
+      "epoch": 0.34139885014683374,
+      "grad_norm": 744.5557250976562,
+      "learning_rate": 8.429155915774839e-06,
+      "loss": 98.6538,
+      "step": 41270
+    },
+    {
+      "epoch": 0.341481573396203,
+      "grad_norm": 1084.3350830078125,
+      "learning_rate": 8.428129369686299e-06,
+      "loss": 96.4803,
+      "step": 41280
+    },
+    {
+      "epoch": 0.34156429664557225,
+      "grad_norm": 887.330078125,
+      "learning_rate": 8.427102550837238e-06,
+      "loss": 111.9868,
+      "step": 41290
+    },
+    {
+      "epoch": 0.34164701989494145,
+      "grad_norm": 869.30908203125,
+      "learning_rate": 8.426075459309356e-06,
+      "loss": 120.3469,
+      "step": 41300
+    },
+    {
+      "epoch": 0.3417297431443107,
+      "grad_norm": 2174.565673828125,
+      "learning_rate": 8.42504809518437e-06,
+      "loss": 115.0281,
+      "step": 41310
+    },
+    {
+      "epoch": 0.34181246639367996,
+      "grad_norm": 939.2781372070312,
+      "learning_rate": 8.42402045854403e-06,
+      "loss": 118.3146,
+      "step": 41320
+    },
+    {
+      "epoch": 0.34189518964304916,
+      "grad_norm": 962.0004272460938,
+      "learning_rate": 8.422992549470094e-06,
+      "loss": 111.6336,
+      "step": 41330
+    },
+    {
+      "epoch": 0.3419779128924184,
+      "grad_norm": 856.4796142578125,
+      "learning_rate": 8.42196436804435e-06,
+      "loss": 89.8677,
+      "step": 41340
+    },
+    {
+      "epoch": 0.34206063614178767,
+      "grad_norm": 882.47509765625,
+      "learning_rate": 8.420935914348607e-06,
+      "loss": 109.2613,
+      "step": 41350
+    },
+    {
+      "epoch": 0.34214335939115686,
+      "grad_norm": 1152.5184326171875,
+      "learning_rate": 8.419907188464691e-06,
+      "loss": 83.5429,
+      "step": 41360
+    },
+    {
+      "epoch": 0.3422260826405261,
+      "grad_norm": 1086.84521484375,
+      "learning_rate": 8.418878190474459e-06,
+      "loss": 107.8546,
+      "step": 41370
+    },
+    {
+      "epoch": 0.3423088058898954,
+      "grad_norm": 1217.004150390625,
+      "learning_rate": 8.417848920459778e-06,
+      "loss": 137.2482,
+      "step": 41380
+    },
+    {
+      "epoch": 0.34239152913926457,
+      "grad_norm": 706.2305297851562,
+      "learning_rate": 8.416819378502543e-06,
+      "loss": 74.3434,
+      "step": 41390
+    },
+    {
+      "epoch": 0.3424742523886338,
+      "grad_norm": 1614.1566162109375,
+      "learning_rate": 8.415789564684673e-06,
+      "loss": 142.1887,
+      "step": 41400
+    },
+    {
+      "epoch": 0.3425569756380031,
+      "grad_norm": 875.3898315429688,
+      "learning_rate": 8.414759479088102e-06,
+      "loss": 97.1488,
+      "step": 41410
+    },
+    {
+      "epoch": 0.3426396988873723,
+      "grad_norm": 1026.1566162109375,
+      "learning_rate": 8.413729121794794e-06,
+      "loss": 130.0628,
+      "step": 41420
+    },
+    {
+      "epoch": 0.34272242213674153,
+      "grad_norm": 977.263427734375,
+      "learning_rate": 8.412698492886723e-06,
+      "loss": 78.8849,
+      "step": 41430
+    },
+    {
+      "epoch": 0.3428051453861108,
+      "grad_norm": 1244.2081298828125,
+      "learning_rate": 8.411667592445898e-06,
+      "loss": 142.179,
+      "step": 41440
+    },
+    {
+      "epoch": 0.34288786863548,
+      "grad_norm": 751.5814819335938,
+      "learning_rate": 8.410636420554337e-06,
+      "loss": 95.5533,
+      "step": 41450
+    },
+    {
+      "epoch": 0.34297059188484924,
+      "grad_norm": 786.3610229492188,
+      "learning_rate": 8.409604977294093e-06,
+      "loss": 103.3417,
+      "step": 41460
+    },
+    {
+      "epoch": 0.3430533151342185,
+      "grad_norm": 931.486572265625,
+      "learning_rate": 8.408573262747225e-06,
+      "loss": 141.6174,
+      "step": 41470
+    },
+    {
+      "epoch": 0.3431360383835877,
+      "grad_norm": 838.9139404296875,
+      "learning_rate": 8.407541276995828e-06,
+      "loss": 105.8545,
+      "step": 41480
+    },
+    {
+      "epoch": 0.34321876163295695,
+      "grad_norm": 654.8321533203125,
+      "learning_rate": 8.40650902012201e-06,
+      "loss": 107.5087,
+      "step": 41490
+    },
+    {
+      "epoch": 0.3433014848823262,
+      "grad_norm": 766.4340209960938,
+      "learning_rate": 8.405476492207902e-06,
+      "loss": 113.7733,
+      "step": 41500
+    },
+    {
+      "epoch": 0.3433842081316954,
+      "grad_norm": 1260.985595703125,
+      "learning_rate": 8.404443693335658e-06,
+      "loss": 108.2389,
+      "step": 41510
+    },
+    {
+      "epoch": 0.34346693138106466,
+      "grad_norm": 777.0344848632812,
+      "learning_rate": 8.403410623587454e-06,
+      "loss": 112.1793,
+      "step": 41520
+    },
+    {
+      "epoch": 0.3435496546304339,
+      "grad_norm": 836.8817749023438,
+      "learning_rate": 8.402377283045487e-06,
+      "loss": 121.241,
+      "step": 41530
+    },
+    {
+      "epoch": 0.3436323778798031,
+      "grad_norm": 675.0514526367188,
+      "learning_rate": 8.401343671791974e-06,
+      "loss": 121.9953,
+      "step": 41540
+    },
+    {
+      "epoch": 0.34371510112917236,
+      "grad_norm": 1098.513916015625,
+      "learning_rate": 8.400309789909155e-06,
+      "loss": 129.7842,
+      "step": 41550
+    },
+    {
+      "epoch": 0.3437978243785416,
+      "grad_norm": 728.864013671875,
+      "learning_rate": 8.399275637479291e-06,
+      "loss": 99.4059,
+      "step": 41560
+    },
+    {
+      "epoch": 0.3438805476279108,
+      "grad_norm": 744.5280151367188,
+      "learning_rate": 8.398241214584666e-06,
+      "loss": 115.1609,
+      "step": 41570
+    },
+    {
+      "epoch": 0.34396327087728007,
+      "grad_norm": 938.07080078125,
+      "learning_rate": 8.397206521307584e-06,
+      "loss": 102.9427,
+      "step": 41580
+    },
+    {
+      "epoch": 0.34404599412664927,
+      "grad_norm": 1658.468994140625,
+      "learning_rate": 8.396171557730369e-06,
+      "loss": 115.574,
+      "step": 41590
+    },
+    {
+      "epoch": 0.3441287173760185,
+      "grad_norm": 634.4130859375,
+      "learning_rate": 8.39513632393537e-06,
+      "loss": 93.7718,
+      "step": 41600
+    },
+    {
+      "epoch": 0.3442114406253878,
+      "grad_norm": 579.6549072265625,
+      "learning_rate": 8.394100820004954e-06,
+      "loss": 92.9008,
+      "step": 41610
+    },
+    {
+      "epoch": 0.344294163874757,
+      "grad_norm": 450.8374938964844,
+      "learning_rate": 8.393065046021513e-06,
+      "loss": 90.1996,
+      "step": 41620
+    },
+    {
+      "epoch": 0.34437688712412623,
+      "grad_norm": 572.2847290039062,
+      "learning_rate": 8.39202900206746e-06,
+      "loss": 147.2724,
+      "step": 41630
+    },
+    {
+      "epoch": 0.3444596103734955,
+      "grad_norm": 2148.787353515625,
+      "learning_rate": 8.390992688225226e-06,
+      "loss": 132.4656,
+      "step": 41640
+    },
+    {
+      "epoch": 0.3445423336228647,
+      "grad_norm": 707.6249389648438,
+      "learning_rate": 8.389956104577265e-06,
+      "loss": 90.7987,
+      "step": 41650
+    },
+    {
+      "epoch": 0.34462505687223394,
+      "grad_norm": 1235.2393798828125,
+      "learning_rate": 8.388919251206054e-06,
+      "loss": 120.5997,
+      "step": 41660
+    },
+    {
+      "epoch": 0.3447077801216032,
+      "grad_norm": 802.1920166015625,
+      "learning_rate": 8.387882128194094e-06,
+      "loss": 127.8519,
+      "step": 41670
+    },
+    {
+      "epoch": 0.3447905033709724,
+      "grad_norm": 1075.386962890625,
+      "learning_rate": 8.3868447356239e-06,
+      "loss": 95.1565,
+      "step": 41680
+    },
+    {
+      "epoch": 0.34487322662034164,
+      "grad_norm": 1524.40283203125,
+      "learning_rate": 8.385807073578014e-06,
+      "loss": 125.0543,
+      "step": 41690
+    },
+    {
+      "epoch": 0.3449559498697109,
+      "grad_norm": 887.3607177734375,
+      "learning_rate": 8.384769142138998e-06,
+      "loss": 86.6021,
+      "step": 41700
+    },
+    {
+      "epoch": 0.3450386731190801,
+      "grad_norm": 932.5927124023438,
+      "learning_rate": 8.383730941389434e-06,
+      "loss": 96.8807,
+      "step": 41710
+    },
+    {
+      "epoch": 0.34512139636844935,
+      "grad_norm": 956.2900390625,
+      "learning_rate": 8.382692471411931e-06,
+      "loss": 97.0625,
+      "step": 41720
+    },
+    {
+      "epoch": 0.3452041196178186,
+      "grad_norm": 646.645263671875,
+      "learning_rate": 8.38165373228911e-06,
+      "loss": 95.7504,
+      "step": 41730
+    },
+    {
+      "epoch": 0.3452868428671878,
+      "grad_norm": 1114.9278564453125,
+      "learning_rate": 8.380614724103622e-06,
+      "loss": 120.9078,
+      "step": 41740
+    },
+    {
+      "epoch": 0.34536956611655706,
+      "grad_norm": 693.3104248046875,
+      "learning_rate": 8.379575446938136e-06,
+      "loss": 106.318,
+      "step": 41750
+    },
+    {
+      "epoch": 0.3454522893659263,
+      "grad_norm": 8694.8857421875,
+      "learning_rate": 8.37853590087534e-06,
+      "loss": 133.577,
+      "step": 41760
+    },
+    {
+      "epoch": 0.3455350126152955,
+      "grad_norm": 645.8015747070312,
+      "learning_rate": 8.377496085997949e-06,
+      "loss": 124.9107,
+      "step": 41770
+    },
+    {
+      "epoch": 0.34561773586466477,
+      "grad_norm": 2443.57421875,
+      "learning_rate": 8.376456002388695e-06,
+      "loss": 115.6421,
+      "step": 41780
+    },
+    {
+      "epoch": 0.345700459114034,
+      "grad_norm": 1185.3404541015625,
+      "learning_rate": 8.375415650130332e-06,
+      "loss": 103.3099,
+      "step": 41790
+    },
+    {
+      "epoch": 0.3457831823634032,
+      "grad_norm": 1205.232666015625,
+      "learning_rate": 8.37437502930564e-06,
+      "loss": 100.6235,
+      "step": 41800
+    },
+    {
+      "epoch": 0.3458659056127725,
+      "grad_norm": 1242.7322998046875,
+      "learning_rate": 8.373334139997409e-06,
+      "loss": 131.52,
+      "step": 41810
+    },
+    {
+      "epoch": 0.3459486288621417,
+      "grad_norm": 747.3458862304688,
+      "learning_rate": 8.372292982288463e-06,
+      "loss": 118.5125,
+      "step": 41820
+    },
+    {
+      "epoch": 0.3460313521115109,
+      "grad_norm": 1096.74462890625,
+      "learning_rate": 8.371251556261642e-06,
+      "loss": 136.9112,
+      "step": 41830
+    },
+    {
+      "epoch": 0.3461140753608802,
+      "grad_norm": 801.9876708984375,
+      "learning_rate": 8.370209861999807e-06,
+      "loss": 106.0218,
+      "step": 41840
+    },
+    {
+      "epoch": 0.34619679861024943,
+      "grad_norm": 909.0194091796875,
+      "learning_rate": 8.36916789958584e-06,
+      "loss": 103.4507,
+      "step": 41850
+    },
+    {
+      "epoch": 0.34627952185961863,
+      "grad_norm": 1055.92041015625,
+      "learning_rate": 8.368125669102645e-06,
+      "loss": 94.5659,
+      "step": 41860
+    },
+    {
+      "epoch": 0.3463622451089879,
+      "grad_norm": 867.0262451171875,
+      "learning_rate": 8.36708317063315e-06,
+      "loss": 135.4051,
+      "step": 41870
+    },
+    {
+      "epoch": 0.34644496835835714,
+      "grad_norm": 871.2109985351562,
+      "learning_rate": 8.366040404260298e-06,
+      "loss": 99.1085,
+      "step": 41880
+    },
+    {
+      "epoch": 0.34652769160772634,
+      "grad_norm": 830.19384765625,
+      "learning_rate": 8.36499737006706e-06,
+      "loss": 76.3817,
+      "step": 41890
+    },
+    {
+      "epoch": 0.3466104148570956,
+      "grad_norm": 951.84326171875,
+      "learning_rate": 8.363954068136424e-06,
+      "loss": 119.5681,
+      "step": 41900
+    },
+    {
+      "epoch": 0.34669313810646485,
+      "grad_norm": 731.0087890625,
+      "learning_rate": 8.362910498551402e-06,
+      "loss": 114.4868,
+      "step": 41910
+    },
+    {
+      "epoch": 0.34677586135583405,
+      "grad_norm": 635.1656494140625,
+      "learning_rate": 8.361866661395024e-06,
+      "loss": 116.8612,
+      "step": 41920
+    },
+    {
+      "epoch": 0.3468585846052033,
+      "grad_norm": 1068.5445556640625,
+      "learning_rate": 8.360822556750345e-06,
+      "loss": 91.3164,
+      "step": 41930
+    },
+    {
+      "epoch": 0.3469413078545725,
+      "grad_norm": 949.19873046875,
+      "learning_rate": 8.35977818470044e-06,
+      "loss": 104.9618,
+      "step": 41940
+    },
+    {
+      "epoch": 0.34702403110394175,
+      "grad_norm": 429.62677001953125,
+      "learning_rate": 8.358733545328404e-06,
+      "loss": 93.7747,
+      "step": 41950
+    },
+    {
+      "epoch": 0.347106754353311,
+      "grad_norm": 893.44189453125,
+      "learning_rate": 8.357688638717354e-06,
+      "loss": 106.5521,
+      "step": 41960
+    },
+    {
+      "epoch": 0.3471894776026802,
+      "grad_norm": 585.1188354492188,
+      "learning_rate": 8.356643464950428e-06,
+      "loss": 80.4151,
+      "step": 41970
+    },
+    {
+      "epoch": 0.34727220085204946,
+      "grad_norm": 1215.0146484375,
+      "learning_rate": 8.355598024110789e-06,
+      "loss": 181.5081,
+      "step": 41980
+    },
+    {
+      "epoch": 0.3473549241014187,
+      "grad_norm": 1458.9388427734375,
+      "learning_rate": 8.354552316281613e-06,
+      "loss": 141.7899,
+      "step": 41990
+    },
+    {
+      "epoch": 0.3474376473507879,
+      "grad_norm": 915.29443359375,
+      "learning_rate": 8.353506341546106e-06,
+      "loss": 108.375,
+      "step": 42000
+    },
+    {
+      "epoch": 0.34752037060015717,
+      "grad_norm": 1018.5140991210938,
+      "learning_rate": 8.352460099987488e-06,
+      "loss": 118.4601,
+      "step": 42010
+    },
+    {
+      "epoch": 0.3476030938495264,
+      "grad_norm": 1653.6846923828125,
+      "learning_rate": 8.351413591689007e-06,
+      "loss": 127.7061,
+      "step": 42020
+    },
+    {
+      "epoch": 0.3476858170988956,
+      "grad_norm": 1303.524169921875,
+      "learning_rate": 8.350366816733927e-06,
+      "loss": 109.428,
+      "step": 42030
+    },
+    {
+      "epoch": 0.3477685403482649,
+      "grad_norm": 1039.2083740234375,
+      "learning_rate": 8.349319775205536e-06,
+      "loss": 120.8401,
+      "step": 42040
+    },
+    {
+      "epoch": 0.34785126359763413,
+      "grad_norm": 1214.731689453125,
+      "learning_rate": 8.34827246718714e-06,
+      "loss": 157.5093,
+      "step": 42050
+    },
+    {
+      "epoch": 0.34793398684700333,
+      "grad_norm": 832.408203125,
+      "learning_rate": 8.347224892762072e-06,
+      "loss": 106.25,
+      "step": 42060
+    },
+    {
+      "epoch": 0.3480167100963726,
+      "grad_norm": 1899.1883544921875,
+      "learning_rate": 8.346177052013681e-06,
+      "loss": 128.2392,
+      "step": 42070
+    },
+    {
+      "epoch": 0.34809943334574184,
+      "grad_norm": 475.4049072265625,
+      "learning_rate": 8.345128945025338e-06,
+      "loss": 128.0041,
+      "step": 42080
+    },
+    {
+      "epoch": 0.34818215659511104,
+      "grad_norm": 483.9809875488281,
+      "learning_rate": 8.344080571880438e-06,
+      "loss": 92.6426,
+      "step": 42090
+    },
+    {
+      "epoch": 0.3482648798444803,
+      "grad_norm": 1139.35302734375,
+      "learning_rate": 8.343031932662394e-06,
+      "loss": 89.6336,
+      "step": 42100
+    },
+    {
+      "epoch": 0.34834760309384954,
+      "grad_norm": 998.1423950195312,
+      "learning_rate": 8.341983027454641e-06,
+      "loss": 148.1835,
+      "step": 42110
+    },
+    {
+      "epoch": 0.34843032634321874,
+      "grad_norm": 1257.747802734375,
+      "learning_rate": 8.340933856340637e-06,
+      "loss": 116.7931,
+      "step": 42120
+    },
+    {
+      "epoch": 0.348513049592588,
+      "grad_norm": 1050.6527099609375,
+      "learning_rate": 8.339884419403857e-06,
+      "loss": 124.4426,
+      "step": 42130
+    },
+    {
+      "epoch": 0.34859577284195725,
+      "grad_norm": 935.9532470703125,
+      "learning_rate": 8.338834716727801e-06,
+      "loss": 100.9498,
+      "step": 42140
+    },
+    {
+      "epoch": 0.34867849609132645,
+      "grad_norm": 1319.49365234375,
+      "learning_rate": 8.337784748395992e-06,
+      "loss": 89.4219,
+      "step": 42150
+    },
+    {
+      "epoch": 0.3487612193406957,
+      "grad_norm": 583.6215209960938,
+      "learning_rate": 8.336734514491968e-06,
+      "loss": 101.7242,
+      "step": 42160
+    },
+    {
+      "epoch": 0.34884394259006496,
+      "grad_norm": 1093.5703125,
+      "learning_rate": 8.335684015099294e-06,
+      "loss": 100.3031,
+      "step": 42170
+    },
+    {
+      "epoch": 0.34892666583943416,
+      "grad_norm": 7167.025390625,
+      "learning_rate": 8.33463325030155e-06,
+      "loss": 142.965,
+      "step": 42180
+    },
+    {
+      "epoch": 0.3490093890888034,
+      "grad_norm": 1053.3721923828125,
+      "learning_rate": 8.333582220182344e-06,
+      "loss": 118.1564,
+      "step": 42190
+    },
+    {
+      "epoch": 0.34909211233817267,
+      "grad_norm": 1960.47705078125,
+      "learning_rate": 8.332530924825297e-06,
+      "loss": 119.301,
+      "step": 42200
+    },
+    {
+      "epoch": 0.34917483558754187,
+      "grad_norm": 973.1529541015625,
+      "learning_rate": 8.33147936431406e-06,
+      "loss": 131.6417,
+      "step": 42210
+    },
+    {
+      "epoch": 0.3492575588369111,
+      "grad_norm": 925.6248168945312,
+      "learning_rate": 8.3304275387323e-06,
+      "loss": 116.9732,
+      "step": 42220
+    },
+    {
+      "epoch": 0.3493402820862804,
+      "grad_norm": 0.0,
+      "learning_rate": 8.329375448163703e-06,
+      "loss": 87.3547,
+      "step": 42230
+    },
+    {
+      "epoch": 0.3494230053356496,
+      "grad_norm": 1005.3858032226562,
+      "learning_rate": 8.328323092691985e-06,
+      "loss": 104.9806,
+      "step": 42240
+    },
+    {
+      "epoch": 0.3495057285850188,
+      "grad_norm": 790.18603515625,
+      "learning_rate": 8.32727047240087e-06,
+      "loss": 141.6189,
+      "step": 42250
+    },
+    {
+      "epoch": 0.3495884518343881,
+      "grad_norm": 885.5327758789062,
+      "learning_rate": 8.326217587374115e-06,
+      "loss": 95.2874,
+      "step": 42260
+    },
+    {
+      "epoch": 0.3496711750837573,
+      "grad_norm": 1333.3345947265625,
+      "learning_rate": 8.325164437695493e-06,
+      "loss": 110.1591,
+      "step": 42270
+    },
+    {
+      "epoch": 0.34975389833312653,
+      "grad_norm": 444.2149963378906,
+      "learning_rate": 8.324111023448795e-06,
+      "loss": 89.4089,
+      "step": 42280
+    },
+    {
+      "epoch": 0.3498366215824958,
+      "grad_norm": 712.4427490234375,
+      "learning_rate": 8.32305734471784e-06,
+      "loss": 104.8016,
+      "step": 42290
+    },
+    {
+      "epoch": 0.349919344831865,
+      "grad_norm": 825.5255737304688,
+      "learning_rate": 8.322003401586463e-06,
+      "loss": 109.8285,
+      "step": 42300
+    },
+    {
+      "epoch": 0.35000206808123424,
+      "grad_norm": 674.3945922851562,
+      "learning_rate": 8.32094919413852e-06,
+      "loss": 118.0758,
+      "step": 42310
+    },
+    {
+      "epoch": 0.35008479133060344,
+      "grad_norm": 1092.6353759765625,
+      "learning_rate": 8.319894722457892e-06,
+      "loss": 100.6579,
+      "step": 42320
+    },
+    {
+      "epoch": 0.3501675145799727,
+      "grad_norm": 791.6581420898438,
+      "learning_rate": 8.318839986628477e-06,
+      "loss": 96.1517,
+      "step": 42330
+    },
+    {
+      "epoch": 0.35025023782934195,
+      "grad_norm": 483.9101867675781,
+      "learning_rate": 8.317784986734194e-06,
+      "loss": 86.3806,
+      "step": 42340
+    },
+    {
+      "epoch": 0.35033296107871115,
+      "grad_norm": 982.6383056640625,
+      "learning_rate": 8.316729722858987e-06,
+      "loss": 142.3889,
+      "step": 42350
+    },
+    {
+      "epoch": 0.3504156843280804,
+      "grad_norm": 1031.0849609375,
+      "learning_rate": 8.31567419508682e-06,
+      "loss": 89.9149,
+      "step": 42360
+    },
+    {
+      "epoch": 0.35049840757744966,
+      "grad_norm": 1122.879150390625,
+      "learning_rate": 8.31461840350167e-06,
+      "loss": 103.2544,
+      "step": 42370
+    },
+    {
+      "epoch": 0.35058113082681885,
+      "grad_norm": 1027.361328125,
+      "learning_rate": 8.313562348187549e-06,
+      "loss": 90.0591,
+      "step": 42380
+    },
+    {
+      "epoch": 0.3506638540761881,
+      "grad_norm": 789.521240234375,
+      "learning_rate": 8.312506029228478e-06,
+      "loss": 90.2937,
+      "step": 42390
+    },
+    {
+      "epoch": 0.35074657732555736,
+      "grad_norm": 616.3732299804688,
+      "learning_rate": 8.311449446708506e-06,
+      "loss": 129.6206,
+      "step": 42400
+    },
+    {
+      "epoch": 0.35082930057492656,
+      "grad_norm": 1325.065185546875,
+      "learning_rate": 8.310392600711698e-06,
+      "loss": 114.9959,
+      "step": 42410
+    },
+    {
+      "epoch": 0.3509120238242958,
+      "grad_norm": 976.2049560546875,
+      "learning_rate": 8.309335491322143e-06,
+      "loss": 95.8001,
+      "step": 42420
+    },
+    {
+      "epoch": 0.35099474707366507,
+      "grad_norm": 1060.470947265625,
+      "learning_rate": 8.30827811862395e-06,
+      "loss": 101.8484,
+      "step": 42430
+    },
+    {
+      "epoch": 0.35107747032303427,
+      "grad_norm": 813.407470703125,
+      "learning_rate": 8.307220482701251e-06,
+      "loss": 119.0382,
+      "step": 42440
+    },
+    {
+      "epoch": 0.3511601935724035,
+      "grad_norm": 769.6190795898438,
+      "learning_rate": 8.306162583638197e-06,
+      "loss": 116.1655,
+      "step": 42450
+    },
+    {
+      "epoch": 0.3512429168217728,
+      "grad_norm": 1182.1982421875,
+      "learning_rate": 8.305104421518959e-06,
+      "loss": 116.6159,
+      "step": 42460
+    },
+    {
+      "epoch": 0.351325640071142,
+      "grad_norm": 1151.38671875,
+      "learning_rate": 8.30404599642773e-06,
+      "loss": 97.3802,
+      "step": 42470
+    },
+    {
+      "epoch": 0.35140836332051123,
+      "grad_norm": 0.0,
+      "learning_rate": 8.302987308448724e-06,
+      "loss": 81.0704,
+      "step": 42480
+    },
+    {
+      "epoch": 0.3514910865698805,
+      "grad_norm": 728.1912231445312,
+      "learning_rate": 8.301928357666178e-06,
+      "loss": 92.2258,
+      "step": 42490
+    },
+    {
+      "epoch": 0.3515738098192497,
+      "grad_norm": 793.99267578125,
+      "learning_rate": 8.300869144164346e-06,
+      "loss": 110.2738,
+      "step": 42500
+    },
+    {
+      "epoch": 0.35165653306861894,
+      "grad_norm": 993.7640991210938,
+      "learning_rate": 8.299809668027505e-06,
+      "loss": 131.7156,
+      "step": 42510
+    },
+    {
+      "epoch": 0.3517392563179882,
+      "grad_norm": 606.397705078125,
+      "learning_rate": 8.298749929339953e-06,
+      "loss": 104.0031,
+      "step": 42520
+    },
+    {
+      "epoch": 0.3518219795673574,
+      "grad_norm": 827.4598388671875,
+      "learning_rate": 8.297689928186009e-06,
+      "loss": 110.1917,
+      "step": 42530
+    },
+    {
+      "epoch": 0.35190470281672664,
+      "grad_norm": 2409.9150390625,
+      "learning_rate": 8.29662966465001e-06,
+      "loss": 135.4566,
+      "step": 42540
+    },
+    {
+      "epoch": 0.3519874260660959,
+      "grad_norm": 613.3368530273438,
+      "learning_rate": 8.295569138816319e-06,
+      "loss": 80.187,
+      "step": 42550
+    },
+    {
+      "epoch": 0.3520701493154651,
+      "grad_norm": 843.9635009765625,
+      "learning_rate": 8.294508350769315e-06,
+      "loss": 109.0294,
+      "step": 42560
+    },
+    {
+      "epoch": 0.35215287256483435,
+      "grad_norm": 622.92578125,
+      "learning_rate": 8.293447300593402e-06,
+      "loss": 127.0855,
+      "step": 42570
+    },
+    {
+      "epoch": 0.3522355958142036,
+      "grad_norm": 657.6416625976562,
+      "learning_rate": 8.292385988373005e-06,
+      "loss": 108.607,
+      "step": 42580
+    },
+    {
+      "epoch": 0.3523183190635728,
+      "grad_norm": 1057.9915771484375,
+      "learning_rate": 8.29132441419256e-06,
+      "loss": 117.3586,
+      "step": 42590
+    },
+    {
+      "epoch": 0.35240104231294206,
+      "grad_norm": 704.32421875,
+      "learning_rate": 8.290262578136541e-06,
+      "loss": 92.2817,
+      "step": 42600
+    },
+    {
+      "epoch": 0.3524837655623113,
+      "grad_norm": 391.53265380859375,
+      "learning_rate": 8.289200480289426e-06,
+      "loss": 91.4089,
+      "step": 42610
+    },
+    {
+      "epoch": 0.3525664888116805,
+      "grad_norm": 1248.0340576171875,
+      "learning_rate": 8.288138120735726e-06,
+      "loss": 94.6713,
+      "step": 42620
+    },
+    {
+      "epoch": 0.35264921206104977,
+      "grad_norm": 1576.25,
+      "learning_rate": 8.287075499559965e-06,
+      "loss": 120.6687,
+      "step": 42630
+    },
+    {
+      "epoch": 0.352731935310419,
+      "grad_norm": 850.7510375976562,
+      "learning_rate": 8.286012616846693e-06,
+      "loss": 68.9104,
+      "step": 42640
+    },
+    {
+      "epoch": 0.3528146585597882,
+      "grad_norm": 1018.4036254882812,
+      "learning_rate": 8.284949472680477e-06,
+      "loss": 96.3005,
+      "step": 42650
+    },
+    {
+      "epoch": 0.3528973818091575,
+      "grad_norm": 769.276611328125,
+      "learning_rate": 8.283886067145908e-06,
+      "loss": 101.5941,
+      "step": 42660
+    },
+    {
+      "epoch": 0.35298010505852667,
+      "grad_norm": 1009.304931640625,
+      "learning_rate": 8.282822400327595e-06,
+      "loss": 100.8695,
+      "step": 42670
+    },
+    {
+      "epoch": 0.3530628283078959,
+      "grad_norm": 965.6119384765625,
+      "learning_rate": 8.28175847231017e-06,
+      "loss": 112.0248,
+      "step": 42680
+    },
+    {
+      "epoch": 0.3531455515572652,
+      "grad_norm": 862.6695556640625,
+      "learning_rate": 8.280694283178285e-06,
+      "loss": 97.2944,
+      "step": 42690
+    },
+    {
+      "epoch": 0.3532282748066344,
+      "grad_norm": 698.0172119140625,
+      "learning_rate": 8.27962983301661e-06,
+      "loss": 66.0908,
+      "step": 42700
+    },
+    {
+      "epoch": 0.35331099805600363,
+      "grad_norm": 680.3596801757812,
+      "learning_rate": 8.278565121909845e-06,
+      "loss": 56.0012,
+      "step": 42710
+    },
+    {
+      "epoch": 0.3533937213053729,
+      "grad_norm": 979.0956420898438,
+      "learning_rate": 8.277500149942697e-06,
+      "loss": 111.6114,
+      "step": 42720
+    },
+    {
+      "epoch": 0.3534764445547421,
+      "grad_norm": 839.169921875,
+      "learning_rate": 8.276434917199904e-06,
+      "loss": 126.5481,
+      "step": 42730
+    },
+    {
+      "epoch": 0.35355916780411134,
+      "grad_norm": 1183.0323486328125,
+      "learning_rate": 8.275369423766222e-06,
+      "loss": 97.7488,
+      "step": 42740
+    },
+    {
+      "epoch": 0.3536418910534806,
+      "grad_norm": 1349.1572265625,
+      "learning_rate": 8.274303669726427e-06,
+      "loss": 110.0713,
+      "step": 42750
+    },
+    {
+      "epoch": 0.3537246143028498,
+      "grad_norm": 916.60107421875,
+      "learning_rate": 8.273237655165314e-06,
+      "loss": 106.8208,
+      "step": 42760
+    },
+    {
+      "epoch": 0.35380733755221905,
+      "grad_norm": 800.146484375,
+      "learning_rate": 8.272171380167705e-06,
+      "loss": 83.875,
+      "step": 42770
+    },
+    {
+      "epoch": 0.3538900608015883,
+      "grad_norm": 1539.4642333984375,
+      "learning_rate": 8.271104844818436e-06,
+      "loss": 130.0894,
+      "step": 42780
+    },
+    {
+      "epoch": 0.3539727840509575,
+      "grad_norm": 1235.24560546875,
+      "learning_rate": 8.270038049202366e-06,
+      "loss": 131.9467,
+      "step": 42790
+    },
+    {
+      "epoch": 0.35405550730032675,
+      "grad_norm": 962.0623168945312,
+      "learning_rate": 8.268970993404377e-06,
+      "loss": 100.8265,
+      "step": 42800
+    },
+    {
+      "epoch": 0.354138230549696,
+      "grad_norm": 836.7759399414062,
+      "learning_rate": 8.267903677509368e-06,
+      "loss": 119.7814,
+      "step": 42810
+    },
+    {
+      "epoch": 0.3542209537990652,
+      "grad_norm": 1205.632080078125,
+      "learning_rate": 8.266836101602263e-06,
+      "loss": 107.5889,
+      "step": 42820
+    },
+    {
+      "epoch": 0.35430367704843446,
+      "grad_norm": 1180.9609375,
+      "learning_rate": 8.265768265767999e-06,
+      "loss": 146.9007,
+      "step": 42830
+    },
+    {
+      "epoch": 0.3543864002978037,
+      "grad_norm": 784.8786010742188,
+      "learning_rate": 8.264700170091543e-06,
+      "loss": 93.1217,
+      "step": 42840
+    },
+    {
+      "epoch": 0.3544691235471729,
+      "grad_norm": 583.6641845703125,
+      "learning_rate": 8.263631814657879e-06,
+      "loss": 127.952,
+      "step": 42850
+    },
+    {
+      "epoch": 0.35455184679654217,
+      "grad_norm": 1015.6278076171875,
+      "learning_rate": 8.262563199552007e-06,
+      "loss": 109.5995,
+      "step": 42860
+    },
+    {
+      "epoch": 0.3546345700459114,
+      "grad_norm": 1008.2879638671875,
+      "learning_rate": 8.261494324858956e-06,
+      "loss": 83.4442,
+      "step": 42870
+    },
+    {
+      "epoch": 0.3547172932952806,
+      "grad_norm": 934.4283447265625,
+      "learning_rate": 8.26042519066377e-06,
+      "loss": 81.4526,
+      "step": 42880
+    },
+    {
+      "epoch": 0.3548000165446499,
+      "grad_norm": 602.0402221679688,
+      "learning_rate": 8.259355797051515e-06,
+      "loss": 87.8508,
+      "step": 42890
+    },
+    {
+      "epoch": 0.35488273979401913,
+      "grad_norm": 952.8324584960938,
+      "learning_rate": 8.258286144107277e-06,
+      "loss": 101.1536,
+      "step": 42900
+    },
+    {
+      "epoch": 0.35496546304338833,
+      "grad_norm": 924.0335693359375,
+      "learning_rate": 8.257216231916162e-06,
+      "loss": 113.0049,
+      "step": 42910
+    },
+    {
+      "epoch": 0.3550481862927576,
+      "grad_norm": 684.9584350585938,
+      "learning_rate": 8.256146060563304e-06,
+      "loss": 117.6641,
+      "step": 42920
+    },
+    {
+      "epoch": 0.35513090954212684,
+      "grad_norm": 1416.010009765625,
+      "learning_rate": 8.255075630133847e-06,
+      "loss": 93.2686,
+      "step": 42930
+    },
+    {
+      "epoch": 0.35521363279149604,
+      "grad_norm": 1231.5831298828125,
+      "learning_rate": 8.254004940712958e-06,
+      "loss": 111.2918,
+      "step": 42940
+    },
+    {
+      "epoch": 0.3552963560408653,
+      "grad_norm": 1368.33935546875,
+      "learning_rate": 8.252933992385833e-06,
+      "loss": 101.9154,
+      "step": 42950
+    },
+    {
+      "epoch": 0.35537907929023455,
+      "grad_norm": 754.49609375,
+      "learning_rate": 8.251862785237676e-06,
+      "loss": 88.8121,
+      "step": 42960
+    },
+    {
+      "epoch": 0.35546180253960374,
+      "grad_norm": 1223.3607177734375,
+      "learning_rate": 8.250791319353723e-06,
+      "loss": 111.0358,
+      "step": 42970
+    },
+    {
+      "epoch": 0.355544525788973,
+      "grad_norm": 1084.2374267578125,
+      "learning_rate": 8.249719594819225e-06,
+      "loss": 107.8028,
+      "step": 42980
+    },
+    {
+      "epoch": 0.35562724903834225,
+      "grad_norm": 1002.8360595703125,
+      "learning_rate": 8.248647611719452e-06,
+      "loss": 87.2639,
+      "step": 42990
+    },
+    {
+      "epoch": 0.35570997228771145,
+      "grad_norm": 882.518310546875,
+      "learning_rate": 8.247575370139695e-06,
+      "loss": 120.7826,
+      "step": 43000
+    },
+    {
+      "epoch": 0.3557926955370807,
+      "grad_norm": 710.4674682617188,
+      "learning_rate": 8.246502870165273e-06,
+      "loss": 130.1348,
+      "step": 43010
+    },
+    {
+      "epoch": 0.3558754187864499,
+      "grad_norm": 705.7244262695312,
+      "learning_rate": 8.245430111881519e-06,
+      "loss": 82.2953,
+      "step": 43020
+    },
+    {
+      "epoch": 0.35595814203581916,
+      "grad_norm": 1012.916748046875,
+      "learning_rate": 8.244357095373783e-06,
+      "loss": 103.8642,
+      "step": 43030
+    },
+    {
+      "epoch": 0.3560408652851884,
+      "grad_norm": 823.5924682617188,
+      "learning_rate": 8.243283820727441e-06,
+      "loss": 120.169,
+      "step": 43040
+    },
+    {
+      "epoch": 0.3561235885345576,
+      "grad_norm": 506.7729187011719,
+      "learning_rate": 8.242210288027893e-06,
+      "loss": 106.6605,
+      "step": 43050
+    },
+    {
+      "epoch": 0.35620631178392687,
+      "grad_norm": 573.303955078125,
+      "learning_rate": 8.241136497360552e-06,
+      "loss": 96.0248,
+      "step": 43060
+    },
+    {
+      "epoch": 0.3562890350332961,
+      "grad_norm": 1349.73974609375,
+      "learning_rate": 8.240062448810853e-06,
+      "loss": 150.305,
+      "step": 43070
+    },
+    {
+      "epoch": 0.3563717582826653,
+      "grad_norm": 898.83349609375,
+      "learning_rate": 8.238988142464254e-06,
+      "loss": 99.9782,
+      "step": 43080
+    },
+    {
+      "epoch": 0.3564544815320346,
+      "grad_norm": 1892.2440185546875,
+      "learning_rate": 8.237913578406236e-06,
+      "loss": 127.4689,
+      "step": 43090
+    },
+    {
+      "epoch": 0.3565372047814038,
+      "grad_norm": 1101.9765625,
+      "learning_rate": 8.236838756722294e-06,
+      "loss": 99.9242,
+      "step": 43100
+    },
+    {
+      "epoch": 0.356619928030773,
+      "grad_norm": 785.5177001953125,
+      "learning_rate": 8.235763677497945e-06,
+      "loss": 102.3918,
+      "step": 43110
+    },
+    {
+      "epoch": 0.3567026512801423,
+      "grad_norm": 1374.7742919921875,
+      "learning_rate": 8.234688340818732e-06,
+      "loss": 113.4264,
+      "step": 43120
+    },
+    {
+      "epoch": 0.35678537452951153,
+      "grad_norm": 606.4083862304688,
+      "learning_rate": 8.233612746770214e-06,
+      "loss": 93.4949,
+      "step": 43130
+    },
+    {
+      "epoch": 0.35686809777888073,
+      "grad_norm": 922.9923095703125,
+      "learning_rate": 8.232536895437968e-06,
+      "loss": 85.8221,
+      "step": 43140
+    },
+    {
+      "epoch": 0.35695082102825,
+      "grad_norm": 790.5139770507812,
+      "learning_rate": 8.231460786907597e-06,
+      "loss": 117.3538,
+      "step": 43150
+    },
+    {
+      "epoch": 0.35703354427761924,
+      "grad_norm": 417.8067626953125,
+      "learning_rate": 8.230384421264722e-06,
+      "loss": 87.6937,
+      "step": 43160
+    },
+    {
+      "epoch": 0.35711626752698844,
+      "grad_norm": 1458.1962890625,
+      "learning_rate": 8.229307798594985e-06,
+      "loss": 133.923,
+      "step": 43170
+    },
+    {
+      "epoch": 0.3571989907763577,
+      "grad_norm": 987.9276123046875,
+      "learning_rate": 8.228230918984046e-06,
+      "loss": 111.6961,
+      "step": 43180
+    },
+    {
+      "epoch": 0.35728171402572695,
+      "grad_norm": 1103.388427734375,
+      "learning_rate": 8.22715378251759e-06,
+      "loss": 93.5427,
+      "step": 43190
+    },
+    {
+      "epoch": 0.35736443727509615,
+      "grad_norm": 697.0567626953125,
+      "learning_rate": 8.226076389281316e-06,
+      "loss": 117.8876,
+      "step": 43200
+    },
+    {
+      "epoch": 0.3574471605244654,
+      "grad_norm": 1062.5294189453125,
+      "learning_rate": 8.22499873936095e-06,
+      "loss": 85.2779,
+      "step": 43210
+    },
+    {
+      "epoch": 0.35752988377383466,
+      "grad_norm": 997.5350341796875,
+      "learning_rate": 8.223920832842236e-06,
+      "loss": 127.3359,
+      "step": 43220
+    },
+    {
+      "epoch": 0.35761260702320385,
+      "grad_norm": 839.9952392578125,
+      "learning_rate": 8.222842669810936e-06,
+      "loss": 112.6368,
+      "step": 43230
+    },
+    {
+      "epoch": 0.3576953302725731,
+      "grad_norm": 1012.621337890625,
+      "learning_rate": 8.221764250352835e-06,
+      "loss": 108.84,
+      "step": 43240
+    },
+    {
+      "epoch": 0.35777805352194236,
+      "grad_norm": 918.2972412109375,
+      "learning_rate": 8.220685574553739e-06,
+      "loss": 85.8282,
+      "step": 43250
+    },
+    {
+      "epoch": 0.35786077677131156,
+      "grad_norm": 776.9376220703125,
+      "learning_rate": 8.219606642499474e-06,
+      "loss": 96.5936,
+      "step": 43260
+    },
+    {
+      "epoch": 0.3579435000206808,
+      "grad_norm": 868.3795166015625,
+      "learning_rate": 8.218527454275884e-06,
+      "loss": 91.7565,
+      "step": 43270
+    },
+    {
+      "epoch": 0.35802622327005007,
+      "grad_norm": 958.1405029296875,
+      "learning_rate": 8.217448009968834e-06,
+      "loss": 110.3028,
+      "step": 43280
+    },
+    {
+      "epoch": 0.35810894651941927,
+      "grad_norm": 1635.0120849609375,
+      "learning_rate": 8.216368309664213e-06,
+      "loss": 115.6983,
+      "step": 43290
+    },
+    {
+      "epoch": 0.3581916697687885,
+      "grad_norm": 495.8813781738281,
+      "learning_rate": 8.215288353447927e-06,
+      "loss": 125.6738,
+      "step": 43300
+    },
+    {
+      "epoch": 0.3582743930181578,
+      "grad_norm": 1934.987548828125,
+      "learning_rate": 8.214208141405903e-06,
+      "loss": 96.0109,
+      "step": 43310
+    },
+    {
+      "epoch": 0.358357116267527,
+      "grad_norm": 1259.3150634765625,
+      "learning_rate": 8.213127673624088e-06,
+      "loss": 98.7009,
+      "step": 43320
+    },
+    {
+      "epoch": 0.35843983951689623,
+      "grad_norm": 985.3812255859375,
+      "learning_rate": 8.212046950188451e-06,
+      "loss": 119.01,
+      "step": 43330
+    },
+    {
+      "epoch": 0.3585225627662655,
+      "grad_norm": 657.3631591796875,
+      "learning_rate": 8.21096597118498e-06,
+      "loss": 102.065,
+      "step": 43340
+    },
+    {
+      "epoch": 0.3586052860156347,
+      "grad_norm": 808.63232421875,
+      "learning_rate": 8.209884736699681e-06,
+      "loss": 86.2247,
+      "step": 43350
+    },
+    {
+      "epoch": 0.35868800926500394,
+      "grad_norm": 963.5828857421875,
+      "learning_rate": 8.208803246818586e-06,
+      "loss": 99.5541,
+      "step": 43360
+    },
+    {
+      "epoch": 0.3587707325143732,
+      "grad_norm": 356.1247253417969,
+      "learning_rate": 8.207721501627743e-06,
+      "loss": 132.9291,
+      "step": 43370
+    },
+    {
+      "epoch": 0.3588534557637424,
+      "grad_norm": 1359.7357177734375,
+      "learning_rate": 8.20663950121322e-06,
+      "loss": 106.6871,
+      "step": 43380
+    },
+    {
+      "epoch": 0.35893617901311164,
+      "grad_norm": 741.9186401367188,
+      "learning_rate": 8.20555724566111e-06,
+      "loss": 110.5265,
+      "step": 43390
+    },
+    {
+      "epoch": 0.35901890226248084,
+      "grad_norm": 804.3803100585938,
+      "learning_rate": 8.204474735057522e-06,
+      "loss": 89.7678,
+      "step": 43400
+    },
+    {
+      "epoch": 0.3591016255118501,
+      "grad_norm": 986.8497924804688,
+      "learning_rate": 8.203391969488586e-06,
+      "loss": 76.5805,
+      "step": 43410
+    },
+    {
+      "epoch": 0.35918434876121935,
+      "grad_norm": 981.0845947265625,
+      "learning_rate": 8.20230894904045e-06,
+      "loss": 108.0519,
+      "step": 43420
+    },
+    {
+      "epoch": 0.35926707201058855,
+      "grad_norm": 967.68310546875,
+      "learning_rate": 8.20122567379929e-06,
+      "loss": 144.2405,
+      "step": 43430
+    },
+    {
+      "epoch": 0.3593497952599578,
+      "grad_norm": 1253.8336181640625,
+      "learning_rate": 8.200142143851295e-06,
+      "loss": 85.0357,
+      "step": 43440
+    },
+    {
+      "epoch": 0.35943251850932706,
+      "grad_norm": 660.7843627929688,
+      "learning_rate": 8.199058359282675e-06,
+      "loss": 110.5242,
+      "step": 43450
+    },
+    {
+      "epoch": 0.35951524175869626,
+      "grad_norm": 876.0247802734375,
+      "learning_rate": 8.197974320179664e-06,
+      "loss": 143.5727,
+      "step": 43460
+    },
+    {
+      "epoch": 0.3595979650080655,
+      "grad_norm": 1184.567626953125,
+      "learning_rate": 8.19689002662851e-06,
+      "loss": 87.2803,
+      "step": 43470
+    },
+    {
+      "epoch": 0.35968068825743477,
+      "grad_norm": 899.4736938476562,
+      "learning_rate": 8.195805478715492e-06,
+      "loss": 94.5841,
+      "step": 43480
+    },
+    {
+      "epoch": 0.35976341150680397,
+      "grad_norm": 1176.9891357421875,
+      "learning_rate": 8.194720676526898e-06,
+      "loss": 105.5688,
+      "step": 43490
+    },
+    {
+      "epoch": 0.3598461347561732,
+      "grad_norm": 1430.493896484375,
+      "learning_rate": 8.193635620149041e-06,
+      "loss": 120.161,
+      "step": 43500
+    },
+    {
+      "epoch": 0.3599288580055425,
+      "grad_norm": 1398.2462158203125,
+      "learning_rate": 8.192550309668254e-06,
+      "loss": 153.4543,
+      "step": 43510
+    },
+    {
+      "epoch": 0.3600115812549117,
+      "grad_norm": 978.09033203125,
+      "learning_rate": 8.191464745170892e-06,
+      "loss": 97.1732,
+      "step": 43520
+    },
+    {
+      "epoch": 0.3600943045042809,
+      "grad_norm": 1013.3282470703125,
+      "learning_rate": 8.190378926743327e-06,
+      "loss": 101.2923,
+      "step": 43530
+    },
+    {
+      "epoch": 0.3601770277536502,
+      "grad_norm": 881.2088012695312,
+      "learning_rate": 8.189292854471953e-06,
+      "loss": 148.854,
+      "step": 43540
+    },
+    {
+      "epoch": 0.3602597510030194,
+      "grad_norm": 870.5662841796875,
+      "learning_rate": 8.188206528443182e-06,
+      "loss": 92.8082,
+      "step": 43550
+    },
+    {
+      "epoch": 0.36034247425238863,
+      "grad_norm": 776.38916015625,
+      "learning_rate": 8.18711994874345e-06,
+      "loss": 82.2745,
+      "step": 43560
+    },
+    {
+      "epoch": 0.3604251975017579,
+      "grad_norm": 904.0106811523438,
+      "learning_rate": 8.186033115459211e-06,
+      "loss": 97.3916,
+      "step": 43570
+    },
+    {
+      "epoch": 0.3605079207511271,
+      "grad_norm": 1215.83349609375,
+      "learning_rate": 8.184946028676937e-06,
+      "loss": 106.3127,
+      "step": 43580
+    },
+    {
+      "epoch": 0.36059064400049634,
+      "grad_norm": 794.0036010742188,
+      "learning_rate": 8.183858688483126e-06,
+      "loss": 91.0681,
+      "step": 43590
+    },
+    {
+      "epoch": 0.3606733672498656,
+      "grad_norm": 1268.9676513671875,
+      "learning_rate": 8.182771094964292e-06,
+      "loss": 123.4264,
+      "step": 43600
+    },
+    {
+      "epoch": 0.3607560904992348,
+      "grad_norm": 777.9105224609375,
+      "learning_rate": 8.181683248206968e-06,
+      "loss": 111.6841,
+      "step": 43610
+    },
+    {
+      "epoch": 0.36083881374860405,
+      "grad_norm": 854.5106201171875,
+      "learning_rate": 8.180595148297709e-06,
+      "loss": 113.4441,
+      "step": 43620
+    },
+    {
+      "epoch": 0.3609215369979733,
+      "grad_norm": 850.8244018554688,
+      "learning_rate": 8.179506795323092e-06,
+      "loss": 135.2171,
+      "step": 43630
+    },
+    {
+      "epoch": 0.3610042602473425,
+      "grad_norm": 1269.437255859375,
+      "learning_rate": 8.17841818936971e-06,
+      "loss": 155.0749,
+      "step": 43640
+    },
+    {
+      "epoch": 0.36108698349671176,
+      "grad_norm": 1103.447509765625,
+      "learning_rate": 8.177329330524182e-06,
+      "loss": 85.1156,
+      "step": 43650
+    },
+    {
+      "epoch": 0.361169706746081,
+      "grad_norm": 800.2313232421875,
+      "learning_rate": 8.17624021887314e-06,
+      "loss": 103.1669,
+      "step": 43660
+    },
+    {
+      "epoch": 0.3612524299954502,
+      "grad_norm": 662.4274291992188,
+      "learning_rate": 8.17515085450324e-06,
+      "loss": 98.6518,
+      "step": 43670
+    },
+    {
+      "epoch": 0.36133515324481946,
+      "grad_norm": 975.4820556640625,
+      "learning_rate": 8.174061237501159e-06,
+      "loss": 120.7466,
+      "step": 43680
+    },
+    {
+      "epoch": 0.3614178764941887,
+      "grad_norm": 1076.44873046875,
+      "learning_rate": 8.172971367953593e-06,
+      "loss": 80.2128,
+      "step": 43690
+    },
+    {
+      "epoch": 0.3615005997435579,
+      "grad_norm": 1007.4608764648438,
+      "learning_rate": 8.171881245947257e-06,
+      "loss": 62.0215,
+      "step": 43700
+    },
+    {
+      "epoch": 0.36158332299292717,
+      "grad_norm": 1051.32177734375,
+      "learning_rate": 8.170790871568887e-06,
+      "loss": 157.7504,
+      "step": 43710
+    },
+    {
+      "epoch": 0.3616660462422964,
+      "grad_norm": 1688.2108154296875,
+      "learning_rate": 8.169700244905239e-06,
+      "loss": 123.6984,
+      "step": 43720
+    },
+    {
+      "epoch": 0.3617487694916656,
+      "grad_norm": 1116.3714599609375,
+      "learning_rate": 8.168609366043089e-06,
+      "loss": 92.2827,
+      "step": 43730
+    },
+    {
+      "epoch": 0.3618314927410349,
+      "grad_norm": 845.093505859375,
+      "learning_rate": 8.167518235069234e-06,
+      "loss": 77.9922,
+      "step": 43740
+    },
+    {
+      "epoch": 0.3619142159904041,
+      "grad_norm": 2042.611572265625,
+      "learning_rate": 8.16642685207049e-06,
+      "loss": 141.047,
+      "step": 43750
+    },
+    {
+      "epoch": 0.36199693923977333,
+      "grad_norm": 602.3466186523438,
+      "learning_rate": 8.165335217133695e-06,
+      "loss": 122.7751,
+      "step": 43760
+    },
+    {
+      "epoch": 0.3620796624891426,
+      "grad_norm": 998.31005859375,
+      "learning_rate": 8.164243330345702e-06,
+      "loss": 95.6849,
+      "step": 43770
+    },
+    {
+      "epoch": 0.3621623857385118,
+      "grad_norm": 939.9814453125,
+      "learning_rate": 8.16315119179339e-06,
+      "loss": 88.2418,
+      "step": 43780
+    },
+    {
+      "epoch": 0.36224510898788104,
+      "grad_norm": 2236.7392578125,
+      "learning_rate": 8.162058801563652e-06,
+      "loss": 116.7937,
+      "step": 43790
+    },
+    {
+      "epoch": 0.3623278322372503,
+      "grad_norm": 973.8187255859375,
+      "learning_rate": 8.160966159743411e-06,
+      "loss": 94.5988,
+      "step": 43800
+    },
+    {
+      "epoch": 0.3624105554866195,
+      "grad_norm": 640.3901977539062,
+      "learning_rate": 8.159873266419598e-06,
+      "loss": 103.685,
+      "step": 43810
+    },
+    {
+      "epoch": 0.36249327873598874,
+      "grad_norm": 1432.554931640625,
+      "learning_rate": 8.15878012167917e-06,
+      "loss": 106.0658,
+      "step": 43820
+    },
+    {
+      "epoch": 0.362576001985358,
+      "grad_norm": 859.1890258789062,
+      "learning_rate": 8.157686725609105e-06,
+      "loss": 87.8233,
+      "step": 43830
+    },
+    {
+      "epoch": 0.3626587252347272,
+      "grad_norm": 1776.3751220703125,
+      "learning_rate": 8.1565930782964e-06,
+      "loss": 114.2181,
+      "step": 43840
+    },
+    {
+      "epoch": 0.36274144848409645,
+      "grad_norm": 926.181884765625,
+      "learning_rate": 8.155499179828068e-06,
+      "loss": 114.7968,
+      "step": 43850
+    },
+    {
+      "epoch": 0.3628241717334657,
+      "grad_norm": 959.3102416992188,
+      "learning_rate": 8.15440503029115e-06,
+      "loss": 105.2323,
+      "step": 43860
+    },
+    {
+      "epoch": 0.3629068949828349,
+      "grad_norm": 1929.3040771484375,
+      "learning_rate": 8.153310629772702e-06,
+      "loss": 131.4064,
+      "step": 43870
+    },
+    {
+      "epoch": 0.36298961823220416,
+      "grad_norm": 1120.273193359375,
+      "learning_rate": 8.152215978359796e-06,
+      "loss": 92.3281,
+      "step": 43880
+    },
+    {
+      "epoch": 0.3630723414815734,
+      "grad_norm": 830.0736694335938,
+      "learning_rate": 8.151121076139534e-06,
+      "loss": 91.5073,
+      "step": 43890
+    },
+    {
+      "epoch": 0.3631550647309426,
+      "grad_norm": 1354.7823486328125,
+      "learning_rate": 8.150025923199027e-06,
+      "loss": 201.7689,
+      "step": 43900
+    },
+    {
+      "epoch": 0.36323778798031187,
+      "grad_norm": 1229.7574462890625,
+      "learning_rate": 8.148930519625417e-06,
+      "loss": 116.0604,
+      "step": 43910
+    },
+    {
+      "epoch": 0.3633205112296811,
+      "grad_norm": 1157.707763671875,
+      "learning_rate": 8.147834865505855e-06,
+      "loss": 118.8252,
+      "step": 43920
+    },
+    {
+      "epoch": 0.3634032344790503,
+      "grad_norm": 391.3431396484375,
+      "learning_rate": 8.14673896092752e-06,
+      "loss": 94.0042,
+      "step": 43930
+    },
+    {
+      "epoch": 0.3634859577284196,
+      "grad_norm": 963.4109497070312,
+      "learning_rate": 8.145642805977608e-06,
+      "loss": 94.306,
+      "step": 43940
+    },
+    {
+      "epoch": 0.3635686809777888,
+      "grad_norm": 680.21826171875,
+      "learning_rate": 8.144546400743334e-06,
+      "loss": 121.9921,
+      "step": 43950
+    },
+    {
+      "epoch": 0.363651404227158,
+      "grad_norm": 763.9702758789062,
+      "learning_rate": 8.143449745311934e-06,
+      "loss": 105.5048,
+      "step": 43960
+    },
+    {
+      "epoch": 0.3637341274765273,
+      "grad_norm": 706.5518188476562,
+      "learning_rate": 8.142352839770663e-06,
+      "loss": 112.0056,
+      "step": 43970
+    },
+    {
+      "epoch": 0.36381685072589653,
+      "grad_norm": 1533.0765380859375,
+      "learning_rate": 8.1412556842068e-06,
+      "loss": 109.1279,
+      "step": 43980
+    },
+    {
+      "epoch": 0.36389957397526573,
+      "grad_norm": 784.7587890625,
+      "learning_rate": 8.140158278707637e-06,
+      "loss": 121.243,
+      "step": 43990
+    },
+    {
+      "epoch": 0.363982297224635,
+      "grad_norm": 886.728759765625,
+      "learning_rate": 8.139060623360494e-06,
+      "loss": 131.5882,
+      "step": 44000
+    },
+    {
+      "epoch": 0.36406502047400424,
+      "grad_norm": 990.3309936523438,
+      "learning_rate": 8.1379627182527e-06,
+      "loss": 95.6152,
+      "step": 44010
+    },
+    {
+      "epoch": 0.36414774372337344,
+      "grad_norm": 1236.8133544921875,
+      "learning_rate": 8.136864563471617e-06,
+      "loss": 104.638,
+      "step": 44020
+    },
+    {
+      "epoch": 0.3642304669727427,
+      "grad_norm": 560.0664672851562,
+      "learning_rate": 8.135766159104615e-06,
+      "loss": 90.362,
+      "step": 44030
+    },
+    {
+      "epoch": 0.36431319022211195,
+      "grad_norm": 796.653076171875,
+      "learning_rate": 8.134667505239092e-06,
+      "loss": 99.4829,
+      "step": 44040
+    },
+    {
+      "epoch": 0.36439591347148115,
+      "grad_norm": 777.6304321289062,
+      "learning_rate": 8.133568601962462e-06,
+      "loss": 106.3731,
+      "step": 44050
+    },
+    {
+      "epoch": 0.3644786367208504,
+      "grad_norm": 746.3777465820312,
+      "learning_rate": 8.132469449362158e-06,
+      "loss": 101.6638,
+      "step": 44060
+    },
+    {
+      "epoch": 0.36456135997021966,
+      "grad_norm": 824.9530029296875,
+      "learning_rate": 8.131370047525637e-06,
+      "loss": 144.5076,
+      "step": 44070
+    },
+    {
+      "epoch": 0.36464408321958885,
+      "grad_norm": 1749.48095703125,
+      "learning_rate": 8.130270396540372e-06,
+      "loss": 107.8925,
+      "step": 44080
+    },
+    {
+      "epoch": 0.3647268064689581,
+      "grad_norm": 1590.762451171875,
+      "learning_rate": 8.129170496493857e-06,
+      "loss": 129.4328,
+      "step": 44090
+    },
+    {
+      "epoch": 0.36480952971832736,
+      "grad_norm": 854.1180419921875,
+      "learning_rate": 8.128070347473609e-06,
+      "loss": 97.3397,
+      "step": 44100
+    },
+    {
+      "epoch": 0.36489225296769656,
+      "grad_norm": 920.0262451171875,
+      "learning_rate": 8.126969949567157e-06,
+      "loss": 90.8626,
+      "step": 44110
+    },
+    {
+      "epoch": 0.3649749762170658,
+      "grad_norm": 802.9644775390625,
+      "learning_rate": 8.125869302862058e-06,
+      "loss": 101.2598,
+      "step": 44120
+    },
+    {
+      "epoch": 0.365057699466435,
+      "grad_norm": 721.7289428710938,
+      "learning_rate": 8.124768407445883e-06,
+      "loss": 91.293,
+      "step": 44130
+    },
+    {
+      "epoch": 0.36514042271580427,
+      "grad_norm": 452.2812805175781,
+      "learning_rate": 8.123667263406228e-06,
+      "loss": 115.1237,
+      "step": 44140
+    },
+    {
+      "epoch": 0.3652231459651735,
+      "grad_norm": 755.5316162109375,
+      "learning_rate": 8.122565870830704e-06,
+      "loss": 95.5803,
+      "step": 44150
+    },
+    {
+      "epoch": 0.3653058692145427,
+      "grad_norm": 583.1863403320312,
+      "learning_rate": 8.121464229806944e-06,
+      "loss": 91.2347,
+      "step": 44160
+    },
+    {
+      "epoch": 0.365388592463912,
+      "grad_norm": 894.2572631835938,
+      "learning_rate": 8.120362340422601e-06,
+      "loss": 101.122,
+      "step": 44170
+    },
+    {
+      "epoch": 0.36547131571328123,
+      "grad_norm": 1000.3782958984375,
+      "learning_rate": 8.119260202765347e-06,
+      "loss": 104.094,
+      "step": 44180
+    },
+    {
+      "epoch": 0.36555403896265043,
+      "grad_norm": 1205.3262939453125,
+      "learning_rate": 8.118157816922874e-06,
+      "loss": 107.1193,
+      "step": 44190
+    },
+    {
+      "epoch": 0.3656367622120197,
+      "grad_norm": 1496.134033203125,
+      "learning_rate": 8.117055182982895e-06,
+      "loss": 84.7695,
+      "step": 44200
+    },
+    {
+      "epoch": 0.36571948546138894,
+      "grad_norm": 1437.298095703125,
+      "learning_rate": 8.115952301033141e-06,
+      "loss": 117.0865,
+      "step": 44210
+    },
+    {
+      "epoch": 0.36580220871075814,
+      "grad_norm": 1193.9093017578125,
+      "learning_rate": 8.11484917116136e-06,
+      "loss": 104.2912,
+      "step": 44220
+    },
+    {
+      "epoch": 0.3658849319601274,
+      "grad_norm": 912.3463745117188,
+      "learning_rate": 8.113745793455328e-06,
+      "loss": 105.9879,
+      "step": 44230
+    },
+    {
+      "epoch": 0.36596765520949665,
+      "grad_norm": 1175.63134765625,
+      "learning_rate": 8.112642168002831e-06,
+      "loss": 95.1146,
+      "step": 44240
+    },
+    {
+      "epoch": 0.36605037845886584,
+      "grad_norm": 957.2208862304688,
+      "learning_rate": 8.111538294891684e-06,
+      "loss": 133.3301,
+      "step": 44250
+    },
+    {
+      "epoch": 0.3661331017082351,
+      "grad_norm": 780.9307861328125,
+      "learning_rate": 8.110434174209714e-06,
+      "loss": 112.0869,
+      "step": 44260
+    },
+    {
+      "epoch": 0.36621582495760435,
+      "grad_norm": 1046.54052734375,
+      "learning_rate": 8.109329806044772e-06,
+      "loss": 122.4268,
+      "step": 44270
+    },
+    {
+      "epoch": 0.36629854820697355,
+      "grad_norm": 593.833984375,
+      "learning_rate": 8.108225190484728e-06,
+      "loss": 134.9322,
+      "step": 44280
+    },
+    {
+      "epoch": 0.3663812714563428,
+      "grad_norm": 1084.296630859375,
+      "learning_rate": 8.107120327617469e-06,
+      "loss": 108.1544,
+      "step": 44290
+    },
+    {
+      "epoch": 0.36646399470571206,
+      "grad_norm": 795.716552734375,
+      "learning_rate": 8.106015217530906e-06,
+      "loss": 104.3661,
+      "step": 44300
+    },
+    {
+      "epoch": 0.36654671795508126,
+      "grad_norm": 969.0634765625,
+      "learning_rate": 8.104909860312968e-06,
+      "loss": 118.4515,
+      "step": 44310
+    },
+    {
+      "epoch": 0.3666294412044505,
+      "grad_norm": 1363.8905029296875,
+      "learning_rate": 8.1038042560516e-06,
+      "loss": 109.1337,
+      "step": 44320
+    },
+    {
+      "epoch": 0.36671216445381977,
+      "grad_norm": 715.2791748046875,
+      "learning_rate": 8.102698404834773e-06,
+      "loss": 79.2838,
+      "step": 44330
+    },
+    {
+      "epoch": 0.36679488770318897,
+      "grad_norm": 1469.4854736328125,
+      "learning_rate": 8.101592306750472e-06,
+      "loss": 110.5569,
+      "step": 44340
+    },
+    {
+      "epoch": 0.3668776109525582,
+      "grad_norm": 992.7304077148438,
+      "learning_rate": 8.100485961886707e-06,
+      "loss": 97.823,
+      "step": 44350
+    },
+    {
+      "epoch": 0.3669603342019275,
+      "grad_norm": 1063.874755859375,
+      "learning_rate": 8.099379370331502e-06,
+      "loss": 112.1215,
+      "step": 44360
+    },
+    {
+      "epoch": 0.3670430574512967,
+      "grad_norm": 726.4131469726562,
+      "learning_rate": 8.098272532172906e-06,
+      "loss": 135.1273,
+      "step": 44370
+    },
+    {
+      "epoch": 0.3671257807006659,
+      "grad_norm": 1017.98193359375,
+      "learning_rate": 8.097165447498985e-06,
+      "loss": 102.8711,
+      "step": 44380
+    },
+    {
+      "epoch": 0.3672085039500352,
+      "grad_norm": 1006.7951049804688,
+      "learning_rate": 8.09605811639782e-06,
+      "loss": 120.6296,
+      "step": 44390
+    },
+    {
+      "epoch": 0.3672912271994044,
+      "grad_norm": 1553.7777099609375,
+      "learning_rate": 8.094950538957523e-06,
+      "loss": 116.9153,
+      "step": 44400
+    },
+    {
+      "epoch": 0.36737395044877363,
+      "grad_norm": 578.3179931640625,
+      "learning_rate": 8.093842715266214e-06,
+      "loss": 86.2257,
+      "step": 44410
+    },
+    {
+      "epoch": 0.3674566736981429,
+      "grad_norm": 1568.3966064453125,
+      "learning_rate": 8.092734645412037e-06,
+      "loss": 103.4828,
+      "step": 44420
+    },
+    {
+      "epoch": 0.3675393969475121,
+      "grad_norm": 1186.5718994140625,
+      "learning_rate": 8.09162632948316e-06,
+      "loss": 117.7764,
+      "step": 44430
+    },
+    {
+      "epoch": 0.36762212019688134,
+      "grad_norm": 1135.4630126953125,
+      "learning_rate": 8.090517767567765e-06,
+      "loss": 95.9603,
+      "step": 44440
+    },
+    {
+      "epoch": 0.3677048434462506,
+      "grad_norm": 959.4806518554688,
+      "learning_rate": 8.089408959754055e-06,
+      "loss": 99.0822,
+      "step": 44450
+    },
+    {
+      "epoch": 0.3677875666956198,
+      "grad_norm": 758.1735229492188,
+      "learning_rate": 8.088299906130252e-06,
+      "loss": 149.9401,
+      "step": 44460
+    },
+    {
+      "epoch": 0.36787028994498905,
+      "grad_norm": 791.1016845703125,
+      "learning_rate": 8.087190606784598e-06,
+      "loss": 79.0925,
+      "step": 44470
+    },
+    {
+      "epoch": 0.36795301319435825,
+      "grad_norm": 640.9752807617188,
+      "learning_rate": 8.086081061805357e-06,
+      "loss": 95.8233,
+      "step": 44480
+    },
+    {
+      "epoch": 0.3680357364437275,
+      "grad_norm": 1189.6053466796875,
+      "learning_rate": 8.084971271280808e-06,
+      "loss": 122.035,
+      "step": 44490
+    },
+    {
+      "epoch": 0.36811845969309676,
+      "grad_norm": 1168.3837890625,
+      "learning_rate": 8.083861235299253e-06,
+      "loss": 100.2019,
+      "step": 44500
+    },
+    {
+      "epoch": 0.36820118294246595,
+      "grad_norm": 891.6553955078125,
+      "learning_rate": 8.082750953949015e-06,
+      "loss": 115.5503,
+      "step": 44510
+    },
+    {
+      "epoch": 0.3682839061918352,
+      "grad_norm": 884.3746337890625,
+      "learning_rate": 8.081640427318429e-06,
+      "loss": 94.7234,
+      "step": 44520
+    },
+    {
+      "epoch": 0.36836662944120446,
+      "grad_norm": 760.7252807617188,
+      "learning_rate": 8.080529655495856e-06,
+      "loss": 126.0468,
+      "step": 44530
+    },
+    {
+      "epoch": 0.36844935269057366,
+      "grad_norm": 831.2738647460938,
+      "learning_rate": 8.079418638569679e-06,
+      "loss": 92.954,
+      "step": 44540
+    },
+    {
+      "epoch": 0.3685320759399429,
+      "grad_norm": 460.5909118652344,
+      "learning_rate": 8.078307376628292e-06,
+      "loss": 83.0053,
+      "step": 44550
+    },
+    {
+      "epoch": 0.36861479918931217,
+      "grad_norm": 795.8884887695312,
+      "learning_rate": 8.077195869760114e-06,
+      "loss": 86.7317,
+      "step": 44560
+    },
+    {
+      "epoch": 0.36869752243868137,
+      "grad_norm": 683.8246459960938,
+      "learning_rate": 8.076084118053584e-06,
+      "loss": 90.0042,
+      "step": 44570
+    },
+    {
+      "epoch": 0.3687802456880506,
+      "grad_norm": 1078.8155517578125,
+      "learning_rate": 8.074972121597158e-06,
+      "loss": 140.497,
+      "step": 44580
+    },
+    {
+      "epoch": 0.3688629689374199,
+      "grad_norm": 447.01214599609375,
+      "learning_rate": 8.073859880479314e-06,
+      "loss": 76.7604,
+      "step": 44590
+    },
+    {
+      "epoch": 0.3689456921867891,
+      "grad_norm": 948.76416015625,
+      "learning_rate": 8.072747394788545e-06,
+      "loss": 83.8023,
+      "step": 44600
+    },
+    {
+      "epoch": 0.36902841543615833,
+      "grad_norm": 1048.3646240234375,
+      "learning_rate": 8.071634664613367e-06,
+      "loss": 102.7217,
+      "step": 44610
+    },
+    {
+      "epoch": 0.3691111386855276,
+      "grad_norm": 859.9012451171875,
+      "learning_rate": 8.070521690042317e-06,
+      "loss": 103.1552,
+      "step": 44620
+    },
+    {
+      "epoch": 0.3691938619348968,
+      "grad_norm": 1066.72314453125,
+      "learning_rate": 8.069408471163947e-06,
+      "loss": 121.4919,
+      "step": 44630
+    },
+    {
+      "epoch": 0.36927658518426604,
+      "grad_norm": 969.5633544921875,
+      "learning_rate": 8.068295008066832e-06,
+      "loss": 128.1989,
+      "step": 44640
+    },
+    {
+      "epoch": 0.3693593084336353,
+      "grad_norm": 1198.6636962890625,
+      "learning_rate": 8.067181300839565e-06,
+      "loss": 161.2369,
+      "step": 44650
+    },
+    {
+      "epoch": 0.3694420316830045,
+      "grad_norm": 700.932861328125,
+      "learning_rate": 8.066067349570757e-06,
+      "loss": 86.4731,
+      "step": 44660
+    },
+    {
+      "epoch": 0.36952475493237374,
+      "grad_norm": 1984.68017578125,
+      "learning_rate": 8.064953154349042e-06,
+      "loss": 85.2434,
+      "step": 44670
+    },
+    {
+      "epoch": 0.369607478181743,
+      "grad_norm": 951.8640747070312,
+      "learning_rate": 8.063838715263072e-06,
+      "loss": 92.2627,
+      "step": 44680
+    },
+    {
+      "epoch": 0.3696902014311122,
+      "grad_norm": 991.0081176757812,
+      "learning_rate": 8.062724032401515e-06,
+      "loss": 82.2411,
+      "step": 44690
+    },
+    {
+      "epoch": 0.36977292468048145,
+      "grad_norm": 915.6588134765625,
+      "learning_rate": 8.061609105853062e-06,
+      "loss": 123.3313,
+      "step": 44700
+    },
+    {
+      "epoch": 0.3698556479298507,
+      "grad_norm": 648.7603759765625,
+      "learning_rate": 8.060493935706425e-06,
+      "loss": 86.3172,
+      "step": 44710
+    },
+    {
+      "epoch": 0.3699383711792199,
+      "grad_norm": 731.9179077148438,
+      "learning_rate": 8.059378522050332e-06,
+      "loss": 142.6297,
+      "step": 44720
+    },
+    {
+      "epoch": 0.37002109442858916,
+      "grad_norm": 738.2081909179688,
+      "learning_rate": 8.05826286497353e-06,
+      "loss": 100.2493,
+      "step": 44730
+    },
+    {
+      "epoch": 0.3701038176779584,
+      "grad_norm": 961.7689208984375,
+      "learning_rate": 8.057146964564786e-06,
+      "loss": 108.2104,
+      "step": 44740
+    },
+    {
+      "epoch": 0.3701865409273276,
+      "grad_norm": 1143.8248291015625,
+      "learning_rate": 8.05603082091289e-06,
+      "loss": 102.1352,
+      "step": 44750
+    },
+    {
+      "epoch": 0.37026926417669687,
+      "grad_norm": 968.6405639648438,
+      "learning_rate": 8.054914434106647e-06,
+      "loss": 89.4334,
+      "step": 44760
+    },
+    {
+      "epoch": 0.3703519874260661,
+      "grad_norm": 800.0516967773438,
+      "learning_rate": 8.053797804234882e-06,
+      "loss": 100.4505,
+      "step": 44770
+    },
+    {
+      "epoch": 0.3704347106754353,
+      "grad_norm": 928.8748168945312,
+      "learning_rate": 8.052680931386441e-06,
+      "loss": 88.3113,
+      "step": 44780
+    },
+    {
+      "epoch": 0.3705174339248046,
+      "grad_norm": 680.3980102539062,
+      "learning_rate": 8.051563815650187e-06,
+      "loss": 93.7541,
+      "step": 44790
+    },
+    {
+      "epoch": 0.3706001571741738,
+      "grad_norm": 988.8343505859375,
+      "learning_rate": 8.050446457115005e-06,
+      "loss": 138.652,
+      "step": 44800
+    },
+    {
+      "epoch": 0.370682880423543,
+      "grad_norm": 5196.66015625,
+      "learning_rate": 8.0493288558698e-06,
+      "loss": 161.6813,
+      "step": 44810
+    },
+    {
+      "epoch": 0.3707656036729123,
+      "grad_norm": 1195.2532958984375,
+      "learning_rate": 8.04821101200349e-06,
+      "loss": 104.2872,
+      "step": 44820
+    },
+    {
+      "epoch": 0.37084832692228153,
+      "grad_norm": 916.4719848632812,
+      "learning_rate": 8.047092925605022e-06,
+      "loss": 72.5952,
+      "step": 44830
+    },
+    {
+      "epoch": 0.37093105017165073,
+      "grad_norm": 1935.22119140625,
+      "learning_rate": 8.045974596763352e-06,
+      "loss": 124.3693,
+      "step": 44840
+    },
+    {
+      "epoch": 0.37101377342102,
+      "grad_norm": 660.0169677734375,
+      "learning_rate": 8.044856025567464e-06,
+      "loss": 118.5706,
+      "step": 44850
+    },
+    {
+      "epoch": 0.3710964966703892,
+      "grad_norm": 1358.344482421875,
+      "learning_rate": 8.043737212106356e-06,
+      "loss": 116.8551,
+      "step": 44860
+    },
+    {
+      "epoch": 0.37117921991975844,
+      "grad_norm": 963.0214233398438,
+      "learning_rate": 8.042618156469045e-06,
+      "loss": 88.5938,
+      "step": 44870
+    },
+    {
+      "epoch": 0.3712619431691277,
+      "grad_norm": 718.8231811523438,
+      "learning_rate": 8.041498858744572e-06,
+      "loss": 93.5438,
+      "step": 44880
+    },
+    {
+      "epoch": 0.3713446664184969,
+      "grad_norm": 884.6295776367188,
+      "learning_rate": 8.040379319021994e-06,
+      "loss": 98.0878,
+      "step": 44890
+    },
+    {
+      "epoch": 0.37142738966786615,
+      "grad_norm": 818.326171875,
+      "learning_rate": 8.039259537390388e-06,
+      "loss": 106.2791,
+      "step": 44900
+    },
+    {
+      "epoch": 0.3715101129172354,
+      "grad_norm": 722.222900390625,
+      "learning_rate": 8.038139513938847e-06,
+      "loss": 101.4262,
+      "step": 44910
+    },
+    {
+      "epoch": 0.3715928361666046,
+      "grad_norm": 706.7052001953125,
+      "learning_rate": 8.037019248756488e-06,
+      "loss": 103.0186,
+      "step": 44920
+    },
+    {
+      "epoch": 0.37167555941597386,
+      "grad_norm": 783.2863159179688,
+      "learning_rate": 8.035898741932447e-06,
+      "loss": 101.9469,
+      "step": 44930
+    },
+    {
+      "epoch": 0.3717582826653431,
+      "grad_norm": 1484.9461669921875,
+      "learning_rate": 8.034777993555875e-06,
+      "loss": 140.325,
+      "step": 44940
+    },
+    {
+      "epoch": 0.3718410059147123,
+      "grad_norm": 669.56640625,
+      "learning_rate": 8.033657003715945e-06,
+      "loss": 116.9738,
+      "step": 44950
+    },
+    {
+      "epoch": 0.37192372916408156,
+      "grad_norm": 705.1654663085938,
+      "learning_rate": 8.032535772501851e-06,
+      "loss": 102.4115,
+      "step": 44960
+    },
+    {
+      "epoch": 0.3720064524134508,
+      "grad_norm": 1263.1771240234375,
+      "learning_rate": 8.031414300002802e-06,
+      "loss": 77.7587,
+      "step": 44970
+    },
+    {
+      "epoch": 0.37208917566282,
+      "grad_norm": 1212.223876953125,
+      "learning_rate": 8.03029258630803e-06,
+      "loss": 112.9885,
+      "step": 44980
+    },
+    {
+      "epoch": 0.37217189891218927,
+      "grad_norm": 951.40185546875,
+      "learning_rate": 8.029170631506785e-06,
+      "loss": 95.1384,
+      "step": 44990
+    },
+    {
+      "epoch": 0.3722546221615585,
+      "grad_norm": 1134.679443359375,
+      "learning_rate": 8.028048435688333e-06,
+      "loss": 106.5822,
+      "step": 45000
+    },
+    {
+      "epoch": 0.3723373454109277,
+      "grad_norm": 932.267822265625,
+      "learning_rate": 8.026925998941965e-06,
+      "loss": 110.1278,
+      "step": 45010
+    },
+    {
+      "epoch": 0.372420068660297,
+      "grad_norm": 457.553955078125,
+      "learning_rate": 8.025803321356989e-06,
+      "loss": 95.6591,
+      "step": 45020
+    },
+    {
+      "epoch": 0.37250279190966623,
+      "grad_norm": 655.8544921875,
+      "learning_rate": 8.024680403022726e-06,
+      "loss": 89.4721,
+      "step": 45030
+    },
+    {
+      "epoch": 0.37258551515903543,
+      "grad_norm": 893.6047973632812,
+      "learning_rate": 8.023557244028526e-06,
+      "loss": 101.3761,
+      "step": 45040
+    },
+    {
+      "epoch": 0.3726682384084047,
+      "grad_norm": 765.0892333984375,
+      "learning_rate": 8.022433844463752e-06,
+      "loss": 85.2344,
+      "step": 45050
+    },
+    {
+      "epoch": 0.37275096165777394,
+      "grad_norm": 955.836181640625,
+      "learning_rate": 8.02131020441779e-06,
+      "loss": 105.1853,
+      "step": 45060
+    },
+    {
+      "epoch": 0.37283368490714314,
+      "grad_norm": 1159.584228515625,
+      "learning_rate": 8.02018632398004e-06,
+      "loss": 82.6549,
+      "step": 45070
+    },
+    {
+      "epoch": 0.3729164081565124,
+      "grad_norm": 827.6931762695312,
+      "learning_rate": 8.019062203239923e-06,
+      "loss": 145.6601,
+      "step": 45080
+    },
+    {
+      "epoch": 0.37299913140588165,
+      "grad_norm": 1674.2239990234375,
+      "learning_rate": 8.017937842286882e-06,
+      "loss": 104.1544,
+      "step": 45090
+    },
+    {
+      "epoch": 0.37308185465525084,
+      "grad_norm": 1010.4392700195312,
+      "learning_rate": 8.01681324121038e-06,
+      "loss": 102.7155,
+      "step": 45100
+    },
+    {
+      "epoch": 0.3731645779046201,
+      "grad_norm": 812.0460205078125,
+      "learning_rate": 8.015688400099893e-06,
+      "loss": 80.9767,
+      "step": 45110
+    },
+    {
+      "epoch": 0.37324730115398935,
+      "grad_norm": 676.9262084960938,
+      "learning_rate": 8.014563319044919e-06,
+      "loss": 107.2131,
+      "step": 45120
+    },
+    {
+      "epoch": 0.37333002440335855,
+      "grad_norm": 1148.935302734375,
+      "learning_rate": 8.013437998134978e-06,
+      "loss": 96.8717,
+      "step": 45130
+    },
+    {
+      "epoch": 0.3734127476527278,
+      "grad_norm": 2210.831298828125,
+      "learning_rate": 8.012312437459604e-06,
+      "loss": 115.9179,
+      "step": 45140
+    },
+    {
+      "epoch": 0.37349547090209706,
+      "grad_norm": 740.3560180664062,
+      "learning_rate": 8.011186637108354e-06,
+      "loss": 95.7675,
+      "step": 45150
+    },
+    {
+      "epoch": 0.37357819415146626,
+      "grad_norm": 1315.1302490234375,
+      "learning_rate": 8.010060597170805e-06,
+      "loss": 108.1328,
+      "step": 45160
+    },
+    {
+      "epoch": 0.3736609174008355,
+      "grad_norm": 1144.85400390625,
+      "learning_rate": 8.008934317736546e-06,
+      "loss": 104.8319,
+      "step": 45170
+    },
+    {
+      "epoch": 0.37374364065020477,
+      "grad_norm": 642.925048828125,
+      "learning_rate": 8.007807798895195e-06,
+      "loss": 114.467,
+      "step": 45180
+    },
+    {
+      "epoch": 0.37382636389957397,
+      "grad_norm": 1433.939208984375,
+      "learning_rate": 8.00668104073638e-06,
+      "loss": 142.9769,
+      "step": 45190
+    },
+    {
+      "epoch": 0.3739090871489432,
+      "grad_norm": 615.9909057617188,
+      "learning_rate": 8.005554043349753e-06,
+      "loss": 77.9182,
+      "step": 45200
+    },
+    {
+      "epoch": 0.3739918103983124,
+      "grad_norm": 1355.69287109375,
+      "learning_rate": 8.004426806824985e-06,
+      "loss": 98.0649,
+      "step": 45210
+    },
+    {
+      "epoch": 0.3740745336476817,
+      "grad_norm": 716.5189208984375,
+      "learning_rate": 8.003299331251764e-06,
+      "loss": 113.2009,
+      "step": 45220
+    },
+    {
+      "epoch": 0.3741572568970509,
+      "grad_norm": 860.948974609375,
+      "learning_rate": 8.002171616719798e-06,
+      "loss": 89.8221,
+      "step": 45230
+    },
+    {
+      "epoch": 0.3742399801464201,
+      "grad_norm": 1081.593994140625,
+      "learning_rate": 8.001043663318815e-06,
+      "loss": 148.4487,
+      "step": 45240
+    },
+    {
+      "epoch": 0.3743227033957894,
+      "grad_norm": 714.3857421875,
+      "learning_rate": 7.999915471138562e-06,
+      "loss": 111.338,
+      "step": 45250
+    },
+    {
+      "epoch": 0.37440542664515863,
+      "grad_norm": 863.04052734375,
+      "learning_rate": 7.9987870402688e-06,
+      "loss": 123.2091,
+      "step": 45260
+    },
+    {
+      "epoch": 0.37448814989452783,
+      "grad_norm": 541.889404296875,
+      "learning_rate": 7.997658370799318e-06,
+      "loss": 122.7542,
+      "step": 45270
+    },
+    {
+      "epoch": 0.3745708731438971,
+      "grad_norm": 948.584716796875,
+      "learning_rate": 7.996529462819915e-06,
+      "loss": 119.426,
+      "step": 45280
+    },
+    {
+      "epoch": 0.37465359639326634,
+      "grad_norm": 736.9681396484375,
+      "learning_rate": 7.995400316420416e-06,
+      "loss": 64.0782,
+      "step": 45290
+    },
+    {
+      "epoch": 0.37473631964263554,
+      "grad_norm": 1551.1883544921875,
+      "learning_rate": 7.994270931690662e-06,
+      "loss": 136.7847,
+      "step": 45300
+    },
+    {
+      "epoch": 0.3748190428920048,
+      "grad_norm": 804.4282836914062,
+      "learning_rate": 7.993141308720511e-06,
+      "loss": 136.6521,
+      "step": 45310
+    },
+    {
+      "epoch": 0.37490176614137405,
+      "grad_norm": 778.236083984375,
+      "learning_rate": 7.99201144759984e-06,
+      "loss": 95.8208,
+      "step": 45320
+    },
+    {
+      "epoch": 0.37498448939074325,
+      "grad_norm": 2299.00048828125,
+      "learning_rate": 7.990881348418554e-06,
+      "loss": 124.4509,
+      "step": 45330
+    },
+    {
+      "epoch": 0.3750672126401125,
+      "grad_norm": 973.8239135742188,
+      "learning_rate": 7.989751011266565e-06,
+      "loss": 112.5193,
+      "step": 45340
+    },
+    {
+      "epoch": 0.37514993588948176,
+      "grad_norm": 745.3870239257812,
+      "learning_rate": 7.988620436233806e-06,
+      "loss": 112.1299,
+      "step": 45350
+    },
+    {
+      "epoch": 0.37523265913885095,
+      "grad_norm": 2419.01953125,
+      "learning_rate": 7.987489623410236e-06,
+      "loss": 130.317,
+      "step": 45360
+    },
+    {
+      "epoch": 0.3753153823882202,
+      "grad_norm": 2003.9739990234375,
+      "learning_rate": 7.986358572885828e-06,
+      "loss": 118.5811,
+      "step": 45370
+    },
+    {
+      "epoch": 0.37539810563758946,
+      "grad_norm": 1251.0047607421875,
+      "learning_rate": 7.985227284750574e-06,
+      "loss": 106.9141,
+      "step": 45380
+    },
+    {
+      "epoch": 0.37548082888695866,
+      "grad_norm": 879.7947998046875,
+      "learning_rate": 7.984095759094485e-06,
+      "loss": 93.4843,
+      "step": 45390
+    },
+    {
+      "epoch": 0.3755635521363279,
+      "grad_norm": 852.9107055664062,
+      "learning_rate": 7.982963996007591e-06,
+      "loss": 141.075,
+      "step": 45400
+    },
+    {
+      "epoch": 0.37564627538569717,
+      "grad_norm": 952.68798828125,
+      "learning_rate": 7.981831995579943e-06,
+      "loss": 91.4658,
+      "step": 45410
+    },
+    {
+      "epoch": 0.37572899863506637,
+      "grad_norm": 1191.1885986328125,
+      "learning_rate": 7.980699757901607e-06,
+      "loss": 80.4354,
+      "step": 45420
+    },
+    {
+      "epoch": 0.3758117218844356,
+      "grad_norm": 813.9268188476562,
+      "learning_rate": 7.97956728306267e-06,
+      "loss": 90.7331,
+      "step": 45430
+    },
+    {
+      "epoch": 0.3758944451338049,
+      "grad_norm": 697.1983032226562,
+      "learning_rate": 7.97843457115324e-06,
+      "loss": 98.3303,
+      "step": 45440
+    },
+    {
+      "epoch": 0.3759771683831741,
+      "grad_norm": 628.31982421875,
+      "learning_rate": 7.97730162226344e-06,
+      "loss": 103.7881,
+      "step": 45450
+    },
+    {
+      "epoch": 0.37605989163254333,
+      "grad_norm": 2688.026123046875,
+      "learning_rate": 7.976168436483415e-06,
+      "loss": 97.9229,
+      "step": 45460
+    },
+    {
+      "epoch": 0.3761426148819126,
+      "grad_norm": 813.2411499023438,
+      "learning_rate": 7.975035013903326e-06,
+      "loss": 94.9314,
+      "step": 45470
+    },
+    {
+      "epoch": 0.3762253381312818,
+      "grad_norm": 712.8557739257812,
+      "learning_rate": 7.973901354613353e-06,
+      "loss": 85.9302,
+      "step": 45480
+    },
+    {
+      "epoch": 0.37630806138065104,
+      "grad_norm": 1530.84033203125,
+      "learning_rate": 7.972767458703697e-06,
+      "loss": 128.0701,
+      "step": 45490
+    },
+    {
+      "epoch": 0.3763907846300203,
+      "grad_norm": 1192.7733154296875,
+      "learning_rate": 7.971633326264581e-06,
+      "loss": 99.9536,
+      "step": 45500
+    },
+    {
+      "epoch": 0.3764735078793895,
+      "grad_norm": 854.33203125,
+      "learning_rate": 7.970498957386237e-06,
+      "loss": 121.3529,
+      "step": 45510
+    },
+    {
+      "epoch": 0.37655623112875874,
+      "grad_norm": 673.5110473632812,
+      "learning_rate": 7.969364352158922e-06,
+      "loss": 102.1387,
+      "step": 45520
+    },
+    {
+      "epoch": 0.376638954378128,
+      "grad_norm": 868.6480712890625,
+      "learning_rate": 7.968229510672915e-06,
+      "loss": 97.5747,
+      "step": 45530
+    },
+    {
+      "epoch": 0.3767216776274972,
+      "grad_norm": 1399.73779296875,
+      "learning_rate": 7.967094433018508e-06,
+      "loss": 125.0937,
+      "step": 45540
+    },
+    {
+      "epoch": 0.37680440087686645,
+      "grad_norm": 1239.8349609375,
+      "learning_rate": 7.965959119286013e-06,
+      "loss": 113.5951,
+      "step": 45550
+    },
+    {
+      "epoch": 0.37688712412623565,
+      "grad_norm": 1359.953125,
+      "learning_rate": 7.964823569565765e-06,
+      "loss": 103.3041,
+      "step": 45560
+    },
+    {
+      "epoch": 0.3769698473756049,
+      "grad_norm": 720.0075073242188,
+      "learning_rate": 7.963687783948111e-06,
+      "loss": 81.487,
+      "step": 45570
+    },
+    {
+      "epoch": 0.37705257062497416,
+      "grad_norm": 1454.417724609375,
+      "learning_rate": 7.96255176252342e-06,
+      "loss": 83.4532,
+      "step": 45580
+    },
+    {
+      "epoch": 0.37713529387434336,
+      "grad_norm": 1103.7496337890625,
+      "learning_rate": 7.961415505382083e-06,
+      "loss": 125.8114,
+      "step": 45590
+    },
+    {
+      "epoch": 0.3772180171237126,
+      "grad_norm": 948.4056396484375,
+      "learning_rate": 7.960279012614508e-06,
+      "loss": 90.2628,
+      "step": 45600
+    },
+    {
+      "epoch": 0.37730074037308187,
+      "grad_norm": 1128.2615966796875,
+      "learning_rate": 7.959142284311115e-06,
+      "loss": 113.3847,
+      "step": 45610
+    },
+    {
+      "epoch": 0.37738346362245107,
+      "grad_norm": 921.497802734375,
+      "learning_rate": 7.958005320562349e-06,
+      "loss": 125.8096,
+      "step": 45620
+    },
+    {
+      "epoch": 0.3774661868718203,
+      "grad_norm": 1018.8650512695312,
+      "learning_rate": 7.95686812145868e-06,
+      "loss": 122.7865,
+      "step": 45630
+    },
+    {
+      "epoch": 0.3775489101211896,
+      "grad_norm": 979.5180053710938,
+      "learning_rate": 7.955730687090582e-06,
+      "loss": 125.0732,
+      "step": 45640
+    },
+    {
+      "epoch": 0.3776316333705588,
+      "grad_norm": 978.5775756835938,
+      "learning_rate": 7.954593017548557e-06,
+      "loss": 87.2099,
+      "step": 45650
+    },
+    {
+      "epoch": 0.377714356619928,
+      "grad_norm": 978.8362426757812,
+      "learning_rate": 7.953455112923127e-06,
+      "loss": 134.838,
+      "step": 45660
+    },
+    {
+      "epoch": 0.3777970798692973,
+      "grad_norm": 961.3007202148438,
+      "learning_rate": 7.952316973304828e-06,
+      "loss": 109.7581,
+      "step": 45670
+    },
+    {
+      "epoch": 0.3778798031186665,
+      "grad_norm": 872.6028442382812,
+      "learning_rate": 7.951178598784217e-06,
+      "loss": 94.058,
+      "step": 45680
+    },
+    {
+      "epoch": 0.37796252636803573,
+      "grad_norm": 603.7721557617188,
+      "learning_rate": 7.950039989451868e-06,
+      "loss": 142.0334,
+      "step": 45690
+    },
+    {
+      "epoch": 0.378045249617405,
+      "grad_norm": 769.1307373046875,
+      "learning_rate": 7.948901145398376e-06,
+      "loss": 101.5725,
+      "step": 45700
+    },
+    {
+      "epoch": 0.3781279728667742,
+      "grad_norm": 523.3740844726562,
+      "learning_rate": 7.947762066714353e-06,
+      "loss": 99.7676,
+      "step": 45710
+    },
+    {
+      "epoch": 0.37821069611614344,
+      "grad_norm": 569.120361328125,
+      "learning_rate": 7.946622753490433e-06,
+      "loss": 73.9602,
+      "step": 45720
+    },
+    {
+      "epoch": 0.3782934193655127,
+      "grad_norm": 850.3909912109375,
+      "learning_rate": 7.945483205817262e-06,
+      "loss": 99.4327,
+      "step": 45730
+    },
+    {
+      "epoch": 0.3783761426148819,
+      "grad_norm": 1302.0482177734375,
+      "learning_rate": 7.94434342378551e-06,
+      "loss": 132.1461,
+      "step": 45740
+    },
+    {
+      "epoch": 0.37845886586425115,
+      "grad_norm": 698.3201293945312,
+      "learning_rate": 7.943203407485864e-06,
+      "loss": 69.9039,
+      "step": 45750
+    },
+    {
+      "epoch": 0.3785415891136204,
+      "grad_norm": 911.8252563476562,
+      "learning_rate": 7.942063157009033e-06,
+      "loss": 102.4791,
+      "step": 45760
+    },
+    {
+      "epoch": 0.3786243123629896,
+      "grad_norm": 991.9946899414062,
+      "learning_rate": 7.940922672445737e-06,
+      "loss": 113.1581,
+      "step": 45770
+    },
+    {
+      "epoch": 0.37870703561235886,
+      "grad_norm": 1038.773681640625,
+      "learning_rate": 7.939781953886722e-06,
+      "loss": 128.2211,
+      "step": 45780
+    },
+    {
+      "epoch": 0.3787897588617281,
+      "grad_norm": 1063.1624755859375,
+      "learning_rate": 7.938641001422747e-06,
+      "loss": 106.1225,
+      "step": 45790
+    },
+    {
+      "epoch": 0.3788724821110973,
+      "grad_norm": 897.7556762695312,
+      "learning_rate": 7.937499815144597e-06,
+      "loss": 102.7701,
+      "step": 45800
+    },
+    {
+      "epoch": 0.37895520536046656,
+      "grad_norm": 454.84552001953125,
+      "learning_rate": 7.936358395143065e-06,
+      "loss": 133.0522,
+      "step": 45810
+    },
+    {
+      "epoch": 0.3790379286098358,
+      "grad_norm": 491.5146484375,
+      "learning_rate": 7.935216741508971e-06,
+      "loss": 131.0702,
+      "step": 45820
+    },
+    {
+      "epoch": 0.379120651859205,
+      "grad_norm": 816.953125,
+      "learning_rate": 7.934074854333153e-06,
+      "loss": 102.6913,
+      "step": 45830
+    },
+    {
+      "epoch": 0.37920337510857427,
+      "grad_norm": 991.1273193359375,
+      "learning_rate": 7.932932733706467e-06,
+      "loss": 96.2913,
+      "step": 45840
+    },
+    {
+      "epoch": 0.3792860983579435,
+      "grad_norm": 2715.989501953125,
+      "learning_rate": 7.931790379719781e-06,
+      "loss": 145.7241,
+      "step": 45850
+    },
+    {
+      "epoch": 0.3793688216073127,
+      "grad_norm": 733.826416015625,
+      "learning_rate": 7.93064779246399e-06,
+      "loss": 79.5076,
+      "step": 45860
+    },
+    {
+      "epoch": 0.379451544856682,
+      "grad_norm": 1925.5067138671875,
+      "learning_rate": 7.929504972030003e-06,
+      "loss": 129.3025,
+      "step": 45870
+    },
+    {
+      "epoch": 0.37953426810605123,
+      "grad_norm": 723.109130859375,
+      "learning_rate": 7.928361918508752e-06,
+      "loss": 124.5502,
+      "step": 45880
+    },
+    {
+      "epoch": 0.37961699135542043,
+      "grad_norm": 942.7910766601562,
+      "learning_rate": 7.927218631991182e-06,
+      "loss": 87.0695,
+      "step": 45890
+    },
+    {
+      "epoch": 0.3796997146047897,
+      "grad_norm": 1327.8336181640625,
+      "learning_rate": 7.92607511256826e-06,
+      "loss": 90.3665,
+      "step": 45900
+    },
+    {
+      "epoch": 0.37978243785415894,
+      "grad_norm": 1204.0654296875,
+      "learning_rate": 7.924931360330968e-06,
+      "loss": 120.6505,
+      "step": 45910
+    },
+    {
+      "epoch": 0.37986516110352814,
+      "grad_norm": 643.5101928710938,
+      "learning_rate": 7.92378737537031e-06,
+      "loss": 84.0929,
+      "step": 45920
+    },
+    {
+      "epoch": 0.3799478843528974,
+      "grad_norm": 1710.7091064453125,
+      "learning_rate": 7.922643157777314e-06,
+      "loss": 94.1093,
+      "step": 45930
+    },
+    {
+      "epoch": 0.3800306076022666,
+      "grad_norm": 1306.8468017578125,
+      "learning_rate": 7.921498707643011e-06,
+      "loss": 110.1609,
+      "step": 45940
+    },
+    {
+      "epoch": 0.38011333085163584,
+      "grad_norm": 1266.775146484375,
+      "learning_rate": 7.920354025058467e-06,
+      "loss": 94.9035,
+      "step": 45950
+    },
+    {
+      "epoch": 0.3801960541010051,
+      "grad_norm": 1168.3138427734375,
+      "learning_rate": 7.919209110114752e-06,
+      "loss": 165.2224,
+      "step": 45960
+    },
+    {
+      "epoch": 0.3802787773503743,
+      "grad_norm": 760.196044921875,
+      "learning_rate": 7.918063962902968e-06,
+      "loss": 118.3697,
+      "step": 45970
+    },
+    {
+      "epoch": 0.38036150059974355,
+      "grad_norm": 1145.4422607421875,
+      "learning_rate": 7.916918583514227e-06,
+      "loss": 102.1873,
+      "step": 45980
+    },
+    {
+      "epoch": 0.3804442238491128,
+      "grad_norm": 523.3143310546875,
+      "learning_rate": 7.91577297203966e-06,
+      "loss": 94.5064,
+      "step": 45990
+    },
+    {
+      "epoch": 0.380526947098482,
+      "grad_norm": 777.9324951171875,
+      "learning_rate": 7.91462712857042e-06,
+      "loss": 99.4327,
+      "step": 46000
+    },
+    {
+      "epoch": 0.38060967034785126,
+      "grad_norm": 1038.9837646484375,
+      "learning_rate": 7.913481053197673e-06,
+      "loss": 120.6933,
+      "step": 46010
+    },
+    {
+      "epoch": 0.3806923935972205,
+      "grad_norm": 702.442138671875,
+      "learning_rate": 7.912334746012613e-06,
+      "loss": 141.1689,
+      "step": 46020
+    },
+    {
+      "epoch": 0.3807751168465897,
+      "grad_norm": 910.6521606445312,
+      "learning_rate": 7.911188207106442e-06,
+      "loss": 84.091,
+      "step": 46030
+    },
+    {
+      "epoch": 0.38085784009595897,
+      "grad_norm": 836.0477905273438,
+      "learning_rate": 7.910041436570386e-06,
+      "loss": 92.7803,
+      "step": 46040
+    },
+    {
+      "epoch": 0.3809405633453282,
+      "grad_norm": 796.7591552734375,
+      "learning_rate": 7.90889443449569e-06,
+      "loss": 93.9601,
+      "step": 46050
+    },
+    {
+      "epoch": 0.3810232865946974,
+      "grad_norm": 1023.9012451171875,
+      "learning_rate": 7.90774720097361e-06,
+      "loss": 125.3338,
+      "step": 46060
+    },
+    {
+      "epoch": 0.3811060098440667,
+      "grad_norm": 1169.42822265625,
+      "learning_rate": 7.906599736095433e-06,
+      "loss": 89.7407,
+      "step": 46070
+    },
+    {
+      "epoch": 0.3811887330934359,
+      "grad_norm": 924.0548706054688,
+      "learning_rate": 7.905452039952453e-06,
+      "loss": 86.3304,
+      "step": 46080
+    },
+    {
+      "epoch": 0.3812714563428051,
+      "grad_norm": 595.2965087890625,
+      "learning_rate": 7.904304112635987e-06,
+      "loss": 106.9174,
+      "step": 46090
+    },
+    {
+      "epoch": 0.3813541795921744,
+      "grad_norm": 472.991943359375,
+      "learning_rate": 7.903155954237375e-06,
+      "loss": 120.8275,
+      "step": 46100
+    },
+    {
+      "epoch": 0.38143690284154363,
+      "grad_norm": 1009.1423950195312,
+      "learning_rate": 7.902007564847967e-06,
+      "loss": 105.188,
+      "step": 46110
+    },
+    {
+      "epoch": 0.38151962609091283,
+      "grad_norm": 851.7573852539062,
+      "learning_rate": 7.900858944559133e-06,
+      "loss": 111.3603,
+      "step": 46120
+    },
+    {
+      "epoch": 0.3816023493402821,
+      "grad_norm": 981.0646362304688,
+      "learning_rate": 7.899710093462267e-06,
+      "loss": 86.6864,
+      "step": 46130
+    },
+    {
+      "epoch": 0.38168507258965134,
+      "grad_norm": 864.4039306640625,
+      "learning_rate": 7.898561011648777e-06,
+      "loss": 155.4255,
+      "step": 46140
+    },
+    {
+      "epoch": 0.38176779583902054,
+      "grad_norm": 1161.7957763671875,
+      "learning_rate": 7.89741169921009e-06,
+      "loss": 92.5417,
+      "step": 46150
+    },
+    {
+      "epoch": 0.3818505190883898,
+      "grad_norm": 1559.1328125,
+      "learning_rate": 7.896262156237652e-06,
+      "loss": 109.0317,
+      "step": 46160
+    },
+    {
+      "epoch": 0.38193324233775905,
+      "grad_norm": 839.4773559570312,
+      "learning_rate": 7.895112382822925e-06,
+      "loss": 124.5884,
+      "step": 46170
+    },
+    {
+      "epoch": 0.38201596558712825,
+      "grad_norm": 1135.74658203125,
+      "learning_rate": 7.893962379057393e-06,
+      "loss": 115.562,
+      "step": 46180
+    },
+    {
+      "epoch": 0.3820986888364975,
+      "grad_norm": 979.9491577148438,
+      "learning_rate": 7.892812145032557e-06,
+      "loss": 118.3164,
+      "step": 46190
+    },
+    {
+      "epoch": 0.38218141208586676,
+      "grad_norm": 4294.3427734375,
+      "learning_rate": 7.891661680839932e-06,
+      "loss": 125.0205,
+      "step": 46200
+    },
+    {
+      "epoch": 0.38226413533523596,
+      "grad_norm": 933.6452026367188,
+      "learning_rate": 7.89051098657106e-06,
+      "loss": 121.8078,
+      "step": 46210
+    },
+    {
+      "epoch": 0.3823468585846052,
+      "grad_norm": 527.7791137695312,
+      "learning_rate": 7.889360062317495e-06,
+      "loss": 96.7531,
+      "step": 46220
+    },
+    {
+      "epoch": 0.38242958183397446,
+      "grad_norm": 777.3706665039062,
+      "learning_rate": 7.888208908170812e-06,
+      "loss": 134.1928,
+      "step": 46230
+    },
+    {
+      "epoch": 0.38251230508334366,
+      "grad_norm": 805.9109497070312,
+      "learning_rate": 7.887057524222596e-06,
+      "loss": 101.3832,
+      "step": 46240
+    },
+    {
+      "epoch": 0.3825950283327129,
+      "grad_norm": 1473.904296875,
+      "learning_rate": 7.885905910564466e-06,
+      "loss": 87.6336,
+      "step": 46250
+    },
+    {
+      "epoch": 0.38267775158208217,
+      "grad_norm": 784.258056640625,
+      "learning_rate": 7.884754067288047e-06,
+      "loss": 94.394,
+      "step": 46260
+    },
+    {
+      "epoch": 0.38276047483145137,
+      "grad_norm": 1091.794677734375,
+      "learning_rate": 7.883601994484986e-06,
+      "loss": 106.1401,
+      "step": 46270
+    },
+    {
+      "epoch": 0.3828431980808206,
+      "grad_norm": 1147.5867919921875,
+      "learning_rate": 7.882449692246948e-06,
+      "loss": 104.8446,
+      "step": 46280
+    },
+    {
+      "epoch": 0.3829259213301898,
+      "grad_norm": 1361.256591796875,
+      "learning_rate": 7.881297160665616e-06,
+      "loss": 97.2336,
+      "step": 46290
+    },
+    {
+      "epoch": 0.3830086445795591,
+      "grad_norm": 1118.20556640625,
+      "learning_rate": 7.880144399832693e-06,
+      "loss": 109.2898,
+      "step": 46300
+    },
+    {
+      "epoch": 0.38309136782892833,
+      "grad_norm": 1180.1666259765625,
+      "learning_rate": 7.878991409839897e-06,
+      "loss": 107.2897,
+      "step": 46310
+    },
+    {
+      "epoch": 0.38317409107829753,
+      "grad_norm": 976.9520874023438,
+      "learning_rate": 7.87783819077897e-06,
+      "loss": 127.2354,
+      "step": 46320
+    },
+    {
+      "epoch": 0.3832568143276668,
+      "grad_norm": 569.0108032226562,
+      "learning_rate": 7.876684742741665e-06,
+      "loss": 89.219,
+      "step": 46330
+    },
+    {
+      "epoch": 0.38333953757703604,
+      "grad_norm": 1490.4720458984375,
+      "learning_rate": 7.875531065819755e-06,
+      "loss": 88.2369,
+      "step": 46340
+    },
+    {
+      "epoch": 0.38342226082640524,
+      "grad_norm": 1101.564697265625,
+      "learning_rate": 7.874377160105037e-06,
+      "loss": 104.7651,
+      "step": 46350
+    },
+    {
+      "epoch": 0.3835049840757745,
+      "grad_norm": 1799.0706787109375,
+      "learning_rate": 7.873223025689319e-06,
+      "loss": 138.4782,
+      "step": 46360
+    },
+    {
+      "epoch": 0.38358770732514375,
+      "grad_norm": 1268.860595703125,
+      "learning_rate": 7.872068662664432e-06,
+      "loss": 91.4783,
+      "step": 46370
+    },
+    {
+      "epoch": 0.38367043057451294,
+      "grad_norm": 1431.305419921875,
+      "learning_rate": 7.870914071122222e-06,
+      "loss": 100.8851,
+      "step": 46380
+    },
+    {
+      "epoch": 0.3837531538238822,
+      "grad_norm": 1034.6007080078125,
+      "learning_rate": 7.869759251154554e-06,
+      "loss": 94.5458,
+      "step": 46390
+    },
+    {
+      "epoch": 0.38383587707325145,
+      "grad_norm": 1329.562255859375,
+      "learning_rate": 7.868604202853314e-06,
+      "loss": 95.9945,
+      "step": 46400
+    },
+    {
+      "epoch": 0.38391860032262065,
+      "grad_norm": 755.7239379882812,
+      "learning_rate": 7.867448926310403e-06,
+      "loss": 106.9036,
+      "step": 46410
+    },
+    {
+      "epoch": 0.3840013235719899,
+      "grad_norm": 1034.0926513671875,
+      "learning_rate": 7.866293421617741e-06,
+      "loss": 94.1212,
+      "step": 46420
+    },
+    {
+      "epoch": 0.38408404682135916,
+      "grad_norm": 1169.0289306640625,
+      "learning_rate": 7.865137688867264e-06,
+      "loss": 76.2746,
+      "step": 46430
+    },
+    {
+      "epoch": 0.38416677007072836,
+      "grad_norm": 1374.6610107421875,
+      "learning_rate": 7.86398172815093e-06,
+      "loss": 141.711,
+      "step": 46440
+    },
+    {
+      "epoch": 0.3842494933200976,
+      "grad_norm": 715.2680053710938,
+      "learning_rate": 7.862825539560716e-06,
+      "loss": 127.8841,
+      "step": 46450
+    },
+    {
+      "epoch": 0.38433221656946687,
+      "grad_norm": 1387.7430419921875,
+      "learning_rate": 7.861669123188613e-06,
+      "loss": 124.0174,
+      "step": 46460
+    },
+    {
+      "epoch": 0.38441493981883607,
+      "grad_norm": 826.4485473632812,
+      "learning_rate": 7.86051247912663e-06,
+      "loss": 110.8221,
+      "step": 46470
+    },
+    {
+      "epoch": 0.3844976630682053,
+      "grad_norm": 807.095703125,
+      "learning_rate": 7.859355607466797e-06,
+      "loss": 87.1465,
+      "step": 46480
+    },
+    {
+      "epoch": 0.3845803863175746,
+      "grad_norm": 1193.6893310546875,
+      "learning_rate": 7.858198508301161e-06,
+      "loss": 133.6182,
+      "step": 46490
+    },
+    {
+      "epoch": 0.3846631095669438,
+      "grad_norm": 527.4124755859375,
+      "learning_rate": 7.857041181721788e-06,
+      "loss": 97.8456,
+      "step": 46500
+    },
+    {
+      "epoch": 0.384745832816313,
+      "grad_norm": 862.8912963867188,
+      "learning_rate": 7.855883627820757e-06,
+      "loss": 134.22,
+      "step": 46510
+    },
+    {
+      "epoch": 0.3848285560656823,
+      "grad_norm": 548.6507568359375,
+      "learning_rate": 7.854725846690175e-06,
+      "loss": 83.1777,
+      "step": 46520
+    },
+    {
+      "epoch": 0.3849112793150515,
+      "grad_norm": 902.584228515625,
+      "learning_rate": 7.85356783842216e-06,
+      "loss": 106.9863,
+      "step": 46530
+    },
+    {
+      "epoch": 0.38499400256442073,
+      "grad_norm": 535.5631103515625,
+      "learning_rate": 7.852409603108845e-06,
+      "loss": 113.9048,
+      "step": 46540
+    },
+    {
+      "epoch": 0.38507672581379,
+      "grad_norm": 938.954345703125,
+      "learning_rate": 7.85125114084239e-06,
+      "loss": 105.1793,
+      "step": 46550
+    },
+    {
+      "epoch": 0.3851594490631592,
+      "grad_norm": 1107.5294189453125,
+      "learning_rate": 7.850092451714967e-06,
+      "loss": 115.7462,
+      "step": 46560
+    },
+    {
+      "epoch": 0.38524217231252844,
+      "grad_norm": 1338.5953369140625,
+      "learning_rate": 7.84893353581877e-06,
+      "loss": 93.1394,
+      "step": 46570
+    },
+    {
+      "epoch": 0.3853248955618977,
+      "grad_norm": 1115.7308349609375,
+      "learning_rate": 7.847774393246005e-06,
+      "loss": 117.0743,
+      "step": 46580
+    },
+    {
+      "epoch": 0.3854076188112669,
+      "grad_norm": 712.0968017578125,
+      "learning_rate": 7.8466150240889e-06,
+      "loss": 92.4161,
+      "step": 46590
+    },
+    {
+      "epoch": 0.38549034206063615,
+      "grad_norm": 1130.7666015625,
+      "learning_rate": 7.845455428439703e-06,
+      "loss": 90.2969,
+      "step": 46600
+    },
+    {
+      "epoch": 0.3855730653100054,
+      "grad_norm": 694.2098388671875,
+      "learning_rate": 7.844295606390675e-06,
+      "loss": 106.4844,
+      "step": 46610
+    },
+    {
+      "epoch": 0.3856557885593746,
+      "grad_norm": 758.242431640625,
+      "learning_rate": 7.843135558034101e-06,
+      "loss": 85.5782,
+      "step": 46620
+    },
+    {
+      "epoch": 0.38573851180874386,
+      "grad_norm": 847.0222778320312,
+      "learning_rate": 7.841975283462278e-06,
+      "loss": 83.5226,
+      "step": 46630
+    },
+    {
+      "epoch": 0.3858212350581131,
+      "grad_norm": 828.2359619140625,
+      "learning_rate": 7.840814782767525e-06,
+      "loss": 74.5017,
+      "step": 46640
+    },
+    {
+      "epoch": 0.3859039583074823,
+      "grad_norm": 575.1474609375,
+      "learning_rate": 7.839654056042176e-06,
+      "loss": 84.7638,
+      "step": 46650
+    },
+    {
+      "epoch": 0.38598668155685156,
+      "grad_norm": 961.8069458007812,
+      "learning_rate": 7.838493103378588e-06,
+      "loss": 105.9289,
+      "step": 46660
+    },
+    {
+      "epoch": 0.38606940480622076,
+      "grad_norm": 323.90765380859375,
+      "learning_rate": 7.83733192486913e-06,
+      "loss": 134.9893,
+      "step": 46670
+    },
+    {
+      "epoch": 0.38615212805559,
+      "grad_norm": 906.62060546875,
+      "learning_rate": 7.836170520606191e-06,
+      "loss": 132.9467,
+      "step": 46680
+    },
+    {
+      "epoch": 0.38623485130495927,
+      "grad_norm": 1547.6925048828125,
+      "learning_rate": 7.83500889068218e-06,
+      "loss": 145.7491,
+      "step": 46690
+    },
+    {
+      "epoch": 0.38631757455432847,
+      "grad_norm": 1164.046142578125,
+      "learning_rate": 7.833847035189524e-06,
+      "loss": 110.9907,
+      "step": 46700
+    },
+    {
+      "epoch": 0.3864002978036977,
+      "grad_norm": 1105.657470703125,
+      "learning_rate": 7.832684954220664e-06,
+      "loss": 124.7499,
+      "step": 46710
+    },
+    {
+      "epoch": 0.386483021053067,
+      "grad_norm": 508.67803955078125,
+      "learning_rate": 7.831522647868064e-06,
+      "loss": 95.8649,
+      "step": 46720
+    },
+    {
+      "epoch": 0.3865657443024362,
+      "grad_norm": 674.171875,
+      "learning_rate": 7.8303601162242e-06,
+      "loss": 99.7405,
+      "step": 46730
+    },
+    {
+      "epoch": 0.38664846755180543,
+      "grad_norm": 1009.1093139648438,
+      "learning_rate": 7.829197359381571e-06,
+      "loss": 111.1325,
+      "step": 46740
+    },
+    {
+      "epoch": 0.3867311908011747,
+      "grad_norm": 1165.2833251953125,
+      "learning_rate": 7.828034377432694e-06,
+      "loss": 111.5451,
+      "step": 46750
+    },
+    {
+      "epoch": 0.3868139140505439,
+      "grad_norm": 542.6980590820312,
+      "learning_rate": 7.826871170470099e-06,
+      "loss": 93.1879,
+      "step": 46760
+    },
+    {
+      "epoch": 0.38689663729991314,
+      "grad_norm": 1133.044921875,
+      "learning_rate": 7.82570773858634e-06,
+      "loss": 90.8988,
+      "step": 46770
+    },
+    {
+      "epoch": 0.3869793605492824,
+      "grad_norm": 684.3965454101562,
+      "learning_rate": 7.824544081873984e-06,
+      "loss": 79.4068,
+      "step": 46780
+    },
+    {
+      "epoch": 0.3870620837986516,
+      "grad_norm": 1281.01513671875,
+      "learning_rate": 7.823380200425618e-06,
+      "loss": 132.8948,
+      "step": 46790
+    },
+    {
+      "epoch": 0.38714480704802084,
+      "grad_norm": 1106.5760498046875,
+      "learning_rate": 7.822216094333847e-06,
+      "loss": 109.8843,
+      "step": 46800
+    },
+    {
+      "epoch": 0.3872275302973901,
+      "grad_norm": 736.9039306640625,
+      "learning_rate": 7.821051763691293e-06,
+      "loss": 95.4937,
+      "step": 46810
+    },
+    {
+      "epoch": 0.3873102535467593,
+      "grad_norm": 983.0117797851562,
+      "learning_rate": 7.819887208590597e-06,
+      "loss": 106.8212,
+      "step": 46820
+    },
+    {
+      "epoch": 0.38739297679612855,
+      "grad_norm": 900.6527099609375,
+      "learning_rate": 7.818722429124418e-06,
+      "loss": 97.8815,
+      "step": 46830
+    },
+    {
+      "epoch": 0.3874757000454978,
+      "grad_norm": 850.8548583984375,
+      "learning_rate": 7.817557425385433e-06,
+      "loss": 81.9495,
+      "step": 46840
+    },
+    {
+      "epoch": 0.387558423294867,
+      "grad_norm": 855.3139038085938,
+      "learning_rate": 7.816392197466333e-06,
+      "loss": 89.8393,
+      "step": 46850
+    },
+    {
+      "epoch": 0.38764114654423626,
+      "grad_norm": 587.1494750976562,
+      "learning_rate": 7.815226745459831e-06,
+      "loss": 106.3622,
+      "step": 46860
+    },
+    {
+      "epoch": 0.3877238697936055,
+      "grad_norm": 3245.526123046875,
+      "learning_rate": 7.814061069458657e-06,
+      "loss": 111.3302,
+      "step": 46870
+    },
+    {
+      "epoch": 0.3878065930429747,
+      "grad_norm": 547.8377685546875,
+      "learning_rate": 7.81289516955556e-06,
+      "loss": 78.6426,
+      "step": 46880
+    },
+    {
+      "epoch": 0.38788931629234397,
+      "grad_norm": 1064.7379150390625,
+      "learning_rate": 7.811729045843303e-06,
+      "loss": 90.7304,
+      "step": 46890
+    },
+    {
+      "epoch": 0.3879720395417132,
+      "grad_norm": 616.6322631835938,
+      "learning_rate": 7.81056269841467e-06,
+      "loss": 107.5491,
+      "step": 46900
+    },
+    {
+      "epoch": 0.3880547627910824,
+      "grad_norm": 1086.90185546875,
+      "learning_rate": 7.80939612736246e-06,
+      "loss": 112.952,
+      "step": 46910
+    },
+    {
+      "epoch": 0.3881374860404517,
+      "grad_norm": 762.38916015625,
+      "learning_rate": 7.808229332779496e-06,
+      "loss": 93.649,
+      "step": 46920
+    },
+    {
+      "epoch": 0.38822020928982093,
+      "grad_norm": 1226.2991943359375,
+      "learning_rate": 7.807062314758612e-06,
+      "loss": 122.8837,
+      "step": 46930
+    },
+    {
+      "epoch": 0.3883029325391901,
+      "grad_norm": 980.0850219726562,
+      "learning_rate": 7.80589507339266e-06,
+      "loss": 118.3508,
+      "step": 46940
+    },
+    {
+      "epoch": 0.3883856557885594,
+      "grad_norm": 899.0253295898438,
+      "learning_rate": 7.804727608774516e-06,
+      "loss": 95.3277,
+      "step": 46950
+    },
+    {
+      "epoch": 0.38846837903792864,
+      "grad_norm": 992.0533447265625,
+      "learning_rate": 7.803559920997067e-06,
+      "loss": 100.9585,
+      "step": 46960
+    },
+    {
+      "epoch": 0.38855110228729783,
+      "grad_norm": 1150.515869140625,
+      "learning_rate": 7.802392010153223e-06,
+      "loss": 106.7416,
+      "step": 46970
+    },
+    {
+      "epoch": 0.3886338255366671,
+      "grad_norm": 392.8724670410156,
+      "learning_rate": 7.801223876335907e-06,
+      "loss": 81.8418,
+      "step": 46980
+    },
+    {
+      "epoch": 0.38871654878603634,
+      "grad_norm": 958.8024291992188,
+      "learning_rate": 7.800055519638064e-06,
+      "loss": 104.2003,
+      "step": 46990
+    },
+    {
+      "epoch": 0.38879927203540554,
+      "grad_norm": 1033.255859375,
+      "learning_rate": 7.798886940152654e-06,
+      "loss": 111.1013,
+      "step": 47000
+    },
+    {
+      "epoch": 0.3888819952847748,
+      "grad_norm": 670.1970825195312,
+      "learning_rate": 7.797718137972654e-06,
+      "loss": 108.9194,
+      "step": 47010
+    },
+    {
+      "epoch": 0.388964718534144,
+      "grad_norm": 927.21630859375,
+      "learning_rate": 7.79654911319106e-06,
+      "loss": 107.1055,
+      "step": 47020
+    },
+    {
+      "epoch": 0.38904744178351325,
+      "grad_norm": 1264.6552734375,
+      "learning_rate": 7.795379865900892e-06,
+      "loss": 131.0568,
+      "step": 47030
+    },
+    {
+      "epoch": 0.3891301650328825,
+      "grad_norm": 1989.842529296875,
+      "learning_rate": 7.794210396195175e-06,
+      "loss": 94.6299,
+      "step": 47040
+    },
+    {
+      "epoch": 0.3892128882822517,
+      "grad_norm": 530.0291748046875,
+      "learning_rate": 7.79304070416696e-06,
+      "loss": 121.8401,
+      "step": 47050
+    },
+    {
+      "epoch": 0.38929561153162096,
+      "grad_norm": 936.9892578125,
+      "learning_rate": 7.791870789909315e-06,
+      "loss": 89.362,
+      "step": 47060
+    },
+    {
+      "epoch": 0.3893783347809902,
+      "grad_norm": 2093.591796875,
+      "learning_rate": 7.790700653515324e-06,
+      "loss": 90.1807,
+      "step": 47070
+    },
+    {
+      "epoch": 0.3894610580303594,
+      "grad_norm": 549.9674682617188,
+      "learning_rate": 7.789530295078089e-06,
+      "loss": 100.1859,
+      "step": 47080
+    },
+    {
+      "epoch": 0.38954378127972866,
+      "grad_norm": 882.615966796875,
+      "learning_rate": 7.788359714690732e-06,
+      "loss": 117.6557,
+      "step": 47090
+    },
+    {
+      "epoch": 0.3896265045290979,
+      "grad_norm": 798.0199584960938,
+      "learning_rate": 7.787188912446389e-06,
+      "loss": 127.0244,
+      "step": 47100
+    },
+    {
+      "epoch": 0.3897092277784671,
+      "grad_norm": 776.5193481445312,
+      "learning_rate": 7.786017888438214e-06,
+      "loss": 126.0215,
+      "step": 47110
+    },
+    {
+      "epoch": 0.38979195102783637,
+      "grad_norm": 654.182373046875,
+      "learning_rate": 7.784846642759383e-06,
+      "loss": 82.0503,
+      "step": 47120
+    },
+    {
+      "epoch": 0.3898746742772056,
+      "grad_norm": 822.113525390625,
+      "learning_rate": 7.783675175503087e-06,
+      "loss": 89.8243,
+      "step": 47130
+    },
+    {
+      "epoch": 0.3899573975265748,
+      "grad_norm": 865.35546875,
+      "learning_rate": 7.78250348676253e-06,
+      "loss": 98.9063,
+      "step": 47140
+    },
+    {
+      "epoch": 0.3900401207759441,
+      "grad_norm": 1154.8814697265625,
+      "learning_rate": 7.781331576630941e-06,
+      "loss": 146.3992,
+      "step": 47150
+    },
+    {
+      "epoch": 0.39012284402531333,
+      "grad_norm": 1378.4947509765625,
+      "learning_rate": 7.780159445201562e-06,
+      "loss": 130.5315,
+      "step": 47160
+    },
+    {
+      "epoch": 0.39020556727468253,
+      "grad_norm": 1242.33251953125,
+      "learning_rate": 7.778987092567658e-06,
+      "loss": 119.4341,
+      "step": 47170
+    },
+    {
+      "epoch": 0.3902882905240518,
+      "grad_norm": 797.1754760742188,
+      "learning_rate": 7.777814518822504e-06,
+      "loss": 67.0835,
+      "step": 47180
+    },
+    {
+      "epoch": 0.39037101377342104,
+      "grad_norm": 607.6759033203125,
+      "learning_rate": 7.776641724059398e-06,
+      "loss": 93.4001,
+      "step": 47190
+    },
+    {
+      "epoch": 0.39045373702279024,
+      "grad_norm": 819.6945190429688,
+      "learning_rate": 7.77546870837165e-06,
+      "loss": 74.3229,
+      "step": 47200
+    },
+    {
+      "epoch": 0.3905364602721595,
+      "grad_norm": 903.440185546875,
+      "learning_rate": 7.774295471852596e-06,
+      "loss": 108.8114,
+      "step": 47210
+    },
+    {
+      "epoch": 0.39061918352152875,
+      "grad_norm": 785.1321411132812,
+      "learning_rate": 7.773122014595584e-06,
+      "loss": 169.5685,
+      "step": 47220
+    },
+    {
+      "epoch": 0.39070190677089794,
+      "grad_norm": 1930.775146484375,
+      "learning_rate": 7.771948336693983e-06,
+      "loss": 108.4483,
+      "step": 47230
+    },
+    {
+      "epoch": 0.3907846300202672,
+      "grad_norm": 515.0844116210938,
+      "learning_rate": 7.770774438241168e-06,
+      "loss": 67.2212,
+      "step": 47240
+    },
+    {
+      "epoch": 0.39086735326963645,
+      "grad_norm": 1046.875,
+      "learning_rate": 7.769600319330553e-06,
+      "loss": 122.4751,
+      "step": 47250
+    },
+    {
+      "epoch": 0.39095007651900565,
+      "grad_norm": 516.0284423828125,
+      "learning_rate": 7.768425980055548e-06,
+      "loss": 87.5364,
+      "step": 47260
+    },
+    {
+      "epoch": 0.3910327997683749,
+      "grad_norm": 1055.4755859375,
+      "learning_rate": 7.767251420509593e-06,
+      "loss": 127.459,
+      "step": 47270
+    },
+    {
+      "epoch": 0.39111552301774416,
+      "grad_norm": 1103.45361328125,
+      "learning_rate": 7.766076640786145e-06,
+      "loss": 91.529,
+      "step": 47280
+    },
+    {
+      "epoch": 0.39119824626711336,
+      "grad_norm": 751.03759765625,
+      "learning_rate": 7.764901640978671e-06,
+      "loss": 97.0965,
+      "step": 47290
+    },
+    {
+      "epoch": 0.3912809695164826,
+      "grad_norm": 849.8720703125,
+      "learning_rate": 7.763726421180664e-06,
+      "loss": 100.8384,
+      "step": 47300
+    },
+    {
+      "epoch": 0.39136369276585187,
+      "grad_norm": 1227.1131591796875,
+      "learning_rate": 7.762550981485629e-06,
+      "loss": 85.1875,
+      "step": 47310
+    },
+    {
+      "epoch": 0.39144641601522107,
+      "grad_norm": 1593.5885009765625,
+      "learning_rate": 7.76137532198709e-06,
+      "loss": 117.0101,
+      "step": 47320
+    },
+    {
+      "epoch": 0.3915291392645903,
+      "grad_norm": 901.8778686523438,
+      "learning_rate": 7.76019944277859e-06,
+      "loss": 132.7024,
+      "step": 47330
+    },
+    {
+      "epoch": 0.3916118625139596,
+      "grad_norm": 851.2544555664062,
+      "learning_rate": 7.759023343953689e-06,
+      "loss": 89.3048,
+      "step": 47340
+    },
+    {
+      "epoch": 0.3916945857633288,
+      "grad_norm": 605.6761474609375,
+      "learning_rate": 7.757847025605963e-06,
+      "loss": 103.3425,
+      "step": 47350
+    },
+    {
+      "epoch": 0.391777309012698,
+      "grad_norm": 1666.5958251953125,
+      "learning_rate": 7.756670487829005e-06,
+      "loss": 112.8522,
+      "step": 47360
+    },
+    {
+      "epoch": 0.3918600322620673,
+      "grad_norm": 928.7994384765625,
+      "learning_rate": 7.755493730716428e-06,
+      "loss": 139.6545,
+      "step": 47370
+    },
+    {
+      "epoch": 0.3919427555114365,
+      "grad_norm": 1118.8592529296875,
+      "learning_rate": 7.75431675436186e-06,
+      "loss": 101.6983,
+      "step": 47380
+    },
+    {
+      "epoch": 0.39202547876080573,
+      "grad_norm": 753.40185546875,
+      "learning_rate": 7.753139558858949e-06,
+      "loss": 111.5847,
+      "step": 47390
+    },
+    {
+      "epoch": 0.39210820201017493,
+      "grad_norm": 1253.97509765625,
+      "learning_rate": 7.751962144301359e-06,
+      "loss": 101.4553,
+      "step": 47400
+    },
+    {
+      "epoch": 0.3921909252595442,
+      "grad_norm": 1033.0867919921875,
+      "learning_rate": 7.75078451078277e-06,
+      "loss": 92.0257,
+      "step": 47410
+    },
+    {
+      "epoch": 0.39227364850891344,
+      "grad_norm": 747.4989624023438,
+      "learning_rate": 7.749606658396883e-06,
+      "loss": 100.6043,
+      "step": 47420
+    },
+    {
+      "epoch": 0.39235637175828264,
+      "grad_norm": 822.0686645507812,
+      "learning_rate": 7.748428587237412e-06,
+      "loss": 80.0977,
+      "step": 47430
+    },
+    {
+      "epoch": 0.3924390950076519,
+      "grad_norm": 1061.1436767578125,
+      "learning_rate": 7.747250297398092e-06,
+      "loss": 120.7229,
+      "step": 47440
+    },
+    {
+      "epoch": 0.39252181825702115,
+      "grad_norm": 497.0961608886719,
+      "learning_rate": 7.746071788972675e-06,
+      "loss": 92.1028,
+      "step": 47450
+    },
+    {
+      "epoch": 0.39260454150639035,
+      "grad_norm": 747.8739624023438,
+      "learning_rate": 7.744893062054928e-06,
+      "loss": 84.091,
+      "step": 47460
+    },
+    {
+      "epoch": 0.3926872647557596,
+      "grad_norm": 1407.6231689453125,
+      "learning_rate": 7.743714116738636e-06,
+      "loss": 96.2375,
+      "step": 47470
+    },
+    {
+      "epoch": 0.39276998800512886,
+      "grad_norm": 1389.636962890625,
+      "learning_rate": 7.742534953117607e-06,
+      "loss": 83.5301,
+      "step": 47480
+    },
+    {
+      "epoch": 0.39285271125449805,
+      "grad_norm": 827.8368530273438,
+      "learning_rate": 7.741355571285656e-06,
+      "loss": 98.6002,
+      "step": 47490
+    },
+    {
+      "epoch": 0.3929354345038673,
+      "grad_norm": 1032.63330078125,
+      "learning_rate": 7.740175971336624e-06,
+      "loss": 111.3695,
+      "step": 47500
+    },
+    {
+      "epoch": 0.39301815775323656,
+      "grad_norm": 938.416015625,
+      "learning_rate": 7.738996153364364e-06,
+      "loss": 128.9613,
+      "step": 47510
+    },
+    {
+      "epoch": 0.39310088100260576,
+      "grad_norm": 960.8037719726562,
+      "learning_rate": 7.737816117462752e-06,
+      "loss": 96.7483,
+      "step": 47520
+    },
+    {
+      "epoch": 0.393183604251975,
+      "grad_norm": 880.8233642578125,
+      "learning_rate": 7.736635863725677e-06,
+      "loss": 112.1702,
+      "step": 47530
+    },
+    {
+      "epoch": 0.39326632750134427,
+      "grad_norm": 643.0890502929688,
+      "learning_rate": 7.735455392247044e-06,
+      "loss": 172.7514,
+      "step": 47540
+    },
+    {
+      "epoch": 0.39334905075071347,
+      "grad_norm": 481.9482421875,
+      "learning_rate": 7.73427470312078e-06,
+      "loss": 98.0854,
+      "step": 47550
+    },
+    {
+      "epoch": 0.3934317740000827,
+      "grad_norm": 7659.08837890625,
+      "learning_rate": 7.733093796440828e-06,
+      "loss": 94.3791,
+      "step": 47560
+    },
+    {
+      "epoch": 0.393514497249452,
+      "grad_norm": 611.7446899414062,
+      "learning_rate": 7.731912672301145e-06,
+      "loss": 121.1342,
+      "step": 47570
+    },
+    {
+      "epoch": 0.3935972204988212,
+      "grad_norm": 1074.301513671875,
+      "learning_rate": 7.730731330795707e-06,
+      "loss": 107.5944,
+      "step": 47580
+    },
+    {
+      "epoch": 0.39367994374819043,
+      "grad_norm": 1130.7880859375,
+      "learning_rate": 7.72954977201851e-06,
+      "loss": 88.7905,
+      "step": 47590
+    },
+    {
+      "epoch": 0.3937626669975597,
+      "grad_norm": 427.6048278808594,
+      "learning_rate": 7.728367996063566e-06,
+      "loss": 95.4467,
+      "step": 47600
+    },
+    {
+      "epoch": 0.3938453902469289,
+      "grad_norm": 1087.6966552734375,
+      "learning_rate": 7.727186003024902e-06,
+      "loss": 116.7486,
+      "step": 47610
+    },
+    {
+      "epoch": 0.39392811349629814,
+      "grad_norm": 677.6015014648438,
+      "learning_rate": 7.726003792996562e-06,
+      "loss": 112.2149,
+      "step": 47620
+    },
+    {
+      "epoch": 0.3940108367456674,
+      "grad_norm": 1281.115478515625,
+      "learning_rate": 7.724821366072612e-06,
+      "loss": 112.6385,
+      "step": 47630
+    },
+    {
+      "epoch": 0.3940935599950366,
+      "grad_norm": 1624.0828857421875,
+      "learning_rate": 7.723638722347132e-06,
+      "loss": 114.2262,
+      "step": 47640
+    },
+    {
+      "epoch": 0.39417628324440585,
+      "grad_norm": 835.2601318359375,
+      "learning_rate": 7.722455861914218e-06,
+      "loss": 87.17,
+      "step": 47650
+    },
+    {
+      "epoch": 0.3942590064937751,
+      "grad_norm": 721.5057983398438,
+      "learning_rate": 7.721272784867983e-06,
+      "loss": 112.5632,
+      "step": 47660
+    },
+    {
+      "epoch": 0.3943417297431443,
+      "grad_norm": 909.7614135742188,
+      "learning_rate": 7.720089491302565e-06,
+      "loss": 105.2768,
+      "step": 47670
+    },
+    {
+      "epoch": 0.39442445299251355,
+      "grad_norm": 641.9271240234375,
+      "learning_rate": 7.718905981312108e-06,
+      "loss": 93.7019,
+      "step": 47680
+    },
+    {
+      "epoch": 0.3945071762418828,
+      "grad_norm": 585.551025390625,
+      "learning_rate": 7.71772225499078e-06,
+      "loss": 80.3911,
+      "step": 47690
+    },
+    {
+      "epoch": 0.394589899491252,
+      "grad_norm": 783.8123168945312,
+      "learning_rate": 7.716538312432767e-06,
+      "loss": 115.101,
+      "step": 47700
+    },
+    {
+      "epoch": 0.39467262274062126,
+      "grad_norm": 644.3832397460938,
+      "learning_rate": 7.715354153732265e-06,
+      "loss": 99.6165,
+      "step": 47710
+    },
+    {
+      "epoch": 0.3947553459899905,
+      "grad_norm": 1999.86572265625,
+      "learning_rate": 7.714169778983496e-06,
+      "loss": 117.3061,
+      "step": 47720
+    },
+    {
+      "epoch": 0.3948380692393597,
+      "grad_norm": 1040.9405517578125,
+      "learning_rate": 7.712985188280694e-06,
+      "loss": 101.7906,
+      "step": 47730
+    },
+    {
+      "epoch": 0.39492079248872897,
+      "grad_norm": 1040.4888916015625,
+      "learning_rate": 7.711800381718111e-06,
+      "loss": 110.293,
+      "step": 47740
+    },
+    {
+      "epoch": 0.39500351573809817,
+      "grad_norm": 1044.0440673828125,
+      "learning_rate": 7.710615359390018e-06,
+      "loss": 94.9559,
+      "step": 47750
+    },
+    {
+      "epoch": 0.3950862389874674,
+      "grad_norm": 749.12939453125,
+      "learning_rate": 7.7094301213907e-06,
+      "loss": 92.2077,
+      "step": 47760
+    },
+    {
+      "epoch": 0.3951689622368367,
+      "grad_norm": 1238.37158203125,
+      "learning_rate": 7.708244667814463e-06,
+      "loss": 100.6183,
+      "step": 47770
+    },
+    {
+      "epoch": 0.3952516854862059,
+      "grad_norm": 821.638916015625,
+      "learning_rate": 7.707058998755626e-06,
+      "loss": 109.0208,
+      "step": 47780
+    },
+    {
+      "epoch": 0.3953344087355751,
+      "grad_norm": 767.8424072265625,
+      "learning_rate": 7.705873114308529e-06,
+      "loss": 85.0486,
+      "step": 47790
+    },
+    {
+      "epoch": 0.3954171319849444,
+      "grad_norm": 756.989501953125,
+      "learning_rate": 7.704687014567524e-06,
+      "loss": 89.8211,
+      "step": 47800
+    },
+    {
+      "epoch": 0.3954998552343136,
+      "grad_norm": 1021.226806640625,
+      "learning_rate": 7.703500699626988e-06,
+      "loss": 90.8781,
+      "step": 47810
+    },
+    {
+      "epoch": 0.39558257848368283,
+      "grad_norm": 749.4614868164062,
+      "learning_rate": 7.702314169581311e-06,
+      "loss": 103.451,
+      "step": 47820
+    },
+    {
+      "epoch": 0.3956653017330521,
+      "grad_norm": 954.7346801757812,
+      "learning_rate": 7.701127424524894e-06,
+      "loss": 97.5412,
+      "step": 47830
+    },
+    {
+      "epoch": 0.3957480249824213,
+      "grad_norm": 704.9465942382812,
+      "learning_rate": 7.699940464552166e-06,
+      "loss": 128.5169,
+      "step": 47840
+    },
+    {
+      "epoch": 0.39583074823179054,
+      "grad_norm": 926.2518310546875,
+      "learning_rate": 7.698753289757565e-06,
+      "loss": 103.6893,
+      "step": 47850
+    },
+    {
+      "epoch": 0.3959134714811598,
+      "grad_norm": 903.693115234375,
+      "learning_rate": 7.69756590023555e-06,
+      "loss": 117.7347,
+      "step": 47860
+    },
+    {
+      "epoch": 0.395996194730529,
+      "grad_norm": 576.5755004882812,
+      "learning_rate": 7.696378296080598e-06,
+      "loss": 77.2486,
+      "step": 47870
+    },
+    {
+      "epoch": 0.39607891797989825,
+      "grad_norm": 1016.8932495117188,
+      "learning_rate": 7.6951904773872e-06,
+      "loss": 102.5683,
+      "step": 47880
+    },
+    {
+      "epoch": 0.3961616412292675,
+      "grad_norm": 649.89013671875,
+      "learning_rate": 7.694002444249863e-06,
+      "loss": 101.7743,
+      "step": 47890
+    },
+    {
+      "epoch": 0.3962443644786367,
+      "grad_norm": 2596.306396484375,
+      "learning_rate": 7.692814196763118e-06,
+      "loss": 125.352,
+      "step": 47900
+    },
+    {
+      "epoch": 0.39632708772800596,
+      "grad_norm": 1052.6181640625,
+      "learning_rate": 7.691625735021505e-06,
+      "loss": 109.4487,
+      "step": 47910
+    },
+    {
+      "epoch": 0.3964098109773752,
+      "grad_norm": 909.6773071289062,
+      "learning_rate": 7.690437059119584e-06,
+      "loss": 132.6711,
+      "step": 47920
+    },
+    {
+      "epoch": 0.3964925342267444,
+      "grad_norm": 468.37310791015625,
+      "learning_rate": 7.689248169151935e-06,
+      "loss": 89.8137,
+      "step": 47930
+    },
+    {
+      "epoch": 0.39657525747611366,
+      "grad_norm": 1061.165771484375,
+      "learning_rate": 7.68805906521315e-06,
+      "loss": 117.4328,
+      "step": 47940
+    },
+    {
+      "epoch": 0.3966579807254829,
+      "grad_norm": 757.1315307617188,
+      "learning_rate": 7.686869747397843e-06,
+      "loss": 101.1938,
+      "step": 47950
+    },
+    {
+      "epoch": 0.3967407039748521,
+      "grad_norm": 811.54052734375,
+      "learning_rate": 7.685680215800639e-06,
+      "loss": 98.3037,
+      "step": 47960
+    },
+    {
+      "epoch": 0.39682342722422137,
+      "grad_norm": 880.4942626953125,
+      "learning_rate": 7.684490470516185e-06,
+      "loss": 105.742,
+      "step": 47970
+    },
+    {
+      "epoch": 0.3969061504735906,
+      "grad_norm": 3485.52685546875,
+      "learning_rate": 7.683300511639149e-06,
+      "loss": 121.3876,
+      "step": 47980
+    },
+    {
+      "epoch": 0.3969888737229598,
+      "grad_norm": 794.6721801757812,
+      "learning_rate": 7.682110339264203e-06,
+      "loss": 99.216,
+      "step": 47990
+    },
+    {
+      "epoch": 0.3970715969723291,
+      "grad_norm": 1388.62255859375,
+      "learning_rate": 7.680919953486047e-06,
+      "loss": 114.4334,
+      "step": 48000
+    },
+    {
+      "epoch": 0.39715432022169833,
+      "grad_norm": 617.267578125,
+      "learning_rate": 7.679729354399395e-06,
+      "loss": 96.2605,
+      "step": 48010
+    },
+    {
+      "epoch": 0.39723704347106753,
+      "grad_norm": 577.44189453125,
+      "learning_rate": 7.678538542098974e-06,
+      "loss": 106.764,
+      "step": 48020
+    },
+    {
+      "epoch": 0.3973197667204368,
+      "grad_norm": 922.736083984375,
+      "learning_rate": 7.677347516679536e-06,
+      "loss": 94.4455,
+      "step": 48030
+    },
+    {
+      "epoch": 0.39740248996980604,
+      "grad_norm": 1115.3370361328125,
+      "learning_rate": 7.676156278235845e-06,
+      "loss": 124.8899,
+      "step": 48040
+    },
+    {
+      "epoch": 0.39748521321917524,
+      "grad_norm": 956.1436157226562,
+      "learning_rate": 7.674964826862679e-06,
+      "loss": 114.437,
+      "step": 48050
+    },
+    {
+      "epoch": 0.3975679364685445,
+      "grad_norm": 1833.231201171875,
+      "learning_rate": 7.673773162654836e-06,
+      "loss": 127.1646,
+      "step": 48060
+    },
+    {
+      "epoch": 0.39765065971791375,
+      "grad_norm": 424.2684631347656,
+      "learning_rate": 7.672581285707135e-06,
+      "loss": 88.6944,
+      "step": 48070
+    },
+    {
+      "epoch": 0.39773338296728294,
+      "grad_norm": 1856.982421875,
+      "learning_rate": 7.67138919611441e-06,
+      "loss": 154.6424,
+      "step": 48080
+    },
+    {
+      "epoch": 0.3978161062166522,
+      "grad_norm": 907.585205078125,
+      "learning_rate": 7.670196893971502e-06,
+      "loss": 121.6254,
+      "step": 48090
+    },
+    {
+      "epoch": 0.3978988294660214,
+      "grad_norm": 1184.2100830078125,
+      "learning_rate": 7.669004379373284e-06,
+      "loss": 86.6673,
+      "step": 48100
+    },
+    {
+      "epoch": 0.39798155271539065,
+      "grad_norm": 1260.71875,
+      "learning_rate": 7.667811652414637e-06,
+      "loss": 87.2874,
+      "step": 48110
+    },
+    {
+      "epoch": 0.3980642759647599,
+      "grad_norm": 986.4711303710938,
+      "learning_rate": 7.666618713190459e-06,
+      "loss": 98.885,
+      "step": 48120
+    },
+    {
+      "epoch": 0.3981469992141291,
+      "grad_norm": 1163.977783203125,
+      "learning_rate": 7.665425561795669e-06,
+      "loss": 86.0785,
+      "step": 48130
+    },
+    {
+      "epoch": 0.39822972246349836,
+      "grad_norm": 851.3588256835938,
+      "learning_rate": 7.664232198325198e-06,
+      "loss": 88.372,
+      "step": 48140
+    },
+    {
+      "epoch": 0.3983124457128676,
+      "grad_norm": 786.80419921875,
+      "learning_rate": 7.663038622873999e-06,
+      "loss": 83.9941,
+      "step": 48150
+    },
+    {
+      "epoch": 0.3983951689622368,
+      "grad_norm": 1034.7119140625,
+      "learning_rate": 7.66184483553704e-06,
+      "loss": 91.0622,
+      "step": 48160
+    },
+    {
+      "epoch": 0.39847789221160607,
+      "grad_norm": 1714.0181884765625,
+      "learning_rate": 7.660650836409302e-06,
+      "loss": 88.6052,
+      "step": 48170
+    },
+    {
+      "epoch": 0.3985606154609753,
+      "grad_norm": 1282.2288818359375,
+      "learning_rate": 7.65945662558579e-06,
+      "loss": 102.4522,
+      "step": 48180
+    },
+    {
+      "epoch": 0.3986433387103445,
+      "grad_norm": 1070.42626953125,
+      "learning_rate": 7.658262203161517e-06,
+      "loss": 120.5579,
+      "step": 48190
+    },
+    {
+      "epoch": 0.3987260619597138,
+      "grad_norm": 1412.6171875,
+      "learning_rate": 7.65706756923152e-06,
+      "loss": 88.6738,
+      "step": 48200
+    },
+    {
+      "epoch": 0.398808785209083,
+      "grad_norm": 745.8485107421875,
+      "learning_rate": 7.655872723890854e-06,
+      "loss": 116.865,
+      "step": 48210
+    },
+    {
+      "epoch": 0.3988915084584522,
+      "grad_norm": 938.5888671875,
+      "learning_rate": 7.654677667234582e-06,
+      "loss": 112.7861,
+      "step": 48220
+    },
+    {
+      "epoch": 0.3989742317078215,
+      "grad_norm": 692.1904296875,
+      "learning_rate": 7.65348239935779e-06,
+      "loss": 93.3395,
+      "step": 48230
+    },
+    {
+      "epoch": 0.39905695495719073,
+      "grad_norm": 892.9244384765625,
+      "learning_rate": 7.652286920355583e-06,
+      "loss": 123.322,
+      "step": 48240
+    },
+    {
+      "epoch": 0.39913967820655993,
+      "grad_norm": 353.6568603515625,
+      "learning_rate": 7.651091230323079e-06,
+      "loss": 75.8525,
+      "step": 48250
+    },
+    {
+      "epoch": 0.3992224014559292,
+      "grad_norm": 1054.718017578125,
+      "learning_rate": 7.649895329355411e-06,
+      "loss": 144.2353,
+      "step": 48260
+    },
+    {
+      "epoch": 0.39930512470529844,
+      "grad_norm": 854.1619873046875,
+      "learning_rate": 7.648699217547733e-06,
+      "loss": 89.6152,
+      "step": 48270
+    },
+    {
+      "epoch": 0.39938784795466764,
+      "grad_norm": 1015.1759033203125,
+      "learning_rate": 7.647502894995215e-06,
+      "loss": 97.5599,
+      "step": 48280
+    },
+    {
+      "epoch": 0.3994705712040369,
+      "grad_norm": 691.9119873046875,
+      "learning_rate": 7.646306361793042e-06,
+      "loss": 95.9693,
+      "step": 48290
+    },
+    {
+      "epoch": 0.39955329445340615,
+      "grad_norm": 545.0184936523438,
+      "learning_rate": 7.645109618036416e-06,
+      "loss": 89.2513,
+      "step": 48300
+    },
+    {
+      "epoch": 0.39963601770277535,
+      "grad_norm": 1180.732666015625,
+      "learning_rate": 7.643912663820559e-06,
+      "loss": 107.0819,
+      "step": 48310
+    },
+    {
+      "epoch": 0.3997187409521446,
+      "grad_norm": 634.8629760742188,
+      "learning_rate": 7.642715499240702e-06,
+      "loss": 93.8869,
+      "step": 48320
+    },
+    {
+      "epoch": 0.39980146420151386,
+      "grad_norm": 878.6117553710938,
+      "learning_rate": 7.641518124392105e-06,
+      "loss": 117.2257,
+      "step": 48330
+    },
+    {
+      "epoch": 0.39988418745088306,
+      "grad_norm": 741.9995727539062,
+      "learning_rate": 7.640320539370032e-06,
+      "loss": 92.1223,
+      "step": 48340
+    },
+    {
+      "epoch": 0.3999669107002523,
+      "grad_norm": 693.643310546875,
+      "learning_rate": 7.63912274426977e-06,
+      "loss": 102.7763,
+      "step": 48350
+    },
+    {
+      "epoch": 0.40004963394962156,
+      "grad_norm": 747.3237915039062,
+      "learning_rate": 7.637924739186624e-06,
+      "loss": 83.4088,
+      "step": 48360
+    },
+    {
+      "epoch": 0.40013235719899076,
+      "grad_norm": 1104.4271240234375,
+      "learning_rate": 7.636726524215913e-06,
+      "loss": 96.2126,
+      "step": 48370
+    },
+    {
+      "epoch": 0.40021508044836,
+      "grad_norm": 1066.0506591796875,
+      "learning_rate": 7.635528099452974e-06,
+      "loss": 137.2111,
+      "step": 48380
+    },
+    {
+      "epoch": 0.40029780369772927,
+      "grad_norm": 764.3858032226562,
+      "learning_rate": 7.634329464993158e-06,
+      "loss": 99.6301,
+      "step": 48390
+    },
+    {
+      "epoch": 0.40038052694709847,
+      "grad_norm": 619.43017578125,
+      "learning_rate": 7.633130620931837e-06,
+      "loss": 95.3814,
+      "step": 48400
+    },
+    {
+      "epoch": 0.4004632501964677,
+      "grad_norm": 576.7614135742188,
+      "learning_rate": 7.631931567364398e-06,
+      "loss": 115.1573,
+      "step": 48410
+    },
+    {
+      "epoch": 0.400545973445837,
+      "grad_norm": 1004.3753662109375,
+      "learning_rate": 7.630732304386244e-06,
+      "loss": 90.6397,
+      "step": 48420
+    },
+    {
+      "epoch": 0.4006286966952062,
+      "grad_norm": 1928.7547607421875,
+      "learning_rate": 7.629532832092792e-06,
+      "loss": 98.6528,
+      "step": 48430
+    },
+    {
+      "epoch": 0.40071141994457543,
+      "grad_norm": 1009.5800170898438,
+      "learning_rate": 7.62833315057948e-06,
+      "loss": 105.3359,
+      "step": 48440
+    },
+    {
+      "epoch": 0.4007941431939447,
+      "grad_norm": 573.4387817382812,
+      "learning_rate": 7.627133259941762e-06,
+      "loss": 95.6264,
+      "step": 48450
+    },
+    {
+      "epoch": 0.4008768664433139,
+      "grad_norm": 1055.6937255859375,
+      "learning_rate": 7.625933160275109e-06,
+      "loss": 98.1826,
+      "step": 48460
+    },
+    {
+      "epoch": 0.40095958969268314,
+      "grad_norm": 636.2737426757812,
+      "learning_rate": 7.6247328516750055e-06,
+      "loss": 106.1883,
+      "step": 48470
+    },
+    {
+      "epoch": 0.40104231294205234,
+      "grad_norm": 624.5562744140625,
+      "learning_rate": 7.623532334236954e-06,
+      "loss": 115.2045,
+      "step": 48480
+    },
+    {
+      "epoch": 0.4011250361914216,
+      "grad_norm": 1036.740234375,
+      "learning_rate": 7.622331608056474e-06,
+      "loss": 100.7731,
+      "step": 48490
+    },
+    {
+      "epoch": 0.40120775944079085,
+      "grad_norm": 1890.8753662109375,
+      "learning_rate": 7.621130673229105e-06,
+      "loss": 98.8333,
+      "step": 48500
+    },
+    {
+      "epoch": 0.40129048269016004,
+      "grad_norm": 967.46044921875,
+      "learning_rate": 7.619929529850397e-06,
+      "loss": 62.4726,
+      "step": 48510
+    },
+    {
+      "epoch": 0.4013732059395293,
+      "grad_norm": 861.5704956054688,
+      "learning_rate": 7.618728178015919e-06,
+      "loss": 94.6595,
+      "step": 48520
+    },
+    {
+      "epoch": 0.40145592918889855,
+      "grad_norm": 1017.3840942382812,
+      "learning_rate": 7.617526617821259e-06,
+      "loss": 107.7875,
+      "step": 48530
+    },
+    {
+      "epoch": 0.40153865243826775,
+      "grad_norm": 2595.431640625,
+      "learning_rate": 7.616324849362019e-06,
+      "loss": 104.2326,
+      "step": 48540
+    },
+    {
+      "epoch": 0.401621375687637,
+      "grad_norm": 767.1851196289062,
+      "learning_rate": 7.615122872733819e-06,
+      "loss": 126.0866,
+      "step": 48550
+    },
+    {
+      "epoch": 0.40170409893700626,
+      "grad_norm": 901.1397705078125,
+      "learning_rate": 7.613920688032293e-06,
+      "loss": 120.4315,
+      "step": 48560
+    },
+    {
+      "epoch": 0.40178682218637546,
+      "grad_norm": 1408.295166015625,
+      "learning_rate": 7.612718295353094e-06,
+      "loss": 104.612,
+      "step": 48570
+    },
+    {
+      "epoch": 0.4018695454357447,
+      "grad_norm": 746.8675537109375,
+      "learning_rate": 7.61151569479189e-06,
+      "loss": 84.4078,
+      "step": 48580
+    },
+    {
+      "epoch": 0.40195226868511397,
+      "grad_norm": 925.1431884765625,
+      "learning_rate": 7.610312886444369e-06,
+      "loss": 89.8368,
+      "step": 48590
+    },
+    {
+      "epoch": 0.40203499193448317,
+      "grad_norm": 757.10693359375,
+      "learning_rate": 7.60910987040623e-06,
+      "loss": 92.3049,
+      "step": 48600
+    },
+    {
+      "epoch": 0.4021177151838524,
+      "grad_norm": 777.216552734375,
+      "learning_rate": 7.607906646773195e-06,
+      "loss": 105.3716,
+      "step": 48610
+    },
+    {
+      "epoch": 0.4022004384332217,
+      "grad_norm": 751.5682983398438,
+      "learning_rate": 7.606703215640995e-06,
+      "loss": 92.2909,
+      "step": 48620
+    },
+    {
+      "epoch": 0.4022831616825909,
+      "grad_norm": 1239.3206787109375,
+      "learning_rate": 7.605499577105382e-06,
+      "loss": 106.9025,
+      "step": 48630
+    },
+    {
+      "epoch": 0.4023658849319601,
+      "grad_norm": 750.9842529296875,
+      "learning_rate": 7.604295731262128e-06,
+      "loss": 115.5005,
+      "step": 48640
+    },
+    {
+      "epoch": 0.4024486081813294,
+      "grad_norm": 1378.44677734375,
+      "learning_rate": 7.603091678207013e-06,
+      "loss": 95.4831,
+      "step": 48650
+    },
+    {
+      "epoch": 0.4025313314306986,
+      "grad_norm": 883.4937133789062,
+      "learning_rate": 7.60188741803584e-06,
+      "loss": 94.2378,
+      "step": 48660
+    },
+    {
+      "epoch": 0.40261405468006783,
+      "grad_norm": 927.4861450195312,
+      "learning_rate": 7.600682950844428e-06,
+      "loss": 97.9098,
+      "step": 48670
+    },
+    {
+      "epoch": 0.4026967779294371,
+      "grad_norm": 999.6773071289062,
+      "learning_rate": 7.599478276728607e-06,
+      "loss": 88.1278,
+      "step": 48680
+    },
+    {
+      "epoch": 0.4027795011788063,
+      "grad_norm": 981.5151977539062,
+      "learning_rate": 7.5982733957842304e-06,
+      "loss": 96.1045,
+      "step": 48690
+    },
+    {
+      "epoch": 0.40286222442817554,
+      "grad_norm": 779.521728515625,
+      "learning_rate": 7.597068308107165e-06,
+      "loss": 96.5194,
+      "step": 48700
+    },
+    {
+      "epoch": 0.4029449476775448,
+      "grad_norm": 723.9231567382812,
+      "learning_rate": 7.595863013793292e-06,
+      "loss": 105.2723,
+      "step": 48710
+    },
+    {
+      "epoch": 0.403027670926914,
+      "grad_norm": 688.5134887695312,
+      "learning_rate": 7.594657512938513e-06,
+      "loss": 73.8923,
+      "step": 48720
+    },
+    {
+      "epoch": 0.40311039417628325,
+      "grad_norm": 1107.6732177734375,
+      "learning_rate": 7.593451805638743e-06,
+      "loss": 84.0132,
+      "step": 48730
+    },
+    {
+      "epoch": 0.4031931174256525,
+      "grad_norm": 971.2855834960938,
+      "learning_rate": 7.592245891989914e-06,
+      "loss": 126.4593,
+      "step": 48740
+    },
+    {
+      "epoch": 0.4032758406750217,
+      "grad_norm": 906.73388671875,
+      "learning_rate": 7.5910397720879785e-06,
+      "loss": 112.5872,
+      "step": 48750
+    },
+    {
+      "epoch": 0.40335856392439096,
+      "grad_norm": 976.5882568359375,
+      "learning_rate": 7.589833446028898e-06,
+      "loss": 113.1635,
+      "step": 48760
+    },
+    {
+      "epoch": 0.4034412871737602,
+      "grad_norm": 517.5700073242188,
+      "learning_rate": 7.5886269139086565e-06,
+      "loss": 95.9183,
+      "step": 48770
+    },
+    {
+      "epoch": 0.4035240104231294,
+      "grad_norm": 1049.484375,
+      "learning_rate": 7.587420175823252e-06,
+      "loss": 114.5341,
+      "step": 48780
+    },
+    {
+      "epoch": 0.40360673367249866,
+      "grad_norm": 1442.658447265625,
+      "learning_rate": 7.586213231868699e-06,
+      "loss": 89.2601,
+      "step": 48790
+    },
+    {
+      "epoch": 0.4036894569218679,
+      "grad_norm": 923.6640625,
+      "learning_rate": 7.585006082141028e-06,
+      "loss": 100.5833,
+      "step": 48800
+    },
+    {
+      "epoch": 0.4037721801712371,
+      "grad_norm": 882.6138916015625,
+      "learning_rate": 7.583798726736286e-06,
+      "loss": 111.9895,
+      "step": 48810
+    },
+    {
+      "epoch": 0.40385490342060637,
+      "grad_norm": 841.87548828125,
+      "learning_rate": 7.5825911657505365e-06,
+      "loss": 110.3644,
+      "step": 48820
+    },
+    {
+      "epoch": 0.40393762666997557,
+      "grad_norm": 686.5569458007812,
+      "learning_rate": 7.581383399279863e-06,
+      "loss": 113.7908,
+      "step": 48830
+    },
+    {
+      "epoch": 0.4040203499193448,
+      "grad_norm": 1229.239501953125,
+      "learning_rate": 7.580175427420358e-06,
+      "loss": 94.2136,
+      "step": 48840
+    },
+    {
+      "epoch": 0.4041030731687141,
+      "grad_norm": 1013.9456176757812,
+      "learning_rate": 7.578967250268137e-06,
+      "loss": 112.4359,
+      "step": 48850
+    },
+    {
+      "epoch": 0.4041857964180833,
+      "grad_norm": 843.6847534179688,
+      "learning_rate": 7.577758867919325e-06,
+      "loss": 119.3482,
+      "step": 48860
+    },
+    {
+      "epoch": 0.40426851966745253,
+      "grad_norm": 660.8492431640625,
+      "learning_rate": 7.576550280470072e-06,
+      "loss": 105.5132,
+      "step": 48870
+    },
+    {
+      "epoch": 0.4043512429168218,
+      "grad_norm": 1110.703125,
+      "learning_rate": 7.5753414880165365e-06,
+      "loss": 117.6248,
+      "step": 48880
+    },
+    {
+      "epoch": 0.404433966166191,
+      "grad_norm": 971.7891235351562,
+      "learning_rate": 7.5741324906548996e-06,
+      "loss": 84.8425,
+      "step": 48890
+    },
+    {
+      "epoch": 0.40451668941556024,
+      "grad_norm": 674.7775268554688,
+      "learning_rate": 7.572923288481355e-06,
+      "loss": 93.9152,
+      "step": 48900
+    },
+    {
+      "epoch": 0.4045994126649295,
+      "grad_norm": 1152.0150146484375,
+      "learning_rate": 7.571713881592109e-06,
+      "loss": 97.321,
+      "step": 48910
+    },
+    {
+      "epoch": 0.4046821359142987,
+      "grad_norm": 813.1100463867188,
+      "learning_rate": 7.570504270083394e-06,
+      "loss": 107.887,
+      "step": 48920
+    },
+    {
+      "epoch": 0.40476485916366794,
+      "grad_norm": 860.7267456054688,
+      "learning_rate": 7.569294454051452e-06,
+      "loss": 93.7367,
+      "step": 48930
+    },
+    {
+      "epoch": 0.4048475824130372,
+      "grad_norm": 641.8156127929688,
+      "learning_rate": 7.568084433592542e-06,
+      "loss": 103.0248,
+      "step": 48940
+    },
+    {
+      "epoch": 0.4049303056624064,
+      "grad_norm": 1134.04736328125,
+      "learning_rate": 7.566874208802939e-06,
+      "loss": 101.0284,
+      "step": 48950
+    },
+    {
+      "epoch": 0.40501302891177565,
+      "grad_norm": 596.593017578125,
+      "learning_rate": 7.5656637797789335e-06,
+      "loss": 80.7004,
+      "step": 48960
+    },
+    {
+      "epoch": 0.4050957521611449,
+      "grad_norm": 791.4570922851562,
+      "learning_rate": 7.564453146616837e-06,
+      "loss": 109.8588,
+      "step": 48970
+    },
+    {
+      "epoch": 0.4051784754105141,
+      "grad_norm": 681.3819580078125,
+      "learning_rate": 7.563242309412975e-06,
+      "loss": 105.6946,
+      "step": 48980
+    },
+    {
+      "epoch": 0.40526119865988336,
+      "grad_norm": 881.9998168945312,
+      "learning_rate": 7.562031268263686e-06,
+      "loss": 104.3682,
+      "step": 48990
+    },
+    {
+      "epoch": 0.4053439219092526,
+      "grad_norm": 985.5377807617188,
+      "learning_rate": 7.5608200232653254e-06,
+      "loss": 146.3788,
+      "step": 49000
+    },
+    {
+      "epoch": 0.4054266451586218,
+      "grad_norm": 547.89697265625,
+      "learning_rate": 7.5596085745142654e-06,
+      "loss": 126.0642,
+      "step": 49010
+    },
+    {
+      "epoch": 0.40550936840799107,
+      "grad_norm": 1244.3304443359375,
+      "learning_rate": 7.558396922106903e-06,
+      "loss": 105.9497,
+      "step": 49020
+    },
+    {
+      "epoch": 0.4055920916573603,
+      "grad_norm": 1159.8284912109375,
+      "learning_rate": 7.557185066139638e-06,
+      "loss": 123.0697,
+      "step": 49030
+    },
+    {
+      "epoch": 0.4056748149067295,
+      "grad_norm": 1442.5506591796875,
+      "learning_rate": 7.555973006708892e-06,
+      "loss": 124.4241,
+      "step": 49040
+    },
+    {
+      "epoch": 0.4057575381560988,
+      "grad_norm": 1344.17236328125,
+      "learning_rate": 7.554760743911104e-06,
+      "loss": 98.3027,
+      "step": 49050
+    },
+    {
+      "epoch": 0.40584026140546803,
+      "grad_norm": 733.8927001953125,
+      "learning_rate": 7.553548277842729e-06,
+      "loss": 90.8391,
+      "step": 49060
+    },
+    {
+      "epoch": 0.4059229846548372,
+      "grad_norm": 1737.365966796875,
+      "learning_rate": 7.5523356086002364e-06,
+      "loss": 128.2919,
+      "step": 49070
+    },
+    {
+      "epoch": 0.4060057079042065,
+      "grad_norm": 1121.77783203125,
+      "learning_rate": 7.551122736280113e-06,
+      "loss": 114.6872,
+      "step": 49080
+    },
+    {
+      "epoch": 0.40608843115357574,
+      "grad_norm": 830.7010498046875,
+      "learning_rate": 7.549909660978863e-06,
+      "loss": 103.0037,
+      "step": 49090
+    },
+    {
+      "epoch": 0.40617115440294493,
+      "grad_norm": 950.9368896484375,
+      "learning_rate": 7.548696382793002e-06,
+      "loss": 76.3594,
+      "step": 49100
+    },
+    {
+      "epoch": 0.4062538776523142,
+      "grad_norm": 721.6538696289062,
+      "learning_rate": 7.547482901819066e-06,
+      "loss": 101.0682,
+      "step": 49110
+    },
+    {
+      "epoch": 0.40633660090168344,
+      "grad_norm": 1179.335693359375,
+      "learning_rate": 7.5462692181536094e-06,
+      "loss": 113.5078,
+      "step": 49120
+    },
+    {
+      "epoch": 0.40641932415105264,
+      "grad_norm": 1521.753173828125,
+      "learning_rate": 7.545055331893195e-06,
+      "loss": 127.0566,
+      "step": 49130
+    },
+    {
+      "epoch": 0.4065020474004219,
+      "grad_norm": 718.9688720703125,
+      "learning_rate": 7.543841243134409e-06,
+      "loss": 77.3431,
+      "step": 49140
+    },
+    {
+      "epoch": 0.40658477064979115,
+      "grad_norm": 1070.0787353515625,
+      "learning_rate": 7.5426269519738495e-06,
+      "loss": 108.8352,
+      "step": 49150
+    },
+    {
+      "epoch": 0.40666749389916035,
+      "grad_norm": 514.2880859375,
+      "learning_rate": 7.541412458508133e-06,
+      "loss": 104.5422,
+      "step": 49160
+    },
+    {
+      "epoch": 0.4067502171485296,
+      "grad_norm": 774.6307983398438,
+      "learning_rate": 7.54019776283389e-06,
+      "loss": 100.5812,
+      "step": 49170
+    },
+    {
+      "epoch": 0.40683294039789886,
+      "grad_norm": 600.9510498046875,
+      "learning_rate": 7.53898286504777e-06,
+      "loss": 75.3506,
+      "step": 49180
+    },
+    {
+      "epoch": 0.40691566364726806,
+      "grad_norm": 843.1857299804688,
+      "learning_rate": 7.537767765246436e-06,
+      "loss": 110.9489,
+      "step": 49190
+    },
+    {
+      "epoch": 0.4069983868966373,
+      "grad_norm": 957.2750854492188,
+      "learning_rate": 7.536552463526565e-06,
+      "loss": 86.3647,
+      "step": 49200
+    },
+    {
+      "epoch": 0.4070811101460065,
+      "grad_norm": 1061.4150390625,
+      "learning_rate": 7.535336959984858e-06,
+      "loss": 127.3531,
+      "step": 49210
+    },
+    {
+      "epoch": 0.40716383339537576,
+      "grad_norm": 2062.2001953125,
+      "learning_rate": 7.5341212547180246e-06,
+      "loss": 111.2139,
+      "step": 49220
+    },
+    {
+      "epoch": 0.407246556644745,
+      "grad_norm": 768.780517578125,
+      "learning_rate": 7.532905347822792e-06,
+      "loss": 116.6353,
+      "step": 49230
+    },
+    {
+      "epoch": 0.4073292798941142,
+      "grad_norm": 780.7527465820312,
+      "learning_rate": 7.5316892393959064e-06,
+      "loss": 91.1807,
+      "step": 49240
+    },
+    {
+      "epoch": 0.40741200314348347,
+      "grad_norm": 487.50665283203125,
+      "learning_rate": 7.530472929534126e-06,
+      "loss": 110.9754,
+      "step": 49250
+    },
+    {
+      "epoch": 0.4074947263928527,
+      "grad_norm": 578.7011108398438,
+      "learning_rate": 7.529256418334228e-06,
+      "loss": 131.2194,
+      "step": 49260
+    },
+    {
+      "epoch": 0.4075774496422219,
+      "grad_norm": 818.4442749023438,
+      "learning_rate": 7.528039705893006e-06,
+      "loss": 98.4813,
+      "step": 49270
+    },
+    {
+      "epoch": 0.4076601728915912,
+      "grad_norm": 555.372314453125,
+      "learning_rate": 7.5268227923072665e-06,
+      "loss": 100.5567,
+      "step": 49280
+    },
+    {
+      "epoch": 0.40774289614096043,
+      "grad_norm": 1478.378173828125,
+      "learning_rate": 7.525605677673831e-06,
+      "loss": 94.0006,
+      "step": 49290
+    },
+    {
+      "epoch": 0.40782561939032963,
+      "grad_norm": 1054.407958984375,
+      "learning_rate": 7.524388362089545e-06,
+      "loss": 110.3638,
+      "step": 49300
+    },
+    {
+      "epoch": 0.4079083426396989,
+      "grad_norm": 936.5081176757812,
+      "learning_rate": 7.523170845651263e-06,
+      "loss": 108.1984,
+      "step": 49310
+    },
+    {
+      "epoch": 0.40799106588906814,
+      "grad_norm": 714.961181640625,
+      "learning_rate": 7.521953128455856e-06,
+      "loss": 109.1958,
+      "step": 49320
+    },
+    {
+      "epoch": 0.40807378913843734,
+      "grad_norm": 850.507080078125,
+      "learning_rate": 7.520735210600213e-06,
+      "loss": 117.9377,
+      "step": 49330
+    },
+    {
+      "epoch": 0.4081565123878066,
+      "grad_norm": 815.2796020507812,
+      "learning_rate": 7.519517092181237e-06,
+      "loss": 128.2985,
+      "step": 49340
+    },
+    {
+      "epoch": 0.40823923563717585,
+      "grad_norm": 1209.8961181640625,
+      "learning_rate": 7.518298773295849e-06,
+      "loss": 108.5808,
+      "step": 49350
+    },
+    {
+      "epoch": 0.40832195888654504,
+      "grad_norm": 1075.5911865234375,
+      "learning_rate": 7.517080254040985e-06,
+      "loss": 87.4483,
+      "step": 49360
+    },
+    {
+      "epoch": 0.4084046821359143,
+      "grad_norm": 952.4523315429688,
+      "learning_rate": 7.5158615345136e-06,
+      "loss": 85.5586,
+      "step": 49370
+    },
+    {
+      "epoch": 0.40848740538528355,
+      "grad_norm": 688.7626342773438,
+      "learning_rate": 7.514642614810655e-06,
+      "loss": 113.6158,
+      "step": 49380
+    },
+    {
+      "epoch": 0.40857012863465275,
+      "grad_norm": 754.0592651367188,
+      "learning_rate": 7.51342349502914e-06,
+      "loss": 95.794,
+      "step": 49390
+    },
+    {
+      "epoch": 0.408652851884022,
+      "grad_norm": 1221.88818359375,
+      "learning_rate": 7.512204175266052e-06,
+      "loss": 99.691,
+      "step": 49400
+    },
+    {
+      "epoch": 0.40873557513339126,
+      "grad_norm": 781.3279418945312,
+      "learning_rate": 7.510984655618407e-06,
+      "loss": 105.6204,
+      "step": 49410
+    },
+    {
+      "epoch": 0.40881829838276046,
+      "grad_norm": 593.7680053710938,
+      "learning_rate": 7.509764936183237e-06,
+      "loss": 82.4356,
+      "step": 49420
+    },
+    {
+      "epoch": 0.4089010216321297,
+      "grad_norm": 2055.203369140625,
+      "learning_rate": 7.5085450170575876e-06,
+      "loss": 120.1463,
+      "step": 49430
+    },
+    {
+      "epoch": 0.40898374488149897,
+      "grad_norm": 1296.7412109375,
+      "learning_rate": 7.5073248983385265e-06,
+      "loss": 130.0576,
+      "step": 49440
+    },
+    {
+      "epoch": 0.40906646813086817,
+      "grad_norm": 754.3132934570312,
+      "learning_rate": 7.50610458012313e-06,
+      "loss": 88.8123,
+      "step": 49450
+    },
+    {
+      "epoch": 0.4091491913802374,
+      "grad_norm": 827.0701904296875,
+      "learning_rate": 7.504884062508493e-06,
+      "loss": 85.6176,
+      "step": 49460
+    },
+    {
+      "epoch": 0.4092319146296067,
+      "grad_norm": 533.4575805664062,
+      "learning_rate": 7.503663345591726e-06,
+      "loss": 99.3742,
+      "step": 49470
+    },
+    {
+      "epoch": 0.4093146378789759,
+      "grad_norm": 914.8536376953125,
+      "learning_rate": 7.502442429469956e-06,
+      "loss": 123.0487,
+      "step": 49480
+    },
+    {
+      "epoch": 0.4093973611283451,
+      "grad_norm": 1090.47998046875,
+      "learning_rate": 7.501221314240329e-06,
+      "loss": 110.9421,
+      "step": 49490
+    },
+    {
+      "epoch": 0.4094800843777144,
+      "grad_norm": 969.5048828125,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 132.2819,
+      "step": 49500
+    },
+    {
+      "epoch": 0.4095628076270836,
+      "grad_norm": 1003.5017700195312,
+      "learning_rate": 7.4987784868461455e-06,
+      "loss": 125.4059,
+      "step": 49510
+    },
+    {
+      "epoch": 0.40964553087645283,
+      "grad_norm": 1233.950927734375,
+      "learning_rate": 7.497556774875953e-06,
+      "loss": 100.2119,
+      "step": 49520
+    },
+    {
+      "epoch": 0.4097282541258221,
+      "grad_norm": 638.7094116210938,
+      "learning_rate": 7.496334864186632e-06,
+      "loss": 133.3458,
+      "step": 49530
+    },
+    {
+      "epoch": 0.4098109773751913,
+      "grad_norm": 448.5832214355469,
+      "learning_rate": 7.4951127548754025e-06,
+      "loss": 95.0754,
+      "step": 49540
+    },
+    {
+      "epoch": 0.40989370062456054,
+      "grad_norm": 1257.3818359375,
+      "learning_rate": 7.4938904470395e-06,
+      "loss": 91.5809,
+      "step": 49550
+    },
+    {
+      "epoch": 0.40997642387392974,
+      "grad_norm": 1006.02294921875,
+      "learning_rate": 7.492667940776182e-06,
+      "loss": 102.9834,
+      "step": 49560
+    },
+    {
+      "epoch": 0.410059147123299,
+      "grad_norm": 1712.72607421875,
+      "learning_rate": 7.491445236182715e-06,
+      "loss": 103.0404,
+      "step": 49570
+    },
+    {
+      "epoch": 0.41014187037266825,
+      "grad_norm": 799.1417236328125,
+      "learning_rate": 7.490222333356384e-06,
+      "loss": 106.2483,
+      "step": 49580
+    },
+    {
+      "epoch": 0.41022459362203745,
+      "grad_norm": 1142.0809326171875,
+      "learning_rate": 7.488999232394492e-06,
+      "loss": 151.4166,
+      "step": 49590
+    },
+    {
+      "epoch": 0.4103073168714067,
+      "grad_norm": 1388.60693359375,
+      "learning_rate": 7.487775933394353e-06,
+      "loss": 103.9434,
+      "step": 49600
+    },
+    {
+      "epoch": 0.41039004012077596,
+      "grad_norm": 835.6099853515625,
+      "learning_rate": 7.4865524364533e-06,
+      "loss": 104.3207,
+      "step": 49610
+    },
+    {
+      "epoch": 0.41047276337014516,
+      "grad_norm": 653.7705688476562,
+      "learning_rate": 7.485328741668683e-06,
+      "loss": 83.743,
+      "step": 49620
+    },
+    {
+      "epoch": 0.4105554866195144,
+      "grad_norm": 870.46484375,
+      "learning_rate": 7.484104849137862e-06,
+      "loss": 108.1679,
+      "step": 49630
+    },
+    {
+      "epoch": 0.41063820986888366,
+      "grad_norm": 839.9007568359375,
+      "learning_rate": 7.482880758958219e-06,
+      "loss": 83.2425,
+      "step": 49640
+    },
+    {
+      "epoch": 0.41072093311825286,
+      "grad_norm": 931.4784545898438,
+      "learning_rate": 7.48165647122715e-06,
+      "loss": 114.039,
+      "step": 49650
+    },
+    {
+      "epoch": 0.4108036563676221,
+      "grad_norm": 776.007080078125,
+      "learning_rate": 7.480431986042065e-06,
+      "loss": 138.1686,
+      "step": 49660
+    },
+    {
+      "epoch": 0.41088637961699137,
+      "grad_norm": 890.8333129882812,
+      "learning_rate": 7.47920730350039e-06,
+      "loss": 129.4647,
+      "step": 49670
+    },
+    {
+      "epoch": 0.41096910286636057,
+      "grad_norm": 1296.5220947265625,
+      "learning_rate": 7.477982423699568e-06,
+      "loss": 120.0857,
+      "step": 49680
+    },
+    {
+      "epoch": 0.4110518261157298,
+      "grad_norm": 1131.2828369140625,
+      "learning_rate": 7.476757346737057e-06,
+      "loss": 112.3677,
+      "step": 49690
+    },
+    {
+      "epoch": 0.4111345493650991,
+      "grad_norm": 1283.157470703125,
+      "learning_rate": 7.47553207271033e-06,
+      "loss": 114.4162,
+      "step": 49700
+    },
+    {
+      "epoch": 0.4112172726144683,
+      "grad_norm": 309.3968200683594,
+      "learning_rate": 7.474306601716877e-06,
+      "loss": 89.349,
+      "step": 49710
+    },
+    {
+      "epoch": 0.41129999586383753,
+      "grad_norm": 961.7044677734375,
+      "learning_rate": 7.473080933854205e-06,
+      "loss": 81.6981,
+      "step": 49720
+    },
+    {
+      "epoch": 0.4113827191132068,
+      "grad_norm": 427.75469970703125,
+      "learning_rate": 7.471855069219831e-06,
+      "loss": 111.079,
+      "step": 49730
+    },
+    {
+      "epoch": 0.411465442362576,
+      "grad_norm": 973.0006103515625,
+      "learning_rate": 7.470629007911294e-06,
+      "loss": 99.5809,
+      "step": 49740
+    },
+    {
+      "epoch": 0.41154816561194524,
+      "grad_norm": 1471.537109375,
+      "learning_rate": 7.469402750026147e-06,
+      "loss": 130.2198,
+      "step": 49750
+    },
+    {
+      "epoch": 0.4116308888613145,
+      "grad_norm": 765.6078491210938,
+      "learning_rate": 7.468176295661955e-06,
+      "loss": 124.7598,
+      "step": 49760
+    },
+    {
+      "epoch": 0.4117136121106837,
+      "grad_norm": 918.6041259765625,
+      "learning_rate": 7.466949644916301e-06,
+      "loss": 103.558,
+      "step": 49770
+    },
+    {
+      "epoch": 0.41179633536005295,
+      "grad_norm": 933.6361694335938,
+      "learning_rate": 7.465722797886788e-06,
+      "loss": 94.5863,
+      "step": 49780
+    },
+    {
+      "epoch": 0.4118790586094222,
+      "grad_norm": 623.9163818359375,
+      "learning_rate": 7.464495754671027e-06,
+      "loss": 86.9486,
+      "step": 49790
+    },
+    {
+      "epoch": 0.4119617818587914,
+      "grad_norm": 592.259033203125,
+      "learning_rate": 7.4632685153666505e-06,
+      "loss": 111.5722,
+      "step": 49800
+    },
+    {
+      "epoch": 0.41204450510816065,
+      "grad_norm": 430.714599609375,
+      "learning_rate": 7.462041080071301e-06,
+      "loss": 81.418,
+      "step": 49810
+    },
+    {
+      "epoch": 0.4121272283575299,
+      "grad_norm": 1098.07373046875,
+      "learning_rate": 7.460813448882643e-06,
+      "loss": 145.5894,
+      "step": 49820
+    },
+    {
+      "epoch": 0.4122099516068991,
+      "grad_norm": 630.3162841796875,
+      "learning_rate": 7.459585621898353e-06,
+      "loss": 88.393,
+      "step": 49830
+    },
+    {
+      "epoch": 0.41229267485626836,
+      "grad_norm": 783.9848022460938,
+      "learning_rate": 7.4583575992161235e-06,
+      "loss": 73.975,
+      "step": 49840
+    },
+    {
+      "epoch": 0.4123753981056376,
+      "grad_norm": 645.27392578125,
+      "learning_rate": 7.457129380933662e-06,
+      "loss": 108.0057,
+      "step": 49850
+    },
+    {
+      "epoch": 0.4124581213550068,
+      "grad_norm": 522.0549926757812,
+      "learning_rate": 7.4559009671486906e-06,
+      "loss": 65.7118,
+      "step": 49860
+    },
+    {
+      "epoch": 0.41254084460437607,
+      "grad_norm": 706.9224853515625,
+      "learning_rate": 7.454672357958951e-06,
+      "loss": 127.6793,
+      "step": 49870
+    },
+    {
+      "epoch": 0.4126235678537453,
+      "grad_norm": 1543.4708251953125,
+      "learning_rate": 7.453443553462198e-06,
+      "loss": 138.2186,
+      "step": 49880
+    },
+    {
+      "epoch": 0.4127062911031145,
+      "grad_norm": 1025.1986083984375,
+      "learning_rate": 7.4522145537562015e-06,
+      "loss": 117.5677,
+      "step": 49890
+    },
+    {
+      "epoch": 0.4127890143524838,
+      "grad_norm": 611.4989013671875,
+      "learning_rate": 7.450985358938747e-06,
+      "loss": 106.0235,
+      "step": 49900
+    },
+    {
+      "epoch": 0.41287173760185303,
+      "grad_norm": 1154.43701171875,
+      "learning_rate": 7.449755969107635e-06,
+      "loss": 113.3347,
+      "step": 49910
+    },
+    {
+      "epoch": 0.4129544608512222,
+      "grad_norm": 1076.0106201171875,
+      "learning_rate": 7.4485263843606835e-06,
+      "loss": 104.9749,
+      "step": 49920
+    },
+    {
+      "epoch": 0.4130371841005915,
+      "grad_norm": 1215.836669921875,
+      "learning_rate": 7.447296604795726e-06,
+      "loss": 119.1512,
+      "step": 49930
+    },
+    {
+      "epoch": 0.4131199073499607,
+      "grad_norm": 1207.3167724609375,
+      "learning_rate": 7.4460666305106084e-06,
+      "loss": 101.0483,
+      "step": 49940
+    },
+    {
+      "epoch": 0.41320263059932993,
+      "grad_norm": 1147.7183837890625,
+      "learning_rate": 7.444836461603195e-06,
+      "loss": 101.2153,
+      "step": 49950
+    },
+    {
+      "epoch": 0.4132853538486992,
+      "grad_norm": 784.3466796875,
+      "learning_rate": 7.443606098171363e-06,
+      "loss": 86.7002,
+      "step": 49960
+    },
+    {
+      "epoch": 0.4133680770980684,
+      "grad_norm": 452.4320068359375,
+      "learning_rate": 7.442375540313012e-06,
+      "loss": 94.4948,
+      "step": 49970
+    },
+    {
+      "epoch": 0.41345080034743764,
+      "grad_norm": 911.47021484375,
+      "learning_rate": 7.441144788126045e-06,
+      "loss": 110.6758,
+      "step": 49980
+    },
+    {
+      "epoch": 0.4135335235968069,
+      "grad_norm": 1147.6248779296875,
+      "learning_rate": 7.4399138417083925e-06,
+      "loss": 95.634,
+      "step": 49990
+    },
+    {
+      "epoch": 0.4136162468461761,
+      "grad_norm": 1253.6624755859375,
+      "learning_rate": 7.438682701157993e-06,
+      "loss": 141.2699,
+      "step": 50000
+    },
+    {
+      "epoch": 0.41369897009554535,
+      "grad_norm": 1440.07080078125,
+      "learning_rate": 7.437451366572803e-06,
+      "loss": 109.6948,
+      "step": 50010
+    },
+    {
+      "epoch": 0.4137816933449146,
+      "grad_norm": 833.8212280273438,
+      "learning_rate": 7.436219838050793e-06,
+      "loss": 106.7741,
+      "step": 50020
+    },
+    {
+      "epoch": 0.4138644165942838,
+      "grad_norm": 999.9931640625,
+      "learning_rate": 7.4349881156899525e-06,
+      "loss": 110.1778,
+      "step": 50030
+    },
+    {
+      "epoch": 0.41394713984365306,
+      "grad_norm": 1517.33251953125,
+      "learning_rate": 7.433756199588282e-06,
+      "loss": 106.7463,
+      "step": 50040
+    },
+    {
+      "epoch": 0.4140298630930223,
+      "grad_norm": 749.0157470703125,
+      "learning_rate": 7.4325240898438e-06,
+      "loss": 86.2277,
+      "step": 50050
+    },
+    {
+      "epoch": 0.4141125863423915,
+      "grad_norm": 1004.859619140625,
+      "learning_rate": 7.4312917865545406e-06,
+      "loss": 101.3041,
+      "step": 50060
+    },
+    {
+      "epoch": 0.41419530959176076,
+      "grad_norm": 2136.435546875,
+      "learning_rate": 7.430059289818552e-06,
+      "loss": 109.2253,
+      "step": 50070
+    },
+    {
+      "epoch": 0.41427803284113,
+      "grad_norm": 1305.9910888671875,
+      "learning_rate": 7.4288265997338985e-06,
+      "loss": 116.3073,
+      "step": 50080
+    },
+    {
+      "epoch": 0.4143607560904992,
+      "grad_norm": 1031.19921875,
+      "learning_rate": 7.427593716398658e-06,
+      "loss": 136.2479,
+      "step": 50090
+    },
+    {
+      "epoch": 0.41444347933986847,
+      "grad_norm": 539.9341430664062,
+      "learning_rate": 7.426360639910927e-06,
+      "loss": 86.8462,
+      "step": 50100
+    },
+    {
+      "epoch": 0.4145262025892377,
+      "grad_norm": 527.5219116210938,
+      "learning_rate": 7.425127370368815e-06,
+      "loss": 104.5311,
+      "step": 50110
+    },
+    {
+      "epoch": 0.4146089258386069,
+      "grad_norm": 836.485107421875,
+      "learning_rate": 7.423893907870449e-06,
+      "loss": 125.8744,
+      "step": 50120
+    },
+    {
+      "epoch": 0.4146916490879762,
+      "grad_norm": 514.3257446289062,
+      "learning_rate": 7.422660252513969e-06,
+      "loss": 100.2241,
+      "step": 50130
+    },
+    {
+      "epoch": 0.41477437233734543,
+      "grad_norm": 773.184814453125,
+      "learning_rate": 7.421426404397531e-06,
+      "loss": 82.5671,
+      "step": 50140
+    },
+    {
+      "epoch": 0.41485709558671463,
+      "grad_norm": 1118.6546630859375,
+      "learning_rate": 7.420192363619305e-06,
+      "loss": 116.5798,
+      "step": 50150
+    },
+    {
+      "epoch": 0.4149398188360839,
+      "grad_norm": 1059.586669921875,
+      "learning_rate": 7.418958130277483e-06,
+      "loss": 94.7955,
+      "step": 50160
+    },
+    {
+      "epoch": 0.41502254208545314,
+      "grad_norm": 468.1589660644531,
+      "learning_rate": 7.417723704470261e-06,
+      "loss": 92.9759,
+      "step": 50170
+    },
+    {
+      "epoch": 0.41510526533482234,
+      "grad_norm": 2084.325927734375,
+      "learning_rate": 7.4164890862958615e-06,
+      "loss": 128.8667,
+      "step": 50180
+    },
+    {
+      "epoch": 0.4151879885841916,
+      "grad_norm": 834.3111572265625,
+      "learning_rate": 7.415254275852515e-06,
+      "loss": 103.7669,
+      "step": 50190
+    },
+    {
+      "epoch": 0.41527071183356085,
+      "grad_norm": 869.2041625976562,
+      "learning_rate": 7.414019273238471e-06,
+      "loss": 88.4897,
+      "step": 50200
+    },
+    {
+      "epoch": 0.41535343508293004,
+      "grad_norm": 1044.08056640625,
+      "learning_rate": 7.4127840785519915e-06,
+      "loss": 98.155,
+      "step": 50210
+    },
+    {
+      "epoch": 0.4154361583322993,
+      "grad_norm": 967.443603515625,
+      "learning_rate": 7.411548691891357e-06,
+      "loss": 115.068,
+      "step": 50220
+    },
+    {
+      "epoch": 0.41551888158166855,
+      "grad_norm": 534.2567138671875,
+      "learning_rate": 7.41031311335486e-06,
+      "loss": 106.5501,
+      "step": 50230
+    },
+    {
+      "epoch": 0.41560160483103775,
+      "grad_norm": 1074.7396240234375,
+      "learning_rate": 7.409077343040809e-06,
+      "loss": 153.6957,
+      "step": 50240
+    },
+    {
+      "epoch": 0.415684328080407,
+      "grad_norm": 657.9391479492188,
+      "learning_rate": 7.407841381047533e-06,
+      "loss": 99.6328,
+      "step": 50250
+    },
+    {
+      "epoch": 0.41576705132977626,
+      "grad_norm": 750.8782958984375,
+      "learning_rate": 7.406605227473367e-06,
+      "loss": 88.5056,
+      "step": 50260
+    },
+    {
+      "epoch": 0.41584977457914546,
+      "grad_norm": 912.8604125976562,
+      "learning_rate": 7.405368882416668e-06,
+      "loss": 127.2815,
+      "step": 50270
+    },
+    {
+      "epoch": 0.4159324978285147,
+      "grad_norm": 722.0275268554688,
+      "learning_rate": 7.404132345975806e-06,
+      "loss": 89.6634,
+      "step": 50280
+    },
+    {
+      "epoch": 0.4160152210778839,
+      "grad_norm": 503.6108093261719,
+      "learning_rate": 7.4028956182491665e-06,
+      "loss": 65.2768,
+      "step": 50290
+    },
+    {
+      "epoch": 0.41609794432725317,
+      "grad_norm": 1528.93212890625,
+      "learning_rate": 7.401658699335151e-06,
+      "loss": 110.4871,
+      "step": 50300
+    },
+    {
+      "epoch": 0.4161806675766224,
+      "grad_norm": 742.7732543945312,
+      "learning_rate": 7.400421589332175e-06,
+      "loss": 85.2059,
+      "step": 50310
+    },
+    {
+      "epoch": 0.4162633908259916,
+      "grad_norm": 1395.8526611328125,
+      "learning_rate": 7.39918428833867e-06,
+      "loss": 94.5206,
+      "step": 50320
+    },
+    {
+      "epoch": 0.4163461140753609,
+      "grad_norm": 978.8078002929688,
+      "learning_rate": 7.397946796453081e-06,
+      "loss": 122.651,
+      "step": 50330
+    },
+    {
+      "epoch": 0.41642883732473013,
+      "grad_norm": 738.7698974609375,
+      "learning_rate": 7.39670911377387e-06,
+      "loss": 128.4523,
+      "step": 50340
+    },
+    {
+      "epoch": 0.4165115605740993,
+      "grad_norm": 924.0703735351562,
+      "learning_rate": 7.395471240399515e-06,
+      "loss": 100.5796,
+      "step": 50350
+    },
+    {
+      "epoch": 0.4165942838234686,
+      "grad_norm": 1406.5579833984375,
+      "learning_rate": 7.394233176428508e-06,
+      "loss": 84.9948,
+      "step": 50360
+    },
+    {
+      "epoch": 0.41667700707283784,
+      "grad_norm": 692.1464233398438,
+      "learning_rate": 7.3929949219593545e-06,
+      "loss": 108.0571,
+      "step": 50370
+    },
+    {
+      "epoch": 0.41675973032220703,
+      "grad_norm": 935.2982177734375,
+      "learning_rate": 7.391756477090577e-06,
+      "loss": 106.4006,
+      "step": 50380
+    },
+    {
+      "epoch": 0.4168424535715763,
+      "grad_norm": 803.556396484375,
+      "learning_rate": 7.3905178419207126e-06,
+      "loss": 91.8345,
+      "step": 50390
+    },
+    {
+      "epoch": 0.41692517682094554,
+      "grad_norm": 170.64480590820312,
+      "learning_rate": 7.3892790165483164e-06,
+      "loss": 86.5613,
+      "step": 50400
+    },
+    {
+      "epoch": 0.41700790007031474,
+      "grad_norm": 846.9085693359375,
+      "learning_rate": 7.388040001071953e-06,
+      "loss": 116.5208,
+      "step": 50410
+    },
+    {
+      "epoch": 0.417090623319684,
+      "grad_norm": 488.96343994140625,
+      "learning_rate": 7.386800795590208e-06,
+      "loss": 84.3048,
+      "step": 50420
+    },
+    {
+      "epoch": 0.41717334656905325,
+      "grad_norm": 2128.975830078125,
+      "learning_rate": 7.385561400201675e-06,
+      "loss": 101.9137,
+      "step": 50430
+    },
+    {
+      "epoch": 0.41725606981842245,
+      "grad_norm": 886.4940795898438,
+      "learning_rate": 7.384321815004971e-06,
+      "loss": 122.7343,
+      "step": 50440
+    },
+    {
+      "epoch": 0.4173387930677917,
+      "grad_norm": 1239.8963623046875,
+      "learning_rate": 7.383082040098723e-06,
+      "loss": 118.5375,
+      "step": 50450
+    },
+    {
+      "epoch": 0.41742151631716096,
+      "grad_norm": 598.2474365234375,
+      "learning_rate": 7.381842075581573e-06,
+      "loss": 82.4002,
+      "step": 50460
+    },
+    {
+      "epoch": 0.41750423956653016,
+      "grad_norm": 927.5972900390625,
+      "learning_rate": 7.380601921552181e-06,
+      "loss": 96.2848,
+      "step": 50470
+    },
+    {
+      "epoch": 0.4175869628158994,
+      "grad_norm": 1360.900146484375,
+      "learning_rate": 7.379361578109218e-06,
+      "loss": 99.3107,
+      "step": 50480
+    },
+    {
+      "epoch": 0.41766968606526866,
+      "grad_norm": 637.6759643554688,
+      "learning_rate": 7.378121045351378e-06,
+      "loss": 115.3234,
+      "step": 50490
+    },
+    {
+      "epoch": 0.41775240931463786,
+      "grad_norm": 1148.364990234375,
+      "learning_rate": 7.376880323377357e-06,
+      "loss": 106.6024,
+      "step": 50500
+    },
+    {
+      "epoch": 0.4178351325640071,
+      "grad_norm": 567.2054443359375,
+      "learning_rate": 7.375639412285877e-06,
+      "loss": 94.4765,
+      "step": 50510
+    },
+    {
+      "epoch": 0.41791785581337637,
+      "grad_norm": 742.42236328125,
+      "learning_rate": 7.374398312175674e-06,
+      "loss": 103.2163,
+      "step": 50520
+    },
+    {
+      "epoch": 0.41800057906274557,
+      "grad_norm": 698.6530151367188,
+      "learning_rate": 7.373157023145493e-06,
+      "loss": 86.0616,
+      "step": 50530
+    },
+    {
+      "epoch": 0.4180833023121148,
+      "grad_norm": 1061.374755859375,
+      "learning_rate": 7.371915545294098e-06,
+      "loss": 74.492,
+      "step": 50540
+    },
+    {
+      "epoch": 0.4181660255614841,
+      "grad_norm": 748.8624877929688,
+      "learning_rate": 7.37067387872027e-06,
+      "loss": 75.0185,
+      "step": 50550
+    },
+    {
+      "epoch": 0.4182487488108533,
+      "grad_norm": 315.02374267578125,
+      "learning_rate": 7.369432023522801e-06,
+      "loss": 77.1597,
+      "step": 50560
+    },
+    {
+      "epoch": 0.41833147206022253,
+      "grad_norm": 981.3681030273438,
+      "learning_rate": 7.3681899798005006e-06,
+      "loss": 118.4615,
+      "step": 50570
+    },
+    {
+      "epoch": 0.4184141953095918,
+      "grad_norm": 1849.541259765625,
+      "learning_rate": 7.366947747652191e-06,
+      "loss": 104.9723,
+      "step": 50580
+    },
+    {
+      "epoch": 0.418496918558961,
+      "grad_norm": 1334.4669189453125,
+      "learning_rate": 7.365705327176713e-06,
+      "loss": 100.1431,
+      "step": 50590
+    },
+    {
+      "epoch": 0.41857964180833024,
+      "grad_norm": 1951.8896484375,
+      "learning_rate": 7.364462718472919e-06,
+      "loss": 142.2957,
+      "step": 50600
+    },
+    {
+      "epoch": 0.4186623650576995,
+      "grad_norm": 659.5794067382812,
+      "learning_rate": 7.363219921639677e-06,
+      "loss": 96.1194,
+      "step": 50610
+    },
+    {
+      "epoch": 0.4187450883070687,
+      "grad_norm": 960.1915893554688,
+      "learning_rate": 7.361976936775872e-06,
+      "loss": 117.1287,
+      "step": 50620
+    },
+    {
+      "epoch": 0.41882781155643795,
+      "grad_norm": 940.3717651367188,
+      "learning_rate": 7.360733763980404e-06,
+      "loss": 110.9005,
+      "step": 50630
+    },
+    {
+      "epoch": 0.4189105348058072,
+      "grad_norm": 741.2780151367188,
+      "learning_rate": 7.3594904033521815e-06,
+      "loss": 86.0149,
+      "step": 50640
+    },
+    {
+      "epoch": 0.4189932580551764,
+      "grad_norm": 1676.6136474609375,
+      "learning_rate": 7.358246854990138e-06,
+      "loss": 119.8522,
+      "step": 50650
+    },
+    {
+      "epoch": 0.41907598130454565,
+      "grad_norm": 1362.4383544921875,
+      "learning_rate": 7.357003118993215e-06,
+      "loss": 125.9308,
+      "step": 50660
+    },
+    {
+      "epoch": 0.41915870455391485,
+      "grad_norm": 1395.822998046875,
+      "learning_rate": 7.355759195460371e-06,
+      "loss": 125.4457,
+      "step": 50670
+    },
+    {
+      "epoch": 0.4192414278032841,
+      "grad_norm": 754.3294067382812,
+      "learning_rate": 7.354515084490579e-06,
+      "loss": 105.6307,
+      "step": 50680
+    },
+    {
+      "epoch": 0.41932415105265336,
+      "grad_norm": 1079.152587890625,
+      "learning_rate": 7.353270786182828e-06,
+      "loss": 95.9782,
+      "step": 50690
+    },
+    {
+      "epoch": 0.41940687430202256,
+      "grad_norm": 713.3341674804688,
+      "learning_rate": 7.352026300636121e-06,
+      "loss": 104.3826,
+      "step": 50700
+    },
+    {
+      "epoch": 0.4194895975513918,
+      "grad_norm": 879.9957275390625,
+      "learning_rate": 7.350781627949475e-06,
+      "loss": 86.2512,
+      "step": 50710
+    },
+    {
+      "epoch": 0.41957232080076107,
+      "grad_norm": 1013.7462158203125,
+      "learning_rate": 7.3495367682219236e-06,
+      "loss": 110.0644,
+      "step": 50720
+    },
+    {
+      "epoch": 0.41965504405013027,
+      "grad_norm": 948.6749877929688,
+      "learning_rate": 7.348291721552514e-06,
+      "loss": 145.048,
+      "step": 50730
+    },
+    {
+      "epoch": 0.4197377672994995,
+      "grad_norm": 2777.333251953125,
+      "learning_rate": 7.3470464880403105e-06,
+      "loss": 116.8931,
+      "step": 50740
+    },
+    {
+      "epoch": 0.4198204905488688,
+      "grad_norm": 1682.1556396484375,
+      "learning_rate": 7.345801067784388e-06,
+      "loss": 94.5308,
+      "step": 50750
+    },
+    {
+      "epoch": 0.419903213798238,
+      "grad_norm": 988.3783569335938,
+      "learning_rate": 7.34455546088384e-06,
+      "loss": 100.0454,
+      "step": 50760
+    },
+    {
+      "epoch": 0.4199859370476072,
+      "grad_norm": 721.0875244140625,
+      "learning_rate": 7.343309667437775e-06,
+      "loss": 116.8833,
+      "step": 50770
+    },
+    {
+      "epoch": 0.4200686602969765,
+      "grad_norm": 832.2625122070312,
+      "learning_rate": 7.3420636875453135e-06,
+      "loss": 77.9705,
+      "step": 50780
+    },
+    {
+      "epoch": 0.4201513835463457,
+      "grad_norm": 995.8257446289062,
+      "learning_rate": 7.340817521305595e-06,
+      "loss": 101.0271,
+      "step": 50790
+    },
+    {
+      "epoch": 0.42023410679571493,
+      "grad_norm": 935.89404296875,
+      "learning_rate": 7.3395711688177676e-06,
+      "loss": 123.7885,
+      "step": 50800
+    },
+    {
+      "epoch": 0.4203168300450842,
+      "grad_norm": 957.3890380859375,
+      "learning_rate": 7.3383246301809985e-06,
+      "loss": 79.4234,
+      "step": 50810
+    },
+    {
+      "epoch": 0.4203995532944534,
+      "grad_norm": 1143.3251953125,
+      "learning_rate": 7.337077905494472e-06,
+      "loss": 79.35,
+      "step": 50820
+    },
+    {
+      "epoch": 0.42048227654382264,
+      "grad_norm": 617.7476806640625,
+      "learning_rate": 7.335830994857382e-06,
+      "loss": 83.5682,
+      "step": 50830
+    },
+    {
+      "epoch": 0.4205649997931919,
+      "grad_norm": 923.8682861328125,
+      "learning_rate": 7.334583898368939e-06,
+      "loss": 83.5606,
+      "step": 50840
+    },
+    {
+      "epoch": 0.4206477230425611,
+      "grad_norm": 456.18902587890625,
+      "learning_rate": 7.333336616128369e-06,
+      "loss": 110.9043,
+      "step": 50850
+    },
+    {
+      "epoch": 0.42073044629193035,
+      "grad_norm": 629.4192504882812,
+      "learning_rate": 7.332089148234913e-06,
+      "loss": 84.0271,
+      "step": 50860
+    },
+    {
+      "epoch": 0.4208131695412996,
+      "grad_norm": 777.4642333984375,
+      "learning_rate": 7.330841494787828e-06,
+      "loss": 94.1915,
+      "step": 50870
+    },
+    {
+      "epoch": 0.4208958927906688,
+      "grad_norm": 1021.780029296875,
+      "learning_rate": 7.329593655886382e-06,
+      "loss": 86.0737,
+      "step": 50880
+    },
+    {
+      "epoch": 0.42097861604003806,
+      "grad_norm": 1032.098876953125,
+      "learning_rate": 7.3283456316298595e-06,
+      "loss": 83.7801,
+      "step": 50890
+    },
+    {
+      "epoch": 0.4210613392894073,
+      "grad_norm": 987.5896606445312,
+      "learning_rate": 7.32709742211756e-06,
+      "loss": 86.3313,
+      "step": 50900
+    },
+    {
+      "epoch": 0.4211440625387765,
+      "grad_norm": 799.8026733398438,
+      "learning_rate": 7.325849027448799e-06,
+      "loss": 71.7253,
+      "step": 50910
+    },
+    {
+      "epoch": 0.42122678578814576,
+      "grad_norm": 675.6878051757812,
+      "learning_rate": 7.324600447722907e-06,
+      "loss": 79.754,
+      "step": 50920
+    },
+    {
+      "epoch": 0.421309509037515,
+      "grad_norm": 578.5625,
+      "learning_rate": 7.323351683039224e-06,
+      "loss": 103.9349,
+      "step": 50930
+    },
+    {
+      "epoch": 0.4213922322868842,
+      "grad_norm": 1150.8740234375,
+      "learning_rate": 7.32210273349711e-06,
+      "loss": 101.4098,
+      "step": 50940
+    },
+    {
+      "epoch": 0.42147495553625347,
+      "grad_norm": 1209.810546875,
+      "learning_rate": 7.32085359919594e-06,
+      "loss": 112.2176,
+      "step": 50950
+    },
+    {
+      "epoch": 0.4215576787856227,
+      "grad_norm": 1046.1680908203125,
+      "learning_rate": 7.3196042802350995e-06,
+      "loss": 81.5384,
+      "step": 50960
+    },
+    {
+      "epoch": 0.4216404020349919,
+      "grad_norm": 1647.843017578125,
+      "learning_rate": 7.3183547767139916e-06,
+      "loss": 106.6991,
+      "step": 50970
+    },
+    {
+      "epoch": 0.4217231252843612,
+      "grad_norm": 1031.4130859375,
+      "learning_rate": 7.317105088732035e-06,
+      "loss": 102.2982,
+      "step": 50980
+    },
+    {
+      "epoch": 0.42180584853373043,
+      "grad_norm": 739.594482421875,
+      "learning_rate": 7.31585521638866e-06,
+      "loss": 83.5706,
+      "step": 50990
+    },
+    {
+      "epoch": 0.42188857178309963,
+      "grad_norm": 1355.9227294921875,
+      "learning_rate": 7.314605159783313e-06,
+      "loss": 155.9577,
+      "step": 51000
+    },
+    {
+      "epoch": 0.4219712950324689,
+      "grad_norm": 752.1953735351562,
+      "learning_rate": 7.313354919015457e-06,
+      "loss": 87.4393,
+      "step": 51010
+    },
+    {
+      "epoch": 0.4220540182818381,
+      "grad_norm": 1794.4256591796875,
+      "learning_rate": 7.312104494184566e-06,
+      "loss": 140.2643,
+      "step": 51020
+    },
+    {
+      "epoch": 0.42213674153120734,
+      "grad_norm": 1220.9268798828125,
+      "learning_rate": 7.310853885390133e-06,
+      "loss": 98.7091,
+      "step": 51030
+    },
+    {
+      "epoch": 0.4222194647805766,
+      "grad_norm": 2305.75,
+      "learning_rate": 7.309603092731661e-06,
+      "loss": 112.4157,
+      "step": 51040
+    },
+    {
+      "epoch": 0.4223021880299458,
+      "grad_norm": 1095.7064208984375,
+      "learning_rate": 7.30835211630867e-06,
+      "loss": 102.6219,
+      "step": 51050
+    },
+    {
+      "epoch": 0.42238491127931505,
+      "grad_norm": 678.8933715820312,
+      "learning_rate": 7.3071009562206965e-06,
+      "loss": 88.1793,
+      "step": 51060
+    },
+    {
+      "epoch": 0.4224676345286843,
+      "grad_norm": 809.6482543945312,
+      "learning_rate": 7.305849612567287e-06,
+      "loss": 99.8888,
+      "step": 51070
+    },
+    {
+      "epoch": 0.4225503577780535,
+      "grad_norm": 1535.9327392578125,
+      "learning_rate": 7.304598085448007e-06,
+      "loss": 119.08,
+      "step": 51080
+    },
+    {
+      "epoch": 0.42263308102742275,
+      "grad_norm": 953.1244506835938,
+      "learning_rate": 7.303346374962433e-06,
+      "loss": 86.5312,
+      "step": 51090
+    },
+    {
+      "epoch": 0.422715804276792,
+      "grad_norm": 1713.26318359375,
+      "learning_rate": 7.302094481210159e-06,
+      "loss": 103.2178,
+      "step": 51100
+    },
+    {
+      "epoch": 0.4227985275261612,
+      "grad_norm": 826.4808959960938,
+      "learning_rate": 7.300842404290792e-06,
+      "loss": 96.0054,
+      "step": 51110
+    },
+    {
+      "epoch": 0.42288125077553046,
+      "grad_norm": 1028.5643310546875,
+      "learning_rate": 7.2995901443039554e-06,
+      "loss": 128.5591,
+      "step": 51120
+    },
+    {
+      "epoch": 0.4229639740248997,
+      "grad_norm": 840.199462890625,
+      "learning_rate": 7.298337701349285e-06,
+      "loss": 116.7384,
+      "step": 51130
+    },
+    {
+      "epoch": 0.4230466972742689,
+      "grad_norm": 672.873779296875,
+      "learning_rate": 7.29708507552643e-06,
+      "loss": 68.3276,
+      "step": 51140
+    },
+    {
+      "epoch": 0.42312942052363817,
+      "grad_norm": 383.77313232421875,
+      "learning_rate": 7.295832266935059e-06,
+      "loss": 84.0062,
+      "step": 51150
+    },
+    {
+      "epoch": 0.4232121437730074,
+      "grad_norm": 1001.8519287109375,
+      "learning_rate": 7.2945792756748505e-06,
+      "loss": 140.7101,
+      "step": 51160
+    },
+    {
+      "epoch": 0.4232948670223766,
+      "grad_norm": 709.9981689453125,
+      "learning_rate": 7.2933261018455005e-06,
+      "loss": 75.1831,
+      "step": 51170
+    },
+    {
+      "epoch": 0.4233775902717459,
+      "grad_norm": 675.74658203125,
+      "learning_rate": 7.292072745546716e-06,
+      "loss": 73.5509,
+      "step": 51180
+    },
+    {
+      "epoch": 0.42346031352111513,
+      "grad_norm": 886.9816284179688,
+      "learning_rate": 7.290819206878223e-06,
+      "loss": 92.8137,
+      "step": 51190
+    },
+    {
+      "epoch": 0.4235430367704843,
+      "grad_norm": 688.6138305664062,
+      "learning_rate": 7.289565485939759e-06,
+      "loss": 109.6997,
+      "step": 51200
+    },
+    {
+      "epoch": 0.4236257600198536,
+      "grad_norm": 1372.497802734375,
+      "learning_rate": 7.288311582831078e-06,
+      "loss": 86.5049,
+      "step": 51210
+    },
+    {
+      "epoch": 0.42370848326922284,
+      "grad_norm": 632.5243530273438,
+      "learning_rate": 7.2870574976519455e-06,
+      "loss": 73.3826,
+      "step": 51220
+    },
+    {
+      "epoch": 0.42379120651859203,
+      "grad_norm": 564.6524658203125,
+      "learning_rate": 7.2858032305021455e-06,
+      "loss": 123.0075,
+      "step": 51230
+    },
+    {
+      "epoch": 0.4238739297679613,
+      "grad_norm": 758.579833984375,
+      "learning_rate": 7.28454878148147e-06,
+      "loss": 97.95,
+      "step": 51240
+    },
+    {
+      "epoch": 0.42395665301733054,
+      "grad_norm": 1073.8216552734375,
+      "learning_rate": 7.283294150689735e-06,
+      "loss": 88.5506,
+      "step": 51250
+    },
+    {
+      "epoch": 0.42403937626669974,
+      "grad_norm": 621.6570434570312,
+      "learning_rate": 7.282039338226763e-06,
+      "loss": 106.7351,
+      "step": 51260
+    },
+    {
+      "epoch": 0.424122099516069,
+      "grad_norm": 1591.994384765625,
+      "learning_rate": 7.280784344192393e-06,
+      "loss": 117.9606,
+      "step": 51270
+    },
+    {
+      "epoch": 0.42420482276543825,
+      "grad_norm": 811.947265625,
+      "learning_rate": 7.279529168686481e-06,
+      "loss": 83.5623,
+      "step": 51280
+    },
+    {
+      "epoch": 0.42428754601480745,
+      "grad_norm": 939.7516479492188,
+      "learning_rate": 7.278273811808894e-06,
+      "loss": 91.11,
+      "step": 51290
+    },
+    {
+      "epoch": 0.4243702692641767,
+      "grad_norm": 1210.847900390625,
+      "learning_rate": 7.2770182736595164e-06,
+      "loss": 130.9754,
+      "step": 51300
+    },
+    {
+      "epoch": 0.42445299251354596,
+      "grad_norm": 1220.7769775390625,
+      "learning_rate": 7.275762554338244e-06,
+      "loss": 76.7486,
+      "step": 51310
+    },
+    {
+      "epoch": 0.42453571576291516,
+      "grad_norm": 1265.53271484375,
+      "learning_rate": 7.2745066539449905e-06,
+      "loss": 95.9141,
+      "step": 51320
+    },
+    {
+      "epoch": 0.4246184390122844,
+      "grad_norm": 875.6475219726562,
+      "learning_rate": 7.27325057257968e-06,
+      "loss": 97.6407,
+      "step": 51330
+    },
+    {
+      "epoch": 0.42470116226165366,
+      "grad_norm": 693.2019653320312,
+      "learning_rate": 7.271994310342254e-06,
+      "loss": 99.6023,
+      "step": 51340
+    },
+    {
+      "epoch": 0.42478388551102286,
+      "grad_norm": 897.9375,
+      "learning_rate": 7.270737867332669e-06,
+      "loss": 104.1033,
+      "step": 51350
+    },
+    {
+      "epoch": 0.4248666087603921,
+      "grad_norm": 1381.2633056640625,
+      "learning_rate": 7.2694812436508934e-06,
+      "loss": 98.0639,
+      "step": 51360
+    },
+    {
+      "epoch": 0.4249493320097613,
+      "grad_norm": 825.0390625,
+      "learning_rate": 7.268224439396909e-06,
+      "loss": 109.9973,
+      "step": 51370
+    },
+    {
+      "epoch": 0.42503205525913057,
+      "grad_norm": 997.9774780273438,
+      "learning_rate": 7.266967454670717e-06,
+      "loss": 115.4315,
+      "step": 51380
+    },
+    {
+      "epoch": 0.4251147785084998,
+      "grad_norm": 1014.7069702148438,
+      "learning_rate": 7.265710289572328e-06,
+      "loss": 122.6427,
+      "step": 51390
+    },
+    {
+      "epoch": 0.425197501757869,
+      "grad_norm": 1186.5941162109375,
+      "learning_rate": 7.264452944201771e-06,
+      "loss": 79.164,
+      "step": 51400
+    },
+    {
+      "epoch": 0.4252802250072383,
+      "grad_norm": 896.6864013671875,
+      "learning_rate": 7.263195418659083e-06,
+      "loss": 94.9668,
+      "step": 51410
+    },
+    {
+      "epoch": 0.42536294825660753,
+      "grad_norm": 1269.0845947265625,
+      "learning_rate": 7.261937713044325e-06,
+      "loss": 111.4878,
+      "step": 51420
+    },
+    {
+      "epoch": 0.42544567150597673,
+      "grad_norm": 737.5647583007812,
+      "learning_rate": 7.260679827457562e-06,
+      "loss": 144.4097,
+      "step": 51430
+    },
+    {
+      "epoch": 0.425528394755346,
+      "grad_norm": 1106.5899658203125,
+      "learning_rate": 7.259421761998881e-06,
+      "loss": 97.2455,
+      "step": 51440
+    },
+    {
+      "epoch": 0.42561111800471524,
+      "grad_norm": 867.489013671875,
+      "learning_rate": 7.2581635167683805e-06,
+      "loss": 105.3224,
+      "step": 51450
+    },
+    {
+      "epoch": 0.42569384125408444,
+      "grad_norm": 693.6383666992188,
+      "learning_rate": 7.256905091866171e-06,
+      "loss": 101.9842,
+      "step": 51460
+    },
+    {
+      "epoch": 0.4257765645034537,
+      "grad_norm": 611.8115844726562,
+      "learning_rate": 7.255646487392382e-06,
+      "loss": 115.6671,
+      "step": 51470
+    },
+    {
+      "epoch": 0.42585928775282295,
+      "grad_norm": 1504.69775390625,
+      "learning_rate": 7.254387703447154e-06,
+      "loss": 95.3933,
+      "step": 51480
+    },
+    {
+      "epoch": 0.42594201100219214,
+      "grad_norm": 600.2556762695312,
+      "learning_rate": 7.2531287401306435e-06,
+      "loss": 92.5239,
+      "step": 51490
+    },
+    {
+      "epoch": 0.4260247342515614,
+      "grad_norm": 407.8628845214844,
+      "learning_rate": 7.251869597543019e-06,
+      "loss": 82.7194,
+      "step": 51500
+    },
+    {
+      "epoch": 0.42610745750093065,
+      "grad_norm": 614.1116943359375,
+      "learning_rate": 7.250610275784464e-06,
+      "loss": 93.6232,
+      "step": 51510
+    },
+    {
+      "epoch": 0.42619018075029985,
+      "grad_norm": 898.9727172851562,
+      "learning_rate": 7.2493507749551795e-06,
+      "loss": 85.4422,
+      "step": 51520
+    },
+    {
+      "epoch": 0.4262729039996691,
+      "grad_norm": 1155.338623046875,
+      "learning_rate": 7.248091095155378e-06,
+      "loss": 118.6169,
+      "step": 51530
+    },
+    {
+      "epoch": 0.42635562724903836,
+      "grad_norm": 1159.13916015625,
+      "learning_rate": 7.246831236485283e-06,
+      "loss": 135.8041,
+      "step": 51540
+    },
+    {
+      "epoch": 0.42643835049840756,
+      "grad_norm": 928.2463989257812,
+      "learning_rate": 7.245571199045139e-06,
+      "loss": 108.4106,
+      "step": 51550
+    },
+    {
+      "epoch": 0.4265210737477768,
+      "grad_norm": 767.41259765625,
+      "learning_rate": 7.244310982935202e-06,
+      "loss": 74.6865,
+      "step": 51560
+    },
+    {
+      "epoch": 0.42660379699714607,
+      "grad_norm": 709.4650268554688,
+      "learning_rate": 7.243050588255738e-06,
+      "loss": 76.4944,
+      "step": 51570
+    },
+    {
+      "epoch": 0.42668652024651527,
+      "grad_norm": 1164.441162109375,
+      "learning_rate": 7.241790015107034e-06,
+      "loss": 88.2155,
+      "step": 51580
+    },
+    {
+      "epoch": 0.4267692434958845,
+      "grad_norm": 1503.4273681640625,
+      "learning_rate": 7.240529263589386e-06,
+      "loss": 126.4441,
+      "step": 51590
+    },
+    {
+      "epoch": 0.4268519667452538,
+      "grad_norm": 760.6532592773438,
+      "learning_rate": 7.239268333803109e-06,
+      "loss": 105.8668,
+      "step": 51600
+    },
+    {
+      "epoch": 0.426934689994623,
+      "grad_norm": 1101.9364013671875,
+      "learning_rate": 7.2380072258485265e-06,
+      "loss": 92.2242,
+      "step": 51610
+    },
+    {
+      "epoch": 0.4270174132439922,
+      "grad_norm": 554.7161254882812,
+      "learning_rate": 7.2367459398259795e-06,
+      "loss": 83.6779,
+      "step": 51620
+    },
+    {
+      "epoch": 0.4271001364933615,
+      "grad_norm": 972.8421020507812,
+      "learning_rate": 7.2354844758358234e-06,
+      "loss": 123.7181,
+      "step": 51630
+    },
+    {
+      "epoch": 0.4271828597427307,
+      "grad_norm": 1161.73828125,
+      "learning_rate": 7.234222833978427e-06,
+      "loss": 106.6433,
+      "step": 51640
+    },
+    {
+      "epoch": 0.42726558299209993,
+      "grad_norm": 1067.09765625,
+      "learning_rate": 7.232961014354175e-06,
+      "loss": 104.9934,
+      "step": 51650
+    },
+    {
+      "epoch": 0.4273483062414692,
+      "grad_norm": 1101.327392578125,
+      "learning_rate": 7.23169901706346e-06,
+      "loss": 106.5562,
+      "step": 51660
+    },
+    {
+      "epoch": 0.4274310294908384,
+      "grad_norm": 1341.46044921875,
+      "learning_rate": 7.2304368422067e-06,
+      "loss": 134.7537,
+      "step": 51670
+    },
+    {
+      "epoch": 0.42751375274020764,
+      "grad_norm": 1436.9295654296875,
+      "learning_rate": 7.2291744898843145e-06,
+      "loss": 98.4549,
+      "step": 51680
+    },
+    {
+      "epoch": 0.4275964759895769,
+      "grad_norm": 540.67236328125,
+      "learning_rate": 7.227911960196746e-06,
+      "loss": 86.389,
+      "step": 51690
+    },
+    {
+      "epoch": 0.4276791992389461,
+      "grad_norm": 1108.2845458984375,
+      "learning_rate": 7.226649253244448e-06,
+      "loss": 74.8376,
+      "step": 51700
+    },
+    {
+      "epoch": 0.42776192248831535,
+      "grad_norm": 964.5218505859375,
+      "learning_rate": 7.225386369127886e-06,
+      "loss": 112.0215,
+      "step": 51710
+    },
+    {
+      "epoch": 0.4278446457376846,
+      "grad_norm": 768.393310546875,
+      "learning_rate": 7.224123307947545e-06,
+      "loss": 94.5367,
+      "step": 51720
+    },
+    {
+      "epoch": 0.4279273689870538,
+      "grad_norm": 708.0656127929688,
+      "learning_rate": 7.2228600698039205e-06,
+      "loss": 108.4423,
+      "step": 51730
+    },
+    {
+      "epoch": 0.42801009223642306,
+      "grad_norm": 1345.224853515625,
+      "learning_rate": 7.221596654797522e-06,
+      "loss": 91.7173,
+      "step": 51740
+    },
+    {
+      "epoch": 0.42809281548579226,
+      "grad_norm": 1059.3037109375,
+      "learning_rate": 7.2203330630288714e-06,
+      "loss": 109.0566,
+      "step": 51750
+    },
+    {
+      "epoch": 0.4281755387351615,
+      "grad_norm": 975.2772216796875,
+      "learning_rate": 7.21906929459851e-06,
+      "loss": 140.6817,
+      "step": 51760
+    },
+    {
+      "epoch": 0.42825826198453076,
+      "grad_norm": 1177.9395751953125,
+      "learning_rate": 7.217805349606988e-06,
+      "loss": 91.187,
+      "step": 51770
+    },
+    {
+      "epoch": 0.42834098523389996,
+      "grad_norm": 685.8370971679688,
+      "learning_rate": 7.216541228154875e-06,
+      "loss": 93.8095,
+      "step": 51780
+    },
+    {
+      "epoch": 0.4284237084832692,
+      "grad_norm": 1138.31689453125,
+      "learning_rate": 7.215276930342747e-06,
+      "loss": 104.4566,
+      "step": 51790
+    },
+    {
+      "epoch": 0.42850643173263847,
+      "grad_norm": 601.1682739257812,
+      "learning_rate": 7.214012456271202e-06,
+      "loss": 130.8603,
+      "step": 51800
+    },
+    {
+      "epoch": 0.42858915498200767,
+      "grad_norm": 1000.6312866210938,
+      "learning_rate": 7.212747806040845e-06,
+      "loss": 123.3161,
+      "step": 51810
+    },
+    {
+      "epoch": 0.4286718782313769,
+      "grad_norm": 1142.5823974609375,
+      "learning_rate": 7.211482979752302e-06,
+      "loss": 85.4368,
+      "step": 51820
+    },
+    {
+      "epoch": 0.4287546014807462,
+      "grad_norm": 816.385986328125,
+      "learning_rate": 7.210217977506207e-06,
+      "loss": 109.9975,
+      "step": 51830
+    },
+    {
+      "epoch": 0.4288373247301154,
+      "grad_norm": 944.0142822265625,
+      "learning_rate": 7.208952799403211e-06,
+      "loss": 108.7334,
+      "step": 51840
+    },
+    {
+      "epoch": 0.42892004797948463,
+      "grad_norm": 824.8801879882812,
+      "learning_rate": 7.207687445543977e-06,
+      "loss": 82.7929,
+      "step": 51850
+    },
+    {
+      "epoch": 0.4290027712288539,
+      "grad_norm": 964.5783081054688,
+      "learning_rate": 7.206421916029187e-06,
+      "loss": 121.5998,
+      "step": 51860
+    },
+    {
+      "epoch": 0.4290854944782231,
+      "grad_norm": 1063.656982421875,
+      "learning_rate": 7.205156210959529e-06,
+      "loss": 90.583,
+      "step": 51870
+    },
+    {
+      "epoch": 0.42916821772759234,
+      "grad_norm": 3292.513671875,
+      "learning_rate": 7.203890330435715e-06,
+      "loss": 105.8095,
+      "step": 51880
+    },
+    {
+      "epoch": 0.4292509409769616,
+      "grad_norm": 703.033447265625,
+      "learning_rate": 7.202624274558458e-06,
+      "loss": 106.7044,
+      "step": 51890
+    },
+    {
+      "epoch": 0.4293336642263308,
+      "grad_norm": 669.6672973632812,
+      "learning_rate": 7.201358043428499e-06,
+      "loss": 89.5573,
+      "step": 51900
+    },
+    {
+      "epoch": 0.42941638747570005,
+      "grad_norm": 647.8473510742188,
+      "learning_rate": 7.200091637146582e-06,
+      "loss": 99.5425,
+      "step": 51910
+    },
+    {
+      "epoch": 0.4294991107250693,
+      "grad_norm": 1656.5313720703125,
+      "learning_rate": 7.198825055813471e-06,
+      "loss": 120.4202,
+      "step": 51920
+    },
+    {
+      "epoch": 0.4295818339744385,
+      "grad_norm": 755.7421875,
+      "learning_rate": 7.197558299529941e-06,
+      "loss": 74.5723,
+      "step": 51930
+    },
+    {
+      "epoch": 0.42966455722380775,
+      "grad_norm": 1194.8021240234375,
+      "learning_rate": 7.196291368396784e-06,
+      "loss": 104.8849,
+      "step": 51940
+    },
+    {
+      "epoch": 0.429747280473177,
+      "grad_norm": 810.8287353515625,
+      "learning_rate": 7.1950242625148e-06,
+      "loss": 109.7433,
+      "step": 51950
+    },
+    {
+      "epoch": 0.4298300037225462,
+      "grad_norm": 5541.013671875,
+      "learning_rate": 7.1937569819848115e-06,
+      "loss": 117.9274,
+      "step": 51960
+    },
+    {
+      "epoch": 0.42991272697191546,
+      "grad_norm": 1193.3897705078125,
+      "learning_rate": 7.192489526907646e-06,
+      "loss": 112.9372,
+      "step": 51970
+    },
+    {
+      "epoch": 0.4299954502212847,
+      "grad_norm": 1110.0302734375,
+      "learning_rate": 7.191221897384153e-06,
+      "loss": 120.9369,
+      "step": 51980
+    },
+    {
+      "epoch": 0.4300781734706539,
+      "grad_norm": 1447.90771484375,
+      "learning_rate": 7.189954093515189e-06,
+      "loss": 142.4958,
+      "step": 51990
+    },
+    {
+      "epoch": 0.43016089672002317,
+      "grad_norm": 1462.6444091796875,
+      "learning_rate": 7.188686115401628e-06,
+      "loss": 127.5024,
+      "step": 52000
+    },
+    {
+      "epoch": 0.4302436199693924,
+      "grad_norm": 1592.32666015625,
+      "learning_rate": 7.187417963144358e-06,
+      "loss": 102.1783,
+      "step": 52010
+    },
+    {
+      "epoch": 0.4303263432187616,
+      "grad_norm": 356.8098449707031,
+      "learning_rate": 7.18614963684428e-06,
+      "loss": 94.6831,
+      "step": 52020
+    },
+    {
+      "epoch": 0.4304090664681309,
+      "grad_norm": 762.9508666992188,
+      "learning_rate": 7.184881136602309e-06,
+      "loss": 117.5537,
+      "step": 52030
+    },
+    {
+      "epoch": 0.43049178971750013,
+      "grad_norm": 614.644775390625,
+      "learning_rate": 7.183612462519371e-06,
+      "loss": 120.3886,
+      "step": 52040
+    },
+    {
+      "epoch": 0.4305745129668693,
+      "grad_norm": 884.6917724609375,
+      "learning_rate": 7.182343614696412e-06,
+      "loss": 87.1297,
+      "step": 52050
+    },
+    {
+      "epoch": 0.4306572362162386,
+      "grad_norm": 1503.93603515625,
+      "learning_rate": 7.181074593234387e-06,
+      "loss": 102.104,
+      "step": 52060
+    },
+    {
+      "epoch": 0.43073995946560784,
+      "grad_norm": 431.5899963378906,
+      "learning_rate": 7.179805398234266e-06,
+      "loss": 93.1396,
+      "step": 52070
+    },
+    {
+      "epoch": 0.43082268271497703,
+      "grad_norm": 1156.7777099609375,
+      "learning_rate": 7.178536029797035e-06,
+      "loss": 113.4172,
+      "step": 52080
+    },
+    {
+      "epoch": 0.4309054059643463,
+      "grad_norm": 680.9636840820312,
+      "learning_rate": 7.177266488023688e-06,
+      "loss": 92.4326,
+      "step": 52090
+    },
+    {
+      "epoch": 0.4309881292137155,
+      "grad_norm": 1016.119873046875,
+      "learning_rate": 7.17599677301524e-06,
+      "loss": 83.3582,
+      "step": 52100
+    },
+    {
+      "epoch": 0.43107085246308474,
+      "grad_norm": 625.1591796875,
+      "learning_rate": 7.174726884872716e-06,
+      "loss": 85.9998,
+      "step": 52110
+    },
+    {
+      "epoch": 0.431153575712454,
+      "grad_norm": 1880.2071533203125,
+      "learning_rate": 7.173456823697154e-06,
+      "loss": 148.4656,
+      "step": 52120
+    },
+    {
+      "epoch": 0.4312362989618232,
+      "grad_norm": 559.7216796875,
+      "learning_rate": 7.172186589589607e-06,
+      "loss": 117.4733,
+      "step": 52130
+    },
+    {
+      "epoch": 0.43131902221119245,
+      "grad_norm": 878.204833984375,
+      "learning_rate": 7.170916182651141e-06,
+      "loss": 72.1657,
+      "step": 52140
+    },
+    {
+      "epoch": 0.4314017454605617,
+      "grad_norm": 785.7388916015625,
+      "learning_rate": 7.1696456029828386e-06,
+      "loss": 78.2875,
+      "step": 52150
+    },
+    {
+      "epoch": 0.4314844687099309,
+      "grad_norm": 960.450439453125,
+      "learning_rate": 7.168374850685794e-06,
+      "loss": 99.7961,
+      "step": 52160
+    },
+    {
+      "epoch": 0.43156719195930016,
+      "grad_norm": 1000.7466430664062,
+      "learning_rate": 7.167103925861113e-06,
+      "loss": 113.2332,
+      "step": 52170
+    },
+    {
+      "epoch": 0.4316499152086694,
+      "grad_norm": 923.0742797851562,
+      "learning_rate": 7.165832828609918e-06,
+      "loss": 108.3951,
+      "step": 52180
+    },
+    {
+      "epoch": 0.4317326384580386,
+      "grad_norm": 1476.0849609375,
+      "learning_rate": 7.164561559033344e-06,
+      "loss": 104.3691,
+      "step": 52190
+    },
+    {
+      "epoch": 0.43181536170740786,
+      "grad_norm": 656.1659545898438,
+      "learning_rate": 7.163290117232542e-06,
+      "loss": 111.1306,
+      "step": 52200
+    },
+    {
+      "epoch": 0.4318980849567771,
+      "grad_norm": 1209.8184814453125,
+      "learning_rate": 7.162018503308674e-06,
+      "loss": 105.6424,
+      "step": 52210
+    },
+    {
+      "epoch": 0.4319808082061463,
+      "grad_norm": 780.0021362304688,
+      "learning_rate": 7.1607467173629145e-06,
+      "loss": 90.2464,
+      "step": 52220
+    },
+    {
+      "epoch": 0.43206353145551557,
+      "grad_norm": 785.8810424804688,
+      "learning_rate": 7.1594747594964564e-06,
+      "loss": 123.5292,
+      "step": 52230
+    },
+    {
+      "epoch": 0.4321462547048848,
+      "grad_norm": 1079.1829833984375,
+      "learning_rate": 7.1582026298105e-06,
+      "loss": 79.3976,
+      "step": 52240
+    },
+    {
+      "epoch": 0.432228977954254,
+      "grad_norm": 1131.198974609375,
+      "learning_rate": 7.156930328406268e-06,
+      "loss": 137.1325,
+      "step": 52250
+    },
+    {
+      "epoch": 0.4323117012036233,
+      "grad_norm": 1154.282470703125,
+      "learning_rate": 7.1556578553849875e-06,
+      "loss": 78.7498,
+      "step": 52260
+    },
+    {
+      "epoch": 0.43239442445299253,
+      "grad_norm": 783.6915283203125,
+      "learning_rate": 7.154385210847905e-06,
+      "loss": 96.9286,
+      "step": 52270
+    },
+    {
+      "epoch": 0.43247714770236173,
+      "grad_norm": 1002.9415893554688,
+      "learning_rate": 7.153112394896279e-06,
+      "loss": 101.9183,
+      "step": 52280
+    },
+    {
+      "epoch": 0.432559870951731,
+      "grad_norm": 805.485595703125,
+      "learning_rate": 7.15183940763138e-06,
+      "loss": 84.4508,
+      "step": 52290
+    },
+    {
+      "epoch": 0.43264259420110024,
+      "grad_norm": 972.188720703125,
+      "learning_rate": 7.150566249154496e-06,
+      "loss": 112.0686,
+      "step": 52300
+    },
+    {
+      "epoch": 0.43272531745046944,
+      "grad_norm": 694.895263671875,
+      "learning_rate": 7.149292919566924e-06,
+      "loss": 86.5633,
+      "step": 52310
+    },
+    {
+      "epoch": 0.4328080406998387,
+      "grad_norm": 507.7707214355469,
+      "learning_rate": 7.148019418969979e-06,
+      "loss": 108.7999,
+      "step": 52320
+    },
+    {
+      "epoch": 0.43289076394920795,
+      "grad_norm": 928.7743530273438,
+      "learning_rate": 7.146745747464987e-06,
+      "loss": 133.0144,
+      "step": 52330
+    },
+    {
+      "epoch": 0.43297348719857714,
+      "grad_norm": 888.1306762695312,
+      "learning_rate": 7.145471905153288e-06,
+      "loss": 74.0938,
+      "step": 52340
+    },
+    {
+      "epoch": 0.4330562104479464,
+      "grad_norm": 602.6085815429688,
+      "learning_rate": 7.1441978921362365e-06,
+      "loss": 106.0087,
+      "step": 52350
+    },
+    {
+      "epoch": 0.43313893369731565,
+      "grad_norm": 739.8253784179688,
+      "learning_rate": 7.142923708515199e-06,
+      "loss": 130.5773,
+      "step": 52360
+    },
+    {
+      "epoch": 0.43322165694668485,
+      "grad_norm": 561.8637084960938,
+      "learning_rate": 7.141649354391556e-06,
+      "loss": 116.5679,
+      "step": 52370
+    },
+    {
+      "epoch": 0.4333043801960541,
+      "grad_norm": 635.5332641601562,
+      "learning_rate": 7.140374829866703e-06,
+      "loss": 102.7501,
+      "step": 52380
+    },
+    {
+      "epoch": 0.43338710344542336,
+      "grad_norm": 849.3636474609375,
+      "learning_rate": 7.1391001350420486e-06,
+      "loss": 112.318,
+      "step": 52390
+    },
+    {
+      "epoch": 0.43346982669479256,
+      "grad_norm": 968.818603515625,
+      "learning_rate": 7.137825270019012e-06,
+      "loss": 116.6222,
+      "step": 52400
+    },
+    {
+      "epoch": 0.4335525499441618,
+      "grad_norm": 879.7605590820312,
+      "learning_rate": 7.1365502348990315e-06,
+      "loss": 72.7146,
+      "step": 52410
+    },
+    {
+      "epoch": 0.43363527319353107,
+      "grad_norm": 974.20947265625,
+      "learning_rate": 7.135275029783554e-06,
+      "loss": 87.4071,
+      "step": 52420
+    },
+    {
+      "epoch": 0.43371799644290027,
+      "grad_norm": 898.2373046875,
+      "learning_rate": 7.133999654774041e-06,
+      "loss": 96.0373,
+      "step": 52430
+    },
+    {
+      "epoch": 0.4338007196922695,
+      "grad_norm": 1289.7435302734375,
+      "learning_rate": 7.13272410997197e-06,
+      "loss": 121.3297,
+      "step": 52440
+    },
+    {
+      "epoch": 0.4338834429416388,
+      "grad_norm": 406.2857666015625,
+      "learning_rate": 7.13144839547883e-06,
+      "loss": 72.3889,
+      "step": 52450
+    },
+    {
+      "epoch": 0.433966166191008,
+      "grad_norm": 1770.5550537109375,
+      "learning_rate": 7.130172511396123e-06,
+      "loss": 157.1078,
+      "step": 52460
+    },
+    {
+      "epoch": 0.43404888944037723,
+      "grad_norm": 924.5420532226562,
+      "learning_rate": 7.128896457825364e-06,
+      "loss": 94.4494,
+      "step": 52470
+    },
+    {
+      "epoch": 0.4341316126897464,
+      "grad_norm": 1168.6419677734375,
+      "learning_rate": 7.127620234868085e-06,
+      "loss": 151.9046,
+      "step": 52480
+    },
+    {
+      "epoch": 0.4342143359391157,
+      "grad_norm": 923.4803466796875,
+      "learning_rate": 7.126343842625828e-06,
+      "loss": 103.1586,
+      "step": 52490
+    },
+    {
+      "epoch": 0.43429705918848494,
+      "grad_norm": 849.2039794921875,
+      "learning_rate": 7.1250672812001505e-06,
+      "loss": 98.9931,
+      "step": 52500
+    },
+    {
+      "epoch": 0.43437978243785413,
+      "grad_norm": 1026.3223876953125,
+      "learning_rate": 7.123790550692624e-06,
+      "loss": 81.8112,
+      "step": 52510
+    },
+    {
+      "epoch": 0.4344625056872234,
+      "grad_norm": 1015.2179565429688,
+      "learning_rate": 7.1225136512048275e-06,
+      "loss": 98.0542,
+      "step": 52520
+    },
+    {
+      "epoch": 0.43454522893659264,
+      "grad_norm": 756.0176391601562,
+      "learning_rate": 7.1212365828383615e-06,
+      "loss": 143.049,
+      "step": 52530
+    },
+    {
+      "epoch": 0.43462795218596184,
+      "grad_norm": 919.13427734375,
+      "learning_rate": 7.119959345694835e-06,
+      "loss": 89.8264,
+      "step": 52540
+    },
+    {
+      "epoch": 0.4347106754353311,
+      "grad_norm": 693.4651489257812,
+      "learning_rate": 7.118681939875875e-06,
+      "loss": 129.897,
+      "step": 52550
+    },
+    {
+      "epoch": 0.43479339868470035,
+      "grad_norm": 1032.639404296875,
+      "learning_rate": 7.117404365483116e-06,
+      "loss": 109.7115,
+      "step": 52560
+    },
+    {
+      "epoch": 0.43487612193406955,
+      "grad_norm": 644.7308959960938,
+      "learning_rate": 7.116126622618207e-06,
+      "loss": 121.4155,
+      "step": 52570
+    },
+    {
+      "epoch": 0.4349588451834388,
+      "grad_norm": 1076.7471923828125,
+      "learning_rate": 7.114848711382816e-06,
+      "loss": 105.8533,
+      "step": 52580
+    },
+    {
+      "epoch": 0.43504156843280806,
+      "grad_norm": 682.4072265625,
+      "learning_rate": 7.1135706318786195e-06,
+      "loss": 126.4976,
+      "step": 52590
+    },
+    {
+      "epoch": 0.43512429168217726,
+      "grad_norm": 334.1803894042969,
+      "learning_rate": 7.112292384207306e-06,
+      "loss": 72.7947,
+      "step": 52600
+    },
+    {
+      "epoch": 0.4352070149315465,
+      "grad_norm": 501.4839172363281,
+      "learning_rate": 7.111013968470581e-06,
+      "loss": 88.0988,
+      "step": 52610
+    },
+    {
+      "epoch": 0.43528973818091576,
+      "grad_norm": 1030.7449951171875,
+      "learning_rate": 7.109735384770166e-06,
+      "loss": 92.1345,
+      "step": 52620
+    },
+    {
+      "epoch": 0.43537246143028496,
+      "grad_norm": 1418.4031982421875,
+      "learning_rate": 7.108456633207787e-06,
+      "loss": 137.2343,
+      "step": 52630
+    },
+    {
+      "epoch": 0.4354551846796542,
+      "grad_norm": 1056.733154296875,
+      "learning_rate": 7.10717771388519e-06,
+      "loss": 122.0539,
+      "step": 52640
+    },
+    {
+      "epoch": 0.43553790792902347,
+      "grad_norm": 567.1400756835938,
+      "learning_rate": 7.105898626904134e-06,
+      "loss": 97.4046,
+      "step": 52650
+    },
+    {
+      "epoch": 0.43562063117839267,
+      "grad_norm": 557.343017578125,
+      "learning_rate": 7.104619372366387e-06,
+      "loss": 97.5606,
+      "step": 52660
+    },
+    {
+      "epoch": 0.4357033544277619,
+      "grad_norm": 663.3614501953125,
+      "learning_rate": 7.103339950373737e-06,
+      "loss": 78.0228,
+      "step": 52670
+    },
+    {
+      "epoch": 0.4357860776771312,
+      "grad_norm": 726.0701293945312,
+      "learning_rate": 7.102060361027981e-06,
+      "loss": 122.9625,
+      "step": 52680
+    },
+    {
+      "epoch": 0.4358688009265004,
+      "grad_norm": 833.1370239257812,
+      "learning_rate": 7.100780604430928e-06,
+      "loss": 92.9005,
+      "step": 52690
+    },
+    {
+      "epoch": 0.43595152417586963,
+      "grad_norm": 1035.490478515625,
+      "learning_rate": 7.099500680684404e-06,
+      "loss": 82.178,
+      "step": 52700
+    },
+    {
+      "epoch": 0.4360342474252389,
+      "grad_norm": 721.7255859375,
+      "learning_rate": 7.0982205898902444e-06,
+      "loss": 134.4474,
+      "step": 52710
+    },
+    {
+      "epoch": 0.4361169706746081,
+      "grad_norm": 872.9884033203125,
+      "learning_rate": 7.096940332150305e-06,
+      "loss": 112.2354,
+      "step": 52720
+    },
+    {
+      "epoch": 0.43619969392397734,
+      "grad_norm": 504.2191162109375,
+      "learning_rate": 7.095659907566446e-06,
+      "loss": 69.3615,
+      "step": 52730
+    },
+    {
+      "epoch": 0.4362824171733466,
+      "grad_norm": 1016.4193725585938,
+      "learning_rate": 7.094379316240545e-06,
+      "loss": 88.9207,
+      "step": 52740
+    },
+    {
+      "epoch": 0.4363651404227158,
+      "grad_norm": 972.1260986328125,
+      "learning_rate": 7.093098558274494e-06,
+      "loss": 104.1136,
+      "step": 52750
+    },
+    {
+      "epoch": 0.43644786367208505,
+      "grad_norm": 1002.4033203125,
+      "learning_rate": 7.091817633770197e-06,
+      "loss": 94.7899,
+      "step": 52760
+    },
+    {
+      "epoch": 0.4365305869214543,
+      "grad_norm": 880.1514892578125,
+      "learning_rate": 7.090536542829571e-06,
+      "loss": 87.8467,
+      "step": 52770
+    },
+    {
+      "epoch": 0.4366133101708235,
+      "grad_norm": 620.6061401367188,
+      "learning_rate": 7.089255285554546e-06,
+      "loss": 136.4645,
+      "step": 52780
+    },
+    {
+      "epoch": 0.43669603342019275,
+      "grad_norm": 1144.162353515625,
+      "learning_rate": 7.087973862047067e-06,
+      "loss": 100.7308,
+      "step": 52790
+    },
+    {
+      "epoch": 0.436778756669562,
+      "grad_norm": 848.2889404296875,
+      "learning_rate": 7.08669227240909e-06,
+      "loss": 87.7345,
+      "step": 52800
+    },
+    {
+      "epoch": 0.4368614799189312,
+      "grad_norm": 812.3153686523438,
+      "learning_rate": 7.085410516742586e-06,
+      "loss": 101.8244,
+      "step": 52810
+    },
+    {
+      "epoch": 0.43694420316830046,
+      "grad_norm": 872.6069946289062,
+      "learning_rate": 7.084128595149538e-06,
+      "loss": 104.2199,
+      "step": 52820
+    },
+    {
+      "epoch": 0.43702692641766966,
+      "grad_norm": 897.1786499023438,
+      "learning_rate": 7.082846507731942e-06,
+      "loss": 96.152,
+      "step": 52830
+    },
+    {
+      "epoch": 0.4371096496670389,
+      "grad_norm": 631.2103271484375,
+      "learning_rate": 7.081564254591809e-06,
+      "loss": 101.3164,
+      "step": 52840
+    },
+    {
+      "epoch": 0.43719237291640817,
+      "grad_norm": 781.96044921875,
+      "learning_rate": 7.08028183583116e-06,
+      "loss": 114.7007,
+      "step": 52850
+    },
+    {
+      "epoch": 0.43727509616577737,
+      "grad_norm": 995.4803466796875,
+      "learning_rate": 7.078999251552034e-06,
+      "loss": 98.927,
+      "step": 52860
+    },
+    {
+      "epoch": 0.4373578194151466,
+      "grad_norm": 841.7239990234375,
+      "learning_rate": 7.077716501856478e-06,
+      "loss": 106.8644,
+      "step": 52870
+    },
+    {
+      "epoch": 0.4374405426645159,
+      "grad_norm": 455.7330322265625,
+      "learning_rate": 7.076433586846555e-06,
+      "loss": 119.5307,
+      "step": 52880
+    },
+    {
+      "epoch": 0.4375232659138851,
+      "grad_norm": 1019.272216796875,
+      "learning_rate": 7.075150506624342e-06,
+      "loss": 120.9018,
+      "step": 52890
+    },
+    {
+      "epoch": 0.4376059891632543,
+      "grad_norm": 783.7548828125,
+      "learning_rate": 7.073867261291926e-06,
+      "loss": 107.6927,
+      "step": 52900
+    },
+    {
+      "epoch": 0.4376887124126236,
+      "grad_norm": 483.0557556152344,
+      "learning_rate": 7.0725838509514115e-06,
+      "loss": 57.1351,
+      "step": 52910
+    },
+    {
+      "epoch": 0.4377714356619928,
+      "grad_norm": 1054.7470703125,
+      "learning_rate": 7.07130027570491e-06,
+      "loss": 99.6627,
+      "step": 52920
+    },
+    {
+      "epoch": 0.43785415891136203,
+      "grad_norm": 716.9881591796875,
+      "learning_rate": 7.070016535654551e-06,
+      "loss": 85.3958,
+      "step": 52930
+    },
+    {
+      "epoch": 0.4379368821607313,
+      "grad_norm": 1354.7027587890625,
+      "learning_rate": 7.068732630902479e-06,
+      "loss": 107.1556,
+      "step": 52940
+    },
+    {
+      "epoch": 0.4380196054101005,
+      "grad_norm": 1029.6153564453125,
+      "learning_rate": 7.067448561550844e-06,
+      "loss": 92.6526,
+      "step": 52950
+    },
+    {
+      "epoch": 0.43810232865946974,
+      "grad_norm": 1133.869873046875,
+      "learning_rate": 7.066164327701815e-06,
+      "loss": 149.094,
+      "step": 52960
+    },
+    {
+      "epoch": 0.438185051908839,
+      "grad_norm": 1119.0980224609375,
+      "learning_rate": 7.064879929457573e-06,
+      "loss": 143.9678,
+      "step": 52970
+    },
+    {
+      "epoch": 0.4382677751582082,
+      "grad_norm": 643.8549194335938,
+      "learning_rate": 7.063595366920314e-06,
+      "loss": 76.5045,
+      "step": 52980
+    },
+    {
+      "epoch": 0.43835049840757745,
+      "grad_norm": 1303.5244140625,
+      "learning_rate": 7.062310640192239e-06,
+      "loss": 111.4981,
+      "step": 52990
+    },
+    {
+      "epoch": 0.4384332216569467,
+      "grad_norm": 862.070556640625,
+      "learning_rate": 7.061025749375572e-06,
+      "loss": 100.7834,
+      "step": 53000
+    },
+    {
+      "epoch": 0.4385159449063159,
+      "grad_norm": 776.7830810546875,
+      "learning_rate": 7.059740694572545e-06,
+      "loss": 103.979,
+      "step": 53010
+    },
+    {
+      "epoch": 0.43859866815568516,
+      "grad_norm": 449.9796447753906,
+      "learning_rate": 7.058455475885405e-06,
+      "loss": 85.2324,
+      "step": 53020
+    },
+    {
+      "epoch": 0.4386813914050544,
+      "grad_norm": 1027.198486328125,
+      "learning_rate": 7.05717009341641e-06,
+      "loss": 93.9646,
+      "step": 53030
+    },
+    {
+      "epoch": 0.4387641146544236,
+      "grad_norm": 1561.4610595703125,
+      "learning_rate": 7.05588454726783e-06,
+      "loss": 98.5875,
+      "step": 53040
+    },
+    {
+      "epoch": 0.43884683790379286,
+      "grad_norm": 886.0679931640625,
+      "learning_rate": 7.054598837541951e-06,
+      "loss": 91.799,
+      "step": 53050
+    },
+    {
+      "epoch": 0.4389295611531621,
+      "grad_norm": 822.8799438476562,
+      "learning_rate": 7.053312964341075e-06,
+      "loss": 96.1325,
+      "step": 53060
+    },
+    {
+      "epoch": 0.4390122844025313,
+      "grad_norm": 1466.96337890625,
+      "learning_rate": 7.052026927767508e-06,
+      "loss": 125.8915,
+      "step": 53070
+    },
+    {
+      "epoch": 0.43909500765190057,
+      "grad_norm": 1202.5977783203125,
+      "learning_rate": 7.050740727923576e-06,
+      "loss": 100.9376,
+      "step": 53080
+    },
+    {
+      "epoch": 0.4391777309012698,
+      "grad_norm": 834.5575561523438,
+      "learning_rate": 7.049454364911615e-06,
+      "loss": 100.5369,
+      "step": 53090
+    },
+    {
+      "epoch": 0.439260454150639,
+      "grad_norm": 589.8638916015625,
+      "learning_rate": 7.048167838833977e-06,
+      "loss": 94.5479,
+      "step": 53100
+    },
+    {
+      "epoch": 0.4393431774000083,
+      "grad_norm": 722.2205200195312,
+      "learning_rate": 7.046881149793026e-06,
+      "loss": 94.6359,
+      "step": 53110
+    },
+    {
+      "epoch": 0.43942590064937753,
+      "grad_norm": 706.8112182617188,
+      "learning_rate": 7.045594297891133e-06,
+      "loss": 80.4411,
+      "step": 53120
+    },
+    {
+      "epoch": 0.43950862389874673,
+      "grad_norm": 1541.61279296875,
+      "learning_rate": 7.04430728323069e-06,
+      "loss": 126.8126,
+      "step": 53130
+    },
+    {
+      "epoch": 0.439591347148116,
+      "grad_norm": 1694.7259521484375,
+      "learning_rate": 7.043020105914098e-06,
+      "loss": 112.2094,
+      "step": 53140
+    },
+    {
+      "epoch": 0.43967407039748524,
+      "grad_norm": 1049.023193359375,
+      "learning_rate": 7.041732766043775e-06,
+      "loss": 96.9257,
+      "step": 53150
+    },
+    {
+      "epoch": 0.43975679364685444,
+      "grad_norm": 928.5380859375,
+      "learning_rate": 7.040445263722145e-06,
+      "loss": 90.1928,
+      "step": 53160
+    },
+    {
+      "epoch": 0.4398395168962237,
+      "grad_norm": 517.4678344726562,
+      "learning_rate": 7.039157599051648e-06,
+      "loss": 100.6094,
+      "step": 53170
+    },
+    {
+      "epoch": 0.43992224014559295,
+      "grad_norm": 1040.879638671875,
+      "learning_rate": 7.037869772134741e-06,
+      "loss": 93.181,
+      "step": 53180
+    },
+    {
+      "epoch": 0.44000496339496215,
+      "grad_norm": 1226.857421875,
+      "learning_rate": 7.036581783073888e-06,
+      "loss": 139.1528,
+      "step": 53190
+    },
+    {
+      "epoch": 0.4400876866443314,
+      "grad_norm": 2283.062744140625,
+      "learning_rate": 7.035293631971569e-06,
+      "loss": 121.1719,
+      "step": 53200
+    },
+    {
+      "epoch": 0.4401704098937006,
+      "grad_norm": 1155.0572509765625,
+      "learning_rate": 7.034005318930277e-06,
+      "loss": 86.4772,
+      "step": 53210
+    },
+    {
+      "epoch": 0.44025313314306985,
+      "grad_norm": 1117.125244140625,
+      "learning_rate": 7.032716844052517e-06,
+      "loss": 111.3512,
+      "step": 53220
+    },
+    {
+      "epoch": 0.4403358563924391,
+      "grad_norm": 1401.2935791015625,
+      "learning_rate": 7.031428207440807e-06,
+      "loss": 122.4468,
+      "step": 53230
+    },
+    {
+      "epoch": 0.4404185796418083,
+      "grad_norm": 719.4153442382812,
+      "learning_rate": 7.030139409197676e-06,
+      "loss": 86.415,
+      "step": 53240
+    },
+    {
+      "epoch": 0.44050130289117756,
+      "grad_norm": 802.9329833984375,
+      "learning_rate": 7.02885044942567e-06,
+      "loss": 161.6186,
+      "step": 53250
+    },
+    {
+      "epoch": 0.4405840261405468,
+      "grad_norm": 720.5238037109375,
+      "learning_rate": 7.027561328227345e-06,
+      "loss": 109.4637,
+      "step": 53260
+    },
+    {
+      "epoch": 0.440666749389916,
+      "grad_norm": 786.7742919921875,
+      "learning_rate": 7.02627204570527e-06,
+      "loss": 113.8812,
+      "step": 53270
+    },
+    {
+      "epoch": 0.44074947263928527,
+      "grad_norm": 449.480712890625,
+      "learning_rate": 7.024982601962027e-06,
+      "loss": 94.8391,
+      "step": 53280
+    },
+    {
+      "epoch": 0.4408321958886545,
+      "grad_norm": 934.8798217773438,
+      "learning_rate": 7.023692997100213e-06,
+      "loss": 79.2688,
+      "step": 53290
+    },
+    {
+      "epoch": 0.4409149191380237,
+      "grad_norm": 1120.0533447265625,
+      "learning_rate": 7.0224032312224345e-06,
+      "loss": 91.2629,
+      "step": 53300
+    },
+    {
+      "epoch": 0.440997642387393,
+      "grad_norm": 1041.491943359375,
+      "learning_rate": 7.021113304431313e-06,
+      "loss": 65.5652,
+      "step": 53310
+    },
+    {
+      "epoch": 0.44108036563676223,
+      "grad_norm": 1426.9508056640625,
+      "learning_rate": 7.01982321682948e-06,
+      "loss": 106.7264,
+      "step": 53320
+    },
+    {
+      "epoch": 0.4411630888861314,
+      "grad_norm": 1245.83154296875,
+      "learning_rate": 7.018532968519584e-06,
+      "loss": 97.5596,
+      "step": 53330
+    },
+    {
+      "epoch": 0.4412458121355007,
+      "grad_norm": 667.5186157226562,
+      "learning_rate": 7.0172425596042846e-06,
+      "loss": 96.0253,
+      "step": 53340
+    },
+    {
+      "epoch": 0.44132853538486994,
+      "grad_norm": 558.8970947265625,
+      "learning_rate": 7.0159519901862515e-06,
+      "loss": 108.9551,
+      "step": 53350
+    },
+    {
+      "epoch": 0.44141125863423913,
+      "grad_norm": 636.3999633789062,
+      "learning_rate": 7.014661260368171e-06,
+      "loss": 84.9733,
+      "step": 53360
+    },
+    {
+      "epoch": 0.4414939818836084,
+      "grad_norm": 927.197265625,
+      "learning_rate": 7.01337037025274e-06,
+      "loss": 111.2972,
+      "step": 53370
+    },
+    {
+      "epoch": 0.44157670513297764,
+      "grad_norm": 786.5078125,
+      "learning_rate": 7.012079319942668e-06,
+      "loss": 95.5135,
+      "step": 53380
+    },
+    {
+      "epoch": 0.44165942838234684,
+      "grad_norm": 807.628173828125,
+      "learning_rate": 7.01078810954068e-06,
+      "loss": 138.3115,
+      "step": 53390
+    },
+    {
+      "epoch": 0.4417421516317161,
+      "grad_norm": 657.76123046875,
+      "learning_rate": 7.0094967391495095e-06,
+      "loss": 96.7088,
+      "step": 53400
+    },
+    {
+      "epoch": 0.44182487488108535,
+      "grad_norm": 414.78240966796875,
+      "learning_rate": 7.008205208871906e-06,
+      "loss": 77.1027,
+      "step": 53410
+    },
+    {
+      "epoch": 0.44190759813045455,
+      "grad_norm": 719.4336547851562,
+      "learning_rate": 7.00691351881063e-06,
+      "loss": 95.9553,
+      "step": 53420
+    },
+    {
+      "epoch": 0.4419903213798238,
+      "grad_norm": 1148.5152587890625,
+      "learning_rate": 7.005621669068456e-06,
+      "loss": 109.5746,
+      "step": 53430
+    },
+    {
+      "epoch": 0.44207304462919306,
+      "grad_norm": 508.8988342285156,
+      "learning_rate": 7.004329659748172e-06,
+      "loss": 123.2771,
+      "step": 53440
+    },
+    {
+      "epoch": 0.44215576787856226,
+      "grad_norm": 777.6555786132812,
+      "learning_rate": 7.003037490952574e-06,
+      "loss": 101.2551,
+      "step": 53450
+    },
+    {
+      "epoch": 0.4422384911279315,
+      "grad_norm": 1049.9522705078125,
+      "learning_rate": 7.0017451627844765e-06,
+      "loss": 137.9618,
+      "step": 53460
+    },
+    {
+      "epoch": 0.44232121437730076,
+      "grad_norm": 666.4544677734375,
+      "learning_rate": 7.0004526753467004e-06,
+      "loss": 109.7146,
+      "step": 53470
+    },
+    {
+      "epoch": 0.44240393762666996,
+      "grad_norm": 1020.0592651367188,
+      "learning_rate": 6.999160028742089e-06,
+      "loss": 113.0266,
+      "step": 53480
+    },
+    {
+      "epoch": 0.4424866608760392,
+      "grad_norm": 646.3621215820312,
+      "learning_rate": 6.997867223073487e-06,
+      "loss": 148.3913,
+      "step": 53490
+    },
+    {
+      "epoch": 0.44256938412540847,
+      "grad_norm": 1154.8201904296875,
+      "learning_rate": 6.996574258443761e-06,
+      "loss": 111.0904,
+      "step": 53500
+    },
+    {
+      "epoch": 0.44265210737477767,
+      "grad_norm": 1100.033935546875,
+      "learning_rate": 6.995281134955784e-06,
+      "loss": 90.6527,
+      "step": 53510
+    },
+    {
+      "epoch": 0.4427348306241469,
+      "grad_norm": 1160.05712890625,
+      "learning_rate": 6.993987852712442e-06,
+      "loss": 113.9817,
+      "step": 53520
+    },
+    {
+      "epoch": 0.4428175538735162,
+      "grad_norm": 1008.1426391601562,
+      "learning_rate": 6.992694411816638e-06,
+      "loss": 91.588,
+      "step": 53530
+    },
+    {
+      "epoch": 0.4429002771228854,
+      "grad_norm": 892.0647583007812,
+      "learning_rate": 6.991400812371287e-06,
+      "loss": 127.6992,
+      "step": 53540
+    },
+    {
+      "epoch": 0.44298300037225463,
+      "grad_norm": 700.9825439453125,
+      "learning_rate": 6.990107054479313e-06,
+      "loss": 85.8635,
+      "step": 53550
+    },
+    {
+      "epoch": 0.44306572362162383,
+      "grad_norm": 452.05950927734375,
+      "learning_rate": 6.988813138243652e-06,
+      "loss": 109.3417,
+      "step": 53560
+    },
+    {
+      "epoch": 0.4431484468709931,
+      "grad_norm": 1181.6788330078125,
+      "learning_rate": 6.987519063767257e-06,
+      "loss": 116.6035,
+      "step": 53570
+    },
+    {
+      "epoch": 0.44323117012036234,
+      "grad_norm": 991.6622924804688,
+      "learning_rate": 6.986224831153092e-06,
+      "loss": 78.8246,
+      "step": 53580
+    },
+    {
+      "epoch": 0.44331389336973154,
+      "grad_norm": 1134.1353759765625,
+      "learning_rate": 6.984930440504134e-06,
+      "loss": 113.1138,
+      "step": 53590
+    },
+    {
+      "epoch": 0.4433966166191008,
+      "grad_norm": 708.4700317382812,
+      "learning_rate": 6.9836358919233695e-06,
+      "loss": 79.0538,
+      "step": 53600
+    },
+    {
+      "epoch": 0.44347933986847005,
+      "grad_norm": 917.70166015625,
+      "learning_rate": 6.982341185513799e-06,
+      "loss": 88.8924,
+      "step": 53610
+    },
+    {
+      "epoch": 0.44356206311783924,
+      "grad_norm": 966.7335815429688,
+      "learning_rate": 6.981046321378441e-06,
+      "loss": 142.0511,
+      "step": 53620
+    },
+    {
+      "epoch": 0.4436447863672085,
+      "grad_norm": 804.2879028320312,
+      "learning_rate": 6.979751299620318e-06,
+      "loss": 75.3757,
+      "step": 53630
+    },
+    {
+      "epoch": 0.44372750961657775,
+      "grad_norm": 1585.9019775390625,
+      "learning_rate": 6.978456120342469e-06,
+      "loss": 95.9753,
+      "step": 53640
+    },
+    {
+      "epoch": 0.44381023286594695,
+      "grad_norm": 1567.5927734375,
+      "learning_rate": 6.977160783647947e-06,
+      "loss": 126.4141,
+      "step": 53650
+    },
+    {
+      "epoch": 0.4438929561153162,
+      "grad_norm": 1687.5382080078125,
+      "learning_rate": 6.975865289639815e-06,
+      "loss": 90.8707,
+      "step": 53660
+    },
+    {
+      "epoch": 0.44397567936468546,
+      "grad_norm": 973.2371215820312,
+      "learning_rate": 6.974569638421151e-06,
+      "loss": 82.7147,
+      "step": 53670
+    },
+    {
+      "epoch": 0.44405840261405466,
+      "grad_norm": 658.38818359375,
+      "learning_rate": 6.973273830095042e-06,
+      "loss": 85.3202,
+      "step": 53680
+    },
+    {
+      "epoch": 0.4441411258634239,
+      "grad_norm": 1075.40673828125,
+      "learning_rate": 6.971977864764591e-06,
+      "loss": 99.2303,
+      "step": 53690
+    },
+    {
+      "epoch": 0.44422384911279317,
+      "grad_norm": 910.6012573242188,
+      "learning_rate": 6.970681742532911e-06,
+      "loss": 129.7319,
+      "step": 53700
+    },
+    {
+      "epoch": 0.44430657236216237,
+      "grad_norm": 1031.2911376953125,
+      "learning_rate": 6.969385463503129e-06,
+      "loss": 96.3028,
+      "step": 53710
+    },
+    {
+      "epoch": 0.4443892956115316,
+      "grad_norm": 679.598876953125,
+      "learning_rate": 6.968089027778384e-06,
+      "loss": 91.5366,
+      "step": 53720
+    },
+    {
+      "epoch": 0.4444720188609009,
+      "grad_norm": 842.2387084960938,
+      "learning_rate": 6.9667924354618275e-06,
+      "loss": 107.6039,
+      "step": 53730
+    },
+    {
+      "epoch": 0.4445547421102701,
+      "grad_norm": 645.2871704101562,
+      "learning_rate": 6.965495686656623e-06,
+      "loss": 125.044,
+      "step": 53740
+    },
+    {
+      "epoch": 0.44463746535963933,
+      "grad_norm": 614.8228149414062,
+      "learning_rate": 6.964198781465948e-06,
+      "loss": 101.3111,
+      "step": 53750
+    },
+    {
+      "epoch": 0.4447201886090086,
+      "grad_norm": 942.0760498046875,
+      "learning_rate": 6.962901719992989e-06,
+      "loss": 89.1372,
+      "step": 53760
+    },
+    {
+      "epoch": 0.4448029118583778,
+      "grad_norm": 577.6919555664062,
+      "learning_rate": 6.961604502340949e-06,
+      "loss": 59.7649,
+      "step": 53770
+    },
+    {
+      "epoch": 0.44488563510774704,
+      "grad_norm": 1402.629638671875,
+      "learning_rate": 6.960307128613042e-06,
+      "loss": 133.4121,
+      "step": 53780
+    },
+    {
+      "epoch": 0.4449683583571163,
+      "grad_norm": 1055.0478515625,
+      "learning_rate": 6.959009598912493e-06,
+      "loss": 127.3038,
+      "step": 53790
+    },
+    {
+      "epoch": 0.4450510816064855,
+      "grad_norm": 1320.6951904296875,
+      "learning_rate": 6.957711913342541e-06,
+      "loss": 86.9509,
+      "step": 53800
+    },
+    {
+      "epoch": 0.44513380485585474,
+      "grad_norm": 1073.6241455078125,
+      "learning_rate": 6.956414072006437e-06,
+      "loss": 122.6924,
+      "step": 53810
+    },
+    {
+      "epoch": 0.445216528105224,
+      "grad_norm": 870.7139282226562,
+      "learning_rate": 6.955116075007443e-06,
+      "loss": 124.2368,
+      "step": 53820
+    },
+    {
+      "epoch": 0.4452992513545932,
+      "grad_norm": 1268.8851318359375,
+      "learning_rate": 6.953817922448837e-06,
+      "loss": 89.0271,
+      "step": 53830
+    },
+    {
+      "epoch": 0.44538197460396245,
+      "grad_norm": 1143.03955078125,
+      "learning_rate": 6.9525196144339055e-06,
+      "loss": 145.819,
+      "step": 53840
+    },
+    {
+      "epoch": 0.4454646978533317,
+      "grad_norm": 1068.4166259765625,
+      "learning_rate": 6.951221151065948e-06,
+      "loss": 131.2076,
+      "step": 53850
+    },
+    {
+      "epoch": 0.4455474211027009,
+      "grad_norm": 797.7089233398438,
+      "learning_rate": 6.949922532448279e-06,
+      "loss": 98.0425,
+      "step": 53860
+    },
+    {
+      "epoch": 0.44563014435207016,
+      "grad_norm": 760.931640625,
+      "learning_rate": 6.948623758684223e-06,
+      "loss": 96.0778,
+      "step": 53870
+    },
+    {
+      "epoch": 0.4457128676014394,
+      "grad_norm": 624.017822265625,
+      "learning_rate": 6.9473248298771176e-06,
+      "loss": 89.5199,
+      "step": 53880
+    },
+    {
+      "epoch": 0.4457955908508086,
+      "grad_norm": 689.559814453125,
+      "learning_rate": 6.946025746130312e-06,
+      "loss": 123.3743,
+      "step": 53890
+    },
+    {
+      "epoch": 0.44587831410017786,
+      "grad_norm": 1045.9923095703125,
+      "learning_rate": 6.944726507547169e-06,
+      "loss": 100.6308,
+      "step": 53900
+    },
+    {
+      "epoch": 0.44596103734954706,
+      "grad_norm": 1568.38330078125,
+      "learning_rate": 6.943427114231064e-06,
+      "loss": 136.4211,
+      "step": 53910
+    },
+    {
+      "epoch": 0.4460437605989163,
+      "grad_norm": 1263.7076416015625,
+      "learning_rate": 6.942127566285382e-06,
+      "loss": 89.5075,
+      "step": 53920
+    },
+    {
+      "epoch": 0.44612648384828557,
+      "grad_norm": 1032.7841796875,
+      "learning_rate": 6.940827863813523e-06,
+      "loss": 124.588,
+      "step": 53930
+    },
+    {
+      "epoch": 0.44620920709765477,
+      "grad_norm": 1023.0800170898438,
+      "learning_rate": 6.9395280069188964e-06,
+      "loss": 127.864,
+      "step": 53940
+    },
+    {
+      "epoch": 0.446291930347024,
+      "grad_norm": 684.29931640625,
+      "learning_rate": 6.9382279957049295e-06,
+      "loss": 106.943,
+      "step": 53950
+    },
+    {
+      "epoch": 0.4463746535963933,
+      "grad_norm": 590.7765502929688,
+      "learning_rate": 6.936927830275055e-06,
+      "loss": 80.368,
+      "step": 53960
+    },
+    {
+      "epoch": 0.4464573768457625,
+      "grad_norm": 953.7996826171875,
+      "learning_rate": 6.935627510732724e-06,
+      "loss": 114.7125,
+      "step": 53970
+    },
+    {
+      "epoch": 0.44654010009513173,
+      "grad_norm": 665.14697265625,
+      "learning_rate": 6.934327037181394e-06,
+      "loss": 107.5566,
+      "step": 53980
+    },
+    {
+      "epoch": 0.446622823344501,
+      "grad_norm": 933.9505615234375,
+      "learning_rate": 6.933026409724538e-06,
+      "loss": 119.2751,
+      "step": 53990
+    },
+    {
+      "epoch": 0.4467055465938702,
+      "grad_norm": 1047.66796875,
+      "learning_rate": 6.931725628465643e-06,
+      "loss": 84.2427,
+      "step": 54000
+    },
+    {
+      "epoch": 0.44678826984323944,
+      "grad_norm": 1026.7486572265625,
+      "learning_rate": 6.9304246935082065e-06,
+      "loss": 80.6261,
+      "step": 54010
+    },
+    {
+      "epoch": 0.4468709930926087,
+      "grad_norm": 948.4655151367188,
+      "learning_rate": 6.929123604955735e-06,
+      "loss": 128.3851,
+      "step": 54020
+    },
+    {
+      "epoch": 0.4469537163419779,
+      "grad_norm": 1473.8193359375,
+      "learning_rate": 6.927822362911753e-06,
+      "loss": 83.6743,
+      "step": 54030
+    },
+    {
+      "epoch": 0.44703643959134715,
+      "grad_norm": 677.3907470703125,
+      "learning_rate": 6.926520967479791e-06,
+      "loss": 96.5376,
+      "step": 54040
+    },
+    {
+      "epoch": 0.4471191628407164,
+      "grad_norm": 759.7684326171875,
+      "learning_rate": 6.9252194187634e-06,
+      "loss": 85.2003,
+      "step": 54050
+    },
+    {
+      "epoch": 0.4472018860900856,
+      "grad_norm": 1008.9971923828125,
+      "learning_rate": 6.923917716866133e-06,
+      "loss": 108.9541,
+      "step": 54060
+    },
+    {
+      "epoch": 0.44728460933945485,
+      "grad_norm": 1046.4508056640625,
+      "learning_rate": 6.922615861891564e-06,
+      "loss": 73.9177,
+      "step": 54070
+    },
+    {
+      "epoch": 0.4473673325888241,
+      "grad_norm": 1061.5517578125,
+      "learning_rate": 6.921313853943275e-06,
+      "loss": 116.9172,
+      "step": 54080
+    },
+    {
+      "epoch": 0.4474500558381933,
+      "grad_norm": 1054.0621337890625,
+      "learning_rate": 6.9200116931248575e-06,
+      "loss": 94.5179,
+      "step": 54090
+    },
+    {
+      "epoch": 0.44753277908756256,
+      "grad_norm": 1365.836181640625,
+      "learning_rate": 6.918709379539924e-06,
+      "loss": 91.8605,
+      "step": 54100
+    },
+    {
+      "epoch": 0.4476155023369318,
+      "grad_norm": 914.6397094726562,
+      "learning_rate": 6.917406913292089e-06,
+      "loss": 95.1237,
+      "step": 54110
+    },
+    {
+      "epoch": 0.447698225586301,
+      "grad_norm": 1938.41064453125,
+      "learning_rate": 6.916104294484988e-06,
+      "loss": 133.195,
+      "step": 54120
+    },
+    {
+      "epoch": 0.44778094883567027,
+      "grad_norm": 728.0489501953125,
+      "learning_rate": 6.91480152322226e-06,
+      "loss": 105.5754,
+      "step": 54130
+    },
+    {
+      "epoch": 0.4478636720850395,
+      "grad_norm": 1188.586181640625,
+      "learning_rate": 6.913498599607563e-06,
+      "loss": 110.4302,
+      "step": 54140
+    },
+    {
+      "epoch": 0.4479463953344087,
+      "grad_norm": 1016.9649047851562,
+      "learning_rate": 6.9121955237445644e-06,
+      "loss": 75.1243,
+      "step": 54150
+    },
+    {
+      "epoch": 0.448029118583778,
+      "grad_norm": 1644.49755859375,
+      "learning_rate": 6.910892295736944e-06,
+      "loss": 87.0271,
+      "step": 54160
+    },
+    {
+      "epoch": 0.44811184183314723,
+      "grad_norm": 885.784423828125,
+      "learning_rate": 6.9095889156883934e-06,
+      "loss": 104.9015,
+      "step": 54170
+    },
+    {
+      "epoch": 0.4481945650825164,
+      "grad_norm": 575.415283203125,
+      "learning_rate": 6.908285383702617e-06,
+      "loss": 90.7336,
+      "step": 54180
+    },
+    {
+      "epoch": 0.4482772883318857,
+      "grad_norm": 876.1041870117188,
+      "learning_rate": 6.906981699883329e-06,
+      "loss": 93.0574,
+      "step": 54190
+    },
+    {
+      "epoch": 0.44836001158125494,
+      "grad_norm": 665.2525024414062,
+      "learning_rate": 6.90567786433426e-06,
+      "loss": 98.4839,
+      "step": 54200
+    },
+    {
+      "epoch": 0.44844273483062413,
+      "grad_norm": 601.7535400390625,
+      "learning_rate": 6.904373877159149e-06,
+      "loss": 88.9101,
+      "step": 54210
+    },
+    {
+      "epoch": 0.4485254580799934,
+      "grad_norm": 761.7291870117188,
+      "learning_rate": 6.903069738461749e-06,
+      "loss": 90.6817,
+      "step": 54220
+    },
+    {
+      "epoch": 0.44860818132936264,
+      "grad_norm": 510.3106994628906,
+      "learning_rate": 6.901765448345823e-06,
+      "loss": 172.6727,
+      "step": 54230
+    },
+    {
+      "epoch": 0.44869090457873184,
+      "grad_norm": 836.3589477539062,
+      "learning_rate": 6.900461006915149e-06,
+      "loss": 107.1047,
+      "step": 54240
+    },
+    {
+      "epoch": 0.4487736278281011,
+      "grad_norm": 981.2500610351562,
+      "learning_rate": 6.899156414273514e-06,
+      "loss": 102.4325,
+      "step": 54250
+    },
+    {
+      "epoch": 0.44885635107747035,
+      "grad_norm": 996.1396484375,
+      "learning_rate": 6.89785167052472e-06,
+      "loss": 76.1564,
+      "step": 54260
+    },
+    {
+      "epoch": 0.44893907432683955,
+      "grad_norm": 1051.859619140625,
+      "learning_rate": 6.896546775772577e-06,
+      "loss": 89.3364,
+      "step": 54270
+    },
+    {
+      "epoch": 0.4490217975762088,
+      "grad_norm": 743.7044677734375,
+      "learning_rate": 6.8952417301209114e-06,
+      "loss": 92.615,
+      "step": 54280
+    },
+    {
+      "epoch": 0.449104520825578,
+      "grad_norm": 1194.3951416015625,
+      "learning_rate": 6.893936533673561e-06,
+      "loss": 87.0885,
+      "step": 54290
+    },
+    {
+      "epoch": 0.44918724407494726,
+      "grad_norm": 855.311767578125,
+      "learning_rate": 6.892631186534371e-06,
+      "loss": 94.2941,
+      "step": 54300
+    },
+    {
+      "epoch": 0.4492699673243165,
+      "grad_norm": 804.5349731445312,
+      "learning_rate": 6.891325688807204e-06,
+      "loss": 119.2308,
+      "step": 54310
+    },
+    {
+      "epoch": 0.4493526905736857,
+      "grad_norm": 1499.4466552734375,
+      "learning_rate": 6.890020040595932e-06,
+      "loss": 117.2243,
+      "step": 54320
+    },
+    {
+      "epoch": 0.44943541382305496,
+      "grad_norm": 907.1102905273438,
+      "learning_rate": 6.88871424200444e-06,
+      "loss": 120.5858,
+      "step": 54330
+    },
+    {
+      "epoch": 0.4495181370724242,
+      "grad_norm": 1519.6060791015625,
+      "learning_rate": 6.887408293136621e-06,
+      "loss": 98.4492,
+      "step": 54340
+    },
+    {
+      "epoch": 0.4496008603217934,
+      "grad_norm": 802.8118896484375,
+      "learning_rate": 6.886102194096389e-06,
+      "loss": 67.4142,
+      "step": 54350
+    },
+    {
+      "epoch": 0.44968358357116267,
+      "grad_norm": 638.83935546875,
+      "learning_rate": 6.884795944987661e-06,
+      "loss": 89.7945,
+      "step": 54360
+    },
+    {
+      "epoch": 0.4497663068205319,
+      "grad_norm": 857.33984375,
+      "learning_rate": 6.8834895459143694e-06,
+      "loss": 103.017,
+      "step": 54370
+    },
+    {
+      "epoch": 0.4498490300699011,
+      "grad_norm": 1263.925537109375,
+      "learning_rate": 6.882182996980457e-06,
+      "loss": 80.3623,
+      "step": 54380
+    },
+    {
+      "epoch": 0.4499317533192704,
+      "grad_norm": 617.5503540039062,
+      "learning_rate": 6.880876298289885e-06,
+      "loss": 90.8478,
+      "step": 54390
+    },
+    {
+      "epoch": 0.45001447656863963,
+      "grad_norm": 978.7340087890625,
+      "learning_rate": 6.879569449946617e-06,
+      "loss": 85.7712,
+      "step": 54400
+    },
+    {
+      "epoch": 0.45009719981800883,
+      "grad_norm": 1923.4696044921875,
+      "learning_rate": 6.878262452054632e-06,
+      "loss": 119.6836,
+      "step": 54410
+    },
+    {
+      "epoch": 0.4501799230673781,
+      "grad_norm": 713.7830200195312,
+      "learning_rate": 6.876955304717925e-06,
+      "loss": 99.3105,
+      "step": 54420
+    },
+    {
+      "epoch": 0.45026264631674734,
+      "grad_norm": 714.7610473632812,
+      "learning_rate": 6.875648008040499e-06,
+      "loss": 67.4421,
+      "step": 54430
+    },
+    {
+      "epoch": 0.45034536956611654,
+      "grad_norm": 1194.353271484375,
+      "learning_rate": 6.874340562126368e-06,
+      "loss": 97.9703,
+      "step": 54440
+    },
+    {
+      "epoch": 0.4504280928154858,
+      "grad_norm": 747.249755859375,
+      "learning_rate": 6.873032967079562e-06,
+      "loss": 113.6194,
+      "step": 54450
+    },
+    {
+      "epoch": 0.45051081606485505,
+      "grad_norm": 1645.705322265625,
+      "learning_rate": 6.871725223004118e-06,
+      "loss": 134.7527,
+      "step": 54460
+    },
+    {
+      "epoch": 0.45059353931422425,
+      "grad_norm": 704.1566162109375,
+      "learning_rate": 6.870417330004086e-06,
+      "loss": 97.6264,
+      "step": 54470
+    },
+    {
+      "epoch": 0.4506762625635935,
+      "grad_norm": 945.1201782226562,
+      "learning_rate": 6.869109288183534e-06,
+      "loss": 82.3648,
+      "step": 54480
+    },
+    {
+      "epoch": 0.45075898581296275,
+      "grad_norm": 752.2423706054688,
+      "learning_rate": 6.867801097646534e-06,
+      "loss": 93.4124,
+      "step": 54490
+    },
+    {
+      "epoch": 0.45084170906233195,
+      "grad_norm": 1171.9190673828125,
+      "learning_rate": 6.866492758497171e-06,
+      "loss": 71.8976,
+      "step": 54500
+    },
+    {
+      "epoch": 0.4509244323117012,
+      "grad_norm": 1028.865478515625,
+      "learning_rate": 6.865184270839546e-06,
+      "loss": 79.6011,
+      "step": 54510
+    },
+    {
+      "epoch": 0.45100715556107046,
+      "grad_norm": 1303.591552734375,
+      "learning_rate": 6.863875634777767e-06,
+      "loss": 111.4603,
+      "step": 54520
+    },
+    {
+      "epoch": 0.45108987881043966,
+      "grad_norm": 702.6375122070312,
+      "learning_rate": 6.86256685041596e-06,
+      "loss": 76.6232,
+      "step": 54530
+    },
+    {
+      "epoch": 0.4511726020598089,
+      "grad_norm": 1701.81005859375,
+      "learning_rate": 6.861257917858257e-06,
+      "loss": 110.8894,
+      "step": 54540
+    },
+    {
+      "epoch": 0.45125532530917817,
+      "grad_norm": 640.2454833984375,
+      "learning_rate": 6.859948837208802e-06,
+      "loss": 83.109,
+      "step": 54550
+    },
+    {
+      "epoch": 0.45133804855854737,
+      "grad_norm": 681.0009155273438,
+      "learning_rate": 6.8586396085717536e-06,
+      "loss": 105.6306,
+      "step": 54560
+    },
+    {
+      "epoch": 0.4514207718079166,
+      "grad_norm": 2675.404052734375,
+      "learning_rate": 6.8573302320512836e-06,
+      "loss": 132.9688,
+      "step": 54570
+    },
+    {
+      "epoch": 0.4515034950572859,
+      "grad_norm": 891.1157836914062,
+      "learning_rate": 6.85602070775157e-06,
+      "loss": 103.1662,
+      "step": 54580
+    },
+    {
+      "epoch": 0.4515862183066551,
+      "grad_norm": 1364.07666015625,
+      "learning_rate": 6.854711035776806e-06,
+      "loss": 99.1324,
+      "step": 54590
+    },
+    {
+      "epoch": 0.45166894155602433,
+      "grad_norm": 524.9562377929688,
+      "learning_rate": 6.853401216231198e-06,
+      "loss": 106.766,
+      "step": 54600
+    },
+    {
+      "epoch": 0.4517516648053936,
+      "grad_norm": 818.6365966796875,
+      "learning_rate": 6.8520912492189605e-06,
+      "loss": 80.276,
+      "step": 54610
+    },
+    {
+      "epoch": 0.4518343880547628,
+      "grad_norm": 1519.1331787109375,
+      "learning_rate": 6.850781134844323e-06,
+      "loss": 69.9319,
+      "step": 54620
+    },
+    {
+      "epoch": 0.45191711130413204,
+      "grad_norm": 697.9396362304688,
+      "learning_rate": 6.8494708732115235e-06,
+      "loss": 123.4269,
+      "step": 54630
+    },
+    {
+      "epoch": 0.45199983455350123,
+      "grad_norm": 990.307373046875,
+      "learning_rate": 6.8481604644248155e-06,
+      "loss": 89.6535,
+      "step": 54640
+    },
+    {
+      "epoch": 0.4520825578028705,
+      "grad_norm": 651.0850830078125,
+      "learning_rate": 6.846849908588461e-06,
+      "loss": 80.7496,
+      "step": 54650
+    },
+    {
+      "epoch": 0.45216528105223974,
+      "grad_norm": 1296.8258056640625,
+      "learning_rate": 6.845539205806735e-06,
+      "loss": 129.8521,
+      "step": 54660
+    },
+    {
+      "epoch": 0.45224800430160894,
+      "grad_norm": 867.2517700195312,
+      "learning_rate": 6.844228356183924e-06,
+      "loss": 76.4351,
+      "step": 54670
+    },
+    {
+      "epoch": 0.4523307275509782,
+      "grad_norm": 878.8154907226562,
+      "learning_rate": 6.842917359824326e-06,
+      "loss": 103.5479,
+      "step": 54680
+    },
+    {
+      "epoch": 0.45241345080034745,
+      "grad_norm": 589.3108520507812,
+      "learning_rate": 6.841606216832253e-06,
+      "loss": 158.1967,
+      "step": 54690
+    },
+    {
+      "epoch": 0.45249617404971665,
+      "grad_norm": 440.4094543457031,
+      "learning_rate": 6.840294927312024e-06,
+      "loss": 84.1425,
+      "step": 54700
+    },
+    {
+      "epoch": 0.4525788972990859,
+      "grad_norm": 773.7758178710938,
+      "learning_rate": 6.838983491367974e-06,
+      "loss": 81.9263,
+      "step": 54710
+    },
+    {
+      "epoch": 0.45266162054845516,
+      "grad_norm": 1917.0274658203125,
+      "learning_rate": 6.837671909104447e-06,
+      "loss": 108.1925,
+      "step": 54720
+    },
+    {
+      "epoch": 0.45274434379782436,
+      "grad_norm": 528.8961791992188,
+      "learning_rate": 6.836360180625801e-06,
+      "loss": 123.8608,
+      "step": 54730
+    },
+    {
+      "epoch": 0.4528270670471936,
+      "grad_norm": 719.6814575195312,
+      "learning_rate": 6.835048306036404e-06,
+      "loss": 108.1355,
+      "step": 54740
+    },
+    {
+      "epoch": 0.45290979029656286,
+      "grad_norm": 1164.0782470703125,
+      "learning_rate": 6.833736285440632e-06,
+      "loss": 83.8529,
+      "step": 54750
+    },
+    {
+      "epoch": 0.45299251354593206,
+      "grad_norm": 540.9407958984375,
+      "learning_rate": 6.832424118942881e-06,
+      "loss": 125.107,
+      "step": 54760
+    },
+    {
+      "epoch": 0.4530752367953013,
+      "grad_norm": 736.422119140625,
+      "learning_rate": 6.831111806647552e-06,
+      "loss": 106.4315,
+      "step": 54770
+    },
+    {
+      "epoch": 0.45315796004467057,
+      "grad_norm": 697.813232421875,
+      "learning_rate": 6.829799348659061e-06,
+      "loss": 105.6159,
+      "step": 54780
+    },
+    {
+      "epoch": 0.45324068329403977,
+      "grad_norm": 925.9052734375,
+      "learning_rate": 6.828486745081835e-06,
+      "loss": 116.2571,
+      "step": 54790
+    },
+    {
+      "epoch": 0.453323406543409,
+      "grad_norm": 2547.846923828125,
+      "learning_rate": 6.8271739960203065e-06,
+      "loss": 142.8061,
+      "step": 54800
+    },
+    {
+      "epoch": 0.4534061297927783,
+      "grad_norm": 1475.7213134765625,
+      "learning_rate": 6.825861101578931e-06,
+      "loss": 81.7697,
+      "step": 54810
+    },
+    {
+      "epoch": 0.4534888530421475,
+      "grad_norm": 1138.5965576171875,
+      "learning_rate": 6.824548061862166e-06,
+      "loss": 92.7645,
+      "step": 54820
+    },
+    {
+      "epoch": 0.45357157629151673,
+      "grad_norm": 851.4307250976562,
+      "learning_rate": 6.823234876974489e-06,
+      "loss": 121.6354,
+      "step": 54830
+    },
+    {
+      "epoch": 0.453654299540886,
+      "grad_norm": 1226.28076171875,
+      "learning_rate": 6.8219215470203756e-06,
+      "loss": 102.6578,
+      "step": 54840
+    },
+    {
+      "epoch": 0.4537370227902552,
+      "grad_norm": 578.7922973632812,
+      "learning_rate": 6.820608072104329e-06,
+      "loss": 102.5517,
+      "step": 54850
+    },
+    {
+      "epoch": 0.45381974603962444,
+      "grad_norm": 673.4541625976562,
+      "learning_rate": 6.819294452330853e-06,
+      "loss": 69.6824,
+      "step": 54860
+    },
+    {
+      "epoch": 0.4539024692889937,
+      "grad_norm": 2711.473876953125,
+      "learning_rate": 6.817980687804467e-06,
+      "loss": 101.1284,
+      "step": 54870
+    },
+    {
+      "epoch": 0.4539851925383629,
+      "grad_norm": 736.4070434570312,
+      "learning_rate": 6.8166667786297e-06,
+      "loss": 96.8542,
+      "step": 54880
+    },
+    {
+      "epoch": 0.45406791578773215,
+      "grad_norm": 760.29150390625,
+      "learning_rate": 6.815352724911095e-06,
+      "loss": 105.462,
+      "step": 54890
+    },
+    {
+      "epoch": 0.4541506390371014,
+      "grad_norm": 692.9434204101562,
+      "learning_rate": 6.814038526753205e-06,
+      "loss": 91.7443,
+      "step": 54900
+    },
+    {
+      "epoch": 0.4542333622864706,
+      "grad_norm": 549.2918090820312,
+      "learning_rate": 6.812724184260596e-06,
+      "loss": 73.3805,
+      "step": 54910
+    },
+    {
+      "epoch": 0.45431608553583985,
+      "grad_norm": 702.6921997070312,
+      "learning_rate": 6.811409697537843e-06,
+      "loss": 84.0114,
+      "step": 54920
+    },
+    {
+      "epoch": 0.4543988087852091,
+      "grad_norm": 787.5999755859375,
+      "learning_rate": 6.810095066689533e-06,
+      "loss": 101.185,
+      "step": 54930
+    },
+    {
+      "epoch": 0.4544815320345783,
+      "grad_norm": 1066.020263671875,
+      "learning_rate": 6.808780291820264e-06,
+      "loss": 99.9101,
+      "step": 54940
+    },
+    {
+      "epoch": 0.45456425528394756,
+      "grad_norm": 1216.9044189453125,
+      "learning_rate": 6.80746537303465e-06,
+      "loss": 120.2617,
+      "step": 54950
+    },
+    {
+      "epoch": 0.4546469785333168,
+      "grad_norm": 1686.6910400390625,
+      "learning_rate": 6.806150310437312e-06,
+      "loss": 121.6102,
+      "step": 54960
+    },
+    {
+      "epoch": 0.454729701782686,
+      "grad_norm": 1484.89111328125,
+      "learning_rate": 6.804835104132883e-06,
+      "loss": 104.0064,
+      "step": 54970
+    },
+    {
+      "epoch": 0.45481242503205527,
+      "grad_norm": 979.0828857421875,
+      "learning_rate": 6.803519754226007e-06,
+      "loss": 94.6825,
+      "step": 54980
+    },
+    {
+      "epoch": 0.4548951482814245,
+      "grad_norm": 864.7905883789062,
+      "learning_rate": 6.80220426082134e-06,
+      "loss": 91.2416,
+      "step": 54990
+    },
+    {
+      "epoch": 0.4549778715307937,
+      "grad_norm": 888.6685791015625,
+      "learning_rate": 6.800888624023552e-06,
+      "loss": 105.9705,
+      "step": 55000
+    },
+    {
+      "epoch": 0.455060594780163,
+      "grad_norm": 497.9424133300781,
+      "learning_rate": 6.799572843937322e-06,
+      "loss": 108.1165,
+      "step": 55010
+    },
+    {
+      "epoch": 0.4551433180295322,
+      "grad_norm": 518.8756713867188,
+      "learning_rate": 6.79825692066734e-06,
+      "loss": 106.614,
+      "step": 55020
+    },
+    {
+      "epoch": 0.4552260412789014,
+      "grad_norm": 1119.955078125,
+      "learning_rate": 6.796940854318306e-06,
+      "loss": 140.4494,
+      "step": 55030
+    },
+    {
+      "epoch": 0.4553087645282707,
+      "grad_norm": 594.2408447265625,
+      "learning_rate": 6.795624644994936e-06,
+      "loss": 77.2674,
+      "step": 55040
+    },
+    {
+      "epoch": 0.4553914877776399,
+      "grad_norm": 797.1567993164062,
+      "learning_rate": 6.794308292801954e-06,
+      "loss": 111.2638,
+      "step": 55050
+    },
+    {
+      "epoch": 0.45547421102700913,
+      "grad_norm": 821.6486206054688,
+      "learning_rate": 6.792991797844095e-06,
+      "loss": 136.5393,
+      "step": 55060
+    },
+    {
+      "epoch": 0.4555569342763784,
+      "grad_norm": 597.4654541015625,
+      "learning_rate": 6.791675160226109e-06,
+      "loss": 96.4142,
+      "step": 55070
+    },
+    {
+      "epoch": 0.4556396575257476,
+      "grad_norm": 592.2135009765625,
+      "learning_rate": 6.790358380052752e-06,
+      "loss": 93.9998,
+      "step": 55080
+    },
+    {
+      "epoch": 0.45572238077511684,
+      "grad_norm": 1021.7890625,
+      "learning_rate": 6.789041457428796e-06,
+      "loss": 124.1362,
+      "step": 55090
+    },
+    {
+      "epoch": 0.4558051040244861,
+      "grad_norm": 486.76080322265625,
+      "learning_rate": 6.7877243924590205e-06,
+      "loss": 77.4204,
+      "step": 55100
+    },
+    {
+      "epoch": 0.4558878272738553,
+      "grad_norm": 875.5304565429688,
+      "learning_rate": 6.7864071852482205e-06,
+      "loss": 104.1027,
+      "step": 55110
+    },
+    {
+      "epoch": 0.45597055052322455,
+      "grad_norm": 738.2188110351562,
+      "learning_rate": 6.7850898359012e-06,
+      "loss": 106.4531,
+      "step": 55120
+    },
+    {
+      "epoch": 0.4560532737725938,
+      "grad_norm": 1422.643798828125,
+      "learning_rate": 6.7837723445227724e-06,
+      "loss": 101.3412,
+      "step": 55130
+    },
+    {
+      "epoch": 0.456135997021963,
+      "grad_norm": 1019.88818359375,
+      "learning_rate": 6.782454711217767e-06,
+      "loss": 104.1804,
+      "step": 55140
+    },
+    {
+      "epoch": 0.45621872027133226,
+      "grad_norm": 1043.4281005859375,
+      "learning_rate": 6.78113693609102e-06,
+      "loss": 159.6309,
+      "step": 55150
+    },
+    {
+      "epoch": 0.4563014435207015,
+      "grad_norm": 733.2887573242188,
+      "learning_rate": 6.77981901924738e-06,
+      "loss": 76.8028,
+      "step": 55160
+    },
+    {
+      "epoch": 0.4563841667700707,
+      "grad_norm": 650.5491943359375,
+      "learning_rate": 6.7785009607917095e-06,
+      "loss": 132.4901,
+      "step": 55170
+    },
+    {
+      "epoch": 0.45646689001943996,
+      "grad_norm": 993.3848876953125,
+      "learning_rate": 6.777182760828881e-06,
+      "loss": 127.0856,
+      "step": 55180
+    },
+    {
+      "epoch": 0.4565496132688092,
+      "grad_norm": 941.2340087890625,
+      "learning_rate": 6.7758644194637755e-06,
+      "loss": 68.8753,
+      "step": 55190
+    },
+    {
+      "epoch": 0.4566323365181784,
+      "grad_norm": 1339.5758056640625,
+      "learning_rate": 6.774545936801289e-06,
+      "loss": 93.002,
+      "step": 55200
+    },
+    {
+      "epoch": 0.45671505976754767,
+      "grad_norm": 605.7567749023438,
+      "learning_rate": 6.773227312946327e-06,
+      "loss": 123.4982,
+      "step": 55210
+    },
+    {
+      "epoch": 0.4567977830169169,
+      "grad_norm": 984.2488403320312,
+      "learning_rate": 6.771908548003803e-06,
+      "loss": 77.7853,
+      "step": 55220
+    },
+    {
+      "epoch": 0.4568805062662861,
+      "grad_norm": 829.1386108398438,
+      "learning_rate": 6.77058964207865e-06,
+      "loss": 213.839,
+      "step": 55230
+    },
+    {
+      "epoch": 0.4569632295156554,
+      "grad_norm": 764.8698120117188,
+      "learning_rate": 6.769270595275804e-06,
+      "loss": 102.3559,
+      "step": 55240
+    },
+    {
+      "epoch": 0.45704595276502463,
+      "grad_norm": 798.2879028320312,
+      "learning_rate": 6.767951407700217e-06,
+      "loss": 90.5174,
+      "step": 55250
+    },
+    {
+      "epoch": 0.45712867601439383,
+      "grad_norm": 856.211181640625,
+      "learning_rate": 6.766632079456852e-06,
+      "loss": 85.0527,
+      "step": 55260
+    },
+    {
+      "epoch": 0.4572113992637631,
+      "grad_norm": 831.584716796875,
+      "learning_rate": 6.765312610650677e-06,
+      "loss": 95.3017,
+      "step": 55270
+    },
+    {
+      "epoch": 0.45729412251313234,
+      "grad_norm": 883.2146606445312,
+      "learning_rate": 6.763993001386681e-06,
+      "loss": 113.933,
+      "step": 55280
+    },
+    {
+      "epoch": 0.45737684576250154,
+      "grad_norm": 954.8955688476562,
+      "learning_rate": 6.762673251769858e-06,
+      "loss": 121.8417,
+      "step": 55290
+    },
+    {
+      "epoch": 0.4574595690118708,
+      "grad_norm": 1161.1273193359375,
+      "learning_rate": 6.761353361905214e-06,
+      "loss": 90.7742,
+      "step": 55300
+    },
+    {
+      "epoch": 0.45754229226124005,
+      "grad_norm": 683.6365966796875,
+      "learning_rate": 6.7600333318977655e-06,
+      "loss": 91.0024,
+      "step": 55310
+    },
+    {
+      "epoch": 0.45762501551060925,
+      "grad_norm": 1601.54638671875,
+      "learning_rate": 6.758713161852541e-06,
+      "loss": 110.4416,
+      "step": 55320
+    },
+    {
+      "epoch": 0.4577077387599785,
+      "grad_norm": 934.130126953125,
+      "learning_rate": 6.757392851874584e-06,
+      "loss": 113.9463,
+      "step": 55330
+    },
+    {
+      "epoch": 0.45779046200934775,
+      "grad_norm": 1528.1434326171875,
+      "learning_rate": 6.756072402068943e-06,
+      "loss": 127.9479,
+      "step": 55340
+    },
+    {
+      "epoch": 0.45787318525871695,
+      "grad_norm": 771.1652221679688,
+      "learning_rate": 6.75475181254068e-06,
+      "loss": 89.5568,
+      "step": 55350
+    },
+    {
+      "epoch": 0.4579559085080862,
+      "grad_norm": 879.8575439453125,
+      "learning_rate": 6.753431083394868e-06,
+      "loss": 99.1166,
+      "step": 55360
+    },
+    {
+      "epoch": 0.4580386317574554,
+      "grad_norm": 684.24609375,
+      "learning_rate": 6.75211021473659e-06,
+      "loss": 86.4435,
+      "step": 55370
+    },
+    {
+      "epoch": 0.45812135500682466,
+      "grad_norm": 1416.9102783203125,
+      "learning_rate": 6.750789206670945e-06,
+      "loss": 95.3575,
+      "step": 55380
+    },
+    {
+      "epoch": 0.4582040782561939,
+      "grad_norm": 889.2215576171875,
+      "learning_rate": 6.749468059303039e-06,
+      "loss": 92.2547,
+      "step": 55390
+    },
+    {
+      "epoch": 0.4582868015055631,
+      "grad_norm": 1061.4666748046875,
+      "learning_rate": 6.748146772737988e-06,
+      "loss": 101.2263,
+      "step": 55400
+    },
+    {
+      "epoch": 0.45836952475493237,
+      "grad_norm": 1019.367919921875,
+      "learning_rate": 6.7468253470809205e-06,
+      "loss": 136.364,
+      "step": 55410
+    },
+    {
+      "epoch": 0.4584522480043016,
+      "grad_norm": 1758.954833984375,
+      "learning_rate": 6.745503782436976e-06,
+      "loss": 124.5927,
+      "step": 55420
+    },
+    {
+      "epoch": 0.4585349712536708,
+      "grad_norm": 763.4608154296875,
+      "learning_rate": 6.7441820789113085e-06,
+      "loss": 105.7726,
+      "step": 55430
+    },
+    {
+      "epoch": 0.4586176945030401,
+      "grad_norm": 946.8983764648438,
+      "learning_rate": 6.7428602366090764e-06,
+      "loss": 112.0909,
+      "step": 55440
+    },
+    {
+      "epoch": 0.45870041775240933,
+      "grad_norm": 932.01318359375,
+      "learning_rate": 6.741538255635454e-06,
+      "loss": 88.3237,
+      "step": 55450
+    },
+    {
+      "epoch": 0.4587831410017785,
+      "grad_norm": 829.2271118164062,
+      "learning_rate": 6.740216136095626e-06,
+      "loss": 77.0077,
+      "step": 55460
+    },
+    {
+      "epoch": 0.4588658642511478,
+      "grad_norm": 386.30364990234375,
+      "learning_rate": 6.738893878094786e-06,
+      "loss": 87.8386,
+      "step": 55470
+    },
+    {
+      "epoch": 0.45894858750051704,
+      "grad_norm": 1005.0819702148438,
+      "learning_rate": 6.737571481738141e-06,
+      "loss": 128.3534,
+      "step": 55480
+    },
+    {
+      "epoch": 0.45903131074988623,
+      "grad_norm": 921.326904296875,
+      "learning_rate": 6.736248947130907e-06,
+      "loss": 107.7665,
+      "step": 55490
+    },
+    {
+      "epoch": 0.4591140339992555,
+      "grad_norm": 1480.9969482421875,
+      "learning_rate": 6.734926274378313e-06,
+      "loss": 163.0793,
+      "step": 55500
+    },
+    {
+      "epoch": 0.45919675724862474,
+      "grad_norm": 955.8799438476562,
+      "learning_rate": 6.733603463585598e-06,
+      "loss": 93.535,
+      "step": 55510
+    },
+    {
+      "epoch": 0.45927948049799394,
+      "grad_norm": 1354.3538818359375,
+      "learning_rate": 6.73228051485801e-06,
+      "loss": 85.5063,
+      "step": 55520
+    },
+    {
+      "epoch": 0.4593622037473632,
+      "grad_norm": 536.1598510742188,
+      "learning_rate": 6.7309574283008125e-06,
+      "loss": 84.9367,
+      "step": 55530
+    },
+    {
+      "epoch": 0.45944492699673245,
+      "grad_norm": 663.9678344726562,
+      "learning_rate": 6.729634204019277e-06,
+      "loss": 104.2453,
+      "step": 55540
+    },
+    {
+      "epoch": 0.45952765024610165,
+      "grad_norm": 690.8705444335938,
+      "learning_rate": 6.7283108421186835e-06,
+      "loss": 108.3504,
+      "step": 55550
+    },
+    {
+      "epoch": 0.4596103734954709,
+      "grad_norm": 1198.9808349609375,
+      "learning_rate": 6.726987342704331e-06,
+      "loss": 83.5574,
+      "step": 55560
+    },
+    {
+      "epoch": 0.45969309674484016,
+      "grad_norm": 780.8455810546875,
+      "learning_rate": 6.72566370588152e-06,
+      "loss": 92.0601,
+      "step": 55570
+    },
+    {
+      "epoch": 0.45977581999420936,
+      "grad_norm": 1550.3758544921875,
+      "learning_rate": 6.724339931755568e-06,
+      "loss": 114.7621,
+      "step": 55580
+    },
+    {
+      "epoch": 0.4598585432435786,
+      "grad_norm": 515.3203735351562,
+      "learning_rate": 6.7230160204318e-06,
+      "loss": 85.6729,
+      "step": 55590
+    },
+    {
+      "epoch": 0.45994126649294786,
+      "grad_norm": 622.2483520507812,
+      "learning_rate": 6.721691972015557e-06,
+      "loss": 91.3313,
+      "step": 55600
+    },
+    {
+      "epoch": 0.46002398974231706,
+      "grad_norm": 1014.8973999023438,
+      "learning_rate": 6.720367786612185e-06,
+      "loss": 71.0175,
+      "step": 55610
+    },
+    {
+      "epoch": 0.4601067129916863,
+      "grad_norm": 2489.36962890625,
+      "learning_rate": 6.719043464327043e-06,
+      "loss": 82.0414,
+      "step": 55620
+    },
+    {
+      "epoch": 0.46018943624105557,
+      "grad_norm": 736.4323120117188,
+      "learning_rate": 6.717719005265502e-06,
+      "loss": 68.9147,
+      "step": 55630
+    },
+    {
+      "epoch": 0.46027215949042477,
+      "grad_norm": 749.0441284179688,
+      "learning_rate": 6.716394409532944e-06,
+      "loss": 117.8003,
+      "step": 55640
+    },
+    {
+      "epoch": 0.460354882739794,
+      "grad_norm": 869.51123046875,
+      "learning_rate": 6.715069677234758e-06,
+      "loss": 116.4776,
+      "step": 55650
+    },
+    {
+      "epoch": 0.4604376059891633,
+      "grad_norm": 686.819091796875,
+      "learning_rate": 6.713744808476349e-06,
+      "loss": 101.6366,
+      "step": 55660
+    },
+    {
+      "epoch": 0.4605203292385325,
+      "grad_norm": 1027.5953369140625,
+      "learning_rate": 6.712419803363132e-06,
+      "loss": 92.2464,
+      "step": 55670
+    },
+    {
+      "epoch": 0.46060305248790173,
+      "grad_norm": 743.2421875,
+      "learning_rate": 6.711094662000529e-06,
+      "loss": 104.2116,
+      "step": 55680
+    },
+    {
+      "epoch": 0.460685775737271,
+      "grad_norm": 992.1659545898438,
+      "learning_rate": 6.709769384493978e-06,
+      "loss": 102.0216,
+      "step": 55690
+    },
+    {
+      "epoch": 0.4607684989866402,
+      "grad_norm": 763.0567626953125,
+      "learning_rate": 6.708443970948923e-06,
+      "loss": 105.3009,
+      "step": 55700
+    },
+    {
+      "epoch": 0.46085122223600944,
+      "grad_norm": 1082.239501953125,
+      "learning_rate": 6.707118421470822e-06,
+      "loss": 82.143,
+      "step": 55710
+    },
+    {
+      "epoch": 0.4609339454853787,
+      "grad_norm": 943.666259765625,
+      "learning_rate": 6.705792736165142e-06,
+      "loss": 117.292,
+      "step": 55720
+    },
+    {
+      "epoch": 0.4610166687347479,
+      "grad_norm": 761.20166015625,
+      "learning_rate": 6.7044669151373645e-06,
+      "loss": 79.007,
+      "step": 55730
+    },
+    {
+      "epoch": 0.46109939198411715,
+      "grad_norm": 676.6871337890625,
+      "learning_rate": 6.7031409584929765e-06,
+      "loss": 96.3533,
+      "step": 55740
+    },
+    {
+      "epoch": 0.46118211523348634,
+      "grad_norm": 851.9503173828125,
+      "learning_rate": 6.701814866337477e-06,
+      "loss": 131.2042,
+      "step": 55750
+    },
+    {
+      "epoch": 0.4612648384828556,
+      "grad_norm": 1339.9852294921875,
+      "learning_rate": 6.700488638776379e-06,
+      "loss": 113.9575,
+      "step": 55760
+    },
+    {
+      "epoch": 0.46134756173222485,
+      "grad_norm": 945.3772583007812,
+      "learning_rate": 6.699162275915208e-06,
+      "loss": 88.1573,
+      "step": 55770
+    },
+    {
+      "epoch": 0.46143028498159405,
+      "grad_norm": 1223.3470458984375,
+      "learning_rate": 6.6978357778594896e-06,
+      "loss": 81.1195,
+      "step": 55780
+    },
+    {
+      "epoch": 0.4615130082309633,
+      "grad_norm": 922.9274291992188,
+      "learning_rate": 6.69650914471477e-06,
+      "loss": 97.8219,
+      "step": 55790
+    },
+    {
+      "epoch": 0.46159573148033256,
+      "grad_norm": 1972.91162109375,
+      "learning_rate": 6.695182376586603e-06,
+      "loss": 99.1887,
+      "step": 55800
+    },
+    {
+      "epoch": 0.46167845472970176,
+      "grad_norm": 1156.863525390625,
+      "learning_rate": 6.6938554735805565e-06,
+      "loss": 99.5834,
+      "step": 55810
+    },
+    {
+      "epoch": 0.461761177979071,
+      "grad_norm": 1650.710205078125,
+      "learning_rate": 6.6925284358022035e-06,
+      "loss": 155.1294,
+      "step": 55820
+    },
+    {
+      "epoch": 0.46184390122844027,
+      "grad_norm": 520.858642578125,
+      "learning_rate": 6.69120126335713e-06,
+      "loss": 122.3727,
+      "step": 55830
+    },
+    {
+      "epoch": 0.46192662447780947,
+      "grad_norm": 916.6767578125,
+      "learning_rate": 6.689873956350932e-06,
+      "loss": 75.5874,
+      "step": 55840
+    },
+    {
+      "epoch": 0.4620093477271787,
+      "grad_norm": 917.3632202148438,
+      "learning_rate": 6.688546514889221e-06,
+      "loss": 82.2032,
+      "step": 55850
+    },
+    {
+      "epoch": 0.462092070976548,
+      "grad_norm": 1109.2620849609375,
+      "learning_rate": 6.687218939077613e-06,
+      "loss": 93.1306,
+      "step": 55860
+    },
+    {
+      "epoch": 0.4621747942259172,
+      "grad_norm": 933.2094116210938,
+      "learning_rate": 6.685891229021736e-06,
+      "loss": 82.2505,
+      "step": 55870
+    },
+    {
+      "epoch": 0.46225751747528643,
+      "grad_norm": 930.94873046875,
+      "learning_rate": 6.6845633848272315e-06,
+      "loss": 113.1939,
+      "step": 55880
+    },
+    {
+      "epoch": 0.4623402407246557,
+      "grad_norm": 808.5699462890625,
+      "learning_rate": 6.68323540659975e-06,
+      "loss": 100.777,
+      "step": 55890
+    },
+    {
+      "epoch": 0.4624229639740249,
+      "grad_norm": 663.3551025390625,
+      "learning_rate": 6.681907294444952e-06,
+      "loss": 67.9267,
+      "step": 55900
+    },
+    {
+      "epoch": 0.46250568722339414,
+      "grad_norm": 1566.524169921875,
+      "learning_rate": 6.6805790484685094e-06,
+      "loss": 93.2308,
+      "step": 55910
+    },
+    {
+      "epoch": 0.4625884104727634,
+      "grad_norm": 1127.6248779296875,
+      "learning_rate": 6.679250668776105e-06,
+      "loss": 140.5565,
+      "step": 55920
+    },
+    {
+      "epoch": 0.4626711337221326,
+      "grad_norm": 1181.0595703125,
+      "learning_rate": 6.677922155473432e-06,
+      "loss": 99.0083,
+      "step": 55930
+    },
+    {
+      "epoch": 0.46275385697150184,
+      "grad_norm": 1022.6056518554688,
+      "learning_rate": 6.676593508666192e-06,
+      "loss": 135.9459,
+      "step": 55940
+    },
+    {
+      "epoch": 0.4628365802208711,
+      "grad_norm": 1002.8577880859375,
+      "learning_rate": 6.675264728460103e-06,
+      "loss": 85.4971,
+      "step": 55950
+    },
+    {
+      "epoch": 0.4629193034702403,
+      "grad_norm": 944.582763671875,
+      "learning_rate": 6.673935814960887e-06,
+      "loss": 107.2265,
+      "step": 55960
+    },
+    {
+      "epoch": 0.46300202671960955,
+      "grad_norm": 1288.594970703125,
+      "learning_rate": 6.672606768274281e-06,
+      "loss": 90.4464,
+      "step": 55970
+    },
+    {
+      "epoch": 0.4630847499689788,
+      "grad_norm": 14770.8916015625,
+      "learning_rate": 6.67127758850603e-06,
+      "loss": 241.7437,
+      "step": 55980
+    },
+    {
+      "epoch": 0.463167473218348,
+      "grad_norm": 957.4649047851562,
+      "learning_rate": 6.669948275761893e-06,
+      "loss": 94.3999,
+      "step": 55990
+    },
+    {
+      "epoch": 0.46325019646771726,
+      "grad_norm": 1285.9119873046875,
+      "learning_rate": 6.668618830147634e-06,
+      "loss": 125.4894,
+      "step": 56000
+    },
+    {
+      "epoch": 0.4633329197170865,
+      "grad_norm": 722.8943481445312,
+      "learning_rate": 6.667289251769033e-06,
+      "loss": 85.9025,
+      "step": 56010
+    },
+    {
+      "epoch": 0.4634156429664557,
+      "grad_norm": 1016.5885009765625,
+      "learning_rate": 6.6659595407318775e-06,
+      "loss": 107.344,
+      "step": 56020
+    },
+    {
+      "epoch": 0.46349836621582496,
+      "grad_norm": 900.4605712890625,
+      "learning_rate": 6.664629697141969e-06,
+      "loss": 111.3321,
+      "step": 56030
+    },
+    {
+      "epoch": 0.4635810894651942,
+      "grad_norm": 971.4387817382812,
+      "learning_rate": 6.663299721105113e-06,
+      "loss": 106.712,
+      "step": 56040
+    },
+    {
+      "epoch": 0.4636638127145634,
+      "grad_norm": 1102.5128173828125,
+      "learning_rate": 6.661969612727133e-06,
+      "loss": 94.0693,
+      "step": 56050
+    },
+    {
+      "epoch": 0.46374653596393267,
+      "grad_norm": 1239.50341796875,
+      "learning_rate": 6.660639372113858e-06,
+      "loss": 109.2637,
+      "step": 56060
+    },
+    {
+      "epoch": 0.4638292592133019,
+      "grad_norm": 986.812255859375,
+      "learning_rate": 6.65930899937113e-06,
+      "loss": 104.8453,
+      "step": 56070
+    },
+    {
+      "epoch": 0.4639119824626711,
+      "grad_norm": 1008.2682495117188,
+      "learning_rate": 6.657978494604799e-06,
+      "loss": 109.9477,
+      "step": 56080
+    },
+    {
+      "epoch": 0.4639947057120404,
+      "grad_norm": 683.7396850585938,
+      "learning_rate": 6.656647857920728e-06,
+      "loss": 109.3742,
+      "step": 56090
+    },
+    {
+      "epoch": 0.4640774289614096,
+      "grad_norm": 1494.6700439453125,
+      "learning_rate": 6.655317089424791e-06,
+      "loss": 106.0912,
+      "step": 56100
+    },
+    {
+      "epoch": 0.46416015221077883,
+      "grad_norm": 1352.005615234375,
+      "learning_rate": 6.6539861892228695e-06,
+      "loss": 96.1662,
+      "step": 56110
+    },
+    {
+      "epoch": 0.4642428754601481,
+      "grad_norm": 787.8734741210938,
+      "learning_rate": 6.652655157420859e-06,
+      "loss": 81.0995,
+      "step": 56120
+    },
+    {
+      "epoch": 0.4643255987095173,
+      "grad_norm": 837.31787109375,
+      "learning_rate": 6.651323994124661e-06,
+      "loss": 93.8052,
+      "step": 56130
+    },
+    {
+      "epoch": 0.46440832195888654,
+      "grad_norm": 1119.1298828125,
+      "learning_rate": 6.649992699440191e-06,
+      "loss": 86.7144,
+      "step": 56140
+    },
+    {
+      "epoch": 0.4644910452082558,
+      "grad_norm": 1033.329345703125,
+      "learning_rate": 6.648661273473375e-06,
+      "loss": 73.4606,
+      "step": 56150
+    },
+    {
+      "epoch": 0.464573768457625,
+      "grad_norm": 1087.544921875,
+      "learning_rate": 6.6473297163301485e-06,
+      "loss": 93.5026,
+      "step": 56160
+    },
+    {
+      "epoch": 0.46465649170699425,
+      "grad_norm": 637.343505859375,
+      "learning_rate": 6.645998028116455e-06,
+      "loss": 106.3616,
+      "step": 56170
+    },
+    {
+      "epoch": 0.4647392149563635,
+      "grad_norm": 1312.9207763671875,
+      "learning_rate": 6.6446662089382545e-06,
+      "loss": 126.186,
+      "step": 56180
+    },
+    {
+      "epoch": 0.4648219382057327,
+      "grad_norm": 756.4004516601562,
+      "learning_rate": 6.643334258901511e-06,
+      "loss": 110.0223,
+      "step": 56190
+    },
+    {
+      "epoch": 0.46490466145510195,
+      "grad_norm": 577.8167724609375,
+      "learning_rate": 6.642002178112202e-06,
+      "loss": 114.2335,
+      "step": 56200
+    },
+    {
+      "epoch": 0.4649873847044712,
+      "grad_norm": 419.95404052734375,
+      "learning_rate": 6.640669966676316e-06,
+      "loss": 88.2521,
+      "step": 56210
+    },
+    {
+      "epoch": 0.4650701079538404,
+      "grad_norm": 503.7681579589844,
+      "learning_rate": 6.6393376246998485e-06,
+      "loss": 105.7174,
+      "step": 56220
+    },
+    {
+      "epoch": 0.46515283120320966,
+      "grad_norm": 721.5469360351562,
+      "learning_rate": 6.638005152288811e-06,
+      "loss": 100.881,
+      "step": 56230
+    },
+    {
+      "epoch": 0.4652355544525789,
+      "grad_norm": 1773.4716796875,
+      "learning_rate": 6.636672549549221e-06,
+      "loss": 115.8908,
+      "step": 56240
+    },
+    {
+      "epoch": 0.4653182777019481,
+      "grad_norm": 960.2298583984375,
+      "learning_rate": 6.635339816587109e-06,
+      "loss": 109.554,
+      "step": 56250
+    },
+    {
+      "epoch": 0.46540100095131737,
+      "grad_norm": 1094.2969970703125,
+      "learning_rate": 6.634006953508512e-06,
+      "loss": 104.5612,
+      "step": 56260
+    },
+    {
+      "epoch": 0.4654837242006866,
+      "grad_norm": 706.4091186523438,
+      "learning_rate": 6.63267396041948e-06,
+      "loss": 113.3086,
+      "step": 56270
+    },
+    {
+      "epoch": 0.4655664474500558,
+      "grad_norm": 1204.162841796875,
+      "learning_rate": 6.631340837426075e-06,
+      "loss": 105.2585,
+      "step": 56280
+    },
+    {
+      "epoch": 0.4656491706994251,
+      "grad_norm": 807.52734375,
+      "learning_rate": 6.630007584634366e-06,
+      "loss": 78.1581,
+      "step": 56290
+    },
+    {
+      "epoch": 0.46573189394879433,
+      "grad_norm": 912.2439575195312,
+      "learning_rate": 6.628674202150434e-06,
+      "loss": 95.7974,
+      "step": 56300
+    },
+    {
+      "epoch": 0.4658146171981635,
+      "grad_norm": 793.3104858398438,
+      "learning_rate": 6.627340690080371e-06,
+      "loss": 94.2195,
+      "step": 56310
+    },
+    {
+      "epoch": 0.4658973404475328,
+      "grad_norm": 1065.8125,
+      "learning_rate": 6.626007048530276e-06,
+      "loss": 72.4793,
+      "step": 56320
+    },
+    {
+      "epoch": 0.46598006369690204,
+      "grad_norm": 565.6997680664062,
+      "learning_rate": 6.624673277606264e-06,
+      "loss": 90.7239,
+      "step": 56330
+    },
+    {
+      "epoch": 0.46606278694627123,
+      "grad_norm": 1088.072998046875,
+      "learning_rate": 6.623339377414456e-06,
+      "loss": 114.9387,
+      "step": 56340
+    },
+    {
+      "epoch": 0.4661455101956405,
+      "grad_norm": 1036.712158203125,
+      "learning_rate": 6.622005348060983e-06,
+      "loss": 98.3773,
+      "step": 56350
+    },
+    {
+      "epoch": 0.46622823344500974,
+      "grad_norm": 657.3320922851562,
+      "learning_rate": 6.620671189651988e-06,
+      "loss": 78.3256,
+      "step": 56360
+    },
+    {
+      "epoch": 0.46631095669437894,
+      "grad_norm": 396.6316223144531,
+      "learning_rate": 6.6193369022936245e-06,
+      "loss": 111.291,
+      "step": 56370
+    },
+    {
+      "epoch": 0.4663936799437482,
+      "grad_norm": 747.0289306640625,
+      "learning_rate": 6.618002486092056e-06,
+      "loss": 131.3509,
+      "step": 56380
+    },
+    {
+      "epoch": 0.46647640319311745,
+      "grad_norm": 645.677978515625,
+      "learning_rate": 6.616667941153456e-06,
+      "loss": 112.265,
+      "step": 56390
+    },
+    {
+      "epoch": 0.46655912644248665,
+      "grad_norm": 2745.71533203125,
+      "learning_rate": 6.615333267584007e-06,
+      "loss": 94.6054,
+      "step": 56400
+    },
+    {
+      "epoch": 0.4666418496918559,
+      "grad_norm": 617.6024780273438,
+      "learning_rate": 6.613998465489902e-06,
+      "loss": 86.2714,
+      "step": 56410
+    },
+    {
+      "epoch": 0.46672457294122516,
+      "grad_norm": 2982.6630859375,
+      "learning_rate": 6.612663534977347e-06,
+      "loss": 158.7063,
+      "step": 56420
+    },
+    {
+      "epoch": 0.46680729619059436,
+      "grad_norm": 372.95379638671875,
+      "learning_rate": 6.611328476152557e-06,
+      "loss": 127.7486,
+      "step": 56430
+    },
+    {
+      "epoch": 0.4668900194399636,
+      "grad_norm": 665.3734130859375,
+      "learning_rate": 6.609993289121753e-06,
+      "loss": 108.4631,
+      "step": 56440
+    },
+    {
+      "epoch": 0.4669727426893328,
+      "grad_norm": 637.3652954101562,
+      "learning_rate": 6.608657973991172e-06,
+      "loss": 84.1843,
+      "step": 56450
+    },
+    {
+      "epoch": 0.46705546593870206,
+      "grad_norm": 962.6121215820312,
+      "learning_rate": 6.607322530867061e-06,
+      "loss": 88.8814,
+      "step": 56460
+    },
+    {
+      "epoch": 0.4671381891880713,
+      "grad_norm": 588.2682495117188,
+      "learning_rate": 6.605986959855672e-06,
+      "loss": 76.9025,
+      "step": 56470
+    },
+    {
+      "epoch": 0.4672209124374405,
+      "grad_norm": 1287.7611083984375,
+      "learning_rate": 6.60465126106327e-06,
+      "loss": 120.8118,
+      "step": 56480
+    },
+    {
+      "epoch": 0.46730363568680977,
+      "grad_norm": 932.3352661132812,
+      "learning_rate": 6.6033154345961314e-06,
+      "loss": 89.3703,
+      "step": 56490
+    },
+    {
+      "epoch": 0.467386358936179,
+      "grad_norm": 577.4403076171875,
+      "learning_rate": 6.601979480560543e-06,
+      "loss": 94.8228,
+      "step": 56500
+    },
+    {
+      "epoch": 0.4674690821855482,
+      "grad_norm": 937.298583984375,
+      "learning_rate": 6.6006433990627985e-06,
+      "loss": 122.7913,
+      "step": 56510
+    },
+    {
+      "epoch": 0.4675518054349175,
+      "grad_norm": 588.9931030273438,
+      "learning_rate": 6.599307190209206e-06,
+      "loss": 112.8304,
+      "step": 56520
+    },
+    {
+      "epoch": 0.46763452868428673,
+      "grad_norm": 1339.0936279296875,
+      "learning_rate": 6.5979708541060796e-06,
+      "loss": 144.9437,
+      "step": 56530
+    },
+    {
+      "epoch": 0.46771725193365593,
+      "grad_norm": 1465.9853515625,
+      "learning_rate": 6.596634390859745e-06,
+      "loss": 91.1747,
+      "step": 56540
+    },
+    {
+      "epoch": 0.4677999751830252,
+      "grad_norm": 890.4679565429688,
+      "learning_rate": 6.59529780057654e-06,
+      "loss": 82.902,
+      "step": 56550
+    },
+    {
+      "epoch": 0.46788269843239444,
+      "grad_norm": 839.3494873046875,
+      "learning_rate": 6.593961083362811e-06,
+      "loss": 85.888,
+      "step": 56560
+    },
+    {
+      "epoch": 0.46796542168176364,
+      "grad_norm": 992.9126586914062,
+      "learning_rate": 6.592624239324914e-06,
+      "loss": 109.3493,
+      "step": 56570
+    },
+    {
+      "epoch": 0.4680481449311329,
+      "grad_norm": 1971.344482421875,
+      "learning_rate": 6.591287268569215e-06,
+      "loss": 99.3406,
+      "step": 56580
+    },
+    {
+      "epoch": 0.46813086818050215,
+      "grad_norm": 1292.9296875,
+      "learning_rate": 6.589950171202092e-06,
+      "loss": 67.5184,
+      "step": 56590
+    },
+    {
+      "epoch": 0.46821359142987135,
+      "grad_norm": 1166.9710693359375,
+      "learning_rate": 6.588612947329929e-06,
+      "loss": 121.0264,
+      "step": 56600
+    },
+    {
+      "epoch": 0.4682963146792406,
+      "grad_norm": 1211.2210693359375,
+      "learning_rate": 6.587275597059125e-06,
+      "loss": 90.2953,
+      "step": 56610
+    },
+    {
+      "epoch": 0.46837903792860985,
+      "grad_norm": 860.5660400390625,
+      "learning_rate": 6.585938120496087e-06,
+      "loss": 110.3769,
+      "step": 56620
+    },
+    {
+      "epoch": 0.46846176117797905,
+      "grad_norm": 1788.8858642578125,
+      "learning_rate": 6.584600517747232e-06,
+      "loss": 111.8886,
+      "step": 56630
+    },
+    {
+      "epoch": 0.4685444844273483,
+      "grad_norm": 826.596923828125,
+      "learning_rate": 6.583262788918985e-06,
+      "loss": 84.6638,
+      "step": 56640
+    },
+    {
+      "epoch": 0.46862720767671756,
+      "grad_norm": 794.5626220703125,
+      "learning_rate": 6.581924934117783e-06,
+      "loss": 108.4789,
+      "step": 56650
+    },
+    {
+      "epoch": 0.46870993092608676,
+      "grad_norm": 1118.61962890625,
+      "learning_rate": 6.580586953450076e-06,
+      "loss": 74.7545,
+      "step": 56660
+    },
+    {
+      "epoch": 0.468792654175456,
+      "grad_norm": 763.598388671875,
+      "learning_rate": 6.579248847022317e-06,
+      "loss": 78.2781,
+      "step": 56670
+    },
+    {
+      "epoch": 0.46887537742482527,
+      "grad_norm": 1182.76806640625,
+      "learning_rate": 6.577910614940978e-06,
+      "loss": 97.6059,
+      "step": 56680
+    },
+    {
+      "epoch": 0.46895810067419447,
+      "grad_norm": 1287.8709716796875,
+      "learning_rate": 6.576572257312531e-06,
+      "loss": 94.5327,
+      "step": 56690
+    },
+    {
+      "epoch": 0.4690408239235637,
+      "grad_norm": 625.0133666992188,
+      "learning_rate": 6.5752337742434644e-06,
+      "loss": 90.6583,
+      "step": 56700
+    },
+    {
+      "epoch": 0.469123547172933,
+      "grad_norm": 352.9053649902344,
+      "learning_rate": 6.573895165840276e-06,
+      "loss": 99.1602,
+      "step": 56710
+    },
+    {
+      "epoch": 0.4692062704223022,
+      "grad_norm": 1227.5992431640625,
+      "learning_rate": 6.5725564322094745e-06,
+      "loss": 106.9007,
+      "step": 56720
+    },
+    {
+      "epoch": 0.46928899367167143,
+      "grad_norm": 1981.240234375,
+      "learning_rate": 6.571217573457573e-06,
+      "loss": 110.456,
+      "step": 56730
+    },
+    {
+      "epoch": 0.4693717169210407,
+      "grad_norm": 915.96728515625,
+      "learning_rate": 6.569878589691101e-06,
+      "loss": 64.4055,
+      "step": 56740
+    },
+    {
+      "epoch": 0.4694544401704099,
+      "grad_norm": 1044.962890625,
+      "learning_rate": 6.568539481016593e-06,
+      "loss": 108.6498,
+      "step": 56750
+    },
+    {
+      "epoch": 0.46953716341977914,
+      "grad_norm": 1632.702880859375,
+      "learning_rate": 6.567200247540599e-06,
+      "loss": 123.9141,
+      "step": 56760
+    },
+    {
+      "epoch": 0.4696198866691484,
+      "grad_norm": 1159.7938232421875,
+      "learning_rate": 6.5658608893696714e-06,
+      "loss": 105.7761,
+      "step": 56770
+    },
+    {
+      "epoch": 0.4697026099185176,
+      "grad_norm": 1023.9876098632812,
+      "learning_rate": 6.564521406610382e-06,
+      "loss": 130.591,
+      "step": 56780
+    },
+    {
+      "epoch": 0.46978533316788684,
+      "grad_norm": 1106.73779296875,
+      "learning_rate": 6.563181799369301e-06,
+      "loss": 89.8389,
+      "step": 56790
+    },
+    {
+      "epoch": 0.4698680564172561,
+      "grad_norm": 636.095458984375,
+      "learning_rate": 6.561842067753021e-06,
+      "loss": 95.9526,
+      "step": 56800
+    },
+    {
+      "epoch": 0.4699507796666253,
+      "grad_norm": 725.4862670898438,
+      "learning_rate": 6.560502211868135e-06,
+      "loss": 71.1143,
+      "step": 56810
+    },
+    {
+      "epoch": 0.47003350291599455,
+      "grad_norm": 946.0494384765625,
+      "learning_rate": 6.55916223182125e-06,
+      "loss": 106.9812,
+      "step": 56820
+    },
+    {
+      "epoch": 0.47011622616536375,
+      "grad_norm": 1510.7655029296875,
+      "learning_rate": 6.55782212771898e-06,
+      "loss": 147.8765,
+      "step": 56830
+    },
+    {
+      "epoch": 0.470198949414733,
+      "grad_norm": 1354.79296875,
+      "learning_rate": 6.5564818996679536e-06,
+      "loss": 121.1624,
+      "step": 56840
+    },
+    {
+      "epoch": 0.47028167266410226,
+      "grad_norm": 1657.0260009765625,
+      "learning_rate": 6.555141547774807e-06,
+      "loss": 153.1369,
+      "step": 56850
+    },
+    {
+      "epoch": 0.47036439591347146,
+      "grad_norm": 432.9583435058594,
+      "learning_rate": 6.553801072146184e-06,
+      "loss": 112.8747,
+      "step": 56860
+    },
+    {
+      "epoch": 0.4704471191628407,
+      "grad_norm": 1424.265380859375,
+      "learning_rate": 6.55246047288874e-06,
+      "loss": 102.5585,
+      "step": 56870
+    },
+    {
+      "epoch": 0.47052984241220996,
+      "grad_norm": 842.8648071289062,
+      "learning_rate": 6.551119750109142e-06,
+      "loss": 95.888,
+      "step": 56880
+    },
+    {
+      "epoch": 0.47061256566157916,
+      "grad_norm": 1254.891845703125,
+      "learning_rate": 6.5497789039140635e-06,
+      "loss": 88.7369,
+      "step": 56890
+    },
+    {
+      "epoch": 0.4706952889109484,
+      "grad_norm": 593.7796020507812,
+      "learning_rate": 6.54843793441019e-06,
+      "loss": 117.424,
+      "step": 56900
+    },
+    {
+      "epoch": 0.47077801216031767,
+      "grad_norm": 5807.5322265625,
+      "learning_rate": 6.547096841704217e-06,
+      "loss": 123.5693,
+      "step": 56910
+    },
+    {
+      "epoch": 0.47086073540968687,
+      "grad_norm": 698.7401733398438,
+      "learning_rate": 6.545755625902848e-06,
+      "loss": 108.5493,
+      "step": 56920
+    },
+    {
+      "epoch": 0.4709434586590561,
+      "grad_norm": 770.432373046875,
+      "learning_rate": 6.544414287112798e-06,
+      "loss": 60.7358,
+      "step": 56930
+    },
+    {
+      "epoch": 0.4710261819084254,
+      "grad_norm": 772.1907348632812,
+      "learning_rate": 6.54307282544079e-06,
+      "loss": 90.2566,
+      "step": 56940
+    },
+    {
+      "epoch": 0.4711089051577946,
+      "grad_norm": 543.7560424804688,
+      "learning_rate": 6.5417312409935606e-06,
+      "loss": 74.9508,
+      "step": 56950
+    },
+    {
+      "epoch": 0.47119162840716383,
+      "grad_norm": 605.939697265625,
+      "learning_rate": 6.540389533877852e-06,
+      "loss": 117.9458,
+      "step": 56960
+    },
+    {
+      "epoch": 0.4712743516565331,
+      "grad_norm": 546.9862670898438,
+      "learning_rate": 6.539047704200417e-06,
+      "loss": 83.1111,
+      "step": 56970
+    },
+    {
+      "epoch": 0.4713570749059023,
+      "grad_norm": 860.575439453125,
+      "learning_rate": 6.53770575206802e-06,
+      "loss": 102.208,
+      "step": 56980
+    },
+    {
+      "epoch": 0.47143979815527154,
+      "grad_norm": 616.0123291015625,
+      "learning_rate": 6.536363677587433e-06,
+      "loss": 101.8752,
+      "step": 56990
+    },
+    {
+      "epoch": 0.4715225214046408,
+      "grad_norm": 912.4677124023438,
+      "learning_rate": 6.535021480865439e-06,
+      "loss": 94.7414,
+      "step": 57000
+    },
+    {
+      "epoch": 0.47160524465401,
+      "grad_norm": 1033.161376953125,
+      "learning_rate": 6.5336791620088306e-06,
+      "loss": 98.8203,
+      "step": 57010
+    },
+    {
+      "epoch": 0.47168796790337925,
+      "grad_norm": 813.3025512695312,
+      "learning_rate": 6.53233672112441e-06,
+      "loss": 111.74,
+      "step": 57020
+    },
+    {
+      "epoch": 0.4717706911527485,
+      "grad_norm": 1582.189208984375,
+      "learning_rate": 6.530994158318988e-06,
+      "loss": 113.1147,
+      "step": 57030
+    },
+    {
+      "epoch": 0.4718534144021177,
+      "grad_norm": 862.9869384765625,
+      "learning_rate": 6.529651473699389e-06,
+      "loss": 85.5126,
+      "step": 57040
+    },
+    {
+      "epoch": 0.47193613765148695,
+      "grad_norm": 1320.765869140625,
+      "learning_rate": 6.528308667372441e-06,
+      "loss": 101.8769,
+      "step": 57050
+    },
+    {
+      "epoch": 0.4720188609008562,
+      "grad_norm": 717.2901611328125,
+      "learning_rate": 6.526965739444988e-06,
+      "loss": 119.4057,
+      "step": 57060
+    },
+    {
+      "epoch": 0.4721015841502254,
+      "grad_norm": 699.390625,
+      "learning_rate": 6.525622690023878e-06,
+      "loss": 105.9801,
+      "step": 57070
+    },
+    {
+      "epoch": 0.47218430739959466,
+      "grad_norm": 1108.2218017578125,
+      "learning_rate": 6.524279519215972e-06,
+      "loss": 105.0386,
+      "step": 57080
+    },
+    {
+      "epoch": 0.4722670306489639,
+      "grad_norm": 1312.4166259765625,
+      "learning_rate": 6.522936227128139e-06,
+      "loss": 116.3358,
+      "step": 57090
+    },
+    {
+      "epoch": 0.4723497538983331,
+      "grad_norm": 881.8489379882812,
+      "learning_rate": 6.521592813867261e-06,
+      "loss": 115.962,
+      "step": 57100
+    },
+    {
+      "epoch": 0.47243247714770237,
+      "grad_norm": 1025.9259033203125,
+      "learning_rate": 6.520249279540227e-06,
+      "loss": 112.2708,
+      "step": 57110
+    },
+    {
+      "epoch": 0.4725152003970716,
+      "grad_norm": 968.8225708007812,
+      "learning_rate": 6.5189056242539325e-06,
+      "loss": 81.8784,
+      "step": 57120
+    },
+    {
+      "epoch": 0.4725979236464408,
+      "grad_norm": 776.3606567382812,
+      "learning_rate": 6.51756184811529e-06,
+      "loss": 97.869,
+      "step": 57130
+    },
+    {
+      "epoch": 0.4726806468958101,
+      "grad_norm": 1188.362060546875,
+      "learning_rate": 6.516217951231215e-06,
+      "loss": 78.5015,
+      "step": 57140
+    },
+    {
+      "epoch": 0.47276337014517933,
+      "grad_norm": 1396.0478515625,
+      "learning_rate": 6.514873933708637e-06,
+      "loss": 115.7227,
+      "step": 57150
+    },
+    {
+      "epoch": 0.47284609339454853,
+      "grad_norm": 701.3724365234375,
+      "learning_rate": 6.513529795654493e-06,
+      "loss": 91.0152,
+      "step": 57160
+    },
+    {
+      "epoch": 0.4729288166439178,
+      "grad_norm": 773.8468627929688,
+      "learning_rate": 6.512185537175727e-06,
+      "loss": 140.7189,
+      "step": 57170
+    },
+    {
+      "epoch": 0.473011539893287,
+      "grad_norm": 948.971923828125,
+      "learning_rate": 6.5108411583793e-06,
+      "loss": 163.1197,
+      "step": 57180
+    },
+    {
+      "epoch": 0.47309426314265624,
+      "grad_norm": 662.3345336914062,
+      "learning_rate": 6.509496659372175e-06,
+      "loss": 92.557,
+      "step": 57190
+    },
+    {
+      "epoch": 0.4731769863920255,
+      "grad_norm": 902.59521484375,
+      "learning_rate": 6.508152040261329e-06,
+      "loss": 98.1511,
+      "step": 57200
+    },
+    {
+      "epoch": 0.4732597096413947,
+      "grad_norm": 912.7888793945312,
+      "learning_rate": 6.506807301153746e-06,
+      "loss": 82.9847,
+      "step": 57210
+    },
+    {
+      "epoch": 0.47334243289076394,
+      "grad_norm": 711.9552001953125,
+      "learning_rate": 6.5054624421564204e-06,
+      "loss": 85.3893,
+      "step": 57220
+    },
+    {
+      "epoch": 0.4734251561401332,
+      "grad_norm": 996.9633178710938,
+      "learning_rate": 6.504117463376358e-06,
+      "loss": 103.8014,
+      "step": 57230
+    },
+    {
+      "epoch": 0.4735078793895024,
+      "grad_norm": 889.486083984375,
+      "learning_rate": 6.502772364920573e-06,
+      "loss": 119.4068,
+      "step": 57240
+    },
+    {
+      "epoch": 0.47359060263887165,
+      "grad_norm": 1359.7982177734375,
+      "learning_rate": 6.501427146896087e-06,
+      "loss": 131.6854,
+      "step": 57250
+    },
+    {
+      "epoch": 0.4736733258882409,
+      "grad_norm": 863.1353759765625,
+      "learning_rate": 6.5000818094099345e-06,
+      "loss": 125.5572,
+      "step": 57260
+    },
+    {
+      "epoch": 0.4737560491376101,
+      "grad_norm": 583.1968383789062,
+      "learning_rate": 6.498736352569155e-06,
+      "loss": 97.2687,
+      "step": 57270
+    },
+    {
+      "epoch": 0.47383877238697936,
+      "grad_norm": 613.0692138671875,
+      "learning_rate": 6.497390776480804e-06,
+      "loss": 83.3367,
+      "step": 57280
+    },
+    {
+      "epoch": 0.4739214956363486,
+      "grad_norm": 894.8236083984375,
+      "learning_rate": 6.49604508125194e-06,
+      "loss": 123.6242,
+      "step": 57290
+    },
+    {
+      "epoch": 0.4740042188857178,
+      "grad_norm": 1123.2113037109375,
+      "learning_rate": 6.4946992669896355e-06,
+      "loss": 90.5414,
+      "step": 57300
+    },
+    {
+      "epoch": 0.47408694213508706,
+      "grad_norm": 634.5689086914062,
+      "learning_rate": 6.493353333800969e-06,
+      "loss": 84.9406,
+      "step": 57310
+    },
+    {
+      "epoch": 0.4741696653844563,
+      "grad_norm": 618.6438598632812,
+      "learning_rate": 6.492007281793032e-06,
+      "loss": 101.8569,
+      "step": 57320
+    },
+    {
+      "epoch": 0.4742523886338255,
+      "grad_norm": 760.6466064453125,
+      "learning_rate": 6.490661111072923e-06,
+      "loss": 98.2763,
+      "step": 57330
+    },
+    {
+      "epoch": 0.47433511188319477,
+      "grad_norm": 642.0409545898438,
+      "learning_rate": 6.489314821747751e-06,
+      "loss": 82.7239,
+      "step": 57340
+    },
+    {
+      "epoch": 0.474417835132564,
+      "grad_norm": 491.7732849121094,
+      "learning_rate": 6.487968413924634e-06,
+      "loss": 82.5276,
+      "step": 57350
+    },
+    {
+      "epoch": 0.4745005583819332,
+      "grad_norm": 847.6077270507812,
+      "learning_rate": 6.486621887710698e-06,
+      "loss": 85.0193,
+      "step": 57360
+    },
+    {
+      "epoch": 0.4745832816313025,
+      "grad_norm": 1501.081787109375,
+      "learning_rate": 6.485275243213081e-06,
+      "loss": 88.0962,
+      "step": 57370
+    },
+    {
+      "epoch": 0.47466600488067173,
+      "grad_norm": 961.3021850585938,
+      "learning_rate": 6.4839284805389305e-06,
+      "loss": 131.3477,
+      "step": 57380
+    },
+    {
+      "epoch": 0.47474872813004093,
+      "grad_norm": 695.2109985351562,
+      "learning_rate": 6.4825815997954e-06,
+      "loss": 99.5357,
+      "step": 57390
+    },
+    {
+      "epoch": 0.4748314513794102,
+      "grad_norm": 478.3403625488281,
+      "learning_rate": 6.481234601089655e-06,
+      "loss": 96.97,
+      "step": 57400
+    },
+    {
+      "epoch": 0.47491417462877944,
+      "grad_norm": 967.1907348632812,
+      "learning_rate": 6.4798874845288725e-06,
+      "loss": 84.3332,
+      "step": 57410
+    },
+    {
+      "epoch": 0.47499689787814864,
+      "grad_norm": 789.9042358398438,
+      "learning_rate": 6.4785402502202345e-06,
+      "loss": 98.0186,
+      "step": 57420
+    },
+    {
+      "epoch": 0.4750796211275179,
+      "grad_norm": 515.9044799804688,
+      "learning_rate": 6.477192898270934e-06,
+      "loss": 98.3447,
+      "step": 57430
+    },
+    {
+      "epoch": 0.47516234437688715,
+      "grad_norm": 777.9580078125,
+      "learning_rate": 6.475845428788173e-06,
+      "loss": 92.8213,
+      "step": 57440
+    },
+    {
+      "epoch": 0.47524506762625635,
+      "grad_norm": 732.5836791992188,
+      "learning_rate": 6.474497841879166e-06,
+      "loss": 100.1301,
+      "step": 57450
+    },
+    {
+      "epoch": 0.4753277908756256,
+      "grad_norm": 905.1466064453125,
+      "learning_rate": 6.473150137651132e-06,
+      "loss": 87.0629,
+      "step": 57460
+    },
+    {
+      "epoch": 0.47541051412499485,
+      "grad_norm": 1096.08935546875,
+      "learning_rate": 6.471802316211302e-06,
+      "loss": 94.4893,
+      "step": 57470
+    },
+    {
+      "epoch": 0.47549323737436405,
+      "grad_norm": 781.8897094726562,
+      "learning_rate": 6.4704543776669174e-06,
+      "loss": 85.6178,
+      "step": 57480
+    },
+    {
+      "epoch": 0.4755759606237333,
+      "grad_norm": 1561.7882080078125,
+      "learning_rate": 6.469106322125227e-06,
+      "loss": 110.5366,
+      "step": 57490
+    },
+    {
+      "epoch": 0.47565868387310256,
+      "grad_norm": 773.6251220703125,
+      "learning_rate": 6.467758149693486e-06,
+      "loss": 94.7606,
+      "step": 57500
+    },
+    {
+      "epoch": 0.47574140712247176,
+      "grad_norm": 696.1630249023438,
+      "learning_rate": 6.466409860478967e-06,
+      "loss": 84.5489,
+      "step": 57510
+    },
+    {
+      "epoch": 0.475824130371841,
+      "grad_norm": 1580.1494140625,
+      "learning_rate": 6.465061454588946e-06,
+      "loss": 114.6306,
+      "step": 57520
+    },
+    {
+      "epoch": 0.47590685362121027,
+      "grad_norm": 584.140380859375,
+      "learning_rate": 6.463712932130708e-06,
+      "loss": 91.1199,
+      "step": 57530
+    },
+    {
+      "epoch": 0.47598957687057947,
+      "grad_norm": 678.8070068359375,
+      "learning_rate": 6.462364293211549e-06,
+      "loss": 80.3412,
+      "step": 57540
+    },
+    {
+      "epoch": 0.4760723001199487,
+      "grad_norm": 671.9398803710938,
+      "learning_rate": 6.4610155379387755e-06,
+      "loss": 116.5642,
+      "step": 57550
+    },
+    {
+      "epoch": 0.4761550233693179,
+      "grad_norm": 1506.737060546875,
+      "learning_rate": 6.459666666419699e-06,
+      "loss": 79.6158,
+      "step": 57560
+    },
+    {
+      "epoch": 0.4762377466186872,
+      "grad_norm": 695.2562255859375,
+      "learning_rate": 6.4583176787616466e-06,
+      "loss": 61.0726,
+      "step": 57570
+    },
+    {
+      "epoch": 0.47632046986805643,
+      "grad_norm": 1047.5999755859375,
+      "learning_rate": 6.456968575071951e-06,
+      "loss": 115.039,
+      "step": 57580
+    },
+    {
+      "epoch": 0.4764031931174256,
+      "grad_norm": 865.7882690429688,
+      "learning_rate": 6.45561935545795e-06,
+      "loss": 111.2492,
+      "step": 57590
+    },
+    {
+      "epoch": 0.4764859163667949,
+      "grad_norm": 991.0267333984375,
+      "learning_rate": 6.454270020026996e-06,
+      "loss": 88.2671,
+      "step": 57600
+    },
+    {
+      "epoch": 0.47656863961616414,
+      "grad_norm": 985.0994873046875,
+      "learning_rate": 6.452920568886452e-06,
+      "loss": 101.4334,
+      "step": 57610
+    },
+    {
+      "epoch": 0.47665136286553333,
+      "grad_norm": 621.8745727539062,
+      "learning_rate": 6.451571002143687e-06,
+      "loss": 80.6865,
+      "step": 57620
+    },
+    {
+      "epoch": 0.4767340861149026,
+      "grad_norm": 851.007080078125,
+      "learning_rate": 6.450221319906079e-06,
+      "loss": 93.8453,
+      "step": 57630
+    },
+    {
+      "epoch": 0.47681680936427184,
+      "grad_norm": 1132.76708984375,
+      "learning_rate": 6.448871522281016e-06,
+      "loss": 85.5419,
+      "step": 57640
+    },
+    {
+      "epoch": 0.47689953261364104,
+      "grad_norm": 739.5919189453125,
+      "learning_rate": 6.447521609375894e-06,
+      "loss": 67.1973,
+      "step": 57650
+    },
+    {
+      "epoch": 0.4769822558630103,
+      "grad_norm": 829.6648559570312,
+      "learning_rate": 6.446171581298123e-06,
+      "loss": 106.6791,
+      "step": 57660
+    },
+    {
+      "epoch": 0.47706497911237955,
+      "grad_norm": 1077.7239990234375,
+      "learning_rate": 6.444821438155115e-06,
+      "loss": 80.1578,
+      "step": 57670
+    },
+    {
+      "epoch": 0.47714770236174875,
+      "grad_norm": 654.5147094726562,
+      "learning_rate": 6.443471180054297e-06,
+      "loss": 69.1088,
+      "step": 57680
+    },
+    {
+      "epoch": 0.477230425611118,
+      "grad_norm": 1064.0953369140625,
+      "learning_rate": 6.442120807103102e-06,
+      "loss": 96.6103,
+      "step": 57690
+    },
+    {
+      "epoch": 0.47731314886048726,
+      "grad_norm": 846.7904663085938,
+      "learning_rate": 6.440770319408971e-06,
+      "loss": 113.4604,
+      "step": 57700
+    },
+    {
+      "epoch": 0.47739587210985646,
+      "grad_norm": 1160.3079833984375,
+      "learning_rate": 6.43941971707936e-06,
+      "loss": 110.0301,
+      "step": 57710
+    },
+    {
+      "epoch": 0.4774785953592257,
+      "grad_norm": 524.7237548828125,
+      "learning_rate": 6.438069000221727e-06,
+      "loss": 92.4454,
+      "step": 57720
+    },
+    {
+      "epoch": 0.47756131860859496,
+      "grad_norm": 363.4929504394531,
+      "learning_rate": 6.4367181689435434e-06,
+      "loss": 107.8367,
+      "step": 57730
+    },
+    {
+      "epoch": 0.47764404185796416,
+      "grad_norm": 953.7645263671875,
+      "learning_rate": 6.435367223352289e-06,
+      "loss": 121.483,
+      "step": 57740
+    },
+    {
+      "epoch": 0.4777267651073334,
+      "grad_norm": 1026.001708984375,
+      "learning_rate": 6.434016163555452e-06,
+      "loss": 91.6397,
+      "step": 57750
+    },
+    {
+      "epoch": 0.47780948835670267,
+      "grad_norm": 811.3062133789062,
+      "learning_rate": 6.432664989660531e-06,
+      "loss": 91.4573,
+      "step": 57760
+    },
+    {
+      "epoch": 0.47789221160607187,
+      "grad_norm": 1303.14111328125,
+      "learning_rate": 6.43131370177503e-06,
+      "loss": 112.111,
+      "step": 57770
+    },
+    {
+      "epoch": 0.4779749348554411,
+      "grad_norm": 754.6259765625,
+      "learning_rate": 6.429962300006468e-06,
+      "loss": 100.0983,
+      "step": 57780
+    },
+    {
+      "epoch": 0.4780576581048104,
+      "grad_norm": 738.0693359375,
+      "learning_rate": 6.428610784462368e-06,
+      "loss": 79.652,
+      "step": 57790
+    },
+    {
+      "epoch": 0.4781403813541796,
+      "grad_norm": 1683.4595947265625,
+      "learning_rate": 6.427259155250265e-06,
+      "loss": 95.6643,
+      "step": 57800
+    },
+    {
+      "epoch": 0.47822310460354883,
+      "grad_norm": 840.048828125,
+      "learning_rate": 6.4259074124777e-06,
+      "loss": 93.0407,
+      "step": 57810
+    },
+    {
+      "epoch": 0.4783058278529181,
+      "grad_norm": 849.6332397460938,
+      "learning_rate": 6.4245555562522265e-06,
+      "loss": 98.129,
+      "step": 57820
+    },
+    {
+      "epoch": 0.4783885511022873,
+      "grad_norm": 495.7574768066406,
+      "learning_rate": 6.423203586681406e-06,
+      "loss": 87.5308,
+      "step": 57830
+    },
+    {
+      "epoch": 0.47847127435165654,
+      "grad_norm": 998.4563598632812,
+      "learning_rate": 6.421851503872807e-06,
+      "loss": 124.0158,
+      "step": 57840
+    },
+    {
+      "epoch": 0.4785539976010258,
+      "grad_norm": 636.4727783203125,
+      "learning_rate": 6.42049930793401e-06,
+      "loss": 74.8035,
+      "step": 57850
+    },
+    {
+      "epoch": 0.478636720850395,
+      "grad_norm": 872.4989013671875,
+      "learning_rate": 6.419146998972602e-06,
+      "loss": 78.3126,
+      "step": 57860
+    },
+    {
+      "epoch": 0.47871944409976425,
+      "grad_norm": 1358.409912109375,
+      "learning_rate": 6.417794577096179e-06,
+      "loss": 98.5134,
+      "step": 57870
+    },
+    {
+      "epoch": 0.4788021673491335,
+      "grad_norm": 1421.047119140625,
+      "learning_rate": 6.41644204241235e-06,
+      "loss": 119.0474,
+      "step": 57880
+    },
+    {
+      "epoch": 0.4788848905985027,
+      "grad_norm": 750.8233642578125,
+      "learning_rate": 6.4150893950287275e-06,
+      "loss": 108.5803,
+      "step": 57890
+    },
+    {
+      "epoch": 0.47896761384787195,
+      "grad_norm": 2111.186767578125,
+      "learning_rate": 6.413736635052936e-06,
+      "loss": 141.4099,
+      "step": 57900
+    },
+    {
+      "epoch": 0.47905033709724115,
+      "grad_norm": 681.4733276367188,
+      "learning_rate": 6.41238376259261e-06,
+      "loss": 63.5274,
+      "step": 57910
+    },
+    {
+      "epoch": 0.4791330603466104,
+      "grad_norm": 805.0604248046875,
+      "learning_rate": 6.411030777755389e-06,
+      "loss": 98.8302,
+      "step": 57920
+    },
+    {
+      "epoch": 0.47921578359597966,
+      "grad_norm": 810.17041015625,
+      "learning_rate": 6.409677680648925e-06,
+      "loss": 113.7874,
+      "step": 57930
+    },
+    {
+      "epoch": 0.47929850684534886,
+      "grad_norm": 827.2803344726562,
+      "learning_rate": 6.4083244713808765e-06,
+      "loss": 119.2214,
+      "step": 57940
+    },
+    {
+      "epoch": 0.4793812300947181,
+      "grad_norm": 1362.2303466796875,
+      "learning_rate": 6.406971150058914e-06,
+      "loss": 92.4825,
+      "step": 57950
+    },
+    {
+      "epoch": 0.47946395334408737,
+      "grad_norm": 855.9775390625,
+      "learning_rate": 6.405617716790714e-06,
+      "loss": 79.9808,
+      "step": 57960
+    },
+    {
+      "epoch": 0.47954667659345657,
+      "grad_norm": 632.9134521484375,
+      "learning_rate": 6.404264171683965e-06,
+      "loss": 87.7965,
+      "step": 57970
+    },
+    {
+      "epoch": 0.4796293998428258,
+      "grad_norm": 731.3795166015625,
+      "learning_rate": 6.402910514846358e-06,
+      "loss": 82.6081,
+      "step": 57980
+    },
+    {
+      "epoch": 0.4797121230921951,
+      "grad_norm": 498.25,
+      "learning_rate": 6.4015567463856e-06,
+      "loss": 82.2082,
+      "step": 57990
+    },
+    {
+      "epoch": 0.4797948463415643,
+      "grad_norm": 1057.318359375,
+      "learning_rate": 6.400202866409405e-06,
+      "loss": 111.3383,
+      "step": 58000
+    },
+    {
+      "epoch": 0.47987756959093353,
+      "grad_norm": 775.5044555664062,
+      "learning_rate": 6.398848875025494e-06,
+      "loss": 88.3121,
+      "step": 58010
+    },
+    {
+      "epoch": 0.4799602928403028,
+      "grad_norm": 1417.432373046875,
+      "learning_rate": 6.3974947723415985e-06,
+      "loss": 95.6965,
+      "step": 58020
+    },
+    {
+      "epoch": 0.480043016089672,
+      "grad_norm": 644.7517700195312,
+      "learning_rate": 6.396140558465456e-06,
+      "loss": 104.0072,
+      "step": 58030
+    },
+    {
+      "epoch": 0.48012573933904124,
+      "grad_norm": 849.2105712890625,
+      "learning_rate": 6.394786233504816e-06,
+      "loss": 86.352,
+      "step": 58040
+    },
+    {
+      "epoch": 0.4802084625884105,
+      "grad_norm": 824.1756591796875,
+      "learning_rate": 6.39343179756744e-06,
+      "loss": 114.2667,
+      "step": 58050
+    },
+    {
+      "epoch": 0.4802911858377797,
+      "grad_norm": 567.4685668945312,
+      "learning_rate": 6.392077250761088e-06,
+      "loss": 88.4801,
+      "step": 58060
+    },
+    {
+      "epoch": 0.48037390908714894,
+      "grad_norm": 910.6168823242188,
+      "learning_rate": 6.390722593193538e-06,
+      "loss": 85.6822,
+      "step": 58070
+    },
+    {
+      "epoch": 0.4804566323365182,
+      "grad_norm": 924.6448364257812,
+      "learning_rate": 6.389367824972575e-06,
+      "loss": 96.7753,
+      "step": 58080
+    },
+    {
+      "epoch": 0.4805393555858874,
+      "grad_norm": 814.3735961914062,
+      "learning_rate": 6.388012946205991e-06,
+      "loss": 90.9101,
+      "step": 58090
+    },
+    {
+      "epoch": 0.48062207883525665,
+      "grad_norm": 1433.7420654296875,
+      "learning_rate": 6.386657957001585e-06,
+      "loss": 105.3125,
+      "step": 58100
+    },
+    {
+      "epoch": 0.4807048020846259,
+      "grad_norm": 841.0825805664062,
+      "learning_rate": 6.38530285746717e-06,
+      "loss": 128.7726,
+      "step": 58110
+    },
+    {
+      "epoch": 0.4807875253339951,
+      "grad_norm": 1228.2952880859375,
+      "learning_rate": 6.383947647710565e-06,
+      "loss": 114.0045,
+      "step": 58120
+    },
+    {
+      "epoch": 0.48087024858336436,
+      "grad_norm": 716.3721313476562,
+      "learning_rate": 6.382592327839596e-06,
+      "loss": 96.0517,
+      "step": 58130
+    },
+    {
+      "epoch": 0.4809529718327336,
+      "grad_norm": 837.0059204101562,
+      "learning_rate": 6.381236897962102e-06,
+      "loss": 112.9432,
+      "step": 58140
+    },
+    {
+      "epoch": 0.4810356950821028,
+      "grad_norm": 1200.560302734375,
+      "learning_rate": 6.379881358185926e-06,
+      "loss": 126.8512,
+      "step": 58150
+    },
+    {
+      "epoch": 0.48111841833147206,
+      "grad_norm": 772.40478515625,
+      "learning_rate": 6.378525708618924e-06,
+      "loss": 75.8569,
+      "step": 58160
+    },
+    {
+      "epoch": 0.4812011415808413,
+      "grad_norm": 980.9589233398438,
+      "learning_rate": 6.377169949368956e-06,
+      "loss": 83.9632,
+      "step": 58170
+    },
+    {
+      "epoch": 0.4812838648302105,
+      "grad_norm": 647.2731323242188,
+      "learning_rate": 6.375814080543899e-06,
+      "loss": 104.6065,
+      "step": 58180
+    },
+    {
+      "epoch": 0.48136658807957977,
+      "grad_norm": 1349.8760986328125,
+      "learning_rate": 6.3744581022516285e-06,
+      "loss": 82.4153,
+      "step": 58190
+    },
+    {
+      "epoch": 0.481449311328949,
+      "grad_norm": 1214.870361328125,
+      "learning_rate": 6.373102014600033e-06,
+      "loss": 106.5302,
+      "step": 58200
+    },
+    {
+      "epoch": 0.4815320345783182,
+      "grad_norm": 1136.8599853515625,
+      "learning_rate": 6.371745817697012e-06,
+      "loss": 110.1129,
+      "step": 58210
+    },
+    {
+      "epoch": 0.4816147578276875,
+      "grad_norm": 2490.23876953125,
+      "learning_rate": 6.370389511650474e-06,
+      "loss": 138.5235,
+      "step": 58220
+    },
+    {
+      "epoch": 0.48169748107705673,
+      "grad_norm": 851.6533203125,
+      "learning_rate": 6.3690330965683304e-06,
+      "loss": 105.1765,
+      "step": 58230
+    },
+    {
+      "epoch": 0.48178020432642593,
+      "grad_norm": 635.8304443359375,
+      "learning_rate": 6.367676572558506e-06,
+      "loss": 138.1114,
+      "step": 58240
+    },
+    {
+      "epoch": 0.4818629275757952,
+      "grad_norm": 747.5317993164062,
+      "learning_rate": 6.366319939728934e-06,
+      "loss": 86.2269,
+      "step": 58250
+    },
+    {
+      "epoch": 0.48194565082516444,
+      "grad_norm": 1157.3438720703125,
+      "learning_rate": 6.364963198187555e-06,
+      "loss": 81.2648,
+      "step": 58260
+    },
+    {
+      "epoch": 0.48202837407453364,
+      "grad_norm": 961.3770141601562,
+      "learning_rate": 6.363606348042318e-06,
+      "loss": 69.4496,
+      "step": 58270
+    },
+    {
+      "epoch": 0.4821110973239029,
+      "grad_norm": 823.779541015625,
+      "learning_rate": 6.362249389401183e-06,
+      "loss": 96.8622,
+      "step": 58280
+    },
+    {
+      "epoch": 0.4821938205732721,
+      "grad_norm": 510.96942138671875,
+      "learning_rate": 6.360892322372115e-06,
+      "loss": 77.7201,
+      "step": 58290
+    },
+    {
+      "epoch": 0.48227654382264135,
+      "grad_norm": 745.046875,
+      "learning_rate": 6.359535147063092e-06,
+      "loss": 81.9999,
+      "step": 58300
+    },
+    {
+      "epoch": 0.4823592670720106,
+      "grad_norm": 611.5297241210938,
+      "learning_rate": 6.358177863582095e-06,
+      "loss": 87.0968,
+      "step": 58310
+    },
+    {
+      "epoch": 0.4824419903213798,
+      "grad_norm": 860.2018432617188,
+      "learning_rate": 6.35682047203712e-06,
+      "loss": 81.5001,
+      "step": 58320
+    },
+    {
+      "epoch": 0.48252471357074905,
+      "grad_norm": 439.9252014160156,
+      "learning_rate": 6.355462972536166e-06,
+      "loss": 88.4102,
+      "step": 58330
+    },
+    {
+      "epoch": 0.4826074368201183,
+      "grad_norm": 922.5792846679688,
+      "learning_rate": 6.354105365187244e-06,
+      "loss": 91.7119,
+      "step": 58340
+    },
+    {
+      "epoch": 0.4826901600694875,
+      "grad_norm": 1041.380615234375,
+      "learning_rate": 6.352747650098373e-06,
+      "loss": 104.9556,
+      "step": 58350
+    },
+    {
+      "epoch": 0.48277288331885676,
+      "grad_norm": 902.4854125976562,
+      "learning_rate": 6.35138982737758e-06,
+      "loss": 104.9955,
+      "step": 58360
+    },
+    {
+      "epoch": 0.482855606568226,
+      "grad_norm": 856.4240112304688,
+      "learning_rate": 6.3500318971329e-06,
+      "loss": 68.9174,
+      "step": 58370
+    },
+    {
+      "epoch": 0.4829383298175952,
+      "grad_norm": 613.3790893554688,
+      "learning_rate": 6.348673859472378e-06,
+      "loss": 104.4474,
+      "step": 58380
+    },
+    {
+      "epoch": 0.48302105306696447,
+      "grad_norm": 625.4129028320312,
+      "learning_rate": 6.347315714504066e-06,
+      "loss": 104.9294,
+      "step": 58390
+    },
+    {
+      "epoch": 0.4831037763163337,
+      "grad_norm": 597.6600341796875,
+      "learning_rate": 6.345957462336026e-06,
+      "loss": 81.4773,
+      "step": 58400
+    },
+    {
+      "epoch": 0.4831864995657029,
+      "grad_norm": 531.10986328125,
+      "learning_rate": 6.344599103076329e-06,
+      "loss": 100.6717,
+      "step": 58410
+    },
+    {
+      "epoch": 0.4832692228150722,
+      "grad_norm": 812.7492065429688,
+      "learning_rate": 6.343240636833051e-06,
+      "loss": 108.6641,
+      "step": 58420
+    },
+    {
+      "epoch": 0.48335194606444143,
+      "grad_norm": 1158.002685546875,
+      "learning_rate": 6.341882063714282e-06,
+      "loss": 125.5026,
+      "step": 58430
+    },
+    {
+      "epoch": 0.4834346693138106,
+      "grad_norm": 846.1316528320312,
+      "learning_rate": 6.340523383828115e-06,
+      "loss": 107.7689,
+      "step": 58440
+    },
+    {
+      "epoch": 0.4835173925631799,
+      "grad_norm": 1098.09130859375,
+      "learning_rate": 6.339164597282652e-06,
+      "loss": 106.0673,
+      "step": 58450
+    },
+    {
+      "epoch": 0.48360011581254914,
+      "grad_norm": 925.63818359375,
+      "learning_rate": 6.337805704186011e-06,
+      "loss": 118.7983,
+      "step": 58460
+    },
+    {
+      "epoch": 0.48368283906191833,
+      "grad_norm": 558.7349853515625,
+      "learning_rate": 6.336446704646307e-06,
+      "loss": 105.5596,
+      "step": 58470
+    },
+    {
+      "epoch": 0.4837655623112876,
+      "grad_norm": 918.0574340820312,
+      "learning_rate": 6.335087598771676e-06,
+      "loss": 102.35,
+      "step": 58480
+    },
+    {
+      "epoch": 0.48384828556065684,
+      "grad_norm": 2873.980712890625,
+      "learning_rate": 6.333728386670249e-06,
+      "loss": 102.2267,
+      "step": 58490
+    },
+    {
+      "epoch": 0.48393100881002604,
+      "grad_norm": 740.7039794921875,
+      "learning_rate": 6.332369068450175e-06,
+      "loss": 92.0805,
+      "step": 58500
+    },
+    {
+      "epoch": 0.4840137320593953,
+      "grad_norm": 1080.8912353515625,
+      "learning_rate": 6.33100964421961e-06,
+      "loss": 99.742,
+      "step": 58510
+    },
+    {
+      "epoch": 0.48409645530876455,
+      "grad_norm": 1029.510009765625,
+      "learning_rate": 6.329650114086717e-06,
+      "loss": 77.3601,
+      "step": 58520
+    },
+    {
+      "epoch": 0.48417917855813375,
+      "grad_norm": 739.2198486328125,
+      "learning_rate": 6.328290478159666e-06,
+      "loss": 108.9343,
+      "step": 58530
+    },
+    {
+      "epoch": 0.484261901807503,
+      "grad_norm": 584.7896118164062,
+      "learning_rate": 6.326930736546637e-06,
+      "loss": 87.4123,
+      "step": 58540
+    },
+    {
+      "epoch": 0.48434462505687226,
+      "grad_norm": 1001.3221435546875,
+      "learning_rate": 6.325570889355819e-06,
+      "loss": 115.3958,
+      "step": 58550
+    },
+    {
+      "epoch": 0.48442734830624146,
+      "grad_norm": 0.0,
+      "learning_rate": 6.32421093669541e-06,
+      "loss": 100.6869,
+      "step": 58560
+    },
+    {
+      "epoch": 0.4845100715556107,
+      "grad_norm": 1452.001953125,
+      "learning_rate": 6.322850878673614e-06,
+      "loss": 127.1429,
+      "step": 58570
+    },
+    {
+      "epoch": 0.48459279480497996,
+      "grad_norm": 841.3821411132812,
+      "learning_rate": 6.321490715398644e-06,
+      "loss": 94.9818,
+      "step": 58580
+    },
+    {
+      "epoch": 0.48467551805434916,
+      "grad_norm": 813.81298828125,
+      "learning_rate": 6.320130446978722e-06,
+      "loss": 76.4308,
+      "step": 58590
+    },
+    {
+      "epoch": 0.4847582413037184,
+      "grad_norm": 1246.0950927734375,
+      "learning_rate": 6.31877007352208e-06,
+      "loss": 111.9752,
+      "step": 58600
+    },
+    {
+      "epoch": 0.48484096455308767,
+      "grad_norm": 822.40283203125,
+      "learning_rate": 6.317409595136956e-06,
+      "loss": 90.058,
+      "step": 58610
+    },
+    {
+      "epoch": 0.48492368780245687,
+      "grad_norm": 1240.24609375,
+      "learning_rate": 6.316049011931595e-06,
+      "loss": 118.0982,
+      "step": 58620
+    },
+    {
+      "epoch": 0.4850064110518261,
+      "grad_norm": 710.9925537109375,
+      "learning_rate": 6.314688324014255e-06,
+      "loss": 99.1123,
+      "step": 58630
+    },
+    {
+      "epoch": 0.4850891343011953,
+      "grad_norm": 1035.0576171875,
+      "learning_rate": 6.3133275314931995e-06,
+      "loss": 123.4674,
+      "step": 58640
+    },
+    {
+      "epoch": 0.4851718575505646,
+      "grad_norm": 811.2945556640625,
+      "learning_rate": 6.311966634476698e-06,
+      "loss": 93.8845,
+      "step": 58650
+    },
+    {
+      "epoch": 0.48525458079993383,
+      "grad_norm": 568.3858642578125,
+      "learning_rate": 6.3106056330730335e-06,
+      "loss": 96.4744,
+      "step": 58660
+    },
+    {
+      "epoch": 0.48533730404930303,
+      "grad_norm": 682.4791259765625,
+      "learning_rate": 6.309244527390493e-06,
+      "loss": 96.0774,
+      "step": 58670
+    },
+    {
+      "epoch": 0.4854200272986723,
+      "grad_norm": 1391.8082275390625,
+      "learning_rate": 6.307883317537375e-06,
+      "loss": 113.0187,
+      "step": 58680
+    },
+    {
+      "epoch": 0.48550275054804154,
+      "grad_norm": 672.600830078125,
+      "learning_rate": 6.306522003621983e-06,
+      "loss": 79.8365,
+      "step": 58690
+    },
+    {
+      "epoch": 0.48558547379741074,
+      "grad_norm": 1114.2835693359375,
+      "learning_rate": 6.305160585752632e-06,
+      "loss": 116.9864,
+      "step": 58700
+    },
+    {
+      "epoch": 0.48566819704678,
+      "grad_norm": 904.7263793945312,
+      "learning_rate": 6.303799064037643e-06,
+      "loss": 95.5957,
+      "step": 58710
+    },
+    {
+      "epoch": 0.48575092029614925,
+      "grad_norm": 1291.3994140625,
+      "learning_rate": 6.302437438585345e-06,
+      "loss": 107.933,
+      "step": 58720
+    },
+    {
+      "epoch": 0.48583364354551845,
+      "grad_norm": 1033.6456298828125,
+      "learning_rate": 6.301075709504077e-06,
+      "loss": 132.5331,
+      "step": 58730
+    },
+    {
+      "epoch": 0.4859163667948877,
+      "grad_norm": 939.7048950195312,
+      "learning_rate": 6.299713876902188e-06,
+      "loss": 93.3385,
+      "step": 58740
+    },
+    {
+      "epoch": 0.48599909004425695,
+      "grad_norm": 1379.9310302734375,
+      "learning_rate": 6.29835194088803e-06,
+      "loss": 93.6808,
+      "step": 58750
+    },
+    {
+      "epoch": 0.48608181329362615,
+      "grad_norm": 850.0859375,
+      "learning_rate": 6.296989901569966e-06,
+      "loss": 106.2699,
+      "step": 58760
+    },
+    {
+      "epoch": 0.4861645365429954,
+      "grad_norm": 845.896728515625,
+      "learning_rate": 6.295627759056368e-06,
+      "loss": 97.0875,
+      "step": 58770
+    },
+    {
+      "epoch": 0.48624725979236466,
+      "grad_norm": 910.60302734375,
+      "learning_rate": 6.294265513455616e-06,
+      "loss": 107.4305,
+      "step": 58780
+    },
+    {
+      "epoch": 0.48632998304173386,
+      "grad_norm": 895.4052734375,
+      "learning_rate": 6.292903164876097e-06,
+      "loss": 99.333,
+      "step": 58790
+    },
+    {
+      "epoch": 0.4864127062911031,
+      "grad_norm": 443.9016418457031,
+      "learning_rate": 6.291540713426206e-06,
+      "loss": 76.0029,
+      "step": 58800
+    },
+    {
+      "epoch": 0.48649542954047237,
+      "grad_norm": 621.6075439453125,
+      "learning_rate": 6.290178159214349e-06,
+      "loss": 88.676,
+      "step": 58810
+    },
+    {
+      "epoch": 0.48657815278984157,
+      "grad_norm": 459.60284423828125,
+      "learning_rate": 6.288815502348935e-06,
+      "loss": 94.1174,
+      "step": 58820
+    },
+    {
+      "epoch": 0.4866608760392108,
+      "grad_norm": 1254.715576171875,
+      "learning_rate": 6.287452742938388e-06,
+      "loss": 91.2785,
+      "step": 58830
+    },
+    {
+      "epoch": 0.4867435992885801,
+      "grad_norm": 1114.0794677734375,
+      "learning_rate": 6.286089881091134e-06,
+      "loss": 139.8125,
+      "step": 58840
+    },
+    {
+      "epoch": 0.4868263225379493,
+      "grad_norm": 853.3452758789062,
+      "learning_rate": 6.284726916915611e-06,
+      "loss": 101.1512,
+      "step": 58850
+    },
+    {
+      "epoch": 0.48690904578731853,
+      "grad_norm": 741.5625610351562,
+      "learning_rate": 6.2833638505202635e-06,
+      "loss": 115.4677,
+      "step": 58860
+    },
+    {
+      "epoch": 0.4869917690366878,
+      "grad_norm": 1342.6466064453125,
+      "learning_rate": 6.282000682013545e-06,
+      "loss": 93.9246,
+      "step": 58870
+    },
+    {
+      "epoch": 0.487074492286057,
+      "grad_norm": 594.8483276367188,
+      "learning_rate": 6.280637411503913e-06,
+      "loss": 108.2251,
+      "step": 58880
+    },
+    {
+      "epoch": 0.48715721553542624,
+      "grad_norm": 899.0263061523438,
+      "learning_rate": 6.279274039099842e-06,
+      "loss": 93.0218,
+      "step": 58890
+    },
+    {
+      "epoch": 0.4872399387847955,
+      "grad_norm": 858.6878662109375,
+      "learning_rate": 6.277910564909806e-06,
+      "loss": 91.2543,
+      "step": 58900
+    },
+    {
+      "epoch": 0.4873226620341647,
+      "grad_norm": 974.9313354492188,
+      "learning_rate": 6.276546989042292e-06,
+      "loss": 91.2285,
+      "step": 58910
+    },
+    {
+      "epoch": 0.48740538528353394,
+      "grad_norm": 653.6936645507812,
+      "learning_rate": 6.275183311605793e-06,
+      "loss": 101.074,
+      "step": 58920
+    },
+    {
+      "epoch": 0.4874881085329032,
+      "grad_norm": 816.3530883789062,
+      "learning_rate": 6.273819532708807e-06,
+      "loss": 94.144,
+      "step": 58930
+    },
+    {
+      "epoch": 0.4875708317822724,
+      "grad_norm": 1098.2479248046875,
+      "learning_rate": 6.27245565245985e-06,
+      "loss": 104.6022,
+      "step": 58940
+    },
+    {
+      "epoch": 0.48765355503164165,
+      "grad_norm": 1253.642333984375,
+      "learning_rate": 6.271091670967437e-06,
+      "loss": 101.674,
+      "step": 58950
+    },
+    {
+      "epoch": 0.4877362782810109,
+      "grad_norm": 577.8956909179688,
+      "learning_rate": 6.269727588340091e-06,
+      "loss": 82.5646,
+      "step": 58960
+    },
+    {
+      "epoch": 0.4878190015303801,
+      "grad_norm": 667.8779296875,
+      "learning_rate": 6.268363404686348e-06,
+      "loss": 116.7945,
+      "step": 58970
+    },
+    {
+      "epoch": 0.48790172477974936,
+      "grad_norm": 716.1468505859375,
+      "learning_rate": 6.26699912011475e-06,
+      "loss": 118.8406,
+      "step": 58980
+    },
+    {
+      "epoch": 0.4879844480291186,
+      "grad_norm": 606.7252197265625,
+      "learning_rate": 6.265634734733848e-06,
+      "loss": 74.8637,
+      "step": 58990
+    },
+    {
+      "epoch": 0.4880671712784878,
+      "grad_norm": 1299.558349609375,
+      "learning_rate": 6.264270248652199e-06,
+      "loss": 110.4696,
+      "step": 59000
+    },
+    {
+      "epoch": 0.48814989452785706,
+      "grad_norm": 913.1611938476562,
+      "learning_rate": 6.262905661978367e-06,
+      "loss": 79.307,
+      "step": 59010
+    },
+    {
+      "epoch": 0.48823261777722626,
+      "grad_norm": 590.417724609375,
+      "learning_rate": 6.261540974820928e-06,
+      "loss": 99.6395,
+      "step": 59020
+    },
+    {
+      "epoch": 0.4883153410265955,
+      "grad_norm": 696.5634155273438,
+      "learning_rate": 6.260176187288463e-06,
+      "loss": 100.7072,
+      "step": 59030
+    },
+    {
+      "epoch": 0.48839806427596477,
+      "grad_norm": 1761.085693359375,
+      "learning_rate": 6.2588112994895636e-06,
+      "loss": 151.4153,
+      "step": 59040
+    },
+    {
+      "epoch": 0.48848078752533397,
+      "grad_norm": 1071.5557861328125,
+      "learning_rate": 6.257446311532824e-06,
+      "loss": 69.3153,
+      "step": 59050
+    },
+    {
+      "epoch": 0.4885635107747032,
+      "grad_norm": 1153.7186279296875,
+      "learning_rate": 6.256081223526854e-06,
+      "loss": 93.4576,
+      "step": 59060
+    },
+    {
+      "epoch": 0.4886462340240725,
+      "grad_norm": 1251.8411865234375,
+      "learning_rate": 6.254716035580264e-06,
+      "loss": 71.2112,
+      "step": 59070
+    },
+    {
+      "epoch": 0.4887289572734417,
+      "grad_norm": 653.9671630859375,
+      "learning_rate": 6.25335074780168e-06,
+      "loss": 101.7776,
+      "step": 59080
+    },
+    {
+      "epoch": 0.48881168052281093,
+      "grad_norm": 917.6962280273438,
+      "learning_rate": 6.251985360299728e-06,
+      "loss": 109.7955,
+      "step": 59090
+    },
+    {
+      "epoch": 0.4888944037721802,
+      "grad_norm": 1049.5726318359375,
+      "learning_rate": 6.250619873183046e-06,
+      "loss": 97.9137,
+      "step": 59100
+    },
+    {
+      "epoch": 0.4889771270215494,
+      "grad_norm": 960.5833129882812,
+      "learning_rate": 6.249254286560281e-06,
+      "loss": 115.5338,
+      "step": 59110
+    },
+    {
+      "epoch": 0.48905985027091864,
+      "grad_norm": 1100.854248046875,
+      "learning_rate": 6.247888600540084e-06,
+      "loss": 102.4374,
+      "step": 59120
+    },
+    {
+      "epoch": 0.4891425735202879,
+      "grad_norm": 858.7109985351562,
+      "learning_rate": 6.246522815231121e-06,
+      "loss": 86.8003,
+      "step": 59130
+    },
+    {
+      "epoch": 0.4892252967696571,
+      "grad_norm": 758.7923583984375,
+      "learning_rate": 6.245156930742057e-06,
+      "loss": 99.9746,
+      "step": 59140
+    },
+    {
+      "epoch": 0.48930802001902635,
+      "grad_norm": 957.5679931640625,
+      "learning_rate": 6.24379094718157e-06,
+      "loss": 105.2421,
+      "step": 59150
+    },
+    {
+      "epoch": 0.4893907432683956,
+      "grad_norm": 932.5010986328125,
+      "learning_rate": 6.2424248646583455e-06,
+      "loss": 109.755,
+      "step": 59160
+    },
+    {
+      "epoch": 0.4894734665177648,
+      "grad_norm": 1039.6802978515625,
+      "learning_rate": 6.241058683281077e-06,
+      "loss": 157.5874,
+      "step": 59170
+    },
+    {
+      "epoch": 0.48955618976713405,
+      "grad_norm": 1249.9261474609375,
+      "learning_rate": 6.239692403158465e-06,
+      "loss": 80.0913,
+      "step": 59180
+    },
+    {
+      "epoch": 0.4896389130165033,
+      "grad_norm": 769.3261108398438,
+      "learning_rate": 6.238326024399217e-06,
+      "loss": 116.0565,
+      "step": 59190
+    },
+    {
+      "epoch": 0.4897216362658725,
+      "grad_norm": 3054.625,
+      "learning_rate": 6.236959547112051e-06,
+      "loss": 135.2796,
+      "step": 59200
+    },
+    {
+      "epoch": 0.48980435951524176,
+      "grad_norm": 1218.1517333984375,
+      "learning_rate": 6.235592971405691e-06,
+      "loss": 96.0189,
+      "step": 59210
+    },
+    {
+      "epoch": 0.489887082764611,
+      "grad_norm": 964.074462890625,
+      "learning_rate": 6.234226297388869e-06,
+      "loss": 135.8784,
+      "step": 59220
+    },
+    {
+      "epoch": 0.4899698060139802,
+      "grad_norm": 822.4147338867188,
+      "learning_rate": 6.232859525170324e-06,
+      "loss": 74.8447,
+      "step": 59230
+    },
+    {
+      "epoch": 0.49005252926334947,
+      "grad_norm": 1401.8765869140625,
+      "learning_rate": 6.231492654858805e-06,
+      "loss": 124.0554,
+      "step": 59240
+    },
+    {
+      "epoch": 0.4901352525127187,
+      "grad_norm": 665.4353637695312,
+      "learning_rate": 6.230125686563068e-06,
+      "loss": 116.1569,
+      "step": 59250
+    },
+    {
+      "epoch": 0.4902179757620879,
+      "grad_norm": 1295.380126953125,
+      "learning_rate": 6.2287586203918745e-06,
+      "loss": 114.117,
+      "step": 59260
+    },
+    {
+      "epoch": 0.4903006990114572,
+      "grad_norm": 1130.935791015625,
+      "learning_rate": 6.227391456453997e-06,
+      "loss": 73.4887,
+      "step": 59270
+    },
+    {
+      "epoch": 0.49038342226082643,
+      "grad_norm": 615.1046142578125,
+      "learning_rate": 6.226024194858214e-06,
+      "loss": 83.7547,
+      "step": 59280
+    },
+    {
+      "epoch": 0.49046614551019563,
+      "grad_norm": 909.8365478515625,
+      "learning_rate": 6.224656835713313e-06,
+      "loss": 99.4262,
+      "step": 59290
+    },
+    {
+      "epoch": 0.4905488687595649,
+      "grad_norm": 768.0503540039062,
+      "learning_rate": 6.223289379128088e-06,
+      "loss": 88.5417,
+      "step": 59300
+    },
+    {
+      "epoch": 0.49063159200893414,
+      "grad_norm": 1546.40771484375,
+      "learning_rate": 6.221921825211342e-06,
+      "loss": 88.7849,
+      "step": 59310
+    },
+    {
+      "epoch": 0.49071431525830334,
+      "grad_norm": 1119.7420654296875,
+      "learning_rate": 6.220554174071884e-06,
+      "loss": 216.5368,
+      "step": 59320
+    },
+    {
+      "epoch": 0.4907970385076726,
+      "grad_norm": 738.0167846679688,
+      "learning_rate": 6.219186425818531e-06,
+      "loss": 111.826,
+      "step": 59330
+    },
+    {
+      "epoch": 0.49087976175704184,
+      "grad_norm": 1079.2379150390625,
+      "learning_rate": 6.217818580560111e-06,
+      "loss": 138.3639,
+      "step": 59340
+    },
+    {
+      "epoch": 0.49096248500641104,
+      "grad_norm": 455.43560791015625,
+      "learning_rate": 6.216450638405454e-06,
+      "loss": 106.5501,
+      "step": 59350
+    },
+    {
+      "epoch": 0.4910452082557803,
+      "grad_norm": 1222.3670654296875,
+      "learning_rate": 6.2150825994634025e-06,
+      "loss": 76.1361,
+      "step": 59360
+    },
+    {
+      "epoch": 0.4911279315051495,
+      "grad_norm": 2011.9967041015625,
+      "learning_rate": 6.2137144638428045e-06,
+      "loss": 113.7866,
+      "step": 59370
+    },
+    {
+      "epoch": 0.49121065475451875,
+      "grad_norm": 1221.827880859375,
+      "learning_rate": 6.21234623165252e-06,
+      "loss": 79.5845,
+      "step": 59380
+    },
+    {
+      "epoch": 0.491293378003888,
+      "grad_norm": 780.6744384765625,
+      "learning_rate": 6.210977903001406e-06,
+      "loss": 88.4106,
+      "step": 59390
+    },
+    {
+      "epoch": 0.4913761012532572,
+      "grad_norm": 755.7965087890625,
+      "learning_rate": 6.209609477998339e-06,
+      "loss": 97.9078,
+      "step": 59400
+    },
+    {
+      "epoch": 0.49145882450262646,
+      "grad_norm": 883.141357421875,
+      "learning_rate": 6.2082409567521975e-06,
+      "loss": 97.257,
+      "step": 59410
+    },
+    {
+      "epoch": 0.4915415477519957,
+      "grad_norm": 1329.85107421875,
+      "learning_rate": 6.206872339371867e-06,
+      "loss": 101.6907,
+      "step": 59420
+    },
+    {
+      "epoch": 0.4916242710013649,
+      "grad_norm": 2107.642578125,
+      "learning_rate": 6.205503625966247e-06,
+      "loss": 109.3981,
+      "step": 59430
+    },
+    {
+      "epoch": 0.49170699425073416,
+      "grad_norm": 663.9820556640625,
+      "learning_rate": 6.204134816644233e-06,
+      "loss": 77.1192,
+      "step": 59440
+    },
+    {
+      "epoch": 0.4917897175001034,
+      "grad_norm": 1082.100341796875,
+      "learning_rate": 6.2027659115147375e-06,
+      "loss": 109.6853,
+      "step": 59450
+    },
+    {
+      "epoch": 0.4918724407494726,
+      "grad_norm": 734.315185546875,
+      "learning_rate": 6.201396910686679e-06,
+      "loss": 92.7889,
+      "step": 59460
+    },
+    {
+      "epoch": 0.49195516399884187,
+      "grad_norm": 825.7645874023438,
+      "learning_rate": 6.200027814268984e-06,
+      "loss": 82.5341,
+      "step": 59470
+    },
+    {
+      "epoch": 0.4920378872482111,
+      "grad_norm": 990.0794677734375,
+      "learning_rate": 6.198658622370582e-06,
+      "loss": 91.5214,
+      "step": 59480
+    },
+    {
+      "epoch": 0.4921206104975803,
+      "grad_norm": 672.1451416015625,
+      "learning_rate": 6.197289335100412e-06,
+      "loss": 99.9061,
+      "step": 59490
+    },
+    {
+      "epoch": 0.4922033337469496,
+      "grad_norm": 663.6240234375,
+      "learning_rate": 6.195919952567426e-06,
+      "loss": 84.1417,
+      "step": 59500
+    },
+    {
+      "epoch": 0.49228605699631883,
+      "grad_norm": 1181.8443603515625,
+      "learning_rate": 6.194550474880579e-06,
+      "loss": 101.2219,
+      "step": 59510
+    },
+    {
+      "epoch": 0.49236878024568803,
+      "grad_norm": 1052.6051025390625,
+      "learning_rate": 6.193180902148833e-06,
+      "loss": 101.8608,
+      "step": 59520
+    },
+    {
+      "epoch": 0.4924515034950573,
+      "grad_norm": 467.2412109375,
+      "learning_rate": 6.1918112344811575e-06,
+      "loss": 64.4284,
+      "step": 59530
+    },
+    {
+      "epoch": 0.49253422674442654,
+      "grad_norm": 756.0142822265625,
+      "learning_rate": 6.190441471986533e-06,
+      "loss": 93.9991,
+      "step": 59540
+    },
+    {
+      "epoch": 0.49261694999379574,
+      "grad_norm": 650.9651489257812,
+      "learning_rate": 6.18907161477394e-06,
+      "loss": 134.6619,
+      "step": 59550
+    },
+    {
+      "epoch": 0.492699673243165,
+      "grad_norm": 845.457275390625,
+      "learning_rate": 6.187701662952381e-06,
+      "loss": 71.9417,
+      "step": 59560
+    },
+    {
+      "epoch": 0.49278239649253425,
+      "grad_norm": 1039.0081787109375,
+      "learning_rate": 6.18633161663085e-06,
+      "loss": 96.217,
+      "step": 59570
+    },
+    {
+      "epoch": 0.49286511974190345,
+      "grad_norm": 1419.552734375,
+      "learning_rate": 6.184961475918355e-06,
+      "loss": 117.6555,
+      "step": 59580
+    },
+    {
+      "epoch": 0.4929478429912727,
+      "grad_norm": 1182.2574462890625,
+      "learning_rate": 6.183591240923914e-06,
+      "loss": 116.7017,
+      "step": 59590
+    },
+    {
+      "epoch": 0.49303056624064195,
+      "grad_norm": 1182.273681640625,
+      "learning_rate": 6.182220911756551e-06,
+      "loss": 96.5914,
+      "step": 59600
+    },
+    {
+      "epoch": 0.49311328949001115,
+      "grad_norm": 1710.057861328125,
+      "learning_rate": 6.1808504885252955e-06,
+      "loss": 116.0724,
+      "step": 59610
+    },
+    {
+      "epoch": 0.4931960127393804,
+      "grad_norm": 738.5501708984375,
+      "learning_rate": 6.179479971339186e-06,
+      "loss": 102.2499,
+      "step": 59620
+    },
+    {
+      "epoch": 0.49327873598874966,
+      "grad_norm": 1123.935791015625,
+      "learning_rate": 6.178109360307267e-06,
+      "loss": 98.9781,
+      "step": 59630
+    },
+    {
+      "epoch": 0.49336145923811886,
+      "grad_norm": 1302.947998046875,
+      "learning_rate": 6.176738655538594e-06,
+      "loss": 86.7837,
+      "step": 59640
+    },
+    {
+      "epoch": 0.4934441824874881,
+      "grad_norm": 1149.8046875,
+      "learning_rate": 6.175367857142227e-06,
+      "loss": 101.772,
+      "step": 59650
+    },
+    {
+      "epoch": 0.49352690573685737,
+      "grad_norm": 313.6778564453125,
+      "learning_rate": 6.173996965227234e-06,
+      "loss": 67.0711,
+      "step": 59660
+    },
+    {
+      "epoch": 0.49360962898622657,
+      "grad_norm": 992.2843627929688,
+      "learning_rate": 6.17262597990269e-06,
+      "loss": 122.4615,
+      "step": 59670
+    },
+    {
+      "epoch": 0.4936923522355958,
+      "grad_norm": 544.5889282226562,
+      "learning_rate": 6.171254901277678e-06,
+      "loss": 118.6765,
+      "step": 59680
+    },
+    {
+      "epoch": 0.4937750754849651,
+      "grad_norm": 760.4612426757812,
+      "learning_rate": 6.169883729461289e-06,
+      "loss": 73.9667,
+      "step": 59690
+    },
+    {
+      "epoch": 0.4938577987343343,
+      "grad_norm": 879.9442749023438,
+      "learning_rate": 6.16851246456262e-06,
+      "loss": 121.3691,
+      "step": 59700
+    },
+    {
+      "epoch": 0.49394052198370353,
+      "grad_norm": 838.228271484375,
+      "learning_rate": 6.167141106690778e-06,
+      "loss": 81.8349,
+      "step": 59710
+    },
+    {
+      "epoch": 0.4940232452330727,
+      "grad_norm": 600.8211059570312,
+      "learning_rate": 6.1657696559548755e-06,
+      "loss": 82.6974,
+      "step": 59720
+    },
+    {
+      "epoch": 0.494105968482442,
+      "grad_norm": 1331.113525390625,
+      "learning_rate": 6.16439811246403e-06,
+      "loss": 91.3955,
+      "step": 59730
+    },
+    {
+      "epoch": 0.49418869173181124,
+      "grad_norm": 1298.560791015625,
+      "learning_rate": 6.163026476327371e-06,
+      "loss": 127.3068,
+      "step": 59740
+    },
+    {
+      "epoch": 0.49427141498118043,
+      "grad_norm": 1139.0467529296875,
+      "learning_rate": 6.161654747654033e-06,
+      "loss": 103.0033,
+      "step": 59750
+    },
+    {
+      "epoch": 0.4943541382305497,
+      "grad_norm": 740.8123779296875,
+      "learning_rate": 6.1602829265531585e-06,
+      "loss": 111.8351,
+      "step": 59760
+    },
+    {
+      "epoch": 0.49443686147991894,
+      "grad_norm": 811.9702758789062,
+      "learning_rate": 6.158911013133896e-06,
+      "loss": 106.3484,
+      "step": 59770
+    },
+    {
+      "epoch": 0.49451958472928814,
+      "grad_norm": 751.2868041992188,
+      "learning_rate": 6.157539007505402e-06,
+      "loss": 81.8818,
+      "step": 59780
+    },
+    {
+      "epoch": 0.4946023079786574,
+      "grad_norm": 596.0902099609375,
+      "learning_rate": 6.156166909776842e-06,
+      "loss": 141.5883,
+      "step": 59790
+    },
+    {
+      "epoch": 0.49468503122802665,
+      "grad_norm": 669.7060546875,
+      "learning_rate": 6.154794720057388e-06,
+      "loss": 92.8498,
+      "step": 59800
+    },
+    {
+      "epoch": 0.49476775447739585,
+      "grad_norm": 769.1928100585938,
+      "learning_rate": 6.153422438456218e-06,
+      "loss": 111.0396,
+      "step": 59810
+    },
+    {
+      "epoch": 0.4948504777267651,
+      "grad_norm": 1065.379638671875,
+      "learning_rate": 6.1520500650825175e-06,
+      "loss": 109.4096,
+      "step": 59820
+    },
+    {
+      "epoch": 0.49493320097613436,
+      "grad_norm": 1148.633056640625,
+      "learning_rate": 6.150677600045479e-06,
+      "loss": 107.9428,
+      "step": 59830
+    },
+    {
+      "epoch": 0.49501592422550356,
+      "grad_norm": 530.499267578125,
+      "learning_rate": 6.1493050434543065e-06,
+      "loss": 80.3538,
+      "step": 59840
+    },
+    {
+      "epoch": 0.4950986474748728,
+      "grad_norm": 888.7738647460938,
+      "learning_rate": 6.1479323954182055e-06,
+      "loss": 82.7131,
+      "step": 59850
+    },
+    {
+      "epoch": 0.49518137072424206,
+      "grad_norm": 1272.56884765625,
+      "learning_rate": 6.146559656046394e-06,
+      "loss": 115.8495,
+      "step": 59860
+    },
+    {
+      "epoch": 0.49526409397361126,
+      "grad_norm": 1251.605712890625,
+      "learning_rate": 6.1451868254480914e-06,
+      "loss": 103.9669,
+      "step": 59870
+    },
+    {
+      "epoch": 0.4953468172229805,
+      "grad_norm": 1469.060546875,
+      "learning_rate": 6.143813903732527e-06,
+      "loss": 98.5221,
+      "step": 59880
+    },
+    {
+      "epoch": 0.49542954047234977,
+      "grad_norm": 613.4310913085938,
+      "learning_rate": 6.142440891008941e-06,
+      "loss": 95.7254,
+      "step": 59890
+    },
+    {
+      "epoch": 0.49551226372171897,
+      "grad_norm": 1487.4815673828125,
+      "learning_rate": 6.141067787386579e-06,
+      "loss": 134.6069,
+      "step": 59900
+    },
+    {
+      "epoch": 0.4955949869710882,
+      "grad_norm": 1083.0224609375,
+      "learning_rate": 6.139694592974687e-06,
+      "loss": 103.1761,
+      "step": 59910
+    },
+    {
+      "epoch": 0.4956777102204575,
+      "grad_norm": 1133.168701171875,
+      "learning_rate": 6.1383213078825275e-06,
+      "loss": 111.6922,
+      "step": 59920
+    },
+    {
+      "epoch": 0.4957604334698267,
+      "grad_norm": 1490.791015625,
+      "learning_rate": 6.136947932219365e-06,
+      "loss": 102.0797,
+      "step": 59930
+    },
+    {
+      "epoch": 0.49584315671919593,
+      "grad_norm": 717.7562866210938,
+      "learning_rate": 6.135574466094475e-06,
+      "loss": 103.1788,
+      "step": 59940
+    },
+    {
+      "epoch": 0.4959258799685652,
+      "grad_norm": 707.7864379882812,
+      "learning_rate": 6.134200909617135e-06,
+      "loss": 84.5769,
+      "step": 59950
+    },
+    {
+      "epoch": 0.4960086032179344,
+      "grad_norm": 607.8535766601562,
+      "learning_rate": 6.132827262896634e-06,
+      "loss": 109.7207,
+      "step": 59960
+    },
+    {
+      "epoch": 0.49609132646730364,
+      "grad_norm": 575.4254150390625,
+      "learning_rate": 6.131453526042267e-06,
+      "loss": 94.0905,
+      "step": 59970
+    },
+    {
+      "epoch": 0.4961740497166729,
+      "grad_norm": 1168.943603515625,
+      "learning_rate": 6.130079699163335e-06,
+      "loss": 123.4409,
+      "step": 59980
+    },
+    {
+      "epoch": 0.4962567729660421,
+      "grad_norm": 806.4592895507812,
+      "learning_rate": 6.128705782369149e-06,
+      "loss": 126.4985,
+      "step": 59990
+    },
+    {
+      "epoch": 0.49633949621541135,
+      "grad_norm": 994.6439208984375,
+      "learning_rate": 6.127331775769023e-06,
+      "loss": 81.3202,
+      "step": 60000
+    },
+    {
+      "epoch": 0.4964222194647806,
+      "grad_norm": 764.6784057617188,
+      "learning_rate": 6.125957679472282e-06,
+      "loss": 98.2079,
+      "step": 60010
+    },
+    {
+      "epoch": 0.4965049427141498,
+      "grad_norm": 963.4498901367188,
+      "learning_rate": 6.124583493588254e-06,
+      "loss": 107.3761,
+      "step": 60020
+    },
+    {
+      "epoch": 0.49658766596351905,
+      "grad_norm": 763.531494140625,
+      "learning_rate": 6.123209218226282e-06,
+      "loss": 86.249,
+      "step": 60030
+    },
+    {
+      "epoch": 0.4966703892128883,
+      "grad_norm": 1130.901611328125,
+      "learning_rate": 6.121834853495704e-06,
+      "loss": 80.7327,
+      "step": 60040
+    },
+    {
+      "epoch": 0.4967531124622575,
+      "grad_norm": 830.3397827148438,
+      "learning_rate": 6.120460399505876e-06,
+      "loss": 79.5612,
+      "step": 60050
+    },
+    {
+      "epoch": 0.49683583571162676,
+      "grad_norm": 1038.9749755859375,
+      "learning_rate": 6.119085856366158e-06,
+      "loss": 104.2222,
+      "step": 60060
+    },
+    {
+      "epoch": 0.496918558960996,
+      "grad_norm": 1272.2034912109375,
+      "learning_rate": 6.117711224185913e-06,
+      "loss": 95.4032,
+      "step": 60070
+    },
+    {
+      "epoch": 0.4970012822103652,
+      "grad_norm": 678.36083984375,
+      "learning_rate": 6.116336503074516e-06,
+      "loss": 111.6325,
+      "step": 60080
+    },
+    {
+      "epoch": 0.49708400545973447,
+      "grad_norm": 902.4628295898438,
+      "learning_rate": 6.114961693141346e-06,
+      "loss": 109.0453,
+      "step": 60090
+    },
+    {
+      "epoch": 0.49716672870910367,
+      "grad_norm": 1161.1533203125,
+      "learning_rate": 6.113586794495792e-06,
+      "loss": 102.0886,
+      "step": 60100
+    },
+    {
+      "epoch": 0.4972494519584729,
+      "grad_norm": 883.01025390625,
+      "learning_rate": 6.112211807247246e-06,
+      "loss": 79.8961,
+      "step": 60110
+    },
+    {
+      "epoch": 0.4973321752078422,
+      "grad_norm": 560.7022094726562,
+      "learning_rate": 6.110836731505112e-06,
+      "loss": 104.4858,
+      "step": 60120
+    },
+    {
+      "epoch": 0.4974148984572114,
+      "grad_norm": 563.2030639648438,
+      "learning_rate": 6.109461567378796e-06,
+      "loss": 76.5894,
+      "step": 60130
+    },
+    {
+      "epoch": 0.49749762170658063,
+      "grad_norm": 567.1194458007812,
+      "learning_rate": 6.108086314977717e-06,
+      "loss": 74.5852,
+      "step": 60140
+    },
+    {
+      "epoch": 0.4975803449559499,
+      "grad_norm": 1001.0436401367188,
+      "learning_rate": 6.106710974411294e-06,
+      "loss": 89.3131,
+      "step": 60150
+    },
+    {
+      "epoch": 0.4976630682053191,
+      "grad_norm": 1134.8504638671875,
+      "learning_rate": 6.105335545788957e-06,
+      "loss": 92.9293,
+      "step": 60160
+    },
+    {
+      "epoch": 0.49774579145468834,
+      "grad_norm": 1836.4036865234375,
+      "learning_rate": 6.103960029220145e-06,
+      "loss": 96.6835,
+      "step": 60170
+    },
+    {
+      "epoch": 0.4978285147040576,
+      "grad_norm": 884.6735229492188,
+      "learning_rate": 6.102584424814299e-06,
+      "loss": 96.8814,
+      "step": 60180
+    },
+    {
+      "epoch": 0.4979112379534268,
+      "grad_norm": 1286.9527587890625,
+      "learning_rate": 6.101208732680872e-06,
+      "loss": 84.8719,
+      "step": 60190
+    },
+    {
+      "epoch": 0.49799396120279604,
+      "grad_norm": 1096.192138671875,
+      "learning_rate": 6.09983295292932e-06,
+      "loss": 133.576,
+      "step": 60200
+    },
+    {
+      "epoch": 0.4980766844521653,
+      "grad_norm": 396.5540771484375,
+      "learning_rate": 6.0984570856691046e-06,
+      "loss": 71.0101,
+      "step": 60210
+    },
+    {
+      "epoch": 0.4981594077015345,
+      "grad_norm": 1324.0350341796875,
+      "learning_rate": 6.097081131009703e-06,
+      "loss": 84.5404,
+      "step": 60220
+    },
+    {
+      "epoch": 0.49824213095090375,
+      "grad_norm": 1569.8370361328125,
+      "learning_rate": 6.095705089060589e-06,
+      "loss": 131.2618,
+      "step": 60230
+    },
+    {
+      "epoch": 0.498324854200273,
+      "grad_norm": 452.867919921875,
+      "learning_rate": 6.094328959931252e-06,
+      "loss": 98.2553,
+      "step": 60240
+    },
+    {
+      "epoch": 0.4984075774496422,
+      "grad_norm": 754.5675659179688,
+      "learning_rate": 6.092952743731179e-06,
+      "loss": 109.5702,
+      "step": 60250
+    },
+    {
+      "epoch": 0.49849030069901146,
+      "grad_norm": 1038.2977294921875,
+      "learning_rate": 6.091576440569873e-06,
+      "loss": 102.3048,
+      "step": 60260
+    },
+    {
+      "epoch": 0.4985730239483807,
+      "grad_norm": 1505.541748046875,
+      "learning_rate": 6.09020005055684e-06,
+      "loss": 93.0384,
+      "step": 60270
+    },
+    {
+      "epoch": 0.4986557471977499,
+      "grad_norm": 714.1956787109375,
+      "learning_rate": 6.088823573801592e-06,
+      "loss": 89.4964,
+      "step": 60280
+    },
+    {
+      "epoch": 0.49873847044711916,
+      "grad_norm": 1760.59033203125,
+      "learning_rate": 6.087447010413651e-06,
+      "loss": 123.4378,
+      "step": 60290
+    },
+    {
+      "epoch": 0.4988211936964884,
+      "grad_norm": 647.3226318359375,
+      "learning_rate": 6.08607036050254e-06,
+      "loss": 97.7962,
+      "step": 60300
+    },
+    {
+      "epoch": 0.4989039169458576,
+      "grad_norm": 670.925048828125,
+      "learning_rate": 6.084693624177794e-06,
+      "loss": 90.9299,
+      "step": 60310
+    },
+    {
+      "epoch": 0.49898664019522687,
+      "grad_norm": 861.1676635742188,
+      "learning_rate": 6.083316801548956e-06,
+      "loss": 96.199,
+      "step": 60320
+    },
+    {
+      "epoch": 0.4990693634445961,
+      "grad_norm": 844.1892700195312,
+      "learning_rate": 6.081939892725572e-06,
+      "loss": 89.0784,
+      "step": 60330
+    },
+    {
+      "epoch": 0.4991520866939653,
+      "grad_norm": 1057.48876953125,
+      "learning_rate": 6.080562897817196e-06,
+      "loss": 80.8512,
+      "step": 60340
+    },
+    {
+      "epoch": 0.4992348099433346,
+      "grad_norm": 547.9786987304688,
+      "learning_rate": 6.079185816933388e-06,
+      "loss": 123.7075,
+      "step": 60350
+    },
+    {
+      "epoch": 0.49931753319270383,
+      "grad_norm": 813.2322998046875,
+      "learning_rate": 6.077808650183718e-06,
+      "loss": 75.0818,
+      "step": 60360
+    },
+    {
+      "epoch": 0.49940025644207303,
+      "grad_norm": 1088.018310546875,
+      "learning_rate": 6.076431397677762e-06,
+      "loss": 99.3224,
+      "step": 60370
+    },
+    {
+      "epoch": 0.4994829796914423,
+      "grad_norm": 443.6618347167969,
+      "learning_rate": 6.0750540595250986e-06,
+      "loss": 92.7872,
+      "step": 60380
+    },
+    {
+      "epoch": 0.49956570294081154,
+      "grad_norm": 1097.77099609375,
+      "learning_rate": 6.073676635835317e-06,
+      "loss": 101.5357,
+      "step": 60390
+    },
+    {
+      "epoch": 0.49964842619018074,
+      "grad_norm": 809.2193603515625,
+      "learning_rate": 6.072299126718012e-06,
+      "loss": 102.8223,
+      "step": 60400
+    },
+    {
+      "epoch": 0.49973114943955,
+      "grad_norm": 1209.04296875,
+      "learning_rate": 6.070921532282788e-06,
+      "loss": 90.7184,
+      "step": 60410
+    },
+    {
+      "epoch": 0.49981387268891925,
+      "grad_norm": 483.89910888671875,
+      "learning_rate": 6.0695438526392536e-06,
+      "loss": 90.6916,
+      "step": 60420
+    },
+    {
+      "epoch": 0.49989659593828845,
+      "grad_norm": 939.6087036132812,
+      "learning_rate": 6.068166087897022e-06,
+      "loss": 95.3435,
+      "step": 60430
+    },
+    {
+      "epoch": 0.4999793191876577,
+      "grad_norm": 1028.2034912109375,
+      "learning_rate": 6.066788238165717e-06,
+      "loss": 122.2318,
+      "step": 60440
+    },
+    {
+      "epoch": 0.5000620424370269,
+      "grad_norm": 871.5949096679688,
+      "learning_rate": 6.0654103035549686e-06,
+      "loss": 107.04,
+      "step": 60450
+    },
+    {
+      "epoch": 0.5001447656863962,
+      "grad_norm": 631.7724609375,
+      "learning_rate": 6.064032284174411e-06,
+      "loss": 107.0457,
+      "step": 60460
+    },
+    {
+      "epoch": 0.5002274889357654,
+      "grad_norm": 1057.8795166015625,
+      "learning_rate": 6.062654180133689e-06,
+      "loss": 95.4936,
+      "step": 60470
+    },
+    {
+      "epoch": 0.5003102121851346,
+      "grad_norm": 625.803466796875,
+      "learning_rate": 6.06127599154245e-06,
+      "loss": 112.727,
+      "step": 60480
+    },
+    {
+      "epoch": 0.5003929354345039,
+      "grad_norm": 957.1911010742188,
+      "learning_rate": 6.059897718510351e-06,
+      "loss": 141.4194,
+      "step": 60490
+    },
+    {
+      "epoch": 0.5004756586838731,
+      "grad_norm": 1082.177734375,
+      "learning_rate": 6.058519361147055e-06,
+      "loss": 106.0662,
+      "step": 60500
+    },
+    {
+      "epoch": 0.5005583819332423,
+      "grad_norm": 961.278076171875,
+      "learning_rate": 6.057140919562231e-06,
+      "loss": 92.1514,
+      "step": 60510
+    },
+    {
+      "epoch": 0.5006411051826116,
+      "grad_norm": 1194.2098388671875,
+      "learning_rate": 6.055762393865555e-06,
+      "loss": 114.0582,
+      "step": 60520
+    },
+    {
+      "epoch": 0.5007238284319808,
+      "grad_norm": 2295.308349609375,
+      "learning_rate": 6.054383784166712e-06,
+      "loss": 122.28,
+      "step": 60530
+    },
+    {
+      "epoch": 0.50080655168135,
+      "grad_norm": 950.94482421875,
+      "learning_rate": 6.05300509057539e-06,
+      "loss": 65.6176,
+      "step": 60540
+    },
+    {
+      "epoch": 0.5008892749307193,
+      "grad_norm": 575.2659301757812,
+      "learning_rate": 6.051626313201285e-06,
+      "loss": 75.2322,
+      "step": 60550
+    },
+    {
+      "epoch": 0.5009719981800885,
+      "grad_norm": 955.7214965820312,
+      "learning_rate": 6.0502474521541014e-06,
+      "loss": 78.4364,
+      "step": 60560
+    },
+    {
+      "epoch": 0.5010547214294577,
+      "grad_norm": 1058.46240234375,
+      "learning_rate": 6.048868507543547e-06,
+      "loss": 76.487,
+      "step": 60570
+    },
+    {
+      "epoch": 0.501137444678827,
+      "grad_norm": 1243.02880859375,
+      "learning_rate": 6.047489479479339e-06,
+      "loss": 94.1711,
+      "step": 60580
+    },
+    {
+      "epoch": 0.5012201679281962,
+      "grad_norm": 539.5478515625,
+      "learning_rate": 6.046110368071201e-06,
+      "loss": 80.0069,
+      "step": 60590
+    },
+    {
+      "epoch": 0.5013028911775654,
+      "grad_norm": 1104.5489501953125,
+      "learning_rate": 6.044731173428862e-06,
+      "loss": 109.4119,
+      "step": 60600
+    },
+    {
+      "epoch": 0.5013856144269347,
+      "grad_norm": 547.853515625,
+      "learning_rate": 6.043351895662059e-06,
+      "loss": 74.3687,
+      "step": 60610
+    },
+    {
+      "epoch": 0.5014683376763039,
+      "grad_norm": 815.2512817382812,
+      "learning_rate": 6.041972534880533e-06,
+      "loss": 98.8586,
+      "step": 60620
+    },
+    {
+      "epoch": 0.5015510609256731,
+      "grad_norm": 934.5771484375,
+      "learning_rate": 6.040593091194035e-06,
+      "loss": 113.7061,
+      "step": 60630
+    },
+    {
+      "epoch": 0.5016337841750423,
+      "grad_norm": 980.511962890625,
+      "learning_rate": 6.039213564712319e-06,
+      "loss": 77.5835,
+      "step": 60640
+    },
+    {
+      "epoch": 0.5017165074244117,
+      "grad_norm": 875.10693359375,
+      "learning_rate": 6.03783395554515e-06,
+      "loss": 90.5567,
+      "step": 60650
+    },
+    {
+      "epoch": 0.5017992306737808,
+      "grad_norm": 564.4214477539062,
+      "learning_rate": 6.036454263802297e-06,
+      "loss": 99.0541,
+      "step": 60660
+    },
+    {
+      "epoch": 0.50188195392315,
+      "grad_norm": 859.4292602539062,
+      "learning_rate": 6.035074489593536e-06,
+      "loss": 81.3836,
+      "step": 60670
+    },
+    {
+      "epoch": 0.5019646771725194,
+      "grad_norm": 1366.642333984375,
+      "learning_rate": 6.033694633028644e-06,
+      "loss": 89.8926,
+      "step": 60680
+    },
+    {
+      "epoch": 0.5020474004218886,
+      "grad_norm": 707.115478515625,
+      "learning_rate": 6.032314694217416e-06,
+      "loss": 119.7774,
+      "step": 60690
+    },
+    {
+      "epoch": 0.5021301236712578,
+      "grad_norm": 509.8358154296875,
+      "learning_rate": 6.030934673269646e-06,
+      "loss": 76.9633,
+      "step": 60700
+    },
+    {
+      "epoch": 0.5022128469206271,
+      "grad_norm": 1127.6051025390625,
+      "learning_rate": 6.029554570295135e-06,
+      "loss": 98.3396,
+      "step": 60710
+    },
+    {
+      "epoch": 0.5022955701699963,
+      "grad_norm": 489.0888366699219,
+      "learning_rate": 6.028174385403693e-06,
+      "loss": 74.8344,
+      "step": 60720
+    },
+    {
+      "epoch": 0.5023782934193655,
+      "grad_norm": 563.7764282226562,
+      "learning_rate": 6.026794118705133e-06,
+      "loss": 123.1201,
+      "step": 60730
+    },
+    {
+      "epoch": 0.5024610166687348,
+      "grad_norm": 922.4900512695312,
+      "learning_rate": 6.025413770309278e-06,
+      "loss": 105.9961,
+      "step": 60740
+    },
+    {
+      "epoch": 0.502543739918104,
+      "grad_norm": 711.1076049804688,
+      "learning_rate": 6.024033340325954e-06,
+      "loss": 138.0339,
+      "step": 60750
+    },
+    {
+      "epoch": 0.5026264631674732,
+      "grad_norm": 1032.06787109375,
+      "learning_rate": 6.022652828864999e-06,
+      "loss": 129.3473,
+      "step": 60760
+    },
+    {
+      "epoch": 0.5027091864168425,
+      "grad_norm": 987.8187255859375,
+      "learning_rate": 6.0212722360362496e-06,
+      "loss": 98.7341,
+      "step": 60770
+    },
+    {
+      "epoch": 0.5027919096662117,
+      "grad_norm": 922.1596069335938,
+      "learning_rate": 6.019891561949554e-06,
+      "loss": 113.0199,
+      "step": 60780
+    },
+    {
+      "epoch": 0.5028746329155809,
+      "grad_norm": 966.7632446289062,
+      "learning_rate": 6.01851080671477e-06,
+      "loss": 95.6403,
+      "step": 60790
+    },
+    {
+      "epoch": 0.5029573561649502,
+      "grad_norm": 626.9385986328125,
+      "learning_rate": 6.017129970441756e-06,
+      "loss": 85.6963,
+      "step": 60800
+    },
+    {
+      "epoch": 0.5030400794143194,
+      "grad_norm": 1487.6356201171875,
+      "learning_rate": 6.015749053240378e-06,
+      "loss": 97.1684,
+      "step": 60810
+    },
+    {
+      "epoch": 0.5031228026636886,
+      "grad_norm": 789.6949462890625,
+      "learning_rate": 6.0143680552205075e-06,
+      "loss": 103.032,
+      "step": 60820
+    },
+    {
+      "epoch": 0.5032055259130579,
+      "grad_norm": 737.3887329101562,
+      "learning_rate": 6.012986976492025e-06,
+      "loss": 86.7182,
+      "step": 60830
+    },
+    {
+      "epoch": 0.5032882491624271,
+      "grad_norm": 600.4498291015625,
+      "learning_rate": 6.011605817164822e-06,
+      "loss": 94.8057,
+      "step": 60840
+    },
+    {
+      "epoch": 0.5033709724117963,
+      "grad_norm": 593.8978271484375,
+      "learning_rate": 6.0102245773487855e-06,
+      "loss": 107.6144,
+      "step": 60850
+    },
+    {
+      "epoch": 0.5034536956611656,
+      "grad_norm": 1065.4730224609375,
+      "learning_rate": 6.008843257153815e-06,
+      "loss": 90.4793,
+      "step": 60860
+    },
+    {
+      "epoch": 0.5035364189105348,
+      "grad_norm": 824.6838989257812,
+      "learning_rate": 6.007461856689815e-06,
+      "loss": 75.2064,
+      "step": 60870
+    },
+    {
+      "epoch": 0.503619142159904,
+      "grad_norm": 450.9040832519531,
+      "learning_rate": 6.0060803760667e-06,
+      "loss": 88.3228,
+      "step": 60880
+    },
+    {
+      "epoch": 0.5037018654092733,
+      "grad_norm": 829.0480346679688,
+      "learning_rate": 6.004698815394389e-06,
+      "loss": 124.8127,
+      "step": 60890
+    },
+    {
+      "epoch": 0.5037845886586425,
+      "grad_norm": 1101.244384765625,
+      "learning_rate": 6.003317174782801e-06,
+      "loss": 77.5046,
+      "step": 60900
+    },
+    {
+      "epoch": 0.5038673119080117,
+      "grad_norm": 1373.401611328125,
+      "learning_rate": 6.001935454341872e-06,
+      "loss": 92.4626,
+      "step": 60910
+    },
+    {
+      "epoch": 0.503950035157381,
+      "grad_norm": 489.91644287109375,
+      "learning_rate": 6.000553654181536e-06,
+      "loss": 92.368,
+      "step": 60920
+    },
+    {
+      "epoch": 0.5040327584067502,
+      "grad_norm": 813.798583984375,
+      "learning_rate": 5.999171774411737e-06,
+      "loss": 96.8709,
+      "step": 60930
+    },
+    {
+      "epoch": 0.5041154816561194,
+      "grad_norm": 1986.8172607421875,
+      "learning_rate": 5.997789815142427e-06,
+      "loss": 83.3268,
+      "step": 60940
+    },
+    {
+      "epoch": 0.5041982049054887,
+      "grad_norm": 721.0343017578125,
+      "learning_rate": 5.99640777648356e-06,
+      "loss": 111.3407,
+      "step": 60950
+    },
+    {
+      "epoch": 0.5042809281548579,
+      "grad_norm": 1814.00146484375,
+      "learning_rate": 5.9950256585450995e-06,
+      "loss": 101.1115,
+      "step": 60960
+    },
+    {
+      "epoch": 0.5043636514042271,
+      "grad_norm": 1539.85009765625,
+      "learning_rate": 5.993643461437013e-06,
+      "loss": 109.2939,
+      "step": 60970
+    },
+    {
+      "epoch": 0.5044463746535964,
+      "grad_norm": 1455.0811767578125,
+      "learning_rate": 5.992261185269278e-06,
+      "loss": 116.5055,
+      "step": 60980
+    },
+    {
+      "epoch": 0.5045290979029656,
+      "grad_norm": 673.0990600585938,
+      "learning_rate": 5.990878830151873e-06,
+      "loss": 84.7869,
+      "step": 60990
+    },
+    {
+      "epoch": 0.5046118211523348,
+      "grad_norm": 990.6354370117188,
+      "learning_rate": 5.989496396194787e-06,
+      "loss": 117.0743,
+      "step": 61000
+    },
+    {
+      "epoch": 0.5046945444017041,
+      "grad_norm": 659.1134033203125,
+      "learning_rate": 5.988113883508016e-06,
+      "loss": 85.4938,
+      "step": 61010
+    },
+    {
+      "epoch": 0.5047772676510733,
+      "grad_norm": 1276.8919677734375,
+      "learning_rate": 5.986731292201555e-06,
+      "loss": 111.5496,
+      "step": 61020
+    },
+    {
+      "epoch": 0.5048599909004425,
+      "grad_norm": 1091.1859130859375,
+      "learning_rate": 5.985348622385415e-06,
+      "loss": 139.6149,
+      "step": 61030
+    },
+    {
+      "epoch": 0.5049427141498118,
+      "grad_norm": 1123.3651123046875,
+      "learning_rate": 5.9839658741696085e-06,
+      "loss": 60.2066,
+      "step": 61040
+    },
+    {
+      "epoch": 0.505025437399181,
+      "grad_norm": 405.6517028808594,
+      "learning_rate": 5.982583047664151e-06,
+      "loss": 89.4623,
+      "step": 61050
+    },
+    {
+      "epoch": 0.5051081606485502,
+      "grad_norm": 489.998046875,
+      "learning_rate": 5.981200142979071e-06,
+      "loss": 70.606,
+      "step": 61060
+    },
+    {
+      "epoch": 0.5051908838979196,
+      "grad_norm": 611.7427368164062,
+      "learning_rate": 5.9798171602244e-06,
+      "loss": 90.265,
+      "step": 61070
+    },
+    {
+      "epoch": 0.5052736071472888,
+      "grad_norm": 500.9188537597656,
+      "learning_rate": 5.978434099510172e-06,
+      "loss": 102.3768,
+      "step": 61080
+    },
+    {
+      "epoch": 0.505356330396658,
+      "grad_norm": 939.5604858398438,
+      "learning_rate": 5.977050960946433e-06,
+      "loss": 95.7826,
+      "step": 61090
+    },
+    {
+      "epoch": 0.5054390536460273,
+      "grad_norm": 1079.417724609375,
+      "learning_rate": 5.975667744643235e-06,
+      "loss": 57.3004,
+      "step": 61100
+    },
+    {
+      "epoch": 0.5055217768953965,
+      "grad_norm": 801.0790405273438,
+      "learning_rate": 5.974284450710631e-06,
+      "loss": 100.4074,
+      "step": 61110
+    },
+    {
+      "epoch": 0.5056045001447657,
+      "grad_norm": 1189.5526123046875,
+      "learning_rate": 5.972901079258685e-06,
+      "loss": 89.206,
+      "step": 61120
+    },
+    {
+      "epoch": 0.505687223394135,
+      "grad_norm": 544.3245239257812,
+      "learning_rate": 5.971517630397465e-06,
+      "loss": 87.3128,
+      "step": 61130
+    },
+    {
+      "epoch": 0.5057699466435042,
+      "grad_norm": 804.9583129882812,
+      "learning_rate": 5.970134104237046e-06,
+      "loss": 118.5412,
+      "step": 61140
+    },
+    {
+      "epoch": 0.5058526698928734,
+      "grad_norm": 878.0859985351562,
+      "learning_rate": 5.96875050088751e-06,
+      "loss": 151.7707,
+      "step": 61150
+    },
+    {
+      "epoch": 0.5059353931422427,
+      "grad_norm": 555.1686401367188,
+      "learning_rate": 5.9673668204589396e-06,
+      "loss": 80.5509,
+      "step": 61160
+    },
+    {
+      "epoch": 0.5060181163916119,
+      "grad_norm": 887.9091186523438,
+      "learning_rate": 5.965983063061432e-06,
+      "loss": 97.7346,
+      "step": 61170
+    },
+    {
+      "epoch": 0.5061008396409811,
+      "grad_norm": 2689.885498046875,
+      "learning_rate": 5.964599228805087e-06,
+      "loss": 109.0754,
+      "step": 61180
+    },
+    {
+      "epoch": 0.5061835628903504,
+      "grad_norm": 851.7760009765625,
+      "learning_rate": 5.963215317800008e-06,
+      "loss": 111.0486,
+      "step": 61190
+    },
+    {
+      "epoch": 0.5062662861397196,
+      "grad_norm": 912.8546142578125,
+      "learning_rate": 5.961831330156306e-06,
+      "loss": 131.0121,
+      "step": 61200
+    },
+    {
+      "epoch": 0.5063490093890888,
+      "grad_norm": 641.365478515625,
+      "learning_rate": 5.960447265984098e-06,
+      "loss": 88.7224,
+      "step": 61210
+    },
+    {
+      "epoch": 0.5064317326384581,
+      "grad_norm": 1167.5374755859375,
+      "learning_rate": 5.95906312539351e-06,
+      "loss": 90.7471,
+      "step": 61220
+    },
+    {
+      "epoch": 0.5065144558878273,
+      "grad_norm": 808.2752685546875,
+      "learning_rate": 5.9576789084946705e-06,
+      "loss": 74.6994,
+      "step": 61230
+    },
+    {
+      "epoch": 0.5065971791371965,
+      "grad_norm": 1379.4791259765625,
+      "learning_rate": 5.956294615397716e-06,
+      "loss": 105.455,
+      "step": 61240
+    },
+    {
+      "epoch": 0.5066799023865658,
+      "grad_norm": 1430.2303466796875,
+      "learning_rate": 5.954910246212787e-06,
+      "loss": 126.3884,
+      "step": 61250
+    },
+    {
+      "epoch": 0.506762625635935,
+      "grad_norm": 1149.9755859375,
+      "learning_rate": 5.953525801050032e-06,
+      "loss": 83.9256,
+      "step": 61260
+    },
+    {
+      "epoch": 0.5068453488853042,
+      "grad_norm": 558.4307861328125,
+      "learning_rate": 5.952141280019605e-06,
+      "loss": 96.2179,
+      "step": 61270
+    },
+    {
+      "epoch": 0.5069280721346735,
+      "grad_norm": 742.656494140625,
+      "learning_rate": 5.950756683231667e-06,
+      "loss": 74.2085,
+      "step": 61280
+    },
+    {
+      "epoch": 0.5070107953840427,
+      "grad_norm": 596.1998291015625,
+      "learning_rate": 5.949372010796384e-06,
+      "loss": 52.8268,
+      "step": 61290
+    },
+    {
+      "epoch": 0.5070935186334119,
+      "grad_norm": 881.3981323242188,
+      "learning_rate": 5.947987262823924e-06,
+      "loss": 107.5389,
+      "step": 61300
+    },
+    {
+      "epoch": 0.5071762418827812,
+      "grad_norm": 675.0526123046875,
+      "learning_rate": 5.94660243942447e-06,
+      "loss": 123.2516,
+      "step": 61310
+    },
+    {
+      "epoch": 0.5072589651321504,
+      "grad_norm": 889.6369018554688,
+      "learning_rate": 5.945217540708206e-06,
+      "loss": 86.9526,
+      "step": 61320
+    },
+    {
+      "epoch": 0.5073416883815196,
+      "grad_norm": 736.0260620117188,
+      "learning_rate": 5.9438325667853185e-06,
+      "loss": 94.7347,
+      "step": 61330
+    },
+    {
+      "epoch": 0.5074244116308889,
+      "grad_norm": 899.1821899414062,
+      "learning_rate": 5.942447517766005e-06,
+      "loss": 106.1979,
+      "step": 61340
+    },
+    {
+      "epoch": 0.5075071348802581,
+      "grad_norm": 760.8129272460938,
+      "learning_rate": 5.941062393760467e-06,
+      "loss": 94.1305,
+      "step": 61350
+    },
+    {
+      "epoch": 0.5075898581296273,
+      "grad_norm": 434.6575622558594,
+      "learning_rate": 5.939677194878915e-06,
+      "loss": 140.0839,
+      "step": 61360
+    },
+    {
+      "epoch": 0.5076725813789965,
+      "grad_norm": 577.9569091796875,
+      "learning_rate": 5.93829192123156e-06,
+      "loss": 74.5071,
+      "step": 61370
+    },
+    {
+      "epoch": 0.5077553046283658,
+      "grad_norm": 694.65625,
+      "learning_rate": 5.936906572928625e-06,
+      "loss": 120.0574,
+      "step": 61380
+    },
+    {
+      "epoch": 0.507838027877735,
+      "grad_norm": 772.7642211914062,
+      "learning_rate": 5.935521150080331e-06,
+      "loss": 75.3494,
+      "step": 61390
+    },
+    {
+      "epoch": 0.5079207511271042,
+      "grad_norm": 790.8641967773438,
+      "learning_rate": 5.934135652796914e-06,
+      "loss": 87.367,
+      "step": 61400
+    },
+    {
+      "epoch": 0.5080034743764735,
+      "grad_norm": 813.5361938476562,
+      "learning_rate": 5.9327500811886095e-06,
+      "loss": 83.0071,
+      "step": 61410
+    },
+    {
+      "epoch": 0.5080861976258427,
+      "grad_norm": 437.74139404296875,
+      "learning_rate": 5.931364435365663e-06,
+      "loss": 115.0982,
+      "step": 61420
+    },
+    {
+      "epoch": 0.5081689208752119,
+      "grad_norm": 890.75927734375,
+      "learning_rate": 5.929978715438322e-06,
+      "loss": 75.1045,
+      "step": 61430
+    },
+    {
+      "epoch": 0.5082516441245812,
+      "grad_norm": 861.6392822265625,
+      "learning_rate": 5.928592921516843e-06,
+      "loss": 87.0132,
+      "step": 61440
+    },
+    {
+      "epoch": 0.5083343673739504,
+      "grad_norm": 401.1084289550781,
+      "learning_rate": 5.9272070537114855e-06,
+      "loss": 111.1549,
+      "step": 61450
+    },
+    {
+      "epoch": 0.5084170906233196,
+      "grad_norm": 886.0931396484375,
+      "learning_rate": 5.92582111213252e-06,
+      "loss": 130.5894,
+      "step": 61460
+    },
+    {
+      "epoch": 0.5084998138726889,
+      "grad_norm": 573.5504150390625,
+      "learning_rate": 5.924435096890216e-06,
+      "loss": 96.3187,
+      "step": 61470
+    },
+    {
+      "epoch": 0.5085825371220581,
+      "grad_norm": 915.1154174804688,
+      "learning_rate": 5.923049008094855e-06,
+      "loss": 102.2801,
+      "step": 61480
+    },
+    {
+      "epoch": 0.5086652603714273,
+      "grad_norm": 789.848876953125,
+      "learning_rate": 5.921662845856719e-06,
+      "loss": 93.1844,
+      "step": 61490
+    },
+    {
+      "epoch": 0.5087479836207967,
+      "grad_norm": 1229.4134521484375,
+      "learning_rate": 5.920276610286102e-06,
+      "loss": 119.4875,
+      "step": 61500
+    },
+    {
+      "epoch": 0.5088307068701658,
+      "grad_norm": 1035.0843505859375,
+      "learning_rate": 5.918890301493299e-06,
+      "loss": 95.7163,
+      "step": 61510
+    },
+    {
+      "epoch": 0.508913430119535,
+      "grad_norm": 880.9725952148438,
+      "learning_rate": 5.91750391958861e-06,
+      "loss": 73.2856,
+      "step": 61520
+    },
+    {
+      "epoch": 0.5089961533689044,
+      "grad_norm": 646.2332763671875,
+      "learning_rate": 5.916117464682346e-06,
+      "loss": 106.2007,
+      "step": 61530
+    },
+    {
+      "epoch": 0.5090788766182736,
+      "grad_norm": 485.7245178222656,
+      "learning_rate": 5.914730936884819e-06,
+      "loss": 112.5809,
+      "step": 61540
+    },
+    {
+      "epoch": 0.5091615998676428,
+      "grad_norm": 1304.14794921875,
+      "learning_rate": 5.91334433630635e-06,
+      "loss": 124.1074,
+      "step": 61550
+    },
+    {
+      "epoch": 0.5092443231170121,
+      "grad_norm": 911.8284912109375,
+      "learning_rate": 5.911957663057264e-06,
+      "loss": 87.838,
+      "step": 61560
+    },
+    {
+      "epoch": 0.5093270463663813,
+      "grad_norm": 863.0823364257812,
+      "learning_rate": 5.910570917247892e-06,
+      "loss": 79.3177,
+      "step": 61570
+    },
+    {
+      "epoch": 0.5094097696157505,
+      "grad_norm": 518.0543823242188,
+      "learning_rate": 5.909184098988571e-06,
+      "loss": 97.5815,
+      "step": 61580
+    },
+    {
+      "epoch": 0.5094924928651198,
+      "grad_norm": 641.208984375,
+      "learning_rate": 5.907797208389644e-06,
+      "loss": 112.5585,
+      "step": 61590
+    },
+    {
+      "epoch": 0.509575216114489,
+      "grad_norm": 644.5621337890625,
+      "learning_rate": 5.906410245561459e-06,
+      "loss": 80.3052,
+      "step": 61600
+    },
+    {
+      "epoch": 0.5096579393638582,
+      "grad_norm": 1210.9039306640625,
+      "learning_rate": 5.90502321061437e-06,
+      "loss": 103.0409,
+      "step": 61610
+    },
+    {
+      "epoch": 0.5097406626132275,
+      "grad_norm": 1110.852294921875,
+      "learning_rate": 5.90363610365874e-06,
+      "loss": 120.3473,
+      "step": 61620
+    },
+    {
+      "epoch": 0.5098233858625967,
+      "grad_norm": 892.0844116210938,
+      "learning_rate": 5.9022489248049295e-06,
+      "loss": 101.4571,
+      "step": 61630
+    },
+    {
+      "epoch": 0.5099061091119659,
+      "grad_norm": 456.140625,
+      "learning_rate": 5.900861674163314e-06,
+      "loss": 74.4327,
+      "step": 61640
+    },
+    {
+      "epoch": 0.5099888323613352,
+      "grad_norm": 755.904541015625,
+      "learning_rate": 5.89947435184427e-06,
+      "loss": 90.8847,
+      "step": 61650
+    },
+    {
+      "epoch": 0.5100715556107044,
+      "grad_norm": 1451.006591796875,
+      "learning_rate": 5.89808695795818e-06,
+      "loss": 101.8353,
+      "step": 61660
+    },
+    {
+      "epoch": 0.5101542788600736,
+      "grad_norm": 1081.45263671875,
+      "learning_rate": 5.896699492615432e-06,
+      "loss": 86.6635,
+      "step": 61670
+    },
+    {
+      "epoch": 0.5102370021094429,
+      "grad_norm": 497.5079040527344,
+      "learning_rate": 5.895311955926419e-06,
+      "loss": 83.6076,
+      "step": 61680
+    },
+    {
+      "epoch": 0.5103197253588121,
+      "grad_norm": 643.5697631835938,
+      "learning_rate": 5.893924348001544e-06,
+      "loss": 97.4833,
+      "step": 61690
+    },
+    {
+      "epoch": 0.5104024486081813,
+      "grad_norm": 956.6340942382812,
+      "learning_rate": 5.8925366689512124e-06,
+      "loss": 103.3445,
+      "step": 61700
+    },
+    {
+      "epoch": 0.5104851718575506,
+      "grad_norm": 665.5700073242188,
+      "learning_rate": 5.891148918885834e-06,
+      "loss": 65.6439,
+      "step": 61710
+    },
+    {
+      "epoch": 0.5105678951069198,
+      "grad_norm": 1049.6845703125,
+      "learning_rate": 5.8897610979158245e-06,
+      "loss": 89.6039,
+      "step": 61720
+    },
+    {
+      "epoch": 0.510650618356289,
+      "grad_norm": 814.806396484375,
+      "learning_rate": 5.888373206151608e-06,
+      "loss": 102.2241,
+      "step": 61730
+    },
+    {
+      "epoch": 0.5107333416056583,
+      "grad_norm": 841.8108520507812,
+      "learning_rate": 5.886985243703612e-06,
+      "loss": 103.3497,
+      "step": 61740
+    },
+    {
+      "epoch": 0.5108160648550275,
+      "grad_norm": 488.0580749511719,
+      "learning_rate": 5.885597210682273e-06,
+      "loss": 117.5541,
+      "step": 61750
+    },
+    {
+      "epoch": 0.5108987881043967,
+      "grad_norm": 780.1454467773438,
+      "learning_rate": 5.884209107198027e-06,
+      "loss": 118.4064,
+      "step": 61760
+    },
+    {
+      "epoch": 0.510981511353766,
+      "grad_norm": 1028.7705078125,
+      "learning_rate": 5.882820933361321e-06,
+      "loss": 127.9731,
+      "step": 61770
+    },
+    {
+      "epoch": 0.5110642346031352,
+      "grad_norm": 684.5718383789062,
+      "learning_rate": 5.881432689282604e-06,
+      "loss": 121.3453,
+      "step": 61780
+    },
+    {
+      "epoch": 0.5111469578525044,
+      "grad_norm": 910.465087890625,
+      "learning_rate": 5.880044375072333e-06,
+      "loss": 135.5772,
+      "step": 61790
+    },
+    {
+      "epoch": 0.5112296811018737,
+      "grad_norm": 1077.9769287109375,
+      "learning_rate": 5.8786559908409715e-06,
+      "loss": 110.6625,
+      "step": 61800
+    },
+    {
+      "epoch": 0.5113124043512429,
+      "grad_norm": 1104.657470703125,
+      "learning_rate": 5.877267536698984e-06,
+      "loss": 91.3742,
+      "step": 61810
+    },
+    {
+      "epoch": 0.5113951276006121,
+      "grad_norm": 924.7949829101562,
+      "learning_rate": 5.875879012756845e-06,
+      "loss": 93.8301,
+      "step": 61820
+    },
+    {
+      "epoch": 0.5114778508499814,
+      "grad_norm": 830.5784301757812,
+      "learning_rate": 5.8744904191250326e-06,
+      "loss": 135.9706,
+      "step": 61830
+    },
+    {
+      "epoch": 0.5115605740993506,
+      "grad_norm": 1287.56884765625,
+      "learning_rate": 5.873101755914031e-06,
+      "loss": 113.9375,
+      "step": 61840
+    },
+    {
+      "epoch": 0.5116432973487198,
+      "grad_norm": 2128.367431640625,
+      "learning_rate": 5.87171302323433e-06,
+      "loss": 91.666,
+      "step": 61850
+    },
+    {
+      "epoch": 0.5117260205980891,
+      "grad_norm": 1100.2708740234375,
+      "learning_rate": 5.870324221196424e-06,
+      "loss": 91.6138,
+      "step": 61860
+    },
+    {
+      "epoch": 0.5118087438474583,
+      "grad_norm": 636.7507934570312,
+      "learning_rate": 5.868935349910814e-06,
+      "loss": 105.8906,
+      "step": 61870
+    },
+    {
+      "epoch": 0.5118914670968275,
+      "grad_norm": 928.61083984375,
+      "learning_rate": 5.867546409488006e-06,
+      "loss": 99.2666,
+      "step": 61880
+    },
+    {
+      "epoch": 0.5119741903461968,
+      "grad_norm": 562.5609130859375,
+      "learning_rate": 5.8661574000385115e-06,
+      "loss": 76.9907,
+      "step": 61890
+    },
+    {
+      "epoch": 0.512056913595566,
+      "grad_norm": 390.1648864746094,
+      "learning_rate": 5.864768321672848e-06,
+      "loss": 76.5244,
+      "step": 61900
+    },
+    {
+      "epoch": 0.5121396368449352,
+      "grad_norm": 944.2666625976562,
+      "learning_rate": 5.863379174501538e-06,
+      "loss": 98.3042,
+      "step": 61910
+    },
+    {
+      "epoch": 0.5122223600943046,
+      "grad_norm": 1686.9356689453125,
+      "learning_rate": 5.861989958635109e-06,
+      "loss": 121.1383,
+      "step": 61920
+    },
+    {
+      "epoch": 0.5123050833436738,
+      "grad_norm": 878.6798095703125,
+      "learning_rate": 5.860600674184096e-06,
+      "loss": 105.9373,
+      "step": 61930
+    },
+    {
+      "epoch": 0.512387806593043,
+      "grad_norm": 528.4129638671875,
+      "learning_rate": 5.859211321259036e-06,
+      "loss": 83.7999,
+      "step": 61940
+    },
+    {
+      "epoch": 0.5124705298424123,
+      "grad_norm": 1257.8131103515625,
+      "learning_rate": 5.857821899970475e-06,
+      "loss": 112.0766,
+      "step": 61950
+    },
+    {
+      "epoch": 0.5125532530917815,
+      "grad_norm": 1426.787353515625,
+      "learning_rate": 5.856432410428963e-06,
+      "loss": 127.3927,
+      "step": 61960
+    },
+    {
+      "epoch": 0.5126359763411507,
+      "grad_norm": 1306.6192626953125,
+      "learning_rate": 5.8550428527450534e-06,
+      "loss": 63.1702,
+      "step": 61970
+    },
+    {
+      "epoch": 0.51271869959052,
+      "grad_norm": 1105.7908935546875,
+      "learning_rate": 5.8536532270293076e-06,
+      "loss": 80.9909,
+      "step": 61980
+    },
+    {
+      "epoch": 0.5128014228398892,
+      "grad_norm": 1267.3624267578125,
+      "learning_rate": 5.852263533392294e-06,
+      "loss": 94.7298,
+      "step": 61990
+    },
+    {
+      "epoch": 0.5128841460892584,
+      "grad_norm": 594.2600708007812,
+      "learning_rate": 5.850873771944581e-06,
+      "loss": 91.9315,
+      "step": 62000
+    },
+    {
+      "epoch": 0.5129668693386277,
+      "grad_norm": 652.3613891601562,
+      "learning_rate": 5.849483942796747e-06,
+      "loss": 87.1766,
+      "step": 62010
+    },
+    {
+      "epoch": 0.5130495925879969,
+      "grad_norm": 602.3596801757812,
+      "learning_rate": 5.848094046059375e-06,
+      "loss": 124.0294,
+      "step": 62020
+    },
+    {
+      "epoch": 0.5131323158373661,
+      "grad_norm": 573.1605224609375,
+      "learning_rate": 5.846704081843052e-06,
+      "loss": 111.6119,
+      "step": 62030
+    },
+    {
+      "epoch": 0.5132150390867354,
+      "grad_norm": 780.5606689453125,
+      "learning_rate": 5.84531405025837e-06,
+      "loss": 82.9944,
+      "step": 62040
+    },
+    {
+      "epoch": 0.5132977623361046,
+      "grad_norm": 913.4967041015625,
+      "learning_rate": 5.843923951415931e-06,
+      "loss": 108.7144,
+      "step": 62050
+    },
+    {
+      "epoch": 0.5133804855854738,
+      "grad_norm": 1110.280029296875,
+      "learning_rate": 5.842533785426334e-06,
+      "loss": 101.928,
+      "step": 62060
+    },
+    {
+      "epoch": 0.5134632088348431,
+      "grad_norm": 1380.97314453125,
+      "learning_rate": 5.84114355240019e-06,
+      "loss": 95.0093,
+      "step": 62070
+    },
+    {
+      "epoch": 0.5135459320842123,
+      "grad_norm": 570.7285766601562,
+      "learning_rate": 5.839753252448115e-06,
+      "loss": 88.7862,
+      "step": 62080
+    },
+    {
+      "epoch": 0.5136286553335815,
+      "grad_norm": 1217.79931640625,
+      "learning_rate": 5.838362885680728e-06,
+      "loss": 94.9879,
+      "step": 62090
+    },
+    {
+      "epoch": 0.5137113785829507,
+      "grad_norm": 1030.6907958984375,
+      "learning_rate": 5.8369724522086545e-06,
+      "loss": 88.0829,
+      "step": 62100
+    },
+    {
+      "epoch": 0.51379410183232,
+      "grad_norm": 2263.951416015625,
+      "learning_rate": 5.835581952142522e-06,
+      "loss": 147.4353,
+      "step": 62110
+    },
+    {
+      "epoch": 0.5138768250816892,
+      "grad_norm": 499.37347412109375,
+      "learning_rate": 5.834191385592969e-06,
+      "loss": 119.7903,
+      "step": 62120
+    },
+    {
+      "epoch": 0.5139595483310584,
+      "grad_norm": 1055.9906005859375,
+      "learning_rate": 5.8328007526706354e-06,
+      "loss": 87.058,
+      "step": 62130
+    },
+    {
+      "epoch": 0.5140422715804277,
+      "grad_norm": 1002.3789672851562,
+      "learning_rate": 5.83141005348617e-06,
+      "loss": 107.1515,
+      "step": 62140
+    },
+    {
+      "epoch": 0.5141249948297969,
+      "grad_norm": 859.0476684570312,
+      "learning_rate": 5.830019288150222e-06,
+      "loss": 90.1526,
+      "step": 62150
+    },
+    {
+      "epoch": 0.5142077180791661,
+      "grad_norm": 765.904052734375,
+      "learning_rate": 5.8286284567734456e-06,
+      "loss": 81.6313,
+      "step": 62160
+    },
+    {
+      "epoch": 0.5142904413285354,
+      "grad_norm": 676.10009765625,
+      "learning_rate": 5.827237559466508e-06,
+      "loss": 84.7288,
+      "step": 62170
+    },
+    {
+      "epoch": 0.5143731645779046,
+      "grad_norm": 1681.114990234375,
+      "learning_rate": 5.825846596340075e-06,
+      "loss": 122.0725,
+      "step": 62180
+    },
+    {
+      "epoch": 0.5144558878272738,
+      "grad_norm": 708.9700927734375,
+      "learning_rate": 5.824455567504817e-06,
+      "loss": 123.205,
+      "step": 62190
+    },
+    {
+      "epoch": 0.5145386110766431,
+      "grad_norm": 853.5513305664062,
+      "learning_rate": 5.823064473071414e-06,
+      "loss": 96.5516,
+      "step": 62200
+    },
+    {
+      "epoch": 0.5146213343260123,
+      "grad_norm": 937.1123046875,
+      "learning_rate": 5.821673313150546e-06,
+      "loss": 98.6421,
+      "step": 62210
+    },
+    {
+      "epoch": 0.5147040575753815,
+      "grad_norm": 802.1367797851562,
+      "learning_rate": 5.820282087852906e-06,
+      "loss": 99.0799,
+      "step": 62220
+    },
+    {
+      "epoch": 0.5147867808247508,
+      "grad_norm": 451.8292236328125,
+      "learning_rate": 5.818890797289185e-06,
+      "loss": 90.0844,
+      "step": 62230
+    },
+    {
+      "epoch": 0.51486950407412,
+      "grad_norm": 1798.909912109375,
+      "learning_rate": 5.81749944157008e-06,
+      "loss": 137.7885,
+      "step": 62240
+    },
+    {
+      "epoch": 0.5149522273234892,
+      "grad_norm": 985.250732421875,
+      "learning_rate": 5.816108020806297e-06,
+      "loss": 97.535,
+      "step": 62250
+    },
+    {
+      "epoch": 0.5150349505728585,
+      "grad_norm": 672.7926635742188,
+      "learning_rate": 5.814716535108545e-06,
+      "loss": 88.2172,
+      "step": 62260
+    },
+    {
+      "epoch": 0.5151176738222277,
+      "grad_norm": 716.781982421875,
+      "learning_rate": 5.813324984587536e-06,
+      "loss": 106.1954,
+      "step": 62270
+    },
+    {
+      "epoch": 0.5152003970715969,
+      "grad_norm": 1432.325439453125,
+      "learning_rate": 5.811933369353992e-06,
+      "loss": 111.049,
+      "step": 62280
+    },
+    {
+      "epoch": 0.5152831203209662,
+      "grad_norm": 760.8700561523438,
+      "learning_rate": 5.810541689518634e-06,
+      "loss": 94.8973,
+      "step": 62290
+    },
+    {
+      "epoch": 0.5153658435703354,
+      "grad_norm": 845.1988525390625,
+      "learning_rate": 5.809149945192194e-06,
+      "loss": 62.3175,
+      "step": 62300
+    },
+    {
+      "epoch": 0.5154485668197046,
+      "grad_norm": 770.4382934570312,
+      "learning_rate": 5.807758136485409e-06,
+      "loss": 106.6364,
+      "step": 62310
+    },
+    {
+      "epoch": 0.5155312900690739,
+      "grad_norm": 1148.9901123046875,
+      "learning_rate": 5.8063662635090136e-06,
+      "loss": 98.8217,
+      "step": 62320
+    },
+    {
+      "epoch": 0.5156140133184431,
+      "grad_norm": 1040.4677734375,
+      "learning_rate": 5.804974326373756e-06,
+      "loss": 92.393,
+      "step": 62330
+    },
+    {
+      "epoch": 0.5156967365678123,
+      "grad_norm": 609.6980590820312,
+      "learning_rate": 5.803582325190387e-06,
+      "loss": 97.1185,
+      "step": 62340
+    },
+    {
+      "epoch": 0.5157794598171817,
+      "grad_norm": 799.3777465820312,
+      "learning_rate": 5.802190260069657e-06,
+      "loss": 87.1282,
+      "step": 62350
+    },
+    {
+      "epoch": 0.5158621830665509,
+      "grad_norm": 968.39990234375,
+      "learning_rate": 5.800798131122332e-06,
+      "loss": 86.2219,
+      "step": 62360
+    },
+    {
+      "epoch": 0.51594490631592,
+      "grad_norm": 1028.2894287109375,
+      "learning_rate": 5.799405938459175e-06,
+      "loss": 76.0357,
+      "step": 62370
+    },
+    {
+      "epoch": 0.5160276295652894,
+      "grad_norm": 1118.3125,
+      "learning_rate": 5.7980136821909565e-06,
+      "loss": 91.065,
+      "step": 62380
+    },
+    {
+      "epoch": 0.5161103528146586,
+      "grad_norm": 817.6798706054688,
+      "learning_rate": 5.79662136242845e-06,
+      "loss": 112.0168,
+      "step": 62390
+    },
+    {
+      "epoch": 0.5161930760640278,
+      "grad_norm": 847.1541748046875,
+      "learning_rate": 5.795228979282439e-06,
+      "loss": 78.081,
+      "step": 62400
+    },
+    {
+      "epoch": 0.5162757993133971,
+      "grad_norm": 474.493896484375,
+      "learning_rate": 5.793836532863707e-06,
+      "loss": 70.5183,
+      "step": 62410
+    },
+    {
+      "epoch": 0.5163585225627663,
+      "grad_norm": 720.2383422851562,
+      "learning_rate": 5.792444023283046e-06,
+      "loss": 87.0467,
+      "step": 62420
+    },
+    {
+      "epoch": 0.5164412458121355,
+      "grad_norm": 847.5111083984375,
+      "learning_rate": 5.791051450651251e-06,
+      "loss": 80.678,
+      "step": 62430
+    },
+    {
+      "epoch": 0.5165239690615048,
+      "grad_norm": 1734.1534423828125,
+      "learning_rate": 5.789658815079121e-06,
+      "loss": 106.6263,
+      "step": 62440
+    },
+    {
+      "epoch": 0.516606692310874,
+      "grad_norm": 418.1882019042969,
+      "learning_rate": 5.788266116677464e-06,
+      "loss": 88.7378,
+      "step": 62450
+    },
+    {
+      "epoch": 0.5166894155602432,
+      "grad_norm": 1164.404541015625,
+      "learning_rate": 5.78687335555709e-06,
+      "loss": 109.2202,
+      "step": 62460
+    },
+    {
+      "epoch": 0.5167721388096125,
+      "grad_norm": 843.881103515625,
+      "learning_rate": 5.785480531828815e-06,
+      "loss": 95.9288,
+      "step": 62470
+    },
+    {
+      "epoch": 0.5168548620589817,
+      "grad_norm": 753.1651000976562,
+      "learning_rate": 5.784087645603459e-06,
+      "loss": 88.32,
+      "step": 62480
+    },
+    {
+      "epoch": 0.5169375853083509,
+      "grad_norm": 488.68487548828125,
+      "learning_rate": 5.782694696991845e-06,
+      "loss": 77.3577,
+      "step": 62490
+    },
+    {
+      "epoch": 0.5170203085577202,
+      "grad_norm": 748.185546875,
+      "learning_rate": 5.781301686104808e-06,
+      "loss": 99.6195,
+      "step": 62500
+    },
+    {
+      "epoch": 0.5171030318070894,
+      "grad_norm": 1028.9573974609375,
+      "learning_rate": 5.779908613053181e-06,
+      "loss": 89.3196,
+      "step": 62510
+    },
+    {
+      "epoch": 0.5171857550564586,
+      "grad_norm": 823.2115478515625,
+      "learning_rate": 5.778515477947807e-06,
+      "loss": 85.4209,
+      "step": 62520
+    },
+    {
+      "epoch": 0.5172684783058279,
+      "grad_norm": 720.9722900390625,
+      "learning_rate": 5.777122280899527e-06,
+      "loss": 87.5316,
+      "step": 62530
+    },
+    {
+      "epoch": 0.5173512015551971,
+      "grad_norm": 393.12646484375,
+      "learning_rate": 5.775729022019193e-06,
+      "loss": 132.2376,
+      "step": 62540
+    },
+    {
+      "epoch": 0.5174339248045663,
+      "grad_norm": 517.8870849609375,
+      "learning_rate": 5.774335701417662e-06,
+      "loss": 74.0599,
+      "step": 62550
+    },
+    {
+      "epoch": 0.5175166480539356,
+      "grad_norm": 779.7863159179688,
+      "learning_rate": 5.7729423192057936e-06,
+      "loss": 104.9648,
+      "step": 62560
+    },
+    {
+      "epoch": 0.5175993713033048,
+      "grad_norm": 1225.2161865234375,
+      "learning_rate": 5.771548875494453e-06,
+      "loss": 73.9343,
+      "step": 62570
+    },
+    {
+      "epoch": 0.517682094552674,
+      "grad_norm": 1787.88232421875,
+      "learning_rate": 5.7701553703945055e-06,
+      "loss": 99.922,
+      "step": 62580
+    },
+    {
+      "epoch": 0.5177648178020433,
+      "grad_norm": 761.9515380859375,
+      "learning_rate": 5.768761804016833e-06,
+      "loss": 126.8296,
+      "step": 62590
+    },
+    {
+      "epoch": 0.5178475410514125,
+      "grad_norm": 700.2348022460938,
+      "learning_rate": 5.767368176472311e-06,
+      "loss": 77.3953,
+      "step": 62600
+    },
+    {
+      "epoch": 0.5179302643007817,
+      "grad_norm": 1287.00732421875,
+      "learning_rate": 5.765974487871826e-06,
+      "loss": 110.9408,
+      "step": 62610
+    },
+    {
+      "epoch": 0.518012987550151,
+      "grad_norm": 1209.4278564453125,
+      "learning_rate": 5.764580738326265e-06,
+      "loss": 122.3806,
+      "step": 62620
+    },
+    {
+      "epoch": 0.5180957107995202,
+      "grad_norm": 972.08251953125,
+      "learning_rate": 5.763186927946523e-06,
+      "loss": 96.5278,
+      "step": 62630
+    },
+    {
+      "epoch": 0.5181784340488894,
+      "grad_norm": 939.2523193359375,
+      "learning_rate": 5.761793056843501e-06,
+      "loss": 96.9909,
+      "step": 62640
+    },
+    {
+      "epoch": 0.5182611572982587,
+      "grad_norm": 729.4473876953125,
+      "learning_rate": 5.760399125128102e-06,
+      "loss": 98.4378,
+      "step": 62650
+    },
+    {
+      "epoch": 0.5183438805476279,
+      "grad_norm": 780.3011474609375,
+      "learning_rate": 5.759005132911233e-06,
+      "loss": 78.8927,
+      "step": 62660
+    },
+    {
+      "epoch": 0.5184266037969971,
+      "grad_norm": 546.9750366210938,
+      "learning_rate": 5.75761108030381e-06,
+      "loss": 97.8188,
+      "step": 62670
+    },
+    {
+      "epoch": 0.5185093270463664,
+      "grad_norm": 898.6033325195312,
+      "learning_rate": 5.756216967416749e-06,
+      "loss": 73.9803,
+      "step": 62680
+    },
+    {
+      "epoch": 0.5185920502957356,
+      "grad_norm": 1270.0867919921875,
+      "learning_rate": 5.754822794360976e-06,
+      "loss": 100.2555,
+      "step": 62690
+    },
+    {
+      "epoch": 0.5186747735451048,
+      "grad_norm": 799.2184448242188,
+      "learning_rate": 5.753428561247416e-06,
+      "loss": 104.4311,
+      "step": 62700
+    },
+    {
+      "epoch": 0.5187574967944741,
+      "grad_norm": 756.2783813476562,
+      "learning_rate": 5.752034268187005e-06,
+      "loss": 90.7448,
+      "step": 62710
+    },
+    {
+      "epoch": 0.5188402200438433,
+      "grad_norm": 673.90576171875,
+      "learning_rate": 5.750639915290677e-06,
+      "loss": 100.3478,
+      "step": 62720
+    },
+    {
+      "epoch": 0.5189229432932125,
+      "grad_norm": 961.5318603515625,
+      "learning_rate": 5.749245502669375e-06,
+      "loss": 88.5219,
+      "step": 62730
+    },
+    {
+      "epoch": 0.5190056665425818,
+      "grad_norm": 700.3480834960938,
+      "learning_rate": 5.747851030434049e-06,
+      "loss": 85.1043,
+      "step": 62740
+    },
+    {
+      "epoch": 0.519088389791951,
+      "grad_norm": 508.70166015625,
+      "learning_rate": 5.746456498695648e-06,
+      "loss": 95.531,
+      "step": 62750
+    },
+    {
+      "epoch": 0.5191711130413202,
+      "grad_norm": 490.99737548828125,
+      "learning_rate": 5.7450619075651305e-06,
+      "loss": 87.5613,
+      "step": 62760
+    },
+    {
+      "epoch": 0.5192538362906896,
+      "grad_norm": 1027.1727294921875,
+      "learning_rate": 5.743667257153454e-06,
+      "loss": 81.3641,
+      "step": 62770
+    },
+    {
+      "epoch": 0.5193365595400588,
+      "grad_norm": 1106.285400390625,
+      "learning_rate": 5.742272547571588e-06,
+      "loss": 113.6713,
+      "step": 62780
+    },
+    {
+      "epoch": 0.519419282789428,
+      "grad_norm": 657.1253662109375,
+      "learning_rate": 5.740877778930503e-06,
+      "loss": 92.9361,
+      "step": 62790
+    },
+    {
+      "epoch": 0.5195020060387973,
+      "grad_norm": 831.34521484375,
+      "learning_rate": 5.739482951341172e-06,
+      "loss": 120.2162,
+      "step": 62800
+    },
+    {
+      "epoch": 0.5195847292881665,
+      "grad_norm": 709.2473754882812,
+      "learning_rate": 5.738088064914576e-06,
+      "loss": 112.4378,
+      "step": 62810
+    },
+    {
+      "epoch": 0.5196674525375357,
+      "grad_norm": 847.8458862304688,
+      "learning_rate": 5.7366931197617e-06,
+      "loss": 85.5061,
+      "step": 62820
+    },
+    {
+      "epoch": 0.5197501757869049,
+      "grad_norm": 1197.6585693359375,
+      "learning_rate": 5.735298115993535e-06,
+      "loss": 93.4385,
+      "step": 62830
+    },
+    {
+      "epoch": 0.5198328990362742,
+      "grad_norm": 1183.1817626953125,
+      "learning_rate": 5.733903053721072e-06,
+      "loss": 123.1538,
+      "step": 62840
+    },
+    {
+      "epoch": 0.5199156222856434,
+      "grad_norm": 1146.695068359375,
+      "learning_rate": 5.732507933055311e-06,
+      "loss": 133.2361,
+      "step": 62850
+    },
+    {
+      "epoch": 0.5199983455350126,
+      "grad_norm": 830.4008178710938,
+      "learning_rate": 5.731112754107257e-06,
+      "loss": 92.5354,
+      "step": 62860
+    },
+    {
+      "epoch": 0.5200810687843819,
+      "grad_norm": 1141.0968017578125,
+      "learning_rate": 5.729717516987916e-06,
+      "loss": 91.7082,
+      "step": 62870
+    },
+    {
+      "epoch": 0.5201637920337511,
+      "grad_norm": 860.5159301757812,
+      "learning_rate": 5.7283222218083e-06,
+      "loss": 84.899,
+      "step": 62880
+    },
+    {
+      "epoch": 0.5202465152831203,
+      "grad_norm": 632.0242309570312,
+      "learning_rate": 5.726926868679429e-06,
+      "loss": 127.0738,
+      "step": 62890
+    },
+    {
+      "epoch": 0.5203292385324896,
+      "grad_norm": 877.5741577148438,
+      "learning_rate": 5.725531457712321e-06,
+      "loss": 98.0383,
+      "step": 62900
+    },
+    {
+      "epoch": 0.5204119617818588,
+      "grad_norm": 955.6021728515625,
+      "learning_rate": 5.724135989018007e-06,
+      "loss": 93.8473,
+      "step": 62910
+    },
+    {
+      "epoch": 0.520494685031228,
+      "grad_norm": 1114.900634765625,
+      "learning_rate": 5.722740462707515e-06,
+      "loss": 97.0482,
+      "step": 62920
+    },
+    {
+      "epoch": 0.5205774082805973,
+      "grad_norm": 1169.9739990234375,
+      "learning_rate": 5.72134487889188e-06,
+      "loss": 75.5919,
+      "step": 62930
+    },
+    {
+      "epoch": 0.5206601315299665,
+      "grad_norm": 1042.204833984375,
+      "learning_rate": 5.719949237682145e-06,
+      "loss": 96.2548,
+      "step": 62940
+    },
+    {
+      "epoch": 0.5207428547793357,
+      "grad_norm": 923.940185546875,
+      "learning_rate": 5.718553539189353e-06,
+      "loss": 120.747,
+      "step": 62950
+    },
+    {
+      "epoch": 0.520825578028705,
+      "grad_norm": 773.2394409179688,
+      "learning_rate": 5.717157783524553e-06,
+      "loss": 107.0883,
+      "step": 62960
+    },
+    {
+      "epoch": 0.5209083012780742,
+      "grad_norm": 1147.978759765625,
+      "learning_rate": 5.7157619707988e-06,
+      "loss": 112.3523,
+      "step": 62970
+    },
+    {
+      "epoch": 0.5209910245274434,
+      "grad_norm": 869.9029541015625,
+      "learning_rate": 5.714366101123152e-06,
+      "loss": 91.1115,
+      "step": 62980
+    },
+    {
+      "epoch": 0.5210737477768127,
+      "grad_norm": 503.1574401855469,
+      "learning_rate": 5.712970174608671e-06,
+      "loss": 62.8698,
+      "step": 62990
+    },
+    {
+      "epoch": 0.5211564710261819,
+      "grad_norm": 567.345947265625,
+      "learning_rate": 5.711574191366427e-06,
+      "loss": 87.9423,
+      "step": 63000
+    },
+    {
+      "epoch": 0.5212391942755511,
+      "grad_norm": 706.41748046875,
+      "learning_rate": 5.710178151507488e-06,
+      "loss": 73.9817,
+      "step": 63010
+    },
+    {
+      "epoch": 0.5213219175249204,
+      "grad_norm": 1259.374755859375,
+      "learning_rate": 5.708782055142934e-06,
+      "loss": 91.3289,
+      "step": 63020
+    },
+    {
+      "epoch": 0.5214046407742896,
+      "grad_norm": 1063.552001953125,
+      "learning_rate": 5.707385902383845e-06,
+      "loss": 120.5428,
+      "step": 63030
+    },
+    {
+      "epoch": 0.5214873640236588,
+      "grad_norm": 910.1624145507812,
+      "learning_rate": 5.7059896933413076e-06,
+      "loss": 83.0945,
+      "step": 63040
+    },
+    {
+      "epoch": 0.5215700872730281,
+      "grad_norm": 2553.767333984375,
+      "learning_rate": 5.7045934281264085e-06,
+      "loss": 102.2294,
+      "step": 63050
+    },
+    {
+      "epoch": 0.5216528105223973,
+      "grad_norm": 1032.9830322265625,
+      "learning_rate": 5.7031971068502425e-06,
+      "loss": 98.3443,
+      "step": 63060
+    },
+    {
+      "epoch": 0.5217355337717665,
+      "grad_norm": 837.3251342773438,
+      "learning_rate": 5.701800729623911e-06,
+      "loss": 105.6823,
+      "step": 63070
+    },
+    {
+      "epoch": 0.5218182570211358,
+      "grad_norm": 470.49658203125,
+      "learning_rate": 5.700404296558518e-06,
+      "loss": 94.3055,
+      "step": 63080
+    },
+    {
+      "epoch": 0.521900980270505,
+      "grad_norm": 722.3120727539062,
+      "learning_rate": 5.699007807765169e-06,
+      "loss": 96.7037,
+      "step": 63090
+    },
+    {
+      "epoch": 0.5219837035198742,
+      "grad_norm": 1043.8006591796875,
+      "learning_rate": 5.6976112633549764e-06,
+      "loss": 63.1547,
+      "step": 63100
+    },
+    {
+      "epoch": 0.5220664267692435,
+      "grad_norm": 559.216552734375,
+      "learning_rate": 5.696214663439055e-06,
+      "loss": 89.2568,
+      "step": 63110
+    },
+    {
+      "epoch": 0.5221491500186127,
+      "grad_norm": 1489.68994140625,
+      "learning_rate": 5.694818008128531e-06,
+      "loss": 106.3069,
+      "step": 63120
+    },
+    {
+      "epoch": 0.5222318732679819,
+      "grad_norm": 1087.22802734375,
+      "learning_rate": 5.693421297534526e-06,
+      "loss": 116.6925,
+      "step": 63130
+    },
+    {
+      "epoch": 0.5223145965173512,
+      "grad_norm": 1571.4100341796875,
+      "learning_rate": 5.69202453176817e-06,
+      "loss": 104.0308,
+      "step": 63140
+    },
+    {
+      "epoch": 0.5223973197667204,
+      "grad_norm": 1171.318359375,
+      "learning_rate": 5.6906277109406e-06,
+      "loss": 85.1603,
+      "step": 63150
+    },
+    {
+      "epoch": 0.5224800430160896,
+      "grad_norm": 1105.7423095703125,
+      "learning_rate": 5.689230835162949e-06,
+      "loss": 92.7724,
+      "step": 63160
+    },
+    {
+      "epoch": 0.5225627662654589,
+      "grad_norm": 2501.460693359375,
+      "learning_rate": 5.687833904546367e-06,
+      "loss": 124.3585,
+      "step": 63170
+    },
+    {
+      "epoch": 0.5226454895148281,
+      "grad_norm": 673.11328125,
+      "learning_rate": 5.686436919201996e-06,
+      "loss": 95.2214,
+      "step": 63180
+    },
+    {
+      "epoch": 0.5227282127641973,
+      "grad_norm": 803.44970703125,
+      "learning_rate": 5.68503987924099e-06,
+      "loss": 96.271,
+      "step": 63190
+    },
+    {
+      "epoch": 0.5228109360135667,
+      "grad_norm": 733.796875,
+      "learning_rate": 5.683642784774506e-06,
+      "loss": 94.5968,
+      "step": 63200
+    },
+    {
+      "epoch": 0.5228936592629359,
+      "grad_norm": 893.05615234375,
+      "learning_rate": 5.682245635913701e-06,
+      "loss": 93.2139,
+      "step": 63210
+    },
+    {
+      "epoch": 0.522976382512305,
+      "grad_norm": 806.1270751953125,
+      "learning_rate": 5.680848432769743e-06,
+      "loss": 83.2208,
+      "step": 63220
+    },
+    {
+      "epoch": 0.5230591057616744,
+      "grad_norm": 950.8448486328125,
+      "learning_rate": 5.6794511754538005e-06,
+      "loss": 102.1387,
+      "step": 63230
+    },
+    {
+      "epoch": 0.5231418290110436,
+      "grad_norm": 994.8265991210938,
+      "learning_rate": 5.6780538640770455e-06,
+      "loss": 94.9941,
+      "step": 63240
+    },
+    {
+      "epoch": 0.5232245522604128,
+      "grad_norm": 805.738037109375,
+      "learning_rate": 5.6766564987506564e-06,
+      "loss": 97.3443,
+      "step": 63250
+    },
+    {
+      "epoch": 0.5233072755097821,
+      "grad_norm": 1825.2171630859375,
+      "learning_rate": 5.675259079585816e-06,
+      "loss": 93.6461,
+      "step": 63260
+    },
+    {
+      "epoch": 0.5233899987591513,
+      "grad_norm": 899.5811767578125,
+      "learning_rate": 5.673861606693708e-06,
+      "loss": 124.3475,
+      "step": 63270
+    },
+    {
+      "epoch": 0.5234727220085205,
+      "grad_norm": 738.1337890625,
+      "learning_rate": 5.672464080185526e-06,
+      "loss": 82.6199,
+      "step": 63280
+    },
+    {
+      "epoch": 0.5235554452578898,
+      "grad_norm": 1002.6314086914062,
+      "learning_rate": 5.671066500172462e-06,
+      "loss": 118.9046,
+      "step": 63290
+    },
+    {
+      "epoch": 0.523638168507259,
+      "grad_norm": 644.3387451171875,
+      "learning_rate": 5.669668866765717e-06,
+      "loss": 127.9811,
+      "step": 63300
+    },
+    {
+      "epoch": 0.5237208917566282,
+      "grad_norm": 654.7725219726562,
+      "learning_rate": 5.6682711800764935e-06,
+      "loss": 93.9227,
+      "step": 63310
+    },
+    {
+      "epoch": 0.5238036150059975,
+      "grad_norm": 973.210205078125,
+      "learning_rate": 5.6668734402159994e-06,
+      "loss": 84.845,
+      "step": 63320
+    },
+    {
+      "epoch": 0.5238863382553667,
+      "grad_norm": 1666.18017578125,
+      "learning_rate": 5.6654756472954464e-06,
+      "loss": 133.6506,
+      "step": 63330
+    },
+    {
+      "epoch": 0.5239690615047359,
+      "grad_norm": 1285.848388671875,
+      "learning_rate": 5.66407780142605e-06,
+      "loss": 102.1989,
+      "step": 63340
+    },
+    {
+      "epoch": 0.5240517847541052,
+      "grad_norm": 1405.201904296875,
+      "learning_rate": 5.66267990271903e-06,
+      "loss": 101.3061,
+      "step": 63350
+    },
+    {
+      "epoch": 0.5241345080034744,
+      "grad_norm": 995.1788940429688,
+      "learning_rate": 5.661281951285613e-06,
+      "loss": 80.424,
+      "step": 63360
+    },
+    {
+      "epoch": 0.5242172312528436,
+      "grad_norm": 1280.261962890625,
+      "learning_rate": 5.6598839472370245e-06,
+      "loss": 129.4751,
+      "step": 63370
+    },
+    {
+      "epoch": 0.5242999545022129,
+      "grad_norm": 1111.2105712890625,
+      "learning_rate": 5.6584858906845e-06,
+      "loss": 111.6386,
+      "step": 63380
+    },
+    {
+      "epoch": 0.5243826777515821,
+      "grad_norm": 438.5060119628906,
+      "learning_rate": 5.657087781739274e-06,
+      "loss": 88.3789,
+      "step": 63390
+    },
+    {
+      "epoch": 0.5244654010009513,
+      "grad_norm": 1246.3992919921875,
+      "learning_rate": 5.6556896205125896e-06,
+      "loss": 113.5103,
+      "step": 63400
+    },
+    {
+      "epoch": 0.5245481242503206,
+      "grad_norm": 867.26220703125,
+      "learning_rate": 5.654291407115692e-06,
+      "loss": 90.075,
+      "step": 63410
+    },
+    {
+      "epoch": 0.5246308474996898,
+      "grad_norm": 922.8386840820312,
+      "learning_rate": 5.652893141659829e-06,
+      "loss": 103.2854,
+      "step": 63420
+    },
+    {
+      "epoch": 0.524713570749059,
+      "grad_norm": 1118.366455078125,
+      "learning_rate": 5.651494824256256e-06,
+      "loss": 92.4039,
+      "step": 63430
+    },
+    {
+      "epoch": 0.5247962939984283,
+      "grad_norm": 1036.559326171875,
+      "learning_rate": 5.650096455016227e-06,
+      "loss": 77.7015,
+      "step": 63440
+    },
+    {
+      "epoch": 0.5248790172477975,
+      "grad_norm": 649.5186157226562,
+      "learning_rate": 5.648698034051009e-06,
+      "loss": 93.3135,
+      "step": 63450
+    },
+    {
+      "epoch": 0.5249617404971667,
+      "grad_norm": 685.0455322265625,
+      "learning_rate": 5.647299561471865e-06,
+      "loss": 88.9663,
+      "step": 63460
+    },
+    {
+      "epoch": 0.525044463746536,
+      "grad_norm": 786.32470703125,
+      "learning_rate": 5.645901037390067e-06,
+      "loss": 108.5513,
+      "step": 63470
+    },
+    {
+      "epoch": 0.5251271869959052,
+      "grad_norm": 863.156005859375,
+      "learning_rate": 5.644502461916886e-06,
+      "loss": 86.3591,
+      "step": 63480
+    },
+    {
+      "epoch": 0.5252099102452744,
+      "grad_norm": 911.6554565429688,
+      "learning_rate": 5.643103835163602e-06,
+      "loss": 92.7152,
+      "step": 63490
+    },
+    {
+      "epoch": 0.5252926334946437,
+      "grad_norm": 1143.3475341796875,
+      "learning_rate": 5.641705157241497e-06,
+      "loss": 126.4204,
+      "step": 63500
+    },
+    {
+      "epoch": 0.5253753567440129,
+      "grad_norm": 1113.570556640625,
+      "learning_rate": 5.64030642826186e-06,
+      "loss": 106.2012,
+      "step": 63510
+    },
+    {
+      "epoch": 0.5254580799933821,
+      "grad_norm": 734.0611572265625,
+      "learning_rate": 5.6389076483359774e-06,
+      "loss": 85.6654,
+      "step": 63520
+    },
+    {
+      "epoch": 0.5255408032427513,
+      "grad_norm": 762.94677734375,
+      "learning_rate": 5.637508817575145e-06,
+      "loss": 111.5038,
+      "step": 63530
+    },
+    {
+      "epoch": 0.5256235264921206,
+      "grad_norm": 730.315673828125,
+      "learning_rate": 5.636109936090661e-06,
+      "loss": 118.9952,
+      "step": 63540
+    },
+    {
+      "epoch": 0.5257062497414898,
+      "grad_norm": 1400.8131103515625,
+      "learning_rate": 5.634711003993832e-06,
+      "loss": 109.5317,
+      "step": 63550
+    },
+    {
+      "epoch": 0.525788972990859,
+      "grad_norm": 822.1387329101562,
+      "learning_rate": 5.633312021395959e-06,
+      "loss": 115.5602,
+      "step": 63560
+    },
+    {
+      "epoch": 0.5258716962402283,
+      "grad_norm": 836.3711547851562,
+      "learning_rate": 5.631912988408356e-06,
+      "loss": 110.6894,
+      "step": 63570
+    },
+    {
+      "epoch": 0.5259544194895975,
+      "grad_norm": 709.7254028320312,
+      "learning_rate": 5.630513905142334e-06,
+      "loss": 92.7573,
+      "step": 63580
+    },
+    {
+      "epoch": 0.5260371427389667,
+      "grad_norm": 633.5755004882812,
+      "learning_rate": 5.629114771709217e-06,
+      "loss": 97.9029,
+      "step": 63590
+    },
+    {
+      "epoch": 0.526119865988336,
+      "grad_norm": 1466.9072265625,
+      "learning_rate": 5.627715588220325e-06,
+      "loss": 82.6277,
+      "step": 63600
+    },
+    {
+      "epoch": 0.5262025892377052,
+      "grad_norm": 1547.8734130859375,
+      "learning_rate": 5.626316354786982e-06,
+      "loss": 113.497,
+      "step": 63610
+    },
+    {
+      "epoch": 0.5262853124870744,
+      "grad_norm": 691.7115478515625,
+      "learning_rate": 5.624917071520524e-06,
+      "loss": 99.4881,
+      "step": 63620
+    },
+    {
+      "epoch": 0.5263680357364438,
+      "grad_norm": 1155.3173828125,
+      "learning_rate": 5.62351773853228e-06,
+      "loss": 85.3007,
+      "step": 63630
+    },
+    {
+      "epoch": 0.526450758985813,
+      "grad_norm": 1123.159423828125,
+      "learning_rate": 5.6221183559335935e-06,
+      "loss": 103.121,
+      "step": 63640
+    },
+    {
+      "epoch": 0.5265334822351821,
+      "grad_norm": 1173.2393798828125,
+      "learning_rate": 5.6207189238358025e-06,
+      "loss": 88.9635,
+      "step": 63650
+    },
+    {
+      "epoch": 0.5266162054845515,
+      "grad_norm": 1155.658935546875,
+      "learning_rate": 5.619319442350256e-06,
+      "loss": 100.2915,
+      "step": 63660
+    },
+    {
+      "epoch": 0.5266989287339207,
+      "grad_norm": 1091.1961669921875,
+      "learning_rate": 5.617919911588304e-06,
+      "loss": 92.6392,
+      "step": 63670
+    },
+    {
+      "epoch": 0.5267816519832899,
+      "grad_norm": 922.4959106445312,
+      "learning_rate": 5.616520331661301e-06,
+      "loss": 112.2287,
+      "step": 63680
+    },
+    {
+      "epoch": 0.5268643752326592,
+      "grad_norm": 765.5256958007812,
+      "learning_rate": 5.615120702680604e-06,
+      "loss": 98.105,
+      "step": 63690
+    },
+    {
+      "epoch": 0.5269470984820284,
+      "grad_norm": 1066.0155029296875,
+      "learning_rate": 5.6137210247575754e-06,
+      "loss": 103.8472,
+      "step": 63700
+    },
+    {
+      "epoch": 0.5270298217313976,
+      "grad_norm": 1942.541015625,
+      "learning_rate": 5.6123212980035825e-06,
+      "loss": 109.4006,
+      "step": 63710
+    },
+    {
+      "epoch": 0.5271125449807669,
+      "grad_norm": 926.7212524414062,
+      "learning_rate": 5.610921522529994e-06,
+      "loss": 103.9244,
+      "step": 63720
+    },
+    {
+      "epoch": 0.5271952682301361,
+      "grad_norm": 811.9002685546875,
+      "learning_rate": 5.609521698448183e-06,
+      "loss": 96.8324,
+      "step": 63730
+    },
+    {
+      "epoch": 0.5272779914795053,
+      "grad_norm": 774.9049682617188,
+      "learning_rate": 5.608121825869528e-06,
+      "loss": 173.0452,
+      "step": 63740
+    },
+    {
+      "epoch": 0.5273607147288746,
+      "grad_norm": 1026.793212890625,
+      "learning_rate": 5.60672190490541e-06,
+      "loss": 94.8755,
+      "step": 63750
+    },
+    {
+      "epoch": 0.5274434379782438,
+      "grad_norm": 405.6708679199219,
+      "learning_rate": 5.6053219356672155e-06,
+      "loss": 86.6626,
+      "step": 63760
+    },
+    {
+      "epoch": 0.527526161227613,
+      "grad_norm": 890.8868408203125,
+      "learning_rate": 5.603921918266332e-06,
+      "loss": 87.5938,
+      "step": 63770
+    },
+    {
+      "epoch": 0.5276088844769823,
+      "grad_norm": 737.17236328125,
+      "learning_rate": 5.602521852814152e-06,
+      "loss": 73.9403,
+      "step": 63780
+    },
+    {
+      "epoch": 0.5276916077263515,
+      "grad_norm": 809.0032348632812,
+      "learning_rate": 5.6011217394220755e-06,
+      "loss": 79.7671,
+      "step": 63790
+    },
+    {
+      "epoch": 0.5277743309757207,
+      "grad_norm": 1906.908203125,
+      "learning_rate": 5.599721578201499e-06,
+      "loss": 82.5073,
+      "step": 63800
+    },
+    {
+      "epoch": 0.52785705422509,
+      "grad_norm": 1029.4927978515625,
+      "learning_rate": 5.59832136926383e-06,
+      "loss": 106.927,
+      "step": 63810
+    },
+    {
+      "epoch": 0.5279397774744592,
+      "grad_norm": 864.1856079101562,
+      "learning_rate": 5.5969211127204744e-06,
+      "loss": 100.7081,
+      "step": 63820
+    },
+    {
+      "epoch": 0.5280225007238284,
+      "grad_norm": 958.0577392578125,
+      "learning_rate": 5.595520808682848e-06,
+      "loss": 81.9101,
+      "step": 63830
+    },
+    {
+      "epoch": 0.5281052239731977,
+      "grad_norm": 1057.2666015625,
+      "learning_rate": 5.594120457262361e-06,
+      "loss": 109.2872,
+      "step": 63840
+    },
+    {
+      "epoch": 0.5281879472225669,
+      "grad_norm": 963.5335083007812,
+      "learning_rate": 5.592720058570438e-06,
+      "loss": 95.8562,
+      "step": 63850
+    },
+    {
+      "epoch": 0.5282706704719361,
+      "grad_norm": 841.0330810546875,
+      "learning_rate": 5.591319612718498e-06,
+      "loss": 76.5216,
+      "step": 63860
+    },
+    {
+      "epoch": 0.5283533937213054,
+      "grad_norm": 1039.90771484375,
+      "learning_rate": 5.589919119817971e-06,
+      "loss": 107.0152,
+      "step": 63870
+    },
+    {
+      "epoch": 0.5284361169706746,
+      "grad_norm": 1642.884033203125,
+      "learning_rate": 5.588518579980288e-06,
+      "loss": 140.7489,
+      "step": 63880
+    },
+    {
+      "epoch": 0.5285188402200438,
+      "grad_norm": 1069.137451171875,
+      "learning_rate": 5.587117993316882e-06,
+      "loss": 101.0652,
+      "step": 63890
+    },
+    {
+      "epoch": 0.5286015634694131,
+      "grad_norm": 494.1618957519531,
+      "learning_rate": 5.585717359939192e-06,
+      "loss": 85.4378,
+      "step": 63900
+    },
+    {
+      "epoch": 0.5286842867187823,
+      "grad_norm": 1175.2783203125,
+      "learning_rate": 5.584316679958659e-06,
+      "loss": 103.4738,
+      "step": 63910
+    },
+    {
+      "epoch": 0.5287670099681515,
+      "grad_norm": 927.4609375,
+      "learning_rate": 5.58291595348673e-06,
+      "loss": 104.5927,
+      "step": 63920
+    },
+    {
+      "epoch": 0.5288497332175208,
+      "grad_norm": 832.366455078125,
+      "learning_rate": 5.581515180634853e-06,
+      "loss": 80.5153,
+      "step": 63930
+    },
+    {
+      "epoch": 0.52893245646689,
+      "grad_norm": 1159.866943359375,
+      "learning_rate": 5.580114361514484e-06,
+      "loss": 90.9927,
+      "step": 63940
+    },
+    {
+      "epoch": 0.5290151797162592,
+      "grad_norm": 938.1612548828125,
+      "learning_rate": 5.5787134962370755e-06,
+      "loss": 67.0976,
+      "step": 63950
+    },
+    {
+      "epoch": 0.5290979029656285,
+      "grad_norm": 965.9879150390625,
+      "learning_rate": 5.57731258491409e-06,
+      "loss": 91.4234,
+      "step": 63960
+    },
+    {
+      "epoch": 0.5291806262149977,
+      "grad_norm": 816.1832885742188,
+      "learning_rate": 5.575911627656993e-06,
+      "loss": 87.2834,
+      "step": 63970
+    },
+    {
+      "epoch": 0.5292633494643669,
+      "grad_norm": 424.278564453125,
+      "learning_rate": 5.5745106245772506e-06,
+      "loss": 82.5088,
+      "step": 63980
+    },
+    {
+      "epoch": 0.5293460727137362,
+      "grad_norm": 1383.386962890625,
+      "learning_rate": 5.573109575786334e-06,
+      "loss": 99.203,
+      "step": 63990
+    },
+    {
+      "epoch": 0.5294287959631054,
+      "grad_norm": 1369.8514404296875,
+      "learning_rate": 5.571708481395719e-06,
+      "loss": 87.6147,
+      "step": 64000
+    },
+    {
+      "epoch": 0.5295115192124746,
+      "grad_norm": 679.2786865234375,
+      "learning_rate": 5.570307341516882e-06,
+      "loss": 109.6216,
+      "step": 64010
+    },
+    {
+      "epoch": 0.529594242461844,
+      "grad_norm": 688.9489135742188,
+      "learning_rate": 5.568906156261309e-06,
+      "loss": 78.812,
+      "step": 64020
+    },
+    {
+      "epoch": 0.5296769657112131,
+      "grad_norm": 716.52880859375,
+      "learning_rate": 5.567504925740484e-06,
+      "loss": 85.4848,
+      "step": 64030
+    },
+    {
+      "epoch": 0.5297596889605823,
+      "grad_norm": 1037.4881591796875,
+      "learning_rate": 5.566103650065897e-06,
+      "loss": 98.9546,
+      "step": 64040
+    },
+    {
+      "epoch": 0.5298424122099517,
+      "grad_norm": 835.7286376953125,
+      "learning_rate": 5.564702329349041e-06,
+      "loss": 104.4599,
+      "step": 64050
+    },
+    {
+      "epoch": 0.5299251354593209,
+      "grad_norm": 1021.035888671875,
+      "learning_rate": 5.56330096370141e-06,
+      "loss": 82.6299,
+      "step": 64060
+    },
+    {
+      "epoch": 0.53000785870869,
+      "grad_norm": 469.7624816894531,
+      "learning_rate": 5.561899553234509e-06,
+      "loss": 90.2693,
+      "step": 64070
+    },
+    {
+      "epoch": 0.5300905819580594,
+      "grad_norm": 698.7485961914062,
+      "learning_rate": 5.560498098059838e-06,
+      "loss": 116.0395,
+      "step": 64080
+    },
+    {
+      "epoch": 0.5301733052074286,
+      "grad_norm": 1172.5970458984375,
+      "learning_rate": 5.559096598288906e-06,
+      "loss": 70.6558,
+      "step": 64090
+    },
+    {
+      "epoch": 0.5302560284567978,
+      "grad_norm": 901.9520263671875,
+      "learning_rate": 5.557695054033223e-06,
+      "loss": 85.0805,
+      "step": 64100
+    },
+    {
+      "epoch": 0.5303387517061671,
+      "grad_norm": 1102.83056640625,
+      "learning_rate": 5.556293465404304e-06,
+      "loss": 104.11,
+      "step": 64110
+    },
+    {
+      "epoch": 0.5304214749555363,
+      "grad_norm": 862.3082275390625,
+      "learning_rate": 5.554891832513668e-06,
+      "loss": 69.0226,
+      "step": 64120
+    },
+    {
+      "epoch": 0.5305041982049055,
+      "grad_norm": 850.2838745117188,
+      "learning_rate": 5.553490155472835e-06,
+      "loss": 88.536,
+      "step": 64130
+    },
+    {
+      "epoch": 0.5305869214542748,
+      "grad_norm": 1342.24951171875,
+      "learning_rate": 5.55208843439333e-06,
+      "loss": 83.8514,
+      "step": 64140
+    },
+    {
+      "epoch": 0.530669644703644,
+      "grad_norm": 704.4754028320312,
+      "learning_rate": 5.550686669386683e-06,
+      "loss": 52.0388,
+      "step": 64150
+    },
+    {
+      "epoch": 0.5307523679530132,
+      "grad_norm": 709.6826171875,
+      "learning_rate": 5.549284860564425e-06,
+      "loss": 151.5818,
+      "step": 64160
+    },
+    {
+      "epoch": 0.5308350912023825,
+      "grad_norm": 891.2298583984375,
+      "learning_rate": 5.547883008038091e-06,
+      "loss": 132.4302,
+      "step": 64170
+    },
+    {
+      "epoch": 0.5309178144517517,
+      "grad_norm": 668.7271728515625,
+      "learning_rate": 5.54648111191922e-06,
+      "loss": 102.1614,
+      "step": 64180
+    },
+    {
+      "epoch": 0.5310005377011209,
+      "grad_norm": 1067.2647705078125,
+      "learning_rate": 5.545079172319355e-06,
+      "loss": 157.2434,
+      "step": 64190
+    },
+    {
+      "epoch": 0.5310832609504902,
+      "grad_norm": 731.7130737304688,
+      "learning_rate": 5.543677189350043e-06,
+      "loss": 86.4254,
+      "step": 64200
+    },
+    {
+      "epoch": 0.5311659841998594,
+      "grad_norm": 895.8660888671875,
+      "learning_rate": 5.542275163122831e-06,
+      "loss": 81.3476,
+      "step": 64210
+    },
+    {
+      "epoch": 0.5312487074492286,
+      "grad_norm": 549.8636474609375,
+      "learning_rate": 5.540873093749274e-06,
+      "loss": 91.9869,
+      "step": 64220
+    },
+    {
+      "epoch": 0.5313314306985979,
+      "grad_norm": 711.4556274414062,
+      "learning_rate": 5.539470981340926e-06,
+      "loss": 80.405,
+      "step": 64230
+    },
+    {
+      "epoch": 0.5314141539479671,
+      "grad_norm": 565.5111083984375,
+      "learning_rate": 5.538068826009349e-06,
+      "loss": 95.2233,
+      "step": 64240
+    },
+    {
+      "epoch": 0.5314968771973363,
+      "grad_norm": 1099.053955078125,
+      "learning_rate": 5.536666627866104e-06,
+      "loss": 78.3078,
+      "step": 64250
+    },
+    {
+      "epoch": 0.5315796004467055,
+      "grad_norm": 1019.4757080078125,
+      "learning_rate": 5.53526438702276e-06,
+      "loss": 100.4306,
+      "step": 64260
+    },
+    {
+      "epoch": 0.5316623236960748,
+      "grad_norm": 684.2471313476562,
+      "learning_rate": 5.533862103590883e-06,
+      "loss": 81.8137,
+      "step": 64270
+    },
+    {
+      "epoch": 0.531745046945444,
+      "grad_norm": 1522.2894287109375,
+      "learning_rate": 5.532459777682051e-06,
+      "loss": 85.9613,
+      "step": 64280
+    },
+    {
+      "epoch": 0.5318277701948132,
+      "grad_norm": 403.1702880859375,
+      "learning_rate": 5.5310574094078365e-06,
+      "loss": 84.0117,
+      "step": 64290
+    },
+    {
+      "epoch": 0.5319104934441825,
+      "grad_norm": 679.674072265625,
+      "learning_rate": 5.529654998879821e-06,
+      "loss": 59.2508,
+      "step": 64300
+    },
+    {
+      "epoch": 0.5319932166935517,
+      "grad_norm": 565.5519409179688,
+      "learning_rate": 5.528252546209588e-06,
+      "loss": 76.0347,
+      "step": 64310
+    },
+    {
+      "epoch": 0.5320759399429209,
+      "grad_norm": 939.0302124023438,
+      "learning_rate": 5.526850051508725e-06,
+      "loss": 84.2198,
+      "step": 64320
+    },
+    {
+      "epoch": 0.5321586631922902,
+      "grad_norm": 1065.8779296875,
+      "learning_rate": 5.525447514888822e-06,
+      "loss": 146.9439,
+      "step": 64330
+    },
+    {
+      "epoch": 0.5322413864416594,
+      "grad_norm": 1845.5435791015625,
+      "learning_rate": 5.52404493646147e-06,
+      "loss": 111.1832,
+      "step": 64340
+    },
+    {
+      "epoch": 0.5323241096910286,
+      "grad_norm": 1164.8885498046875,
+      "learning_rate": 5.522642316338268e-06,
+      "loss": 89.9883,
+      "step": 64350
+    },
+    {
+      "epoch": 0.5324068329403979,
+      "grad_norm": 661.2352294921875,
+      "learning_rate": 5.521239654630816e-06,
+      "loss": 92.0326,
+      "step": 64360
+    },
+    {
+      "epoch": 0.5324895561897671,
+      "grad_norm": 995.6377563476562,
+      "learning_rate": 5.519836951450716e-06,
+      "loss": 120.3884,
+      "step": 64370
+    },
+    {
+      "epoch": 0.5325722794391363,
+      "grad_norm": 664.2215576171875,
+      "learning_rate": 5.518434206909577e-06,
+      "loss": 102.107,
+      "step": 64380
+    },
+    {
+      "epoch": 0.5326550026885056,
+      "grad_norm": 934.103759765625,
+      "learning_rate": 5.517031421119006e-06,
+      "loss": 73.7082,
+      "step": 64390
+    },
+    {
+      "epoch": 0.5327377259378748,
+      "grad_norm": 702.5838012695312,
+      "learning_rate": 5.5156285941906175e-06,
+      "loss": 93.3802,
+      "step": 64400
+    },
+    {
+      "epoch": 0.532820449187244,
+      "grad_norm": 627.2533569335938,
+      "learning_rate": 5.51422572623603e-06,
+      "loss": 101.0479,
+      "step": 64410
+    },
+    {
+      "epoch": 0.5329031724366133,
+      "grad_norm": 362.7051696777344,
+      "learning_rate": 5.512822817366859e-06,
+      "loss": 79.3406,
+      "step": 64420
+    },
+    {
+      "epoch": 0.5329858956859825,
+      "grad_norm": 895.7844848632812,
+      "learning_rate": 5.511419867694733e-06,
+      "loss": 91.442,
+      "step": 64430
+    },
+    {
+      "epoch": 0.5330686189353517,
+      "grad_norm": 726.357421875,
+      "learning_rate": 5.510016877331271e-06,
+      "loss": 93.3173,
+      "step": 64440
+    },
+    {
+      "epoch": 0.533151342184721,
+      "grad_norm": 779.744873046875,
+      "learning_rate": 5.50861384638811e-06,
+      "loss": 88.6204,
+      "step": 64450
+    },
+    {
+      "epoch": 0.5332340654340902,
+      "grad_norm": 591.9418334960938,
+      "learning_rate": 5.50721077497688e-06,
+      "loss": 72.3884,
+      "step": 64460
+    },
+    {
+      "epoch": 0.5333167886834594,
+      "grad_norm": 1127.8675537109375,
+      "learning_rate": 5.505807663209215e-06,
+      "loss": 124.7344,
+      "step": 64470
+    },
+    {
+      "epoch": 0.5333995119328288,
+      "grad_norm": 932.3980102539062,
+      "learning_rate": 5.504404511196755e-06,
+      "loss": 89.4976,
+      "step": 64480
+    },
+    {
+      "epoch": 0.533482235182198,
+      "grad_norm": 972.6319580078125,
+      "learning_rate": 5.503001319051142e-06,
+      "loss": 84.1025,
+      "step": 64490
+    },
+    {
+      "epoch": 0.5335649584315671,
+      "grad_norm": 898.1183471679688,
+      "learning_rate": 5.5015980868840254e-06,
+      "loss": 126.5537,
+      "step": 64500
+    },
+    {
+      "epoch": 0.5336476816809365,
+      "grad_norm": 1008.814208984375,
+      "learning_rate": 5.500194814807051e-06,
+      "loss": 94.5552,
+      "step": 64510
+    },
+    {
+      "epoch": 0.5337304049303057,
+      "grad_norm": 1275.499267578125,
+      "learning_rate": 5.498791502931868e-06,
+      "loss": 98.7025,
+      "step": 64520
+    },
+    {
+      "epoch": 0.5338131281796749,
+      "grad_norm": 939.2733154296875,
+      "learning_rate": 5.497388151370136e-06,
+      "loss": 120.4424,
+      "step": 64530
+    },
+    {
+      "epoch": 0.5338958514290442,
+      "grad_norm": 2313.487548828125,
+      "learning_rate": 5.495984760233511e-06,
+      "loss": 104.2078,
+      "step": 64540
+    },
+    {
+      "epoch": 0.5339785746784134,
+      "grad_norm": 2875.86474609375,
+      "learning_rate": 5.494581329633656e-06,
+      "loss": 118.5582,
+      "step": 64550
+    },
+    {
+      "epoch": 0.5340612979277826,
+      "grad_norm": 1355.4534912109375,
+      "learning_rate": 5.493177859682234e-06,
+      "loss": 119.3427,
+      "step": 64560
+    },
+    {
+      "epoch": 0.5341440211771519,
+      "grad_norm": 643.4739990234375,
+      "learning_rate": 5.491774350490912e-06,
+      "loss": 76.2629,
+      "step": 64570
+    },
+    {
+      "epoch": 0.5342267444265211,
+      "grad_norm": 814.0101928710938,
+      "learning_rate": 5.490370802171362e-06,
+      "loss": 102.4891,
+      "step": 64580
+    },
+    {
+      "epoch": 0.5343094676758903,
+      "grad_norm": 653.1177978515625,
+      "learning_rate": 5.488967214835259e-06,
+      "loss": 104.6328,
+      "step": 64590
+    },
+    {
+      "epoch": 0.5343921909252596,
+      "grad_norm": 650.8169555664062,
+      "learning_rate": 5.487563588594278e-06,
+      "loss": 106.7863,
+      "step": 64600
+    },
+    {
+      "epoch": 0.5344749141746288,
+      "grad_norm": 875.9237060546875,
+      "learning_rate": 5.4861599235601e-06,
+      "loss": 78.3624,
+      "step": 64610
+    },
+    {
+      "epoch": 0.534557637423998,
+      "grad_norm": 1083.6778564453125,
+      "learning_rate": 5.484756219844408e-06,
+      "loss": 117.263,
+      "step": 64620
+    },
+    {
+      "epoch": 0.5346403606733673,
+      "grad_norm": 751.8464965820312,
+      "learning_rate": 5.483352477558889e-06,
+      "loss": 113.5279,
+      "step": 64630
+    },
+    {
+      "epoch": 0.5347230839227365,
+      "grad_norm": 1062.31982421875,
+      "learning_rate": 5.48194869681523e-06,
+      "loss": 103.9293,
+      "step": 64640
+    },
+    {
+      "epoch": 0.5348058071721057,
+      "grad_norm": 515.4086303710938,
+      "learning_rate": 5.480544877725127e-06,
+      "loss": 97.1701,
+      "step": 64650
+    },
+    {
+      "epoch": 0.534888530421475,
+      "grad_norm": 725.5648193359375,
+      "learning_rate": 5.479141020400271e-06,
+      "loss": 76.7176,
+      "step": 64660
+    },
+    {
+      "epoch": 0.5349712536708442,
+      "grad_norm": 719.1748657226562,
+      "learning_rate": 5.477737124952366e-06,
+      "loss": 96.5835,
+      "step": 64670
+    },
+    {
+      "epoch": 0.5350539769202134,
+      "grad_norm": 916.6452026367188,
+      "learning_rate": 5.476333191493108e-06,
+      "loss": 85.3995,
+      "step": 64680
+    },
+    {
+      "epoch": 0.5351367001695827,
+      "grad_norm": 436.0997619628906,
+      "learning_rate": 5.474929220134205e-06,
+      "loss": 86.9272,
+      "step": 64690
+    },
+    {
+      "epoch": 0.5352194234189519,
+      "grad_norm": 1398.6942138671875,
+      "learning_rate": 5.473525210987363e-06,
+      "loss": 91.1325,
+      "step": 64700
+    },
+    {
+      "epoch": 0.5353021466683211,
+      "grad_norm": 1058.8551025390625,
+      "learning_rate": 5.472121164164295e-06,
+      "loss": 80.0483,
+      "step": 64710
+    },
+    {
+      "epoch": 0.5353848699176904,
+      "grad_norm": 784.1776123046875,
+      "learning_rate": 5.47071707977671e-06,
+      "loss": 115.0421,
+      "step": 64720
+    },
+    {
+      "epoch": 0.5354675931670596,
+      "grad_norm": 1069.0155029296875,
+      "learning_rate": 5.46931295793633e-06,
+      "loss": 97.4913,
+      "step": 64730
+    },
+    {
+      "epoch": 0.5355503164164288,
+      "grad_norm": 1227.8470458984375,
+      "learning_rate": 5.46790879875487e-06,
+      "loss": 98.9259,
+      "step": 64740
+    },
+    {
+      "epoch": 0.5356330396657981,
+      "grad_norm": 632.8845825195312,
+      "learning_rate": 5.466504602344055e-06,
+      "loss": 74.2294,
+      "step": 64750
+    },
+    {
+      "epoch": 0.5357157629151673,
+      "grad_norm": 1411.2496337890625,
+      "learning_rate": 5.465100368815609e-06,
+      "loss": 109.9609,
+      "step": 64760
+    },
+    {
+      "epoch": 0.5357984861645365,
+      "grad_norm": 819.7656860351562,
+      "learning_rate": 5.463696098281262e-06,
+      "loss": 113.5461,
+      "step": 64770
+    },
+    {
+      "epoch": 0.5358812094139058,
+      "grad_norm": 576.60546875,
+      "learning_rate": 5.462291790852744e-06,
+      "loss": 83.2793,
+      "step": 64780
+    },
+    {
+      "epoch": 0.535963932663275,
+      "grad_norm": 563.931640625,
+      "learning_rate": 5.46088744664179e-06,
+      "loss": 112.9462,
+      "step": 64790
+    },
+    {
+      "epoch": 0.5360466559126442,
+      "grad_norm": 1462.51416015625,
+      "learning_rate": 5.459483065760138e-06,
+      "loss": 93.2692,
+      "step": 64800
+    },
+    {
+      "epoch": 0.5361293791620135,
+      "grad_norm": 717.8792114257812,
+      "learning_rate": 5.458078648319526e-06,
+      "loss": 72.0933,
+      "step": 64810
+    },
+    {
+      "epoch": 0.5362121024113827,
+      "grad_norm": 873.50341796875,
+      "learning_rate": 5.456674194431698e-06,
+      "loss": 112.2229,
+      "step": 64820
+    },
+    {
+      "epoch": 0.5362948256607519,
+      "grad_norm": 710.7631225585938,
+      "learning_rate": 5.455269704208401e-06,
+      "loss": 66.035,
+      "step": 64830
+    },
+    {
+      "epoch": 0.5363775489101212,
+      "grad_norm": 909.28564453125,
+      "learning_rate": 5.453865177761384e-06,
+      "loss": 100.6332,
+      "step": 64840
+    },
+    {
+      "epoch": 0.5364602721594904,
+      "grad_norm": 1102.3038330078125,
+      "learning_rate": 5.4524606152023975e-06,
+      "loss": 100.337,
+      "step": 64850
+    },
+    {
+      "epoch": 0.5365429954088596,
+      "grad_norm": 2068.275390625,
+      "learning_rate": 5.4510560166431935e-06,
+      "loss": 116.3672,
+      "step": 64860
+    },
+    {
+      "epoch": 0.536625718658229,
+      "grad_norm": 770.456787109375,
+      "learning_rate": 5.449651382195535e-06,
+      "loss": 93.339,
+      "step": 64870
+    },
+    {
+      "epoch": 0.5367084419075981,
+      "grad_norm": 614.5863037109375,
+      "learning_rate": 5.448246711971178e-06,
+      "loss": 112.4163,
+      "step": 64880
+    },
+    {
+      "epoch": 0.5367911651569673,
+      "grad_norm": 853.5326538085938,
+      "learning_rate": 5.44684200608189e-06,
+      "loss": 110.5457,
+      "step": 64890
+    },
+    {
+      "epoch": 0.5368738884063367,
+      "grad_norm": 718.2283935546875,
+      "learning_rate": 5.445437264639433e-06,
+      "loss": 94.2823,
+      "step": 64900
+    },
+    {
+      "epoch": 0.5369566116557059,
+      "grad_norm": 723.5383911132812,
+      "learning_rate": 5.444032487755575e-06,
+      "loss": 93.2752,
+      "step": 64910
+    },
+    {
+      "epoch": 0.537039334905075,
+      "grad_norm": 694.3563842773438,
+      "learning_rate": 5.442627675542092e-06,
+      "loss": 79.1228,
+      "step": 64920
+    },
+    {
+      "epoch": 0.5371220581544444,
+      "grad_norm": 1094.6685791015625,
+      "learning_rate": 5.441222828110756e-06,
+      "loss": 97.7163,
+      "step": 64930
+    },
+    {
+      "epoch": 0.5372047814038136,
+      "grad_norm": 809.086669921875,
+      "learning_rate": 5.439817945573345e-06,
+      "loss": 78.5984,
+      "step": 64940
+    },
+    {
+      "epoch": 0.5372875046531828,
+      "grad_norm": 1463.13037109375,
+      "learning_rate": 5.438413028041637e-06,
+      "loss": 137.7098,
+      "step": 64950
+    },
+    {
+      "epoch": 0.5373702279025521,
+      "grad_norm": 686.939697265625,
+      "learning_rate": 5.4370080756274155e-06,
+      "loss": 119.1664,
+      "step": 64960
+    },
+    {
+      "epoch": 0.5374529511519213,
+      "grad_norm": 1107.8900146484375,
+      "learning_rate": 5.435603088442471e-06,
+      "loss": 95.0079,
+      "step": 64970
+    },
+    {
+      "epoch": 0.5375356744012905,
+      "grad_norm": 705.9915161132812,
+      "learning_rate": 5.434198066598585e-06,
+      "loss": 78.4787,
+      "step": 64980
+    },
+    {
+      "epoch": 0.5376183976506597,
+      "grad_norm": 954.3821411132812,
+      "learning_rate": 5.4327930102075525e-06,
+      "loss": 89.3804,
+      "step": 64990
+    },
+    {
+      "epoch": 0.537701120900029,
+      "grad_norm": 947.5150146484375,
+      "learning_rate": 5.431387919381166e-06,
+      "loss": 84.6999,
+      "step": 65000
+    },
+    {
+      "epoch": 0.5377838441493982,
+      "grad_norm": 356.02276611328125,
+      "learning_rate": 5.429982794231221e-06,
+      "loss": 107.2157,
+      "step": 65010
+    },
+    {
+      "epoch": 0.5378665673987674,
+      "grad_norm": 1482.7928466796875,
+      "learning_rate": 5.428577634869521e-06,
+      "loss": 92.7534,
+      "step": 65020
+    },
+    {
+      "epoch": 0.5379492906481367,
+      "grad_norm": 966.0228271484375,
+      "learning_rate": 5.427172441407864e-06,
+      "loss": 112.3402,
+      "step": 65030
+    },
+    {
+      "epoch": 0.5380320138975059,
+      "grad_norm": 688.48583984375,
+      "learning_rate": 5.425767213958057e-06,
+      "loss": 67.1227,
+      "step": 65040
+    },
+    {
+      "epoch": 0.5381147371468751,
+      "grad_norm": 847.5924072265625,
+      "learning_rate": 5.424361952631907e-06,
+      "loss": 125.7834,
+      "step": 65050
+    },
+    {
+      "epoch": 0.5381974603962444,
+      "grad_norm": 2318.634033203125,
+      "learning_rate": 5.422956657541224e-06,
+      "loss": 103.7072,
+      "step": 65060
+    },
+    {
+      "epoch": 0.5382801836456136,
+      "grad_norm": 1350.766357421875,
+      "learning_rate": 5.421551328797821e-06,
+      "loss": 96.6112,
+      "step": 65070
+    },
+    {
+      "epoch": 0.5383629068949828,
+      "grad_norm": 693.9425659179688,
+      "learning_rate": 5.420145966513513e-06,
+      "loss": 59.4491,
+      "step": 65080
+    },
+    {
+      "epoch": 0.5384456301443521,
+      "grad_norm": 1209.9505615234375,
+      "learning_rate": 5.418740570800117e-06,
+      "loss": 100.2799,
+      "step": 65090
+    },
+    {
+      "epoch": 0.5385283533937213,
+      "grad_norm": 918.7635498046875,
+      "learning_rate": 5.4173351417694575e-06,
+      "loss": 98.6169,
+      "step": 65100
+    },
+    {
+      "epoch": 0.5386110766430905,
+      "grad_norm": 681.4931640625,
+      "learning_rate": 5.415929679533356e-06,
+      "loss": 95.7693,
+      "step": 65110
+    },
+    {
+      "epoch": 0.5386937998924598,
+      "grad_norm": 583.2526245117188,
+      "learning_rate": 5.414524184203638e-06,
+      "loss": 91.5053,
+      "step": 65120
+    },
+    {
+      "epoch": 0.538776523141829,
+      "grad_norm": 2256.79736328125,
+      "learning_rate": 5.4131186558921335e-06,
+      "loss": 111.124,
+      "step": 65130
+    },
+    {
+      "epoch": 0.5388592463911982,
+      "grad_norm": 1016.837890625,
+      "learning_rate": 5.411713094710673e-06,
+      "loss": 98.1128,
+      "step": 65140
+    },
+    {
+      "epoch": 0.5389419696405675,
+      "grad_norm": 914.6116943359375,
+      "learning_rate": 5.410307500771092e-06,
+      "loss": 74.292,
+      "step": 65150
+    },
+    {
+      "epoch": 0.5390246928899367,
+      "grad_norm": 1791.11962890625,
+      "learning_rate": 5.4089018741852264e-06,
+      "loss": 103.5447,
+      "step": 65160
+    },
+    {
+      "epoch": 0.5391074161393059,
+      "grad_norm": 1446.0499267578125,
+      "learning_rate": 5.407496215064915e-06,
+      "loss": 91.7238,
+      "step": 65170
+    },
+    {
+      "epoch": 0.5391901393886752,
+      "grad_norm": 865.380859375,
+      "learning_rate": 5.406090523521999e-06,
+      "loss": 127.9576,
+      "step": 65180
+    },
+    {
+      "epoch": 0.5392728626380444,
+      "grad_norm": 1105.864013671875,
+      "learning_rate": 5.404684799668325e-06,
+      "loss": 94.1973,
+      "step": 65190
+    },
+    {
+      "epoch": 0.5393555858874136,
+      "grad_norm": 675.471435546875,
+      "learning_rate": 5.403279043615738e-06,
+      "loss": 110.8662,
+      "step": 65200
+    },
+    {
+      "epoch": 0.5394383091367829,
+      "grad_norm": 0.0,
+      "learning_rate": 5.4018732554760875e-06,
+      "loss": 74.0708,
+      "step": 65210
+    },
+    {
+      "epoch": 0.5395210323861521,
+      "grad_norm": 609.0394897460938,
+      "learning_rate": 5.400467435361227e-06,
+      "loss": 110.9384,
+      "step": 65220
+    },
+    {
+      "epoch": 0.5396037556355213,
+      "grad_norm": 674.0571899414062,
+      "learning_rate": 5.399061583383013e-06,
+      "loss": 129.5516,
+      "step": 65230
+    },
+    {
+      "epoch": 0.5396864788848906,
+      "grad_norm": 1786.3568115234375,
+      "learning_rate": 5.3976556996532965e-06,
+      "loss": 100.5763,
+      "step": 65240
+    },
+    {
+      "epoch": 0.5397692021342598,
+      "grad_norm": 1053.052001953125,
+      "learning_rate": 5.396249784283943e-06,
+      "loss": 79.2251,
+      "step": 65250
+    },
+    {
+      "epoch": 0.539851925383629,
+      "grad_norm": 659.5496215820312,
+      "learning_rate": 5.394843837386812e-06,
+      "loss": 93.9248,
+      "step": 65260
+    },
+    {
+      "epoch": 0.5399346486329983,
+      "grad_norm": 1476.619140625,
+      "learning_rate": 5.39343785907377e-06,
+      "loss": 69.9831,
+      "step": 65270
+    },
+    {
+      "epoch": 0.5400173718823675,
+      "grad_norm": 873.2113647460938,
+      "learning_rate": 5.392031849456683e-06,
+      "loss": 97.2353,
+      "step": 65280
+    },
+    {
+      "epoch": 0.5401000951317367,
+      "grad_norm": 674.3704223632812,
+      "learning_rate": 5.39062580864742e-06,
+      "loss": 83.2568,
+      "step": 65290
+    },
+    {
+      "epoch": 0.540182818381106,
+      "grad_norm": 2080.84765625,
+      "learning_rate": 5.3892197367578535e-06,
+      "loss": 76.2265,
+      "step": 65300
+    },
+    {
+      "epoch": 0.5402655416304752,
+      "grad_norm": 442.13330078125,
+      "learning_rate": 5.38781363389986e-06,
+      "loss": 95.2125,
+      "step": 65310
+    },
+    {
+      "epoch": 0.5403482648798444,
+      "grad_norm": 638.2518920898438,
+      "learning_rate": 5.386407500185316e-06,
+      "loss": 72.7291,
+      "step": 65320
+    },
+    {
+      "epoch": 0.5404309881292138,
+      "grad_norm": 841.8578491210938,
+      "learning_rate": 5.3850013357261e-06,
+      "loss": 85.5195,
+      "step": 65330
+    },
+    {
+      "epoch": 0.540513711378583,
+      "grad_norm": 486.00140380859375,
+      "learning_rate": 5.383595140634093e-06,
+      "loss": 103.8048,
+      "step": 65340
+    },
+    {
+      "epoch": 0.5405964346279521,
+      "grad_norm": 945.3947143554688,
+      "learning_rate": 5.382188915021182e-06,
+      "loss": 105.8814,
+      "step": 65350
+    },
+    {
+      "epoch": 0.5406791578773215,
+      "grad_norm": 620.21435546875,
+      "learning_rate": 5.380782658999256e-06,
+      "loss": 61.2339,
+      "step": 65360
+    },
+    {
+      "epoch": 0.5407618811266907,
+      "grad_norm": 1061.0386962890625,
+      "learning_rate": 5.379376372680199e-06,
+      "loss": 87.4281,
+      "step": 65370
+    },
+    {
+      "epoch": 0.5408446043760599,
+      "grad_norm": 906.3760986328125,
+      "learning_rate": 5.377970056175905e-06,
+      "loss": 96.8989,
+      "step": 65380
+    },
+    {
+      "epoch": 0.5409273276254292,
+      "grad_norm": 621.90283203125,
+      "learning_rate": 5.376563709598267e-06,
+      "loss": 70.7825,
+      "step": 65390
+    },
+    {
+      "epoch": 0.5410100508747984,
+      "grad_norm": 818.9396362304688,
+      "learning_rate": 5.3751573330591855e-06,
+      "loss": 101.6279,
+      "step": 65400
+    },
+    {
+      "epoch": 0.5410927741241676,
+      "grad_norm": 859.1585693359375,
+      "learning_rate": 5.3737509266705555e-06,
+      "loss": 107.53,
+      "step": 65410
+    },
+    {
+      "epoch": 0.5411754973735369,
+      "grad_norm": 423.7478942871094,
+      "learning_rate": 5.37234449054428e-06,
+      "loss": 88.1217,
+      "step": 65420
+    },
+    {
+      "epoch": 0.5412582206229061,
+      "grad_norm": 1409.0037841796875,
+      "learning_rate": 5.370938024792262e-06,
+      "loss": 91.726,
+      "step": 65430
+    },
+    {
+      "epoch": 0.5413409438722753,
+      "grad_norm": 1655.0394287109375,
+      "learning_rate": 5.369531529526406e-06,
+      "loss": 95.8715,
+      "step": 65440
+    },
+    {
+      "epoch": 0.5414236671216446,
+      "grad_norm": 825.59326171875,
+      "learning_rate": 5.3681250048586246e-06,
+      "loss": 93.8956,
+      "step": 65450
+    },
+    {
+      "epoch": 0.5415063903710138,
+      "grad_norm": 1459.590576171875,
+      "learning_rate": 5.366718450900825e-06,
+      "loss": 122.376,
+      "step": 65460
+    },
+    {
+      "epoch": 0.541589113620383,
+      "grad_norm": 586.8179321289062,
+      "learning_rate": 5.365311867764922e-06,
+      "loss": 80.8746,
+      "step": 65470
+    },
+    {
+      "epoch": 0.5416718368697523,
+      "grad_norm": 973.2362060546875,
+      "learning_rate": 5.363905255562828e-06,
+      "loss": 163.7341,
+      "step": 65480
+    },
+    {
+      "epoch": 0.5417545601191215,
+      "grad_norm": 664.8019409179688,
+      "learning_rate": 5.362498614406466e-06,
+      "loss": 145.3861,
+      "step": 65490
+    },
+    {
+      "epoch": 0.5418372833684907,
+      "grad_norm": 776.0678100585938,
+      "learning_rate": 5.361091944407751e-06,
+      "loss": 91.7549,
+      "step": 65500
+    },
+    {
+      "epoch": 0.54192000661786,
+      "grad_norm": 998.9888305664062,
+      "learning_rate": 5.3596852456786075e-06,
+      "loss": 107.9666,
+      "step": 65510
+    },
+    {
+      "epoch": 0.5420027298672292,
+      "grad_norm": 866.9566040039062,
+      "learning_rate": 5.35827851833096e-06,
+      "loss": 93.1625,
+      "step": 65520
+    },
+    {
+      "epoch": 0.5420854531165984,
+      "grad_norm": 372.51800537109375,
+      "learning_rate": 5.356871762476735e-06,
+      "loss": 89.3893,
+      "step": 65530
+    },
+    {
+      "epoch": 0.5421681763659677,
+      "grad_norm": 1260.4573974609375,
+      "learning_rate": 5.355464978227861e-06,
+      "loss": 117.8481,
+      "step": 65540
+    },
+    {
+      "epoch": 0.5422508996153369,
+      "grad_norm": 911.54296875,
+      "learning_rate": 5.354058165696271e-06,
+      "loss": 107.3121,
+      "step": 65550
+    },
+    {
+      "epoch": 0.5423336228647061,
+      "grad_norm": 1118.6298828125,
+      "learning_rate": 5.352651324993897e-06,
+      "loss": 120.4601,
+      "step": 65560
+    },
+    {
+      "epoch": 0.5424163461140754,
+      "grad_norm": 873.2643432617188,
+      "learning_rate": 5.351244456232676e-06,
+      "loss": 133.8039,
+      "step": 65570
+    },
+    {
+      "epoch": 0.5424990693634446,
+      "grad_norm": 803.547119140625,
+      "learning_rate": 5.349837559524546e-06,
+      "loss": 113.3396,
+      "step": 65580
+    },
+    {
+      "epoch": 0.5425817926128138,
+      "grad_norm": 838.561767578125,
+      "learning_rate": 5.3484306349814455e-06,
+      "loss": 106.6709,
+      "step": 65590
+    },
+    {
+      "epoch": 0.5426645158621831,
+      "grad_norm": 929.80029296875,
+      "learning_rate": 5.34702368271532e-06,
+      "loss": 93.5446,
+      "step": 65600
+    },
+    {
+      "epoch": 0.5427472391115523,
+      "grad_norm": 1133.1529541015625,
+      "learning_rate": 5.345616702838111e-06,
+      "loss": 77.5028,
+      "step": 65610
+    },
+    {
+      "epoch": 0.5428299623609215,
+      "grad_norm": 945.4187622070312,
+      "learning_rate": 5.344209695461768e-06,
+      "loss": 85.2792,
+      "step": 65620
+    },
+    {
+      "epoch": 0.5429126856102908,
+      "grad_norm": 798.3004150390625,
+      "learning_rate": 5.3428026606982396e-06,
+      "loss": 75.1708,
+      "step": 65630
+    },
+    {
+      "epoch": 0.54299540885966,
+      "grad_norm": 840.9337158203125,
+      "learning_rate": 5.341395598659477e-06,
+      "loss": 139.3945,
+      "step": 65640
+    },
+    {
+      "epoch": 0.5430781321090292,
+      "grad_norm": 7416.7548828125,
+      "learning_rate": 5.339988509457432e-06,
+      "loss": 101.7888,
+      "step": 65650
+    },
+    {
+      "epoch": 0.5431608553583985,
+      "grad_norm": 914.0125732421875,
+      "learning_rate": 5.338581393204064e-06,
+      "loss": 94.0368,
+      "step": 65660
+    },
+    {
+      "epoch": 0.5432435786077677,
+      "grad_norm": 1436.60888671875,
+      "learning_rate": 5.337174250011326e-06,
+      "loss": 123.4113,
+      "step": 65670
+    },
+    {
+      "epoch": 0.5433263018571369,
+      "grad_norm": 1032.3555908203125,
+      "learning_rate": 5.3357670799911805e-06,
+      "loss": 87.8645,
+      "step": 65680
+    },
+    {
+      "epoch": 0.5434090251065062,
+      "grad_norm": 958.6307983398438,
+      "learning_rate": 5.334359883255591e-06,
+      "loss": 83.8339,
+      "step": 65690
+    },
+    {
+      "epoch": 0.5434917483558754,
+      "grad_norm": 968.6178588867188,
+      "learning_rate": 5.33295265991652e-06,
+      "loss": 102.6183,
+      "step": 65700
+    },
+    {
+      "epoch": 0.5435744716052446,
+      "grad_norm": 601.3204345703125,
+      "learning_rate": 5.331545410085933e-06,
+      "loss": 74.607,
+      "step": 65710
+    },
+    {
+      "epoch": 0.5436571948546138,
+      "grad_norm": 1049.9119873046875,
+      "learning_rate": 5.330138133875799e-06,
+      "loss": 78.9104,
+      "step": 65720
+    },
+    {
+      "epoch": 0.5437399181039831,
+      "grad_norm": 879.0147705078125,
+      "learning_rate": 5.328730831398089e-06,
+      "loss": 92.8446,
+      "step": 65730
+    },
+    {
+      "epoch": 0.5438226413533523,
+      "grad_norm": 1260.738525390625,
+      "learning_rate": 5.3273235027647764e-06,
+      "loss": 81.406,
+      "step": 65740
+    },
+    {
+      "epoch": 0.5439053646027215,
+      "grad_norm": 807.532958984375,
+      "learning_rate": 5.3259161480878354e-06,
+      "loss": 80.0483,
+      "step": 65750
+    },
+    {
+      "epoch": 0.5439880878520909,
+      "grad_norm": 628.903076171875,
+      "learning_rate": 5.324508767479239e-06,
+      "loss": 111.094,
+      "step": 65760
+    },
+    {
+      "epoch": 0.54407081110146,
+      "grad_norm": 1162.60888671875,
+      "learning_rate": 5.323101361050972e-06,
+      "loss": 102.2003,
+      "step": 65770
+    },
+    {
+      "epoch": 0.5441535343508292,
+      "grad_norm": 840.9710693359375,
+      "learning_rate": 5.321693928915012e-06,
+      "loss": 85.5873,
+      "step": 65780
+    },
+    {
+      "epoch": 0.5442362576001986,
+      "grad_norm": 588.482666015625,
+      "learning_rate": 5.320286471183343e-06,
+      "loss": 88.2541,
+      "step": 65790
+    },
+    {
+      "epoch": 0.5443189808495678,
+      "grad_norm": 826.7291259765625,
+      "learning_rate": 5.3188789879679496e-06,
+      "loss": 113.2529,
+      "step": 65800
+    },
+    {
+      "epoch": 0.544401704098937,
+      "grad_norm": 988.3714599609375,
+      "learning_rate": 5.317471479380816e-06,
+      "loss": 92.3975,
+      "step": 65810
+    },
+    {
+      "epoch": 0.5444844273483063,
+      "grad_norm": 1229.006103515625,
+      "learning_rate": 5.3160639455339355e-06,
+      "loss": 99.8077,
+      "step": 65820
+    },
+    {
+      "epoch": 0.5445671505976755,
+      "grad_norm": 558.47802734375,
+      "learning_rate": 5.314656386539298e-06,
+      "loss": 98.0959,
+      "step": 65830
+    },
+    {
+      "epoch": 0.5446498738470447,
+      "grad_norm": 1046.564697265625,
+      "learning_rate": 5.313248802508896e-06,
+      "loss": 94.008,
+      "step": 65840
+    },
+    {
+      "epoch": 0.544732597096414,
+      "grad_norm": 943.0609741210938,
+      "learning_rate": 5.311841193554723e-06,
+      "loss": 94.6098,
+      "step": 65850
+    },
+    {
+      "epoch": 0.5448153203457832,
+      "grad_norm": 465.5363464355469,
+      "learning_rate": 5.310433559788778e-06,
+      "loss": 100.7524,
+      "step": 65860
+    },
+    {
+      "epoch": 0.5448980435951524,
+      "grad_norm": 902.67333984375,
+      "learning_rate": 5.309025901323059e-06,
+      "loss": 67.7871,
+      "step": 65870
+    },
+    {
+      "epoch": 0.5449807668445217,
+      "grad_norm": 641.2943725585938,
+      "learning_rate": 5.307618218269569e-06,
+      "loss": 89.4344,
+      "step": 65880
+    },
+    {
+      "epoch": 0.5450634900938909,
+      "grad_norm": 730.5675048828125,
+      "learning_rate": 5.306210510740307e-06,
+      "loss": 78.7692,
+      "step": 65890
+    },
+    {
+      "epoch": 0.5451462133432601,
+      "grad_norm": 562.7182006835938,
+      "learning_rate": 5.304802778847281e-06,
+      "loss": 62.6107,
+      "step": 65900
+    },
+    {
+      "epoch": 0.5452289365926294,
+      "grad_norm": 776.1548461914062,
+      "learning_rate": 5.303395022702495e-06,
+      "loss": 83.2003,
+      "step": 65910
+    },
+    {
+      "epoch": 0.5453116598419986,
+      "grad_norm": 716.1808471679688,
+      "learning_rate": 5.301987242417963e-06,
+      "loss": 77.2049,
+      "step": 65920
+    },
+    {
+      "epoch": 0.5453943830913678,
+      "grad_norm": 1019.53759765625,
+      "learning_rate": 5.300579438105689e-06,
+      "loss": 108.0206,
+      "step": 65930
+    },
+    {
+      "epoch": 0.5454771063407371,
+      "grad_norm": 559.3324584960938,
+      "learning_rate": 5.29917160987769e-06,
+      "loss": 83.0377,
+      "step": 65940
+    },
+    {
+      "epoch": 0.5455598295901063,
+      "grad_norm": 785.0572509765625,
+      "learning_rate": 5.297763757845979e-06,
+      "loss": 88.7534,
+      "step": 65950
+    },
+    {
+      "epoch": 0.5456425528394755,
+      "grad_norm": 606.4642944335938,
+      "learning_rate": 5.296355882122572e-06,
+      "loss": 82.7635,
+      "step": 65960
+    },
+    {
+      "epoch": 0.5457252760888448,
+      "grad_norm": 1158.042236328125,
+      "learning_rate": 5.294947982819488e-06,
+      "loss": 108.4545,
+      "step": 65970
+    },
+    {
+      "epoch": 0.545807999338214,
+      "grad_norm": 412.3054504394531,
+      "learning_rate": 5.293540060048746e-06,
+      "loss": 116.0526,
+      "step": 65980
+    },
+    {
+      "epoch": 0.5458907225875832,
+      "grad_norm": 661.3358154296875,
+      "learning_rate": 5.292132113922369e-06,
+      "loss": 74.4473,
+      "step": 65990
+    },
+    {
+      "epoch": 0.5459734458369525,
+      "grad_norm": 637.5306396484375,
+      "learning_rate": 5.290724144552379e-06,
+      "loss": 104.4344,
+      "step": 66000
+    },
+    {
+      "epoch": 0.5460561690863217,
+      "grad_norm": 1194.7322998046875,
+      "learning_rate": 5.2893161520508055e-06,
+      "loss": 93.1337,
+      "step": 66010
+    },
+    {
+      "epoch": 0.5461388923356909,
+      "grad_norm": 1273.340576171875,
+      "learning_rate": 5.287908136529671e-06,
+      "loss": 102.8886,
+      "step": 66020
+    },
+    {
+      "epoch": 0.5462216155850602,
+      "grad_norm": 923.6551513671875,
+      "learning_rate": 5.28650009810101e-06,
+      "loss": 125.1421,
+      "step": 66030
+    },
+    {
+      "epoch": 0.5463043388344294,
+      "grad_norm": 996.6854248046875,
+      "learning_rate": 5.28509203687685e-06,
+      "loss": 93.7317,
+      "step": 66040
+    },
+    {
+      "epoch": 0.5463870620837986,
+      "grad_norm": 655.3250122070312,
+      "learning_rate": 5.283683952969224e-06,
+      "loss": 98.5502,
+      "step": 66050
+    },
+    {
+      "epoch": 0.5464697853331679,
+      "grad_norm": 724.341064453125,
+      "learning_rate": 5.282275846490169e-06,
+      "loss": 76.2478,
+      "step": 66060
+    },
+    {
+      "epoch": 0.5465525085825371,
+      "grad_norm": 665.1072998046875,
+      "learning_rate": 5.280867717551719e-06,
+      "loss": 90.0001,
+      "step": 66070
+    },
+    {
+      "epoch": 0.5466352318319063,
+      "grad_norm": 774.87451171875,
+      "learning_rate": 5.279459566265915e-06,
+      "loss": 83.016,
+      "step": 66080
+    },
+    {
+      "epoch": 0.5467179550812756,
+      "grad_norm": 1207.277587890625,
+      "learning_rate": 5.278051392744796e-06,
+      "loss": 115.3367,
+      "step": 66090
+    },
+    {
+      "epoch": 0.5468006783306448,
+      "grad_norm": 846.2992553710938,
+      "learning_rate": 5.2766431971004025e-06,
+      "loss": 76.7401,
+      "step": 66100
+    },
+    {
+      "epoch": 0.546883401580014,
+      "grad_norm": 794.3187866210938,
+      "learning_rate": 5.275234979444781e-06,
+      "loss": 104.4632,
+      "step": 66110
+    },
+    {
+      "epoch": 0.5469661248293833,
+      "grad_norm": 609.81787109375,
+      "learning_rate": 5.273826739889975e-06,
+      "loss": 93.3402,
+      "step": 66120
+    },
+    {
+      "epoch": 0.5470488480787525,
+      "grad_norm": 818.2113647460938,
+      "learning_rate": 5.272418478548031e-06,
+      "loss": 77.1904,
+      "step": 66130
+    },
+    {
+      "epoch": 0.5471315713281217,
+      "grad_norm": 2090.50390625,
+      "learning_rate": 5.271010195530999e-06,
+      "loss": 101.5279,
+      "step": 66140
+    },
+    {
+      "epoch": 0.547214294577491,
+      "grad_norm": 1038.06103515625,
+      "learning_rate": 5.26960189095093e-06,
+      "loss": 95.1865,
+      "step": 66150
+    },
+    {
+      "epoch": 0.5472970178268602,
+      "grad_norm": 984.3367309570312,
+      "learning_rate": 5.268193564919876e-06,
+      "loss": 91.2734,
+      "step": 66160
+    },
+    {
+      "epoch": 0.5473797410762294,
+      "grad_norm": 1194.6197509765625,
+      "learning_rate": 5.26678521754989e-06,
+      "loss": 116.3886,
+      "step": 66170
+    },
+    {
+      "epoch": 0.5474624643255988,
+      "grad_norm": 978.132080078125,
+      "learning_rate": 5.265376848953031e-06,
+      "loss": 114.3536,
+      "step": 66180
+    },
+    {
+      "epoch": 0.547545187574968,
+      "grad_norm": 1035.0242919921875,
+      "learning_rate": 5.263968459241351e-06,
+      "loss": 89.8195,
+      "step": 66190
+    },
+    {
+      "epoch": 0.5476279108243371,
+      "grad_norm": 1031.2353515625,
+      "learning_rate": 5.262560048526913e-06,
+      "loss": 80.5158,
+      "step": 66200
+    },
+    {
+      "epoch": 0.5477106340737065,
+      "grad_norm": 308.54156494140625,
+      "learning_rate": 5.261151616921778e-06,
+      "loss": 95.6141,
+      "step": 66210
+    },
+    {
+      "epoch": 0.5477933573230757,
+      "grad_norm": 1179.9146728515625,
+      "learning_rate": 5.259743164538008e-06,
+      "loss": 115.8417,
+      "step": 66220
+    },
+    {
+      "epoch": 0.5478760805724449,
+      "grad_norm": 1147.329345703125,
+      "learning_rate": 5.2583346914876655e-06,
+      "loss": 81.3131,
+      "step": 66230
+    },
+    {
+      "epoch": 0.5479588038218142,
+      "grad_norm": 823.7957153320312,
+      "learning_rate": 5.2569261978828155e-06,
+      "loss": 151.2819,
+      "step": 66240
+    },
+    {
+      "epoch": 0.5480415270711834,
+      "grad_norm": 466.2208557128906,
+      "learning_rate": 5.255517683835528e-06,
+      "loss": 84.0521,
+      "step": 66250
+    },
+    {
+      "epoch": 0.5481242503205526,
+      "grad_norm": 1068.723876953125,
+      "learning_rate": 5.254109149457873e-06,
+      "loss": 93.1827,
+      "step": 66260
+    },
+    {
+      "epoch": 0.5482069735699219,
+      "grad_norm": 1123.1505126953125,
+      "learning_rate": 5.252700594861918e-06,
+      "loss": 107.1272,
+      "step": 66270
+    },
+    {
+      "epoch": 0.5482896968192911,
+      "grad_norm": 1002.3335571289062,
+      "learning_rate": 5.251292020159736e-06,
+      "loss": 88.9466,
+      "step": 66280
+    },
+    {
+      "epoch": 0.5483724200686603,
+      "grad_norm": 658.9024047851562,
+      "learning_rate": 5.2498834254634005e-06,
+      "loss": 87.6717,
+      "step": 66290
+    },
+    {
+      "epoch": 0.5484551433180296,
+      "grad_norm": 1128.9300537109375,
+      "learning_rate": 5.248474810884988e-06,
+      "loss": 105.05,
+      "step": 66300
+    },
+    {
+      "epoch": 0.5485378665673988,
+      "grad_norm": 853.0302734375,
+      "learning_rate": 5.247066176536577e-06,
+      "loss": 136.7064,
+      "step": 66310
+    },
+    {
+      "epoch": 0.548620589816768,
+      "grad_norm": 1292.0511474609375,
+      "learning_rate": 5.245657522530243e-06,
+      "loss": 88.9545,
+      "step": 66320
+    },
+    {
+      "epoch": 0.5487033130661373,
+      "grad_norm": 1117.95947265625,
+      "learning_rate": 5.244248848978067e-06,
+      "loss": 74.3471,
+      "step": 66330
+    },
+    {
+      "epoch": 0.5487860363155065,
+      "grad_norm": 1050.1473388671875,
+      "learning_rate": 5.242840155992131e-06,
+      "loss": 104.2194,
+      "step": 66340
+    },
+    {
+      "epoch": 0.5488687595648757,
+      "grad_norm": 462.470458984375,
+      "learning_rate": 5.24143144368452e-06,
+      "loss": 78.9187,
+      "step": 66350
+    },
+    {
+      "epoch": 0.548951482814245,
+      "grad_norm": 1257.4759521484375,
+      "learning_rate": 5.240022712167315e-06,
+      "loss": 83.4133,
+      "step": 66360
+    },
+    {
+      "epoch": 0.5490342060636142,
+      "grad_norm": 1386.3408203125,
+      "learning_rate": 5.2386139615526046e-06,
+      "loss": 116.1773,
+      "step": 66370
+    },
+    {
+      "epoch": 0.5491169293129834,
+      "grad_norm": 742.0828247070312,
+      "learning_rate": 5.237205191952477e-06,
+      "loss": 102.3286,
+      "step": 66380
+    },
+    {
+      "epoch": 0.5491996525623527,
+      "grad_norm": 1096.52392578125,
+      "learning_rate": 5.235796403479021e-06,
+      "loss": 80.9894,
+      "step": 66390
+    },
+    {
+      "epoch": 0.5492823758117219,
+      "grad_norm": 466.41204833984375,
+      "learning_rate": 5.2343875962443255e-06,
+      "loss": 75.4198,
+      "step": 66400
+    },
+    {
+      "epoch": 0.5493650990610911,
+      "grad_norm": 1372.77685546875,
+      "learning_rate": 5.2329787703604875e-06,
+      "loss": 86.9037,
+      "step": 66410
+    },
+    {
+      "epoch": 0.5494478223104604,
+      "grad_norm": 733.8797607421875,
+      "learning_rate": 5.231569925939596e-06,
+      "loss": 84.0945,
+      "step": 66420
+    },
+    {
+      "epoch": 0.5495305455598296,
+      "grad_norm": 815.2667236328125,
+      "learning_rate": 5.230161063093749e-06,
+      "loss": 83.4252,
+      "step": 66430
+    },
+    {
+      "epoch": 0.5496132688091988,
+      "grad_norm": 1322.669921875,
+      "learning_rate": 5.228752181935042e-06,
+      "loss": 100.3188,
+      "step": 66440
+    },
+    {
+      "epoch": 0.549695992058568,
+      "grad_norm": 1326.2945556640625,
+      "learning_rate": 5.227343282575574e-06,
+      "loss": 90.2418,
+      "step": 66450
+    },
+    {
+      "epoch": 0.5497787153079373,
+      "grad_norm": 1287.527099609375,
+      "learning_rate": 5.225934365127445e-06,
+      "loss": 82.3157,
+      "step": 66460
+    },
+    {
+      "epoch": 0.5498614385573065,
+      "grad_norm": 564.8834228515625,
+      "learning_rate": 5.224525429702755e-06,
+      "loss": 67.5519,
+      "step": 66470
+    },
+    {
+      "epoch": 0.5499441618066757,
+      "grad_norm": 1590.8048095703125,
+      "learning_rate": 5.223116476413606e-06,
+      "loss": 145.7727,
+      "step": 66480
+    },
+    {
+      "epoch": 0.550026885056045,
+      "grad_norm": 424.3537292480469,
+      "learning_rate": 5.221707505372105e-06,
+      "loss": 89.2432,
+      "step": 66490
+    },
+    {
+      "epoch": 0.5501096083054142,
+      "grad_norm": 562.13330078125,
+      "learning_rate": 5.220298516690353e-06,
+      "loss": 91.3321,
+      "step": 66500
+    },
+    {
+      "epoch": 0.5501923315547834,
+      "grad_norm": 416.3981018066406,
+      "learning_rate": 5.21888951048046e-06,
+      "loss": 114.0917,
+      "step": 66510
+    },
+    {
+      "epoch": 0.5502750548041527,
+      "grad_norm": 935.5563354492188,
+      "learning_rate": 5.217480486854534e-06,
+      "loss": 112.0527,
+      "step": 66520
+    },
+    {
+      "epoch": 0.5503577780535219,
+      "grad_norm": 448.1026916503906,
+      "learning_rate": 5.216071445924683e-06,
+      "loss": 71.1966,
+      "step": 66530
+    },
+    {
+      "epoch": 0.5504405013028911,
+      "grad_norm": 715.922119140625,
+      "learning_rate": 5.214662387803019e-06,
+      "loss": 69.2816,
+      "step": 66540
+    },
+    {
+      "epoch": 0.5505232245522604,
+      "grad_norm": 816.540771484375,
+      "learning_rate": 5.213253312601654e-06,
+      "loss": 115.1411,
+      "step": 66550
+    },
+    {
+      "epoch": 0.5506059478016296,
+      "grad_norm": 613.8909912109375,
+      "learning_rate": 5.211844220432702e-06,
+      "loss": 74.3888,
+      "step": 66560
+    },
+    {
+      "epoch": 0.5506886710509988,
+      "grad_norm": 1052.91064453125,
+      "learning_rate": 5.210435111408276e-06,
+      "loss": 91.0357,
+      "step": 66570
+    },
+    {
+      "epoch": 0.5507713943003681,
+      "grad_norm": 1757.2353515625,
+      "learning_rate": 5.209025985640496e-06,
+      "loss": 101.9064,
+      "step": 66580
+    },
+    {
+      "epoch": 0.5508541175497373,
+      "grad_norm": 1139.6336669921875,
+      "learning_rate": 5.207616843241476e-06,
+      "loss": 102.2799,
+      "step": 66590
+    },
+    {
+      "epoch": 0.5509368407991065,
+      "grad_norm": 1313.9873046875,
+      "learning_rate": 5.206207684323337e-06,
+      "loss": 93.977,
+      "step": 66600
+    },
+    {
+      "epoch": 0.5510195640484759,
+      "grad_norm": 574.5587158203125,
+      "learning_rate": 5.2047985089982e-06,
+      "loss": 88.6549,
+      "step": 66610
+    },
+    {
+      "epoch": 0.551102287297845,
+      "grad_norm": 881.9164428710938,
+      "learning_rate": 5.203389317378183e-06,
+      "loss": 108.2257,
+      "step": 66620
+    },
+    {
+      "epoch": 0.5511850105472142,
+      "grad_norm": 702.6726684570312,
+      "learning_rate": 5.201980109575414e-06,
+      "loss": 102.8949,
+      "step": 66630
+    },
+    {
+      "epoch": 0.5512677337965836,
+      "grad_norm": 975.09375,
+      "learning_rate": 5.200570885702013e-06,
+      "loss": 95.4903,
+      "step": 66640
+    },
+    {
+      "epoch": 0.5513504570459528,
+      "grad_norm": 502.3410339355469,
+      "learning_rate": 5.19916164587011e-06,
+      "loss": 116.9168,
+      "step": 66650
+    },
+    {
+      "epoch": 0.551433180295322,
+      "grad_norm": 2496.73193359375,
+      "learning_rate": 5.197752390191827e-06,
+      "loss": 95.1221,
+      "step": 66660
+    },
+    {
+      "epoch": 0.5515159035446913,
+      "grad_norm": 1161.61572265625,
+      "learning_rate": 5.196343118779292e-06,
+      "loss": 103.2815,
+      "step": 66670
+    },
+    {
+      "epoch": 0.5515986267940605,
+      "grad_norm": 501.6972961425781,
+      "learning_rate": 5.194933831744637e-06,
+      "loss": 89.6646,
+      "step": 66680
+    },
+    {
+      "epoch": 0.5516813500434297,
+      "grad_norm": 808.2440795898438,
+      "learning_rate": 5.1935245291999945e-06,
+      "loss": 132.3228,
+      "step": 66690
+    },
+    {
+      "epoch": 0.551764073292799,
+      "grad_norm": 602.8800048828125,
+      "learning_rate": 5.192115211257491e-06,
+      "loss": 100.7651,
+      "step": 66700
+    },
+    {
+      "epoch": 0.5518467965421682,
+      "grad_norm": 705.0828247070312,
+      "learning_rate": 5.19070587802926e-06,
+      "loss": 94.9037,
+      "step": 66710
+    },
+    {
+      "epoch": 0.5519295197915374,
+      "grad_norm": 685.058349609375,
+      "learning_rate": 5.189296529627441e-06,
+      "loss": 101.691,
+      "step": 66720
+    },
+    {
+      "epoch": 0.5520122430409067,
+      "grad_norm": 765.4905395507812,
+      "learning_rate": 5.187887166164165e-06,
+      "loss": 89.7552,
+      "step": 66730
+    },
+    {
+      "epoch": 0.5520949662902759,
+      "grad_norm": 695.8983154296875,
+      "learning_rate": 5.186477787751569e-06,
+      "loss": 77.7547,
+      "step": 66740
+    },
+    {
+      "epoch": 0.5521776895396451,
+      "grad_norm": 633.330810546875,
+      "learning_rate": 5.185068394501791e-06,
+      "loss": 93.5725,
+      "step": 66750
+    },
+    {
+      "epoch": 0.5522604127890144,
+      "grad_norm": 566.3089599609375,
+      "learning_rate": 5.183658986526969e-06,
+      "loss": 86.7824,
+      "step": 66760
+    },
+    {
+      "epoch": 0.5523431360383836,
+      "grad_norm": 870.3569946289062,
+      "learning_rate": 5.1822495639392465e-06,
+      "loss": 137.4197,
+      "step": 66770
+    },
+    {
+      "epoch": 0.5524258592877528,
+      "grad_norm": 1036.458251953125,
+      "learning_rate": 5.180840126850764e-06,
+      "loss": 80.6906,
+      "step": 66780
+    },
+    {
+      "epoch": 0.5525085825371221,
+      "grad_norm": 792.98779296875,
+      "learning_rate": 5.179430675373659e-06,
+      "loss": 91.3037,
+      "step": 66790
+    },
+    {
+      "epoch": 0.5525913057864913,
+      "grad_norm": 1176.4154052734375,
+      "learning_rate": 5.17802120962008e-06,
+      "loss": 129.9155,
+      "step": 66800
+    },
+    {
+      "epoch": 0.5526740290358605,
+      "grad_norm": 1051.582275390625,
+      "learning_rate": 5.17661172970217e-06,
+      "loss": 73.2745,
+      "step": 66810
+    },
+    {
+      "epoch": 0.5527567522852298,
+      "grad_norm": 1047.6094970703125,
+      "learning_rate": 5.175202235732077e-06,
+      "loss": 99.0051,
+      "step": 66820
+    },
+    {
+      "epoch": 0.552839475534599,
+      "grad_norm": 871.4730834960938,
+      "learning_rate": 5.1737927278219446e-06,
+      "loss": 86.3209,
+      "step": 66830
+    },
+    {
+      "epoch": 0.5529221987839682,
+      "grad_norm": 644.1160888671875,
+      "learning_rate": 5.1723832060839216e-06,
+      "loss": 90.1693,
+      "step": 66840
+    },
+    {
+      "epoch": 0.5530049220333375,
+      "grad_norm": 688.15576171875,
+      "learning_rate": 5.170973670630159e-06,
+      "loss": 126.0678,
+      "step": 66850
+    },
+    {
+      "epoch": 0.5530876452827067,
+      "grad_norm": 737.1720581054688,
+      "learning_rate": 5.169564121572806e-06,
+      "loss": 83.4195,
+      "step": 66860
+    },
+    {
+      "epoch": 0.5531703685320759,
+      "grad_norm": 719.3773803710938,
+      "learning_rate": 5.168154559024014e-06,
+      "loss": 91.4589,
+      "step": 66870
+    },
+    {
+      "epoch": 0.5532530917814452,
+      "grad_norm": 1081.61572265625,
+      "learning_rate": 5.166744983095937e-06,
+      "loss": 115.1463,
+      "step": 66880
+    },
+    {
+      "epoch": 0.5533358150308144,
+      "grad_norm": 863.379150390625,
+      "learning_rate": 5.165335393900726e-06,
+      "loss": 91.273,
+      "step": 66890
+    },
+    {
+      "epoch": 0.5534185382801836,
+      "grad_norm": 1196.47509765625,
+      "learning_rate": 5.163925791550536e-06,
+      "loss": 74.5763,
+      "step": 66900
+    },
+    {
+      "epoch": 0.5535012615295529,
+      "grad_norm": 754.5050659179688,
+      "learning_rate": 5.162516176157523e-06,
+      "loss": 78.6246,
+      "step": 66910
+    },
+    {
+      "epoch": 0.5535839847789221,
+      "grad_norm": 332.3321228027344,
+      "learning_rate": 5.161106547833843e-06,
+      "loss": 78.1392,
+      "step": 66920
+    },
+    {
+      "epoch": 0.5536667080282913,
+      "grad_norm": 459.50701904296875,
+      "learning_rate": 5.159696906691656e-06,
+      "loss": 78.0986,
+      "step": 66930
+    },
+    {
+      "epoch": 0.5537494312776606,
+      "grad_norm": 1238.329833984375,
+      "learning_rate": 5.158287252843118e-06,
+      "loss": 105.3073,
+      "step": 66940
+    },
+    {
+      "epoch": 0.5538321545270298,
+      "grad_norm": 885.58251953125,
+      "learning_rate": 5.1568775864003894e-06,
+      "loss": 102.8519,
+      "step": 66950
+    },
+    {
+      "epoch": 0.553914877776399,
+      "grad_norm": 634.6365966796875,
+      "learning_rate": 5.155467907475632e-06,
+      "loss": 114.2997,
+      "step": 66960
+    },
+    {
+      "epoch": 0.5539976010257683,
+      "grad_norm": 957.9181518554688,
+      "learning_rate": 5.154058216181007e-06,
+      "loss": 86.0915,
+      "step": 66970
+    },
+    {
+      "epoch": 0.5540803242751375,
+      "grad_norm": 1111.42919921875,
+      "learning_rate": 5.1526485126286766e-06,
+      "loss": 98.9156,
+      "step": 66980
+    },
+    {
+      "epoch": 0.5541630475245067,
+      "grad_norm": 1166.4410400390625,
+      "learning_rate": 5.151238796930804e-06,
+      "loss": 125.5735,
+      "step": 66990
+    },
+    {
+      "epoch": 0.554245770773876,
+      "grad_norm": 650.7413330078125,
+      "learning_rate": 5.149829069199555e-06,
+      "loss": 83.7539,
+      "step": 67000
+    },
+    {
+      "epoch": 0.5543284940232452,
+      "grad_norm": 654.6986694335938,
+      "learning_rate": 5.148419329547094e-06,
+      "loss": 85.525,
+      "step": 67010
+    },
+    {
+      "epoch": 0.5544112172726144,
+      "grad_norm": 713.3673706054688,
+      "learning_rate": 5.147009578085589e-06,
+      "loss": 96.1334,
+      "step": 67020
+    },
+    {
+      "epoch": 0.5544939405219838,
+      "grad_norm": 1134.079345703125,
+      "learning_rate": 5.145599814927205e-06,
+      "loss": 106.5868,
+      "step": 67030
+    },
+    {
+      "epoch": 0.554576663771353,
+      "grad_norm": 782.4326171875,
+      "learning_rate": 5.144190040184114e-06,
+      "loss": 81.4991,
+      "step": 67040
+    },
+    {
+      "epoch": 0.5546593870207221,
+      "grad_norm": 885.5360107421875,
+      "learning_rate": 5.142780253968481e-06,
+      "loss": 106.0122,
+      "step": 67050
+    },
+    {
+      "epoch": 0.5547421102700915,
+      "grad_norm": 761.8356323242188,
+      "learning_rate": 5.14137045639248e-06,
+      "loss": 106.5318,
+      "step": 67060
+    },
+    {
+      "epoch": 0.5548248335194607,
+      "grad_norm": 796.8743896484375,
+      "learning_rate": 5.13996064756828e-06,
+      "loss": 70.0908,
+      "step": 67070
+    },
+    {
+      "epoch": 0.5549075567688299,
+      "grad_norm": 1602.5904541015625,
+      "learning_rate": 5.138550827608055e-06,
+      "loss": 97.1062,
+      "step": 67080
+    },
+    {
+      "epoch": 0.5549902800181992,
+      "grad_norm": 1278.1387939453125,
+      "learning_rate": 5.137140996623975e-06,
+      "loss": 85.0867,
+      "step": 67090
+    },
+    {
+      "epoch": 0.5550730032675684,
+      "grad_norm": 1092.7744140625,
+      "learning_rate": 5.135731154728215e-06,
+      "loss": 95.7226,
+      "step": 67100
+    },
+    {
+      "epoch": 0.5551557265169376,
+      "grad_norm": 897.9187622070312,
+      "learning_rate": 5.134321302032951e-06,
+      "loss": 118.3043,
+      "step": 67110
+    },
+    {
+      "epoch": 0.5552384497663069,
+      "grad_norm": 1308.584716796875,
+      "learning_rate": 5.1329114386503585e-06,
+      "loss": 136.3629,
+      "step": 67120
+    },
+    {
+      "epoch": 0.5553211730156761,
+      "grad_norm": 1024.6846923828125,
+      "learning_rate": 5.131501564692611e-06,
+      "loss": 121.1511,
+      "step": 67130
+    },
+    {
+      "epoch": 0.5554038962650453,
+      "grad_norm": 894.513671875,
+      "learning_rate": 5.130091680271887e-06,
+      "loss": 98.0504,
+      "step": 67140
+    },
+    {
+      "epoch": 0.5554866195144146,
+      "grad_norm": 1355.4970703125,
+      "learning_rate": 5.128681785500365e-06,
+      "loss": 91.5048,
+      "step": 67150
+    },
+    {
+      "epoch": 0.5555693427637838,
+      "grad_norm": 660.9675903320312,
+      "learning_rate": 5.127271880490227e-06,
+      "loss": 107.9526,
+      "step": 67160
+    },
+    {
+      "epoch": 0.555652066013153,
+      "grad_norm": 639.190185546875,
+      "learning_rate": 5.125861965353647e-06,
+      "loss": 95.762,
+      "step": 67170
+    },
+    {
+      "epoch": 0.5557347892625222,
+      "grad_norm": 1066.5399169921875,
+      "learning_rate": 5.124452040202809e-06,
+      "loss": 72.5684,
+      "step": 67180
+    },
+    {
+      "epoch": 0.5558175125118915,
+      "grad_norm": 1595.59033203125,
+      "learning_rate": 5.1230421051498914e-06,
+      "loss": 101.4106,
+      "step": 67190
+    },
+    {
+      "epoch": 0.5559002357612607,
+      "grad_norm": 700.05419921875,
+      "learning_rate": 5.121632160307078e-06,
+      "loss": 110.3626,
+      "step": 67200
+    },
+    {
+      "epoch": 0.5559829590106299,
+      "grad_norm": 1288.469482421875,
+      "learning_rate": 5.120222205786556e-06,
+      "loss": 134.359,
+      "step": 67210
+    },
+    {
+      "epoch": 0.5560656822599992,
+      "grad_norm": 1454.6519775390625,
+      "learning_rate": 5.118812241700501e-06,
+      "loss": 92.2554,
+      "step": 67220
+    },
+    {
+      "epoch": 0.5561484055093684,
+      "grad_norm": 937.5780029296875,
+      "learning_rate": 5.117402268161103e-06,
+      "loss": 78.8586,
+      "step": 67230
+    },
+    {
+      "epoch": 0.5562311287587376,
+      "grad_norm": 1452.914306640625,
+      "learning_rate": 5.115992285280543e-06,
+      "loss": 90.6466,
+      "step": 67240
+    },
+    {
+      "epoch": 0.5563138520081069,
+      "grad_norm": 808.0802612304688,
+      "learning_rate": 5.114582293171012e-06,
+      "loss": 84.3542,
+      "step": 67250
+    },
+    {
+      "epoch": 0.5563965752574761,
+      "grad_norm": 1291.392333984375,
+      "learning_rate": 5.113172291944693e-06,
+      "loss": 80.4222,
+      "step": 67260
+    },
+    {
+      "epoch": 0.5564792985068453,
+      "grad_norm": 0.0,
+      "learning_rate": 5.111762281713773e-06,
+      "loss": 96.9373,
+      "step": 67270
+    },
+    {
+      "epoch": 0.5565620217562146,
+      "grad_norm": 460.95623779296875,
+      "learning_rate": 5.110352262590442e-06,
+      "loss": 98.4892,
+      "step": 67280
+    },
+    {
+      "epoch": 0.5566447450055838,
+      "grad_norm": 921.4959106445312,
+      "learning_rate": 5.108942234686889e-06,
+      "loss": 85.5142,
+      "step": 67290
+    },
+    {
+      "epoch": 0.556727468254953,
+      "grad_norm": 909.4285278320312,
+      "learning_rate": 5.1075321981153014e-06,
+      "loss": 79.041,
+      "step": 67300
+    },
+    {
+      "epoch": 0.5568101915043223,
+      "grad_norm": 786.9864501953125,
+      "learning_rate": 5.106122152987869e-06,
+      "loss": 131.3443,
+      "step": 67310
+    },
+    {
+      "epoch": 0.5568929147536915,
+      "grad_norm": 837.8712158203125,
+      "learning_rate": 5.1047120994167855e-06,
+      "loss": 111.7526,
+      "step": 67320
+    },
+    {
+      "epoch": 0.5569756380030607,
+      "grad_norm": 748.6630859375,
+      "learning_rate": 5.103302037514241e-06,
+      "loss": 83.1395,
+      "step": 67330
+    },
+    {
+      "epoch": 0.55705836125243,
+      "grad_norm": 422.8838195800781,
+      "learning_rate": 5.101891967392426e-06,
+      "loss": 132.004,
+      "step": 67340
+    },
+    {
+      "epoch": 0.5571410845017992,
+      "grad_norm": 784.1287231445312,
+      "learning_rate": 5.100481889163535e-06,
+      "loss": 84.0244,
+      "step": 67350
+    },
+    {
+      "epoch": 0.5572238077511684,
+      "grad_norm": 871.6853637695312,
+      "learning_rate": 5.099071802939763e-06,
+      "loss": 96.8194,
+      "step": 67360
+    },
+    {
+      "epoch": 0.5573065310005377,
+      "grad_norm": 1200.704833984375,
+      "learning_rate": 5.097661708833302e-06,
+      "loss": 101.7852,
+      "step": 67370
+    },
+    {
+      "epoch": 0.5573892542499069,
+      "grad_norm": 605.1507568359375,
+      "learning_rate": 5.096251606956345e-06,
+      "loss": 114.6154,
+      "step": 67380
+    },
+    {
+      "epoch": 0.5574719774992761,
+      "grad_norm": 1629.7506103515625,
+      "learning_rate": 5.0948414974210906e-06,
+      "loss": 108.8764,
+      "step": 67390
+    },
+    {
+      "epoch": 0.5575547007486454,
+      "grad_norm": 963.1193237304688,
+      "learning_rate": 5.093431380339734e-06,
+      "loss": 83.0829,
+      "step": 67400
+    },
+    {
+      "epoch": 0.5576374239980146,
+      "grad_norm": 984.2836303710938,
+      "learning_rate": 5.092021255824471e-06,
+      "loss": 90.3172,
+      "step": 67410
+    },
+    {
+      "epoch": 0.5577201472473838,
+      "grad_norm": 846.0245971679688,
+      "learning_rate": 5.090611123987498e-06,
+      "loss": 75.3123,
+      "step": 67420
+    },
+    {
+      "epoch": 0.5578028704967531,
+      "grad_norm": 880.2362670898438,
+      "learning_rate": 5.089200984941014e-06,
+      "loss": 86.5572,
+      "step": 67430
+    },
+    {
+      "epoch": 0.5578855937461223,
+      "grad_norm": 1226.6094970703125,
+      "learning_rate": 5.087790838797217e-06,
+      "loss": 76.6975,
+      "step": 67440
+    },
+    {
+      "epoch": 0.5579683169954915,
+      "grad_norm": 1174.342529296875,
+      "learning_rate": 5.0863806856683076e-06,
+      "loss": 97.9613,
+      "step": 67450
+    },
+    {
+      "epoch": 0.5580510402448609,
+      "grad_norm": 832.6563110351562,
+      "learning_rate": 5.084970525666481e-06,
+      "loss": 77.5357,
+      "step": 67460
+    },
+    {
+      "epoch": 0.55813376349423,
+      "grad_norm": 1207.7735595703125,
+      "learning_rate": 5.083560358903942e-06,
+      "loss": 108.6904,
+      "step": 67470
+    },
+    {
+      "epoch": 0.5582164867435992,
+      "grad_norm": 1580.60791015625,
+      "learning_rate": 5.082150185492887e-06,
+      "loss": 123.0784,
+      "step": 67480
+    },
+    {
+      "epoch": 0.5582992099929686,
+      "grad_norm": 911.670654296875,
+      "learning_rate": 5.080740005545519e-06,
+      "loss": 81.2963,
+      "step": 67490
+    },
+    {
+      "epoch": 0.5583819332423378,
+      "grad_norm": 1346.0203857421875,
+      "learning_rate": 5.07932981917404e-06,
+      "loss": 105.0922,
+      "step": 67500
+    },
+    {
+      "epoch": 0.558464656491707,
+      "grad_norm": 1385.7469482421875,
+      "learning_rate": 5.077919626490651e-06,
+      "loss": 111.0049,
+      "step": 67510
+    },
+    {
+      "epoch": 0.5585473797410763,
+      "grad_norm": 2228.890380859375,
+      "learning_rate": 5.076509427607555e-06,
+      "loss": 89.5055,
+      "step": 67520
+    },
+    {
+      "epoch": 0.5586301029904455,
+      "grad_norm": 507.3185119628906,
+      "learning_rate": 5.075099222636954e-06,
+      "loss": 78.911,
+      "step": 67530
+    },
+    {
+      "epoch": 0.5587128262398147,
+      "grad_norm": 551.4053955078125,
+      "learning_rate": 5.073689011691054e-06,
+      "loss": 150.896,
+      "step": 67540
+    },
+    {
+      "epoch": 0.558795549489184,
+      "grad_norm": 350.7704162597656,
+      "learning_rate": 5.072278794882058e-06,
+      "loss": 78.5772,
+      "step": 67550
+    },
+    {
+      "epoch": 0.5588782727385532,
+      "grad_norm": 682.6192626953125,
+      "learning_rate": 5.07086857232217e-06,
+      "loss": 93.3245,
+      "step": 67560
+    },
+    {
+      "epoch": 0.5589609959879224,
+      "grad_norm": 699.5905151367188,
+      "learning_rate": 5.069458344123592e-06,
+      "loss": 109.2068,
+      "step": 67570
+    },
+    {
+      "epoch": 0.5590437192372917,
+      "grad_norm": 1194.6055908203125,
+      "learning_rate": 5.068048110398535e-06,
+      "loss": 103.4686,
+      "step": 67580
+    },
+    {
+      "epoch": 0.5591264424866609,
+      "grad_norm": 671.2708740234375,
+      "learning_rate": 5.066637871259201e-06,
+      "loss": 69.1066,
+      "step": 67590
+    },
+    {
+      "epoch": 0.5592091657360301,
+      "grad_norm": 817.7105102539062,
+      "learning_rate": 5.065227626817798e-06,
+      "loss": 105.476,
+      "step": 67600
+    },
+    {
+      "epoch": 0.5592918889853994,
+      "grad_norm": 664.9489135742188,
+      "learning_rate": 5.063817377186531e-06,
+      "loss": 67.6632,
+      "step": 67610
+    },
+    {
+      "epoch": 0.5593746122347686,
+      "grad_norm": 1241.629638671875,
+      "learning_rate": 5.062407122477609e-06,
+      "loss": 87.1826,
+      "step": 67620
+    },
+    {
+      "epoch": 0.5594573354841378,
+      "grad_norm": 1491.961181640625,
+      "learning_rate": 5.060996862803239e-06,
+      "loss": 78.7851,
+      "step": 67630
+    },
+    {
+      "epoch": 0.5595400587335071,
+      "grad_norm": 981.352294921875,
+      "learning_rate": 5.0595865982756284e-06,
+      "loss": 106.6009,
+      "step": 67640
+    },
+    {
+      "epoch": 0.5596227819828763,
+      "grad_norm": 695.9772338867188,
+      "learning_rate": 5.0581763290069865e-06,
+      "loss": 85.9514,
+      "step": 67650
+    },
+    {
+      "epoch": 0.5597055052322455,
+      "grad_norm": 1229.713134765625,
+      "learning_rate": 5.05676605510952e-06,
+      "loss": 105.0807,
+      "step": 67660
+    },
+    {
+      "epoch": 0.5597882284816148,
+      "grad_norm": 534.1005859375,
+      "learning_rate": 5.055355776695437e-06,
+      "loss": 93.8244,
+      "step": 67670
+    },
+    {
+      "epoch": 0.559870951730984,
+      "grad_norm": 911.0359497070312,
+      "learning_rate": 5.0539454938769525e-06,
+      "loss": 101.3288,
+      "step": 67680
+    },
+    {
+      "epoch": 0.5599536749803532,
+      "grad_norm": 918.2195434570312,
+      "learning_rate": 5.052535206766271e-06,
+      "loss": 89.4984,
+      "step": 67690
+    },
+    {
+      "epoch": 0.5600363982297225,
+      "grad_norm": 1059.2137451171875,
+      "learning_rate": 5.051124915475604e-06,
+      "loss": 99.4896,
+      "step": 67700
+    },
+    {
+      "epoch": 0.5601191214790917,
+      "grad_norm": 1157.1102294921875,
+      "learning_rate": 5.049714620117162e-06,
+      "loss": 105.2878,
+      "step": 67710
+    },
+    {
+      "epoch": 0.5602018447284609,
+      "grad_norm": 666.695068359375,
+      "learning_rate": 5.0483043208031575e-06,
+      "loss": 109.8217,
+      "step": 67720
+    },
+    {
+      "epoch": 0.5602845679778302,
+      "grad_norm": 1238.229736328125,
+      "learning_rate": 5.0468940176458e-06,
+      "loss": 94.0387,
+      "step": 67730
+    },
+    {
+      "epoch": 0.5603672912271994,
+      "grad_norm": 2138.0751953125,
+      "learning_rate": 5.045483710757298e-06,
+      "loss": 95.2673,
+      "step": 67740
+    },
+    {
+      "epoch": 0.5604500144765686,
+      "grad_norm": 914.434326171875,
+      "learning_rate": 5.044073400249867e-06,
+      "loss": 92.1393,
+      "step": 67750
+    },
+    {
+      "epoch": 0.5605327377259379,
+      "grad_norm": 1934.4398193359375,
+      "learning_rate": 5.0426630862357176e-06,
+      "loss": 142.6514,
+      "step": 67760
+    },
+    {
+      "epoch": 0.5606154609753071,
+      "grad_norm": 1084.5147705078125,
+      "learning_rate": 5.041252768827064e-06,
+      "loss": 133.2949,
+      "step": 67770
+    },
+    {
+      "epoch": 0.5606981842246763,
+      "grad_norm": 748.2398071289062,
+      "learning_rate": 5.039842448136115e-06,
+      "loss": 69.7304,
+      "step": 67780
+    },
+    {
+      "epoch": 0.5607809074740456,
+      "grad_norm": 804.4089965820312,
+      "learning_rate": 5.038432124275087e-06,
+      "loss": 88.9984,
+      "step": 67790
+    },
+    {
+      "epoch": 0.5608636307234148,
+      "grad_norm": 535.5092163085938,
+      "learning_rate": 5.03702179735619e-06,
+      "loss": 88.9917,
+      "step": 67800
+    },
+    {
+      "epoch": 0.560946353972784,
+      "grad_norm": 635.5907592773438,
+      "learning_rate": 5.035611467491638e-06,
+      "loss": 92.2593,
+      "step": 67810
+    },
+    {
+      "epoch": 0.5610290772221533,
+      "grad_norm": 851.3953247070312,
+      "learning_rate": 5.034201134793646e-06,
+      "loss": 126.7755,
+      "step": 67820
+    },
+    {
+      "epoch": 0.5611118004715225,
+      "grad_norm": 780.0504760742188,
+      "learning_rate": 5.032790799374426e-06,
+      "loss": 131.121,
+      "step": 67830
+    },
+    {
+      "epoch": 0.5611945237208917,
+      "grad_norm": 1291.3529052734375,
+      "learning_rate": 5.0313804613461925e-06,
+      "loss": 110.4924,
+      "step": 67840
+    },
+    {
+      "epoch": 0.561277246970261,
+      "grad_norm": 860.24072265625,
+      "learning_rate": 5.0299701208211605e-06,
+      "loss": 83.46,
+      "step": 67850
+    },
+    {
+      "epoch": 0.5613599702196302,
+      "grad_norm": 1199.1927490234375,
+      "learning_rate": 5.028559777911543e-06,
+      "loss": 107.3534,
+      "step": 67860
+    },
+    {
+      "epoch": 0.5614426934689994,
+      "grad_norm": 613.6046752929688,
+      "learning_rate": 5.027149432729555e-06,
+      "loss": 88.3334,
+      "step": 67870
+    },
+    {
+      "epoch": 0.5615254167183688,
+      "grad_norm": 1034.6842041015625,
+      "learning_rate": 5.025739085387411e-06,
+      "loss": 106.7181,
+      "step": 67880
+    },
+    {
+      "epoch": 0.561608139967738,
+      "grad_norm": 968.4409790039062,
+      "learning_rate": 5.024328735997327e-06,
+      "loss": 73.4767,
+      "step": 67890
+    },
+    {
+      "epoch": 0.5616908632171072,
+      "grad_norm": 636.9981689453125,
+      "learning_rate": 5.0229183846715154e-06,
+      "loss": 80.4445,
+      "step": 67900
+    },
+    {
+      "epoch": 0.5617735864664763,
+      "grad_norm": 762.04931640625,
+      "learning_rate": 5.021508031522195e-06,
+      "loss": 106.2328,
+      "step": 67910
+    },
+    {
+      "epoch": 0.5618563097158457,
+      "grad_norm": 1687.2122802734375,
+      "learning_rate": 5.0200976766615785e-06,
+      "loss": 103.8904,
+      "step": 67920
+    },
+    {
+      "epoch": 0.5619390329652149,
+      "grad_norm": 1144.8087158203125,
+      "learning_rate": 5.018687320201882e-06,
+      "loss": 88.8534,
+      "step": 67930
+    },
+    {
+      "epoch": 0.5620217562145841,
+      "grad_norm": 1060.1658935546875,
+      "learning_rate": 5.017276962255323e-06,
+      "loss": 117.6395,
+      "step": 67940
+    },
+    {
+      "epoch": 0.5621044794639534,
+      "grad_norm": 541.5599975585938,
+      "learning_rate": 5.015866602934112e-06,
+      "loss": 81.2837,
+      "step": 67950
+    },
+    {
+      "epoch": 0.5621872027133226,
+      "grad_norm": 1171.819091796875,
+      "learning_rate": 5.01445624235047e-06,
+      "loss": 122.8472,
+      "step": 67960
+    },
+    {
+      "epoch": 0.5622699259626918,
+      "grad_norm": 569.9471435546875,
+      "learning_rate": 5.013045880616612e-06,
+      "loss": 99.2828,
+      "step": 67970
+    },
+    {
+      "epoch": 0.5623526492120611,
+      "grad_norm": 857.0740356445312,
+      "learning_rate": 5.011635517844753e-06,
+      "loss": 79.8719,
+      "step": 67980
+    },
+    {
+      "epoch": 0.5624353724614303,
+      "grad_norm": 1649.099853515625,
+      "learning_rate": 5.010225154147107e-06,
+      "loss": 116.2859,
+      "step": 67990
+    },
+    {
+      "epoch": 0.5625180957107995,
+      "grad_norm": 991.4071044921875,
+      "learning_rate": 5.008814789635894e-06,
+      "loss": 84.7862,
+      "step": 68000
+    },
+    {
+      "epoch": 0.5626008189601688,
+      "grad_norm": 752.6597900390625,
+      "learning_rate": 5.007404424423329e-06,
+      "loss": 98.8501,
+      "step": 68010
+    },
+    {
+      "epoch": 0.562683542209538,
+      "grad_norm": 720.8966064453125,
+      "learning_rate": 5.0059940586216284e-06,
+      "loss": 105.6753,
+      "step": 68020
+    },
+    {
+      "epoch": 0.5627662654589072,
+      "grad_norm": 720.338134765625,
+      "learning_rate": 5.004583692343007e-06,
+      "loss": 104.9102,
+      "step": 68030
+    },
+    {
+      "epoch": 0.5628489887082765,
+      "grad_norm": 3147.158447265625,
+      "learning_rate": 5.003173325699682e-06,
+      "loss": 115.8981,
+      "step": 68040
+    },
+    {
+      "epoch": 0.5629317119576457,
+      "grad_norm": 649.7725830078125,
+      "learning_rate": 5.00176295880387e-06,
+      "loss": 99.7788,
+      "step": 68050
+    },
+    {
+      "epoch": 0.5630144352070149,
+      "grad_norm": 523.4342651367188,
+      "learning_rate": 5.000352591767787e-06,
+      "loss": 82.1472,
+      "step": 68060
+    },
+    {
+      "epoch": 0.5630971584563842,
+      "grad_norm": 704.3622436523438,
+      "learning_rate": 4.998942224703651e-06,
+      "loss": 95.6916,
+      "step": 68070
+    },
+    {
+      "epoch": 0.5631798817057534,
+      "grad_norm": 487.66796875,
+      "learning_rate": 4.997531857723678e-06,
+      "loss": 77.8491,
+      "step": 68080
+    },
+    {
+      "epoch": 0.5632626049551226,
+      "grad_norm": 959.2645874023438,
+      "learning_rate": 4.996121490940084e-06,
+      "loss": 106.0061,
+      "step": 68090
+    },
+    {
+      "epoch": 0.5633453282044919,
+      "grad_norm": 1130.561279296875,
+      "learning_rate": 4.994711124465084e-06,
+      "loss": 86.5281,
+      "step": 68100
+    },
+    {
+      "epoch": 0.5634280514538611,
+      "grad_norm": 675.5780639648438,
+      "learning_rate": 4.993300758410895e-06,
+      "loss": 110.0674,
+      "step": 68110
+    },
+    {
+      "epoch": 0.5635107747032303,
+      "grad_norm": 833.6633911132812,
+      "learning_rate": 4.991890392889735e-06,
+      "loss": 93.0616,
+      "step": 68120
+    },
+    {
+      "epoch": 0.5635934979525996,
+      "grad_norm": 497.2891845703125,
+      "learning_rate": 4.990480028013818e-06,
+      "loss": 102.913,
+      "step": 68130
+    },
+    {
+      "epoch": 0.5636762212019688,
+      "grad_norm": 976.9085693359375,
+      "learning_rate": 4.989069663895361e-06,
+      "loss": 94.7653,
+      "step": 68140
+    },
+    {
+      "epoch": 0.563758944451338,
+      "grad_norm": 1526.7978515625,
+      "learning_rate": 4.9876593006465825e-06,
+      "loss": 74.5033,
+      "step": 68150
+    },
+    {
+      "epoch": 0.5638416677007073,
+      "grad_norm": 903.107177734375,
+      "learning_rate": 4.986248938379696e-06,
+      "loss": 80.8739,
+      "step": 68160
+    },
+    {
+      "epoch": 0.5639243909500765,
+      "grad_norm": 1482.0433349609375,
+      "learning_rate": 4.984838577206921e-06,
+      "loss": 106.9815,
+      "step": 68170
+    },
+    {
+      "epoch": 0.5640071141994457,
+      "grad_norm": 4656.763671875,
+      "learning_rate": 4.9834282172404665e-06,
+      "loss": 115.4343,
+      "step": 68180
+    },
+    {
+      "epoch": 0.564089837448815,
+      "grad_norm": 661.7854614257812,
+      "learning_rate": 4.982017858592555e-06,
+      "loss": 98.6977,
+      "step": 68190
+    },
+    {
+      "epoch": 0.5641725606981842,
+      "grad_norm": 654.1882934570312,
+      "learning_rate": 4.980607501375399e-06,
+      "loss": 113.0739,
+      "step": 68200
+    },
+    {
+      "epoch": 0.5642552839475534,
+      "grad_norm": 1355.635498046875,
+      "learning_rate": 4.979197145701216e-06,
+      "loss": 125.4591,
+      "step": 68210
+    },
+    {
+      "epoch": 0.5643380071969227,
+      "grad_norm": 793.7406005859375,
+      "learning_rate": 4.977786791682221e-06,
+      "loss": 94.5731,
+      "step": 68220
+    },
+    {
+      "epoch": 0.5644207304462919,
+      "grad_norm": 819.9609985351562,
+      "learning_rate": 4.976376439430627e-06,
+      "loss": 94.1121,
+      "step": 68230
+    },
+    {
+      "epoch": 0.5645034536956611,
+      "grad_norm": 687.6693725585938,
+      "learning_rate": 4.974966089058652e-06,
+      "loss": 91.6705,
+      "step": 68240
+    },
+    {
+      "epoch": 0.5645861769450304,
+      "grad_norm": 645.421875,
+      "learning_rate": 4.973555740678512e-06,
+      "loss": 115.3419,
+      "step": 68250
+    },
+    {
+      "epoch": 0.5646689001943996,
+      "grad_norm": 675.6929931640625,
+      "learning_rate": 4.972145394402421e-06,
+      "loss": 83.4908,
+      "step": 68260
+    },
+    {
+      "epoch": 0.5647516234437688,
+      "grad_norm": 1000.2984008789062,
+      "learning_rate": 4.9707350503425905e-06,
+      "loss": 93.1049,
+      "step": 68270
+    },
+    {
+      "epoch": 0.5648343466931381,
+      "grad_norm": 996.8165283203125,
+      "learning_rate": 4.969324708611239e-06,
+      "loss": 87.6225,
+      "step": 68280
+    },
+    {
+      "epoch": 0.5649170699425073,
+      "grad_norm": 605.3995361328125,
+      "learning_rate": 4.9679143693205785e-06,
+      "loss": 92.8545,
+      "step": 68290
+    },
+    {
+      "epoch": 0.5649997931918765,
+      "grad_norm": 1098.21044921875,
+      "learning_rate": 4.966504032582826e-06,
+      "loss": 129.6778,
+      "step": 68300
+    },
+    {
+      "epoch": 0.5650825164412459,
+      "grad_norm": 822.2824096679688,
+      "learning_rate": 4.965093698510192e-06,
+      "loss": 96.9237,
+      "step": 68310
+    },
+    {
+      "epoch": 0.565165239690615,
+      "grad_norm": 818.29931640625,
+      "learning_rate": 4.963683367214895e-06,
+      "loss": 88.808,
+      "step": 68320
+    },
+    {
+      "epoch": 0.5652479629399842,
+      "grad_norm": 1037.8428955078125,
+      "learning_rate": 4.962273038809143e-06,
+      "loss": 108.7637,
+      "step": 68330
+    },
+    {
+      "epoch": 0.5653306861893536,
+      "grad_norm": 614.8590698242188,
+      "learning_rate": 4.960862713405153e-06,
+      "loss": 89.6115,
+      "step": 68340
+    },
+    {
+      "epoch": 0.5654134094387228,
+      "grad_norm": 389.6104736328125,
+      "learning_rate": 4.95945239111514e-06,
+      "loss": 93.0896,
+      "step": 68350
+    },
+    {
+      "epoch": 0.565496132688092,
+      "grad_norm": 1135.781005859375,
+      "learning_rate": 4.9580420720513115e-06,
+      "loss": 151.4895,
+      "step": 68360
+    },
+    {
+      "epoch": 0.5655788559374613,
+      "grad_norm": 816.8753051757812,
+      "learning_rate": 4.956631756325882e-06,
+      "loss": 98.8892,
+      "step": 68370
+    },
+    {
+      "epoch": 0.5656615791868305,
+      "grad_norm": 782.5609741210938,
+      "learning_rate": 4.955221444051066e-06,
+      "loss": 77.5715,
+      "step": 68380
+    },
+    {
+      "epoch": 0.5657443024361997,
+      "grad_norm": 545.4894409179688,
+      "learning_rate": 4.953811135339073e-06,
+      "loss": 98.956,
+      "step": 68390
+    },
+    {
+      "epoch": 0.565827025685569,
+      "grad_norm": 789.0388793945312,
+      "learning_rate": 4.952400830302117e-06,
+      "loss": 128.0591,
+      "step": 68400
+    },
+    {
+      "epoch": 0.5659097489349382,
+      "grad_norm": 688.3363037109375,
+      "learning_rate": 4.950990529052409e-06,
+      "loss": 79.7653,
+      "step": 68410
+    },
+    {
+      "epoch": 0.5659924721843074,
+      "grad_norm": 1568.041259765625,
+      "learning_rate": 4.949580231702158e-06,
+      "loss": 100.4565,
+      "step": 68420
+    },
+    {
+      "epoch": 0.5660751954336767,
+      "grad_norm": 828.0899658203125,
+      "learning_rate": 4.94816993836358e-06,
+      "loss": 102.3955,
+      "step": 68430
+    },
+    {
+      "epoch": 0.5661579186830459,
+      "grad_norm": 532.2860107421875,
+      "learning_rate": 4.946759649148879e-06,
+      "loss": 77.4463,
+      "step": 68440
+    },
+    {
+      "epoch": 0.5662406419324151,
+      "grad_norm": 925.8582763671875,
+      "learning_rate": 4.945349364170269e-06,
+      "loss": 89.3156,
+      "step": 68450
+    },
+    {
+      "epoch": 0.5663233651817844,
+      "grad_norm": 2182.206787109375,
+      "learning_rate": 4.94393908353996e-06,
+      "loss": 113.8303,
+      "step": 68460
+    },
+    {
+      "epoch": 0.5664060884311536,
+      "grad_norm": 785.1996459960938,
+      "learning_rate": 4.942528807370158e-06,
+      "loss": 98.9256,
+      "step": 68470
+    },
+    {
+      "epoch": 0.5664888116805228,
+      "grad_norm": 636.3436279296875,
+      "learning_rate": 4.941118535773078e-06,
+      "loss": 89.0144,
+      "step": 68480
+    },
+    {
+      "epoch": 0.5665715349298921,
+      "grad_norm": 1138.821044921875,
+      "learning_rate": 4.9397082688609245e-06,
+      "loss": 65.4779,
+      "step": 68490
+    },
+    {
+      "epoch": 0.5666542581792613,
+      "grad_norm": 704.7152099609375,
+      "learning_rate": 4.938298006745909e-06,
+      "loss": 97.2847,
+      "step": 68500
+    },
+    {
+      "epoch": 0.5667369814286305,
+      "grad_norm": 1183.618896484375,
+      "learning_rate": 4.936887749540236e-06,
+      "loss": 85.178,
+      "step": 68510
+    },
+    {
+      "epoch": 0.5668197046779998,
+      "grad_norm": 960.0731201171875,
+      "learning_rate": 4.935477497356118e-06,
+      "loss": 87.2744,
+      "step": 68520
+    },
+    {
+      "epoch": 0.566902427927369,
+      "grad_norm": 1119.224853515625,
+      "learning_rate": 4.934067250305757e-06,
+      "loss": 104.8437,
+      "step": 68530
+    },
+    {
+      "epoch": 0.5669851511767382,
+      "grad_norm": 606.4401245117188,
+      "learning_rate": 4.932657008501362e-06,
+      "loss": 114.7352,
+      "step": 68540
+    },
+    {
+      "epoch": 0.5670678744261075,
+      "grad_norm": 829.0963745117188,
+      "learning_rate": 4.931246772055141e-06,
+      "loss": 112.2022,
+      "step": 68550
+    },
+    {
+      "epoch": 0.5671505976754767,
+      "grad_norm": 1182.5452880859375,
+      "learning_rate": 4.9298365410792985e-06,
+      "loss": 118.4558,
+      "step": 68560
+    },
+    {
+      "epoch": 0.5672333209248459,
+      "grad_norm": 873.345947265625,
+      "learning_rate": 4.928426315686039e-06,
+      "loss": 66.0995,
+      "step": 68570
+    },
+    {
+      "epoch": 0.5673160441742152,
+      "grad_norm": 1000.1868896484375,
+      "learning_rate": 4.92701609598757e-06,
+      "loss": 99.3208,
+      "step": 68580
+    },
+    {
+      "epoch": 0.5673987674235844,
+      "grad_norm": 782.4760131835938,
+      "learning_rate": 4.925605882096096e-06,
+      "loss": 91.0728,
+      "step": 68590
+    },
+    {
+      "epoch": 0.5674814906729536,
+      "grad_norm": 884.9342651367188,
+      "learning_rate": 4.924195674123821e-06,
+      "loss": 89.9195,
+      "step": 68600
+    },
+    {
+      "epoch": 0.5675642139223229,
+      "grad_norm": 813.310546875,
+      "learning_rate": 4.922785472182948e-06,
+      "loss": 72.9367,
+      "step": 68610
+    },
+    {
+      "epoch": 0.5676469371716921,
+      "grad_norm": 566.2291259765625,
+      "learning_rate": 4.92137527638568e-06,
+      "loss": 67.9247,
+      "step": 68620
+    },
+    {
+      "epoch": 0.5677296604210613,
+      "grad_norm": 667.0711059570312,
+      "learning_rate": 4.919965086844221e-06,
+      "loss": 102.4399,
+      "step": 68630
+    },
+    {
+      "epoch": 0.5678123836704305,
+      "grad_norm": 680.4052124023438,
+      "learning_rate": 4.9185549036707715e-06,
+      "loss": 74.7348,
+      "step": 68640
+    },
+    {
+      "epoch": 0.5678951069197998,
+      "grad_norm": 801.6771240234375,
+      "learning_rate": 4.917144726977535e-06,
+      "loss": 124.5145,
+      "step": 68650
+    },
+    {
+      "epoch": 0.567977830169169,
+      "grad_norm": 318.88824462890625,
+      "learning_rate": 4.915734556876713e-06,
+      "loss": 72.6641,
+      "step": 68660
+    },
+    {
+      "epoch": 0.5680605534185382,
+      "grad_norm": 822.7666625976562,
+      "learning_rate": 4.914324393480504e-06,
+      "loss": 89.2944,
+      "step": 68670
+    },
+    {
+      "epoch": 0.5681432766679075,
+      "grad_norm": 956.2263793945312,
+      "learning_rate": 4.9129142369011105e-06,
+      "loss": 106.6015,
+      "step": 68680
+    },
+    {
+      "epoch": 0.5682259999172767,
+      "grad_norm": 811.5741577148438,
+      "learning_rate": 4.911504087250735e-06,
+      "loss": 95.9006,
+      "step": 68690
+    },
+    {
+      "epoch": 0.5683087231666459,
+      "grad_norm": 606.1180419921875,
+      "learning_rate": 4.910093944641569e-06,
+      "loss": 75.9035,
+      "step": 68700
+    },
+    {
+      "epoch": 0.5683914464160152,
+      "grad_norm": 671.3385620117188,
+      "learning_rate": 4.9086838091858155e-06,
+      "loss": 130.6404,
+      "step": 68710
+    },
+    {
+      "epoch": 0.5684741696653844,
+      "grad_norm": 775.1796875,
+      "learning_rate": 4.9072736809956735e-06,
+      "loss": 73.8619,
+      "step": 68720
+    },
+    {
+      "epoch": 0.5685568929147536,
+      "grad_norm": 940.5592651367188,
+      "learning_rate": 4.9058635601833384e-06,
+      "loss": 64.6535,
+      "step": 68730
+    },
+    {
+      "epoch": 0.568639616164123,
+      "grad_norm": 385.8565673828125,
+      "learning_rate": 4.904453446861008e-06,
+      "loss": 84.9902,
+      "step": 68740
+    },
+    {
+      "epoch": 0.5687223394134922,
+      "grad_norm": 1663.734375,
+      "learning_rate": 4.903043341140879e-06,
+      "loss": 100.3515,
+      "step": 68750
+    },
+    {
+      "epoch": 0.5688050626628613,
+      "grad_norm": 715.17626953125,
+      "learning_rate": 4.901633243135144e-06,
+      "loss": 88.3827,
+      "step": 68760
+    },
+    {
+      "epoch": 0.5688877859122307,
+      "grad_norm": 985.6260375976562,
+      "learning_rate": 4.900223152956003e-06,
+      "loss": 103.8463,
+      "step": 68770
+    },
+    {
+      "epoch": 0.5689705091615999,
+      "grad_norm": 981.5747680664062,
+      "learning_rate": 4.898813070715649e-06,
+      "loss": 105.5598,
+      "step": 68780
+    },
+    {
+      "epoch": 0.5690532324109691,
+      "grad_norm": 541.0383911132812,
+      "learning_rate": 4.897402996526273e-06,
+      "loss": 91.828,
+      "step": 68790
+    },
+    {
+      "epoch": 0.5691359556603384,
+      "grad_norm": 378.74261474609375,
+      "learning_rate": 4.895992930500068e-06,
+      "loss": 94.0054,
+      "step": 68800
+    },
+    {
+      "epoch": 0.5692186789097076,
+      "grad_norm": 524.0851440429688,
+      "learning_rate": 4.894582872749229e-06,
+      "loss": 80.3705,
+      "step": 68810
+    },
+    {
+      "epoch": 0.5693014021590768,
+      "grad_norm": 640.1279907226562,
+      "learning_rate": 4.893172823385947e-06,
+      "loss": 74.0081,
+      "step": 68820
+    },
+    {
+      "epoch": 0.5693841254084461,
+      "grad_norm": 1020.3170776367188,
+      "learning_rate": 4.8917627825224146e-06,
+      "loss": 96.6822,
+      "step": 68830
+    },
+    {
+      "epoch": 0.5694668486578153,
+      "grad_norm": 511.52783203125,
+      "learning_rate": 4.89035275027082e-06,
+      "loss": 93.7879,
+      "step": 68840
+    },
+    {
+      "epoch": 0.5695495719071845,
+      "grad_norm": 881.3643188476562,
+      "learning_rate": 4.888942726743353e-06,
+      "loss": 113.874,
+      "step": 68850
+    },
+    {
+      "epoch": 0.5696322951565538,
+      "grad_norm": 638.2870483398438,
+      "learning_rate": 4.887532712052206e-06,
+      "loss": 95.4349,
+      "step": 68860
+    },
+    {
+      "epoch": 0.569715018405923,
+      "grad_norm": 943.398681640625,
+      "learning_rate": 4.886122706309563e-06,
+      "loss": 90.6776,
+      "step": 68870
+    },
+    {
+      "epoch": 0.5697977416552922,
+      "grad_norm": 827.296142578125,
+      "learning_rate": 4.884712709627614e-06,
+      "loss": 106.1661,
+      "step": 68880
+    },
+    {
+      "epoch": 0.5698804649046615,
+      "grad_norm": 942.6498413085938,
+      "learning_rate": 4.8833027221185455e-06,
+      "loss": 101.3283,
+      "step": 68890
+    },
+    {
+      "epoch": 0.5699631881540307,
+      "grad_norm": 805.362060546875,
+      "learning_rate": 4.881892743894543e-06,
+      "loss": 63.2547,
+      "step": 68900
+    },
+    {
+      "epoch": 0.5700459114033999,
+      "grad_norm": 812.5333251953125,
+      "learning_rate": 4.880482775067794e-06,
+      "loss": 80.0272,
+      "step": 68910
+    },
+    {
+      "epoch": 0.5701286346527692,
+      "grad_norm": 1515.912109375,
+      "learning_rate": 4.879072815750481e-06,
+      "loss": 102.9636,
+      "step": 68920
+    },
+    {
+      "epoch": 0.5702113579021384,
+      "grad_norm": 705.251953125,
+      "learning_rate": 4.87766286605479e-06,
+      "loss": 78.7777,
+      "step": 68930
+    },
+    {
+      "epoch": 0.5702940811515076,
+      "grad_norm": 1174.0625,
+      "learning_rate": 4.876252926092903e-06,
+      "loss": 75.163,
+      "step": 68940
+    },
+    {
+      "epoch": 0.5703768044008769,
+      "grad_norm": 731.8579711914062,
+      "learning_rate": 4.874842995977004e-06,
+      "loss": 100.2167,
+      "step": 68950
+    },
+    {
+      "epoch": 0.5704595276502461,
+      "grad_norm": 615.9132690429688,
+      "learning_rate": 4.873433075819272e-06,
+      "loss": 74.5286,
+      "step": 68960
+    },
+    {
+      "epoch": 0.5705422508996153,
+      "grad_norm": 1427.2105712890625,
+      "learning_rate": 4.87202316573189e-06,
+      "loss": 115.191,
+      "step": 68970
+    },
+    {
+      "epoch": 0.5706249741489846,
+      "grad_norm": 464.9676208496094,
+      "learning_rate": 4.870613265827037e-06,
+      "loss": 107.5646,
+      "step": 68980
+    },
+    {
+      "epoch": 0.5707076973983538,
+      "grad_norm": 664.8630981445312,
+      "learning_rate": 4.869203376216891e-06,
+      "loss": 89.4376,
+      "step": 68990
+    },
+    {
+      "epoch": 0.570790420647723,
+      "grad_norm": 952.9496459960938,
+      "learning_rate": 4.867793497013634e-06,
+      "loss": 87.7081,
+      "step": 69000
+    },
+    {
+      "epoch": 0.5708731438970923,
+      "grad_norm": 864.7783203125,
+      "learning_rate": 4.866383628329442e-06,
+      "loss": 103.678,
+      "step": 69010
+    },
+    {
+      "epoch": 0.5709558671464615,
+      "grad_norm": 616.1771850585938,
+      "learning_rate": 4.86497377027649e-06,
+      "loss": 74.9962,
+      "step": 69020
+    },
+    {
+      "epoch": 0.5710385903958307,
+      "grad_norm": 1259.6622314453125,
+      "learning_rate": 4.863563922966957e-06,
+      "loss": 98.8335,
+      "step": 69030
+    },
+    {
+      "epoch": 0.5711213136452,
+      "grad_norm": 562.259765625,
+      "learning_rate": 4.862154086513016e-06,
+      "loss": 79.6309,
+      "step": 69040
+    },
+    {
+      "epoch": 0.5712040368945692,
+      "grad_norm": 941.865966796875,
+      "learning_rate": 4.860744261026841e-06,
+      "loss": 91.4525,
+      "step": 69050
+    },
+    {
+      "epoch": 0.5712867601439384,
+      "grad_norm": 1558.767578125,
+      "learning_rate": 4.8593344466206075e-06,
+      "loss": 102.5667,
+      "step": 69060
+    },
+    {
+      "epoch": 0.5713694833933077,
+      "grad_norm": 677.3783569335938,
+      "learning_rate": 4.857924643406485e-06,
+      "loss": 121.3402,
+      "step": 69070
+    },
+    {
+      "epoch": 0.5714522066426769,
+      "grad_norm": 1102.9102783203125,
+      "learning_rate": 4.856514851496647e-06,
+      "loss": 89.2696,
+      "step": 69080
+    },
+    {
+      "epoch": 0.5715349298920461,
+      "grad_norm": 510.9609375,
+      "learning_rate": 4.8551050710032625e-06,
+      "loss": 120.3483,
+      "step": 69090
+    },
+    {
+      "epoch": 0.5716176531414154,
+      "grad_norm": 624.8123779296875,
+      "learning_rate": 4.853695302038504e-06,
+      "loss": 99.4614,
+      "step": 69100
+    },
+    {
+      "epoch": 0.5717003763907846,
+      "grad_norm": 953.24267578125,
+      "learning_rate": 4.8522855447145385e-06,
+      "loss": 99.8863,
+      "step": 69110
+    },
+    {
+      "epoch": 0.5717830996401538,
+      "grad_norm": 415.95648193359375,
+      "learning_rate": 4.850875799143537e-06,
+      "loss": 79.1049,
+      "step": 69120
+    },
+    {
+      "epoch": 0.5718658228895231,
+      "grad_norm": 1333.311279296875,
+      "learning_rate": 4.84946606543766e-06,
+      "loss": 77.1295,
+      "step": 69130
+    },
+    {
+      "epoch": 0.5719485461388923,
+      "grad_norm": 1068.9039306640625,
+      "learning_rate": 4.848056343709079e-06,
+      "loss": 63.2696,
+      "step": 69140
+    },
+    {
+      "epoch": 0.5720312693882615,
+      "grad_norm": 2413.59814453125,
+      "learning_rate": 4.846646634069957e-06,
+      "loss": 98.2298,
+      "step": 69150
+    },
+    {
+      "epoch": 0.5721139926376309,
+      "grad_norm": 919.4255981445312,
+      "learning_rate": 4.845236936632458e-06,
+      "loss": 105.757,
+      "step": 69160
+    },
+    {
+      "epoch": 0.572196715887,
+      "grad_norm": 1202.3345947265625,
+      "learning_rate": 4.843827251508747e-06,
+      "loss": 101.2527,
+      "step": 69170
+    },
+    {
+      "epoch": 0.5722794391363693,
+      "grad_norm": 699.7549438476562,
+      "learning_rate": 4.842417578810984e-06,
+      "loss": 76.5603,
+      "step": 69180
+    },
+    {
+      "epoch": 0.5723621623857386,
+      "grad_norm": 579.6343383789062,
+      "learning_rate": 4.841007918651329e-06,
+      "loss": 106.7701,
+      "step": 69190
+    },
+    {
+      "epoch": 0.5724448856351078,
+      "grad_norm": 625.5684204101562,
+      "learning_rate": 4.839598271141947e-06,
+      "loss": 91.3166,
+      "step": 69200
+    },
+    {
+      "epoch": 0.572527608884477,
+      "grad_norm": 773.6903076171875,
+      "learning_rate": 4.8381886363949956e-06,
+      "loss": 75.625,
+      "step": 69210
+    },
+    {
+      "epoch": 0.5726103321338463,
+      "grad_norm": 980.545166015625,
+      "learning_rate": 4.83677901452263e-06,
+      "loss": 93.6299,
+      "step": 69220
+    },
+    {
+      "epoch": 0.5726930553832155,
+      "grad_norm": 606.6036987304688,
+      "learning_rate": 4.835369405637009e-06,
+      "loss": 102.3185,
+      "step": 69230
+    },
+    {
+      "epoch": 0.5727757786325847,
+      "grad_norm": 922.7852783203125,
+      "learning_rate": 4.833959809850288e-06,
+      "loss": 89.2718,
+      "step": 69240
+    },
+    {
+      "epoch": 0.572858501881954,
+      "grad_norm": 918.3201293945312,
+      "learning_rate": 4.832550227274624e-06,
+      "loss": 95.1504,
+      "step": 69250
+    },
+    {
+      "epoch": 0.5729412251313232,
+      "grad_norm": 972.9816284179688,
+      "learning_rate": 4.83114065802217e-06,
+      "loss": 106.1438,
+      "step": 69260
+    },
+    {
+      "epoch": 0.5730239483806924,
+      "grad_norm": 1341.721435546875,
+      "learning_rate": 4.829731102205079e-06,
+      "loss": 99.2695,
+      "step": 69270
+    },
+    {
+      "epoch": 0.5731066716300617,
+      "grad_norm": 927.3786010742188,
+      "learning_rate": 4.828321559935502e-06,
+      "loss": 92.5324,
+      "step": 69280
+    },
+    {
+      "epoch": 0.5731893948794309,
+      "grad_norm": 835.2514038085938,
+      "learning_rate": 4.826912031325592e-06,
+      "loss": 79.8504,
+      "step": 69290
+    },
+    {
+      "epoch": 0.5732721181288001,
+      "grad_norm": 480.0574951171875,
+      "learning_rate": 4.825502516487497e-06,
+      "loss": 68.6633,
+      "step": 69300
+    },
+    {
+      "epoch": 0.5733548413781694,
+      "grad_norm": 1109.6693115234375,
+      "learning_rate": 4.824093015533365e-06,
+      "loss": 91.0974,
+      "step": 69310
+    },
+    {
+      "epoch": 0.5734375646275386,
+      "grad_norm": 1023.72314453125,
+      "learning_rate": 4.822683528575344e-06,
+      "loss": 81.7075,
+      "step": 69320
+    },
+    {
+      "epoch": 0.5735202878769078,
+      "grad_norm": 689.5050048828125,
+      "learning_rate": 4.8212740557255815e-06,
+      "loss": 88.6784,
+      "step": 69330
+    },
+    {
+      "epoch": 0.573603011126277,
+      "grad_norm": 864.6280517578125,
+      "learning_rate": 4.819864597096222e-06,
+      "loss": 98.9668,
+      "step": 69340
+    },
+    {
+      "epoch": 0.5736857343756463,
+      "grad_norm": 502.53045654296875,
+      "learning_rate": 4.81845515279941e-06,
+      "loss": 80.9407,
+      "step": 69350
+    },
+    {
+      "epoch": 0.5737684576250155,
+      "grad_norm": 1652.8223876953125,
+      "learning_rate": 4.817045722947288e-06,
+      "loss": 88.5993,
+      "step": 69360
+    },
+    {
+      "epoch": 0.5738511808743847,
+      "grad_norm": 417.58795166015625,
+      "learning_rate": 4.815636307651998e-06,
+      "loss": 64.912,
+      "step": 69370
+    },
+    {
+      "epoch": 0.573933904123754,
+      "grad_norm": 2181.54443359375,
+      "learning_rate": 4.814226907025683e-06,
+      "loss": 93.8224,
+      "step": 69380
+    },
+    {
+      "epoch": 0.5740166273731232,
+      "grad_norm": 961.0506591796875,
+      "learning_rate": 4.812817521180479e-06,
+      "loss": 94.1438,
+      "step": 69390
+    },
+    {
+      "epoch": 0.5740993506224924,
+      "grad_norm": 783.2133178710938,
+      "learning_rate": 4.811408150228526e-06,
+      "loss": 79.3694,
+      "step": 69400
+    },
+    {
+      "epoch": 0.5741820738718617,
+      "grad_norm": 614.4639892578125,
+      "learning_rate": 4.80999879428196e-06,
+      "loss": 96.5799,
+      "step": 69410
+    },
+    {
+      "epoch": 0.5742647971212309,
+      "grad_norm": 1084.011474609375,
+      "learning_rate": 4.808589453452918e-06,
+      "loss": 112.4365,
+      "step": 69420
+    },
+    {
+      "epoch": 0.5743475203706001,
+      "grad_norm": 781.2046508789062,
+      "learning_rate": 4.807180127853535e-06,
+      "loss": 82.4409,
+      "step": 69430
+    },
+    {
+      "epoch": 0.5744302436199694,
+      "grad_norm": 884.5288696289062,
+      "learning_rate": 4.8057708175959446e-06,
+      "loss": 113.9509,
+      "step": 69440
+    },
+    {
+      "epoch": 0.5745129668693386,
+      "grad_norm": 927.2460327148438,
+      "learning_rate": 4.804361522792278e-06,
+      "loss": 82.8712,
+      "step": 69450
+    },
+    {
+      "epoch": 0.5745956901187078,
+      "grad_norm": 778.8930053710938,
+      "learning_rate": 4.8029522435546695e-06,
+      "loss": 71.4773,
+      "step": 69460
+    },
+    {
+      "epoch": 0.5746784133680771,
+      "grad_norm": 631.3436889648438,
+      "learning_rate": 4.801542979995245e-06,
+      "loss": 81.125,
+      "step": 69470
+    },
+    {
+      "epoch": 0.5747611366174463,
+      "grad_norm": 812.0714111328125,
+      "learning_rate": 4.800133732226135e-06,
+      "loss": 82.7905,
+      "step": 69480
+    },
+    {
+      "epoch": 0.5748438598668155,
+      "grad_norm": 738.4343872070312,
+      "learning_rate": 4.798724500359467e-06,
+      "loss": 109.1766,
+      "step": 69490
+    },
+    {
+      "epoch": 0.5749265831161848,
+      "grad_norm": 1364.8973388671875,
+      "learning_rate": 4.7973152845073666e-06,
+      "loss": 74.5933,
+      "step": 69500
+    },
+    {
+      "epoch": 0.575009306365554,
+      "grad_norm": 1043.5596923828125,
+      "learning_rate": 4.795906084781958e-06,
+      "loss": 78.0541,
+      "step": 69510
+    },
+    {
+      "epoch": 0.5750920296149232,
+      "grad_norm": 1055.900146484375,
+      "learning_rate": 4.7944969012953656e-06,
+      "loss": 67.5411,
+      "step": 69520
+    },
+    {
+      "epoch": 0.5751747528642925,
+      "grad_norm": 1076.0845947265625,
+      "learning_rate": 4.793087734159711e-06,
+      "loss": 84.8314,
+      "step": 69530
+    },
+    {
+      "epoch": 0.5752574761136617,
+      "grad_norm": 731.822265625,
+      "learning_rate": 4.791678583487118e-06,
+      "loss": 116.2365,
+      "step": 69540
+    },
+    {
+      "epoch": 0.5753401993630309,
+      "grad_norm": 490.7128601074219,
+      "learning_rate": 4.790269449389703e-06,
+      "loss": 113.4962,
+      "step": 69550
+    },
+    {
+      "epoch": 0.5754229226124002,
+      "grad_norm": 4792.376953125,
+      "learning_rate": 4.788860331979586e-06,
+      "loss": 195.9267,
+      "step": 69560
+    },
+    {
+      "epoch": 0.5755056458617694,
+      "grad_norm": 718.0413208007812,
+      "learning_rate": 4.787451231368883e-06,
+      "loss": 83.7023,
+      "step": 69570
+    },
+    {
+      "epoch": 0.5755883691111386,
+      "grad_norm": 1221.1470947265625,
+      "learning_rate": 4.786042147669709e-06,
+      "loss": 99.3645,
+      "step": 69580
+    },
+    {
+      "epoch": 0.575671092360508,
+      "grad_norm": 642.6139526367188,
+      "learning_rate": 4.784633080994181e-06,
+      "loss": 74.7277,
+      "step": 69590
+    },
+    {
+      "epoch": 0.5757538156098772,
+      "grad_norm": 725.33349609375,
+      "learning_rate": 4.783224031454409e-06,
+      "loss": 108.2193,
+      "step": 69600
+    },
+    {
+      "epoch": 0.5758365388592463,
+      "grad_norm": 512.4448852539062,
+      "learning_rate": 4.781814999162507e-06,
+      "loss": 100.0694,
+      "step": 69610
+    },
+    {
+      "epoch": 0.5759192621086157,
+      "grad_norm": 540.7772216796875,
+      "learning_rate": 4.780405984230582e-06,
+      "loss": 76.5325,
+      "step": 69620
+    },
+    {
+      "epoch": 0.5760019853579849,
+      "grad_norm": 1100.6224365234375,
+      "learning_rate": 4.778996986770747e-06,
+      "loss": 125.3361,
+      "step": 69630
+    },
+    {
+      "epoch": 0.5760847086073541,
+      "grad_norm": 704.91748046875,
+      "learning_rate": 4.777588006895109e-06,
+      "loss": 84.2594,
+      "step": 69640
+    },
+    {
+      "epoch": 0.5761674318567234,
+      "grad_norm": 976.7535400390625,
+      "learning_rate": 4.77617904471577e-06,
+      "loss": 86.525,
+      "step": 69650
+    },
+    {
+      "epoch": 0.5762501551060926,
+      "grad_norm": 1758.328857421875,
+      "learning_rate": 4.774770100344838e-06,
+      "loss": 105.5501,
+      "step": 69660
+    },
+    {
+      "epoch": 0.5763328783554618,
+      "grad_norm": 717.3983764648438,
+      "learning_rate": 4.7733611738944155e-06,
+      "loss": 78.1173,
+      "step": 69670
+    },
+    {
+      "epoch": 0.5764156016048311,
+      "grad_norm": 1015.1489868164062,
+      "learning_rate": 4.7719522654766044e-06,
+      "loss": 93.4123,
+      "step": 69680
+    },
+    {
+      "epoch": 0.5764983248542003,
+      "grad_norm": 2092.849365234375,
+      "learning_rate": 4.7705433752035045e-06,
+      "loss": 120.9727,
+      "step": 69690
+    },
+    {
+      "epoch": 0.5765810481035695,
+      "grad_norm": 491.3244323730469,
+      "learning_rate": 4.7691345031872156e-06,
+      "loss": 70.0393,
+      "step": 69700
+    },
+    {
+      "epoch": 0.5766637713529388,
+      "grad_norm": 954.6812133789062,
+      "learning_rate": 4.767725649539833e-06,
+      "loss": 87.8252,
+      "step": 69710
+    },
+    {
+      "epoch": 0.576746494602308,
+      "grad_norm": 780.2823486328125,
+      "learning_rate": 4.766316814373458e-06,
+      "loss": 125.7246,
+      "step": 69720
+    },
+    {
+      "epoch": 0.5768292178516772,
+      "grad_norm": 1444.184326171875,
+      "learning_rate": 4.76490799780018e-06,
+      "loss": 94.4206,
+      "step": 69730
+    },
+    {
+      "epoch": 0.5769119411010465,
+      "grad_norm": 838.7236328125,
+      "learning_rate": 4.763499199932093e-06,
+      "loss": 126.1872,
+      "step": 69740
+    },
+    {
+      "epoch": 0.5769946643504157,
+      "grad_norm": 1014.107421875,
+      "learning_rate": 4.762090420881289e-06,
+      "loss": 108.3225,
+      "step": 69750
+    },
+    {
+      "epoch": 0.5770773875997849,
+      "grad_norm": 613.5991821289062,
+      "learning_rate": 4.760681660759859e-06,
+      "loss": 101.5193,
+      "step": 69760
+    },
+    {
+      "epoch": 0.5771601108491542,
+      "grad_norm": 661.6652221679688,
+      "learning_rate": 4.7592729196798905e-06,
+      "loss": 78.5238,
+      "step": 69770
+    },
+    {
+      "epoch": 0.5772428340985234,
+      "grad_norm": 917.331298828125,
+      "learning_rate": 4.757864197753472e-06,
+      "loss": 94.9886,
+      "step": 69780
+    },
+    {
+      "epoch": 0.5773255573478926,
+      "grad_norm": 750.0388793945312,
+      "learning_rate": 4.7564554950926876e-06,
+      "loss": 68.4733,
+      "step": 69790
+    },
+    {
+      "epoch": 0.5774082805972619,
+      "grad_norm": 754.461181640625,
+      "learning_rate": 4.755046811809621e-06,
+      "loss": 68.0218,
+      "step": 69800
+    },
+    {
+      "epoch": 0.5774910038466311,
+      "grad_norm": 1071.869384765625,
+      "learning_rate": 4.7536381480163575e-06,
+      "loss": 147.5622,
+      "step": 69810
+    },
+    {
+      "epoch": 0.5775737270960003,
+      "grad_norm": 809.9874877929688,
+      "learning_rate": 4.752229503824974e-06,
+      "loss": 68.0273,
+      "step": 69820
+    },
+    {
+      "epoch": 0.5776564503453696,
+      "grad_norm": 1067.3671875,
+      "learning_rate": 4.7508208793475515e-06,
+      "loss": 77.8083,
+      "step": 69830
+    },
+    {
+      "epoch": 0.5777391735947388,
+      "grad_norm": 757.71435546875,
+      "learning_rate": 4.749412274696169e-06,
+      "loss": 62.3041,
+      "step": 69840
+    },
+    {
+      "epoch": 0.577821896844108,
+      "grad_norm": 619.0784912109375,
+      "learning_rate": 4.748003689982901e-06,
+      "loss": 79.2833,
+      "step": 69850
+    },
+    {
+      "epoch": 0.5779046200934773,
+      "grad_norm": 631.8236083984375,
+      "learning_rate": 4.746595125319823e-06,
+      "loss": 99.2251,
+      "step": 69860
+    },
+    {
+      "epoch": 0.5779873433428465,
+      "grad_norm": 1292.0946044921875,
+      "learning_rate": 4.745186580819008e-06,
+      "loss": 108.7097,
+      "step": 69870
+    },
+    {
+      "epoch": 0.5780700665922157,
+      "grad_norm": 2003.7862548828125,
+      "learning_rate": 4.743778056592528e-06,
+      "loss": 101.937,
+      "step": 69880
+    },
+    {
+      "epoch": 0.578152789841585,
+      "grad_norm": 978.4115600585938,
+      "learning_rate": 4.742369552752453e-06,
+      "loss": 106.5378,
+      "step": 69890
+    },
+    {
+      "epoch": 0.5782355130909542,
+      "grad_norm": 603.2467041015625,
+      "learning_rate": 4.740961069410848e-06,
+      "loss": 93.0756,
+      "step": 69900
+    },
+    {
+      "epoch": 0.5783182363403234,
+      "grad_norm": 951.097900390625,
+      "learning_rate": 4.7395526066797835e-06,
+      "loss": 85.2565,
+      "step": 69910
+    },
+    {
+      "epoch": 0.5784009595896927,
+      "grad_norm": 813.0645751953125,
+      "learning_rate": 4.738144164671322e-06,
+      "loss": 123.9502,
+      "step": 69920
+    },
+    {
+      "epoch": 0.5784836828390619,
+      "grad_norm": 1287.8382568359375,
+      "learning_rate": 4.736735743497528e-06,
+      "loss": 98.9144,
+      "step": 69930
+    },
+    {
+      "epoch": 0.5785664060884311,
+      "grad_norm": 976.73974609375,
+      "learning_rate": 4.735327343270461e-06,
+      "loss": 102.1547,
+      "step": 69940
+    },
+    {
+      "epoch": 0.5786491293378004,
+      "grad_norm": 874.163330078125,
+      "learning_rate": 4.733918964102185e-06,
+      "loss": 108.0046,
+      "step": 69950
+    },
+    {
+      "epoch": 0.5787318525871696,
+      "grad_norm": 493.4253845214844,
+      "learning_rate": 4.732510606104754e-06,
+      "loss": 81.0991,
+      "step": 69960
+    },
+    {
+      "epoch": 0.5788145758365388,
+      "grad_norm": 481.89520263671875,
+      "learning_rate": 4.731102269390227e-06,
+      "loss": 72.0266,
+      "step": 69970
+    },
+    {
+      "epoch": 0.5788972990859081,
+      "grad_norm": 1048.3861083984375,
+      "learning_rate": 4.729693954070661e-06,
+      "loss": 101.3536,
+      "step": 69980
+    },
+    {
+      "epoch": 0.5789800223352773,
+      "grad_norm": 514.9654541015625,
+      "learning_rate": 4.728285660258104e-06,
+      "loss": 84.173,
+      "step": 69990
+    },
+    {
+      "epoch": 0.5790627455846465,
+      "grad_norm": 1212.3302001953125,
+      "learning_rate": 4.726877388064609e-06,
+      "loss": 95.1656,
+      "step": 70000
+    },
+    {
+      "epoch": 0.5791454688340159,
+      "grad_norm": 594.7462768554688,
+      "learning_rate": 4.725469137602229e-06,
+      "loss": 77.2462,
+      "step": 70010
+    },
+    {
+      "epoch": 0.579228192083385,
+      "grad_norm": 1222.4024658203125,
+      "learning_rate": 4.724060908983008e-06,
+      "loss": 77.3384,
+      "step": 70020
+    },
+    {
+      "epoch": 0.5793109153327543,
+      "grad_norm": 953.1253051757812,
+      "learning_rate": 4.7226527023189954e-06,
+      "loss": 69.5169,
+      "step": 70030
+    },
+    {
+      "epoch": 0.5793936385821236,
+      "grad_norm": 474.82373046875,
+      "learning_rate": 4.721244517722233e-06,
+      "loss": 62.3261,
+      "step": 70040
+    },
+    {
+      "epoch": 0.5794763618314928,
+      "grad_norm": 958.0091552734375,
+      "learning_rate": 4.719836355304766e-06,
+      "loss": 114.5203,
+      "step": 70050
+    },
+    {
+      "epoch": 0.579559085080862,
+      "grad_norm": 977.0399780273438,
+      "learning_rate": 4.718428215178634e-06,
+      "loss": 92.8079,
+      "step": 70060
+    },
+    {
+      "epoch": 0.5796418083302312,
+      "grad_norm": 889.9998779296875,
+      "learning_rate": 4.717020097455879e-06,
+      "loss": 118.2158,
+      "step": 70070
+    },
+    {
+      "epoch": 0.5797245315796005,
+      "grad_norm": 730.516357421875,
+      "learning_rate": 4.715612002248533e-06,
+      "loss": 90.555,
+      "step": 70080
+    },
+    {
+      "epoch": 0.5798072548289697,
+      "grad_norm": 905.8795776367188,
+      "learning_rate": 4.714203929668637e-06,
+      "loss": 82.7224,
+      "step": 70090
+    },
+    {
+      "epoch": 0.5798899780783389,
+      "grad_norm": 925.4937744140625,
+      "learning_rate": 4.712795879828221e-06,
+      "loss": 84.4116,
+      "step": 70100
+    },
+    {
+      "epoch": 0.5799727013277082,
+      "grad_norm": 1022.7550659179688,
+      "learning_rate": 4.71138785283932e-06,
+      "loss": 79.103,
+      "step": 70110
+    },
+    {
+      "epoch": 0.5800554245770774,
+      "grad_norm": 765.399169921875,
+      "learning_rate": 4.709979848813963e-06,
+      "loss": 88.3323,
+      "step": 70120
+    },
+    {
+      "epoch": 0.5801381478264466,
+      "grad_norm": 806.2305297851562,
+      "learning_rate": 4.7085718678641776e-06,
+      "loss": 117.2993,
+      "step": 70130
+    },
+    {
+      "epoch": 0.5802208710758159,
+      "grad_norm": 770.2034301757812,
+      "learning_rate": 4.70716391010199e-06,
+      "loss": 96.3582,
+      "step": 70140
+    },
+    {
+      "epoch": 0.5803035943251851,
+      "grad_norm": 934.9890747070312,
+      "learning_rate": 4.70575597563943e-06,
+      "loss": 85.1314,
+      "step": 70150
+    },
+    {
+      "epoch": 0.5803863175745543,
+      "grad_norm": 1144.4893798828125,
+      "learning_rate": 4.704348064588514e-06,
+      "loss": 95.0485,
+      "step": 70160
+    },
+    {
+      "epoch": 0.5804690408239236,
+      "grad_norm": 441.04132080078125,
+      "learning_rate": 4.702940177061266e-06,
+      "loss": 59.0982,
+      "step": 70170
+    },
+    {
+      "epoch": 0.5805517640732928,
+      "grad_norm": 883.7022705078125,
+      "learning_rate": 4.7015323131697035e-06,
+      "loss": 163.2424,
+      "step": 70180
+    },
+    {
+      "epoch": 0.580634487322662,
+      "grad_norm": 1098.4498291015625,
+      "learning_rate": 4.700124473025846e-06,
+      "loss": 87.0862,
+      "step": 70190
+    },
+    {
+      "epoch": 0.5807172105720313,
+      "grad_norm": 1009.8033447265625,
+      "learning_rate": 4.6987166567417085e-06,
+      "loss": 78.1907,
+      "step": 70200
+    },
+    {
+      "epoch": 0.5807999338214005,
+      "grad_norm": 1105.3592529296875,
+      "learning_rate": 4.697308864429303e-06,
+      "loss": 98.8104,
+      "step": 70210
+    },
+    {
+      "epoch": 0.5808826570707697,
+      "grad_norm": 786.5978393554688,
+      "learning_rate": 4.695901096200643e-06,
+      "loss": 79.2503,
+      "step": 70220
+    },
+    {
+      "epoch": 0.580965380320139,
+      "grad_norm": 480.7000427246094,
+      "learning_rate": 4.694493352167736e-06,
+      "loss": 107.3822,
+      "step": 70230
+    },
+    {
+      "epoch": 0.5810481035695082,
+      "grad_norm": 647.6439208984375,
+      "learning_rate": 4.693085632442593e-06,
+      "loss": 55.4946,
+      "step": 70240
+    },
+    {
+      "epoch": 0.5811308268188774,
+      "grad_norm": 1218.6649169921875,
+      "learning_rate": 4.691677937137217e-06,
+      "loss": 116.4978,
+      "step": 70250
+    },
+    {
+      "epoch": 0.5812135500682467,
+      "grad_norm": 511.2562561035156,
+      "learning_rate": 4.690270266363612e-06,
+      "loss": 120.063,
+      "step": 70260
+    },
+    {
+      "epoch": 0.5812962733176159,
+      "grad_norm": 822.0852661132812,
+      "learning_rate": 4.688862620233779e-06,
+      "loss": 100.5438,
+      "step": 70270
+    },
+    {
+      "epoch": 0.5813789965669851,
+      "grad_norm": 1118.2822265625,
+      "learning_rate": 4.687454998859721e-06,
+      "loss": 99.0379,
+      "step": 70280
+    },
+    {
+      "epoch": 0.5814617198163544,
+      "grad_norm": 733.8569946289062,
+      "learning_rate": 4.686047402353433e-06,
+      "loss": 78.6712,
+      "step": 70290
+    },
+    {
+      "epoch": 0.5815444430657236,
+      "grad_norm": 691.5249633789062,
+      "learning_rate": 4.684639830826913e-06,
+      "loss": 99.4106,
+      "step": 70300
+    },
+    {
+      "epoch": 0.5816271663150928,
+      "grad_norm": 542.6314086914062,
+      "learning_rate": 4.683232284392155e-06,
+      "loss": 96.8427,
+      "step": 70310
+    },
+    {
+      "epoch": 0.5817098895644621,
+      "grad_norm": 357.3701477050781,
+      "learning_rate": 4.681824763161151e-06,
+      "loss": 96.8162,
+      "step": 70320
+    },
+    {
+      "epoch": 0.5817926128138313,
+      "grad_norm": 951.8953857421875,
+      "learning_rate": 4.6804172672458905e-06,
+      "loss": 111.6947,
+      "step": 70330
+    },
+    {
+      "epoch": 0.5818753360632005,
+      "grad_norm": 743.7538452148438,
+      "learning_rate": 4.67900979675836e-06,
+      "loss": 97.8226,
+      "step": 70340
+    },
+    {
+      "epoch": 0.5819580593125698,
+      "grad_norm": 659.83056640625,
+      "learning_rate": 4.677602351810547e-06,
+      "loss": 96.5504,
+      "step": 70350
+    },
+    {
+      "epoch": 0.582040782561939,
+      "grad_norm": 643.2234497070312,
+      "learning_rate": 4.676194932514435e-06,
+      "loss": 70.3423,
+      "step": 70360
+    },
+    {
+      "epoch": 0.5821235058113082,
+      "grad_norm": 760.5200805664062,
+      "learning_rate": 4.674787538982006e-06,
+      "loss": 100.9573,
+      "step": 70370
+    },
+    {
+      "epoch": 0.5822062290606775,
+      "grad_norm": 504.2164306640625,
+      "learning_rate": 4.6733801713252405e-06,
+      "loss": 85.723,
+      "step": 70380
+    },
+    {
+      "epoch": 0.5822889523100467,
+      "grad_norm": 458.2579650878906,
+      "learning_rate": 4.671972829656116e-06,
+      "loss": 67.0866,
+      "step": 70390
+    },
+    {
+      "epoch": 0.5823716755594159,
+      "grad_norm": 1060.4288330078125,
+      "learning_rate": 4.670565514086607e-06,
+      "loss": 86.1434,
+      "step": 70400
+    },
+    {
+      "epoch": 0.5824543988087852,
+      "grad_norm": 478.5907897949219,
+      "learning_rate": 4.669158224728691e-06,
+      "loss": 87.8763,
+      "step": 70410
+    },
+    {
+      "epoch": 0.5825371220581544,
+      "grad_norm": 1053.1619873046875,
+      "learning_rate": 4.667750961694334e-06,
+      "loss": 95.2859,
+      "step": 70420
+    },
+    {
+      "epoch": 0.5826198453075236,
+      "grad_norm": 752.842041015625,
+      "learning_rate": 4.666343725095509e-06,
+      "loss": 100.9934,
+      "step": 70430
+    },
+    {
+      "epoch": 0.582702568556893,
+      "grad_norm": 687.784912109375,
+      "learning_rate": 4.6649365150441825e-06,
+      "loss": 76.4099,
+      "step": 70440
+    },
+    {
+      "epoch": 0.5827852918062622,
+      "grad_norm": 836.1194458007812,
+      "learning_rate": 4.66352933165232e-06,
+      "loss": 91.4804,
+      "step": 70450
+    },
+    {
+      "epoch": 0.5828680150556313,
+      "grad_norm": 739.9501953125,
+      "learning_rate": 4.6621221750318835e-06,
+      "loss": 71.6738,
+      "step": 70460
+    },
+    {
+      "epoch": 0.5829507383050007,
+      "grad_norm": 392.7385559082031,
+      "learning_rate": 4.660715045294834e-06,
+      "loss": 110.5148,
+      "step": 70470
+    },
+    {
+      "epoch": 0.5830334615543699,
+      "grad_norm": 821.631103515625,
+      "learning_rate": 4.659307942553133e-06,
+      "loss": 89.2418,
+      "step": 70480
+    },
+    {
+      "epoch": 0.5831161848037391,
+      "grad_norm": 978.71142578125,
+      "learning_rate": 4.657900866918735e-06,
+      "loss": 94.3087,
+      "step": 70490
+    },
+    {
+      "epoch": 0.5831989080531084,
+      "grad_norm": 1593.7996826171875,
+      "learning_rate": 4.6564938185035954e-06,
+      "loss": 103.8715,
+      "step": 70500
+    },
+    {
+      "epoch": 0.5832816313024776,
+      "grad_norm": 825.0999145507812,
+      "learning_rate": 4.655086797419666e-06,
+      "loss": 99.5245,
+      "step": 70510
+    },
+    {
+      "epoch": 0.5833643545518468,
+      "grad_norm": 541.9385375976562,
+      "learning_rate": 4.653679803778897e-06,
+      "loss": 87.7526,
+      "step": 70520
+    },
+    {
+      "epoch": 0.5834470778012161,
+      "grad_norm": 1196.7366943359375,
+      "learning_rate": 4.652272837693237e-06,
+      "loss": 104.002,
+      "step": 70530
+    },
+    {
+      "epoch": 0.5835298010505853,
+      "grad_norm": 718.0545043945312,
+      "learning_rate": 4.650865899274632e-06,
+      "loss": 125.1911,
+      "step": 70540
+    },
+    {
+      "epoch": 0.5836125242999545,
+      "grad_norm": 612.04736328125,
+      "learning_rate": 4.649458988635023e-06,
+      "loss": 75.9752,
+      "step": 70550
+    },
+    {
+      "epoch": 0.5836952475493238,
+      "grad_norm": 1365.9921875,
+      "learning_rate": 4.6480521058863546e-06,
+      "loss": 114.9041,
+      "step": 70560
+    },
+    {
+      "epoch": 0.583777970798693,
+      "grad_norm": 1484.0291748046875,
+      "learning_rate": 4.646645251140564e-06,
+      "loss": 102.265,
+      "step": 70570
+    },
+    {
+      "epoch": 0.5838606940480622,
+      "grad_norm": 756.23974609375,
+      "learning_rate": 4.6452384245095924e-06,
+      "loss": 82.9127,
+      "step": 70580
+    },
+    {
+      "epoch": 0.5839434172974315,
+      "grad_norm": 841.9822387695312,
+      "learning_rate": 4.643831626105369e-06,
+      "loss": 107.5036,
+      "step": 70590
+    },
+    {
+      "epoch": 0.5840261405468007,
+      "grad_norm": 947.1768798828125,
+      "learning_rate": 4.642424856039827e-06,
+      "loss": 89.3748,
+      "step": 70600
+    },
+    {
+      "epoch": 0.5841088637961699,
+      "grad_norm": 939.9810791015625,
+      "learning_rate": 4.6410181144249e-06,
+      "loss": 113.3098,
+      "step": 70610
+    },
+    {
+      "epoch": 0.5841915870455392,
+      "grad_norm": 819.6731567382812,
+      "learning_rate": 4.639611401372514e-06,
+      "loss": 75.1481,
+      "step": 70620
+    },
+    {
+      "epoch": 0.5842743102949084,
+      "grad_norm": 1223.1297607421875,
+      "learning_rate": 4.638204716994594e-06,
+      "loss": 76.0569,
+      "step": 70630
+    },
+    {
+      "epoch": 0.5843570335442776,
+      "grad_norm": 775.81689453125,
+      "learning_rate": 4.636798061403065e-06,
+      "loss": 79.0008,
+      "step": 70640
+    },
+    {
+      "epoch": 0.5844397567936469,
+      "grad_norm": 796.52978515625,
+      "learning_rate": 4.635391434709847e-06,
+      "loss": 87.8786,
+      "step": 70650
+    },
+    {
+      "epoch": 0.5845224800430161,
+      "grad_norm": 4887.2314453125,
+      "learning_rate": 4.6339848370268585e-06,
+      "loss": 128.6168,
+      "step": 70660
+    },
+    {
+      "epoch": 0.5846052032923853,
+      "grad_norm": 644.0697021484375,
+      "learning_rate": 4.63257826846602e-06,
+      "loss": 83.8805,
+      "step": 70670
+    },
+    {
+      "epoch": 0.5846879265417546,
+      "grad_norm": 866.1959838867188,
+      "learning_rate": 4.6311717291392396e-06,
+      "loss": 101.9503,
+      "step": 70680
+    },
+    {
+      "epoch": 0.5847706497911238,
+      "grad_norm": 1018.76025390625,
+      "learning_rate": 4.629765219158433e-06,
+      "loss": 82.65,
+      "step": 70690
+    },
+    {
+      "epoch": 0.584853373040493,
+      "grad_norm": 1179.1864013671875,
+      "learning_rate": 4.628358738635507e-06,
+      "loss": 134.8187,
+      "step": 70700
+    },
+    {
+      "epoch": 0.5849360962898623,
+      "grad_norm": 656.47900390625,
+      "learning_rate": 4.626952287682372e-06,
+      "loss": 117.6582,
+      "step": 70710
+    },
+    {
+      "epoch": 0.5850188195392315,
+      "grad_norm": 1300.9158935546875,
+      "learning_rate": 4.6255458664109306e-06,
+      "loss": 121.145,
+      "step": 70720
+    },
+    {
+      "epoch": 0.5851015427886007,
+      "grad_norm": 1486.7906494140625,
+      "learning_rate": 4.624139474933087e-06,
+      "loss": 94.2786,
+      "step": 70730
+    },
+    {
+      "epoch": 0.58518426603797,
+      "grad_norm": 759.8827514648438,
+      "learning_rate": 4.62273311336074e-06,
+      "loss": 88.7782,
+      "step": 70740
+    },
+    {
+      "epoch": 0.5852669892873392,
+      "grad_norm": 746.8486938476562,
+      "learning_rate": 4.62132678180579e-06,
+      "loss": 78.3264,
+      "step": 70750
+    },
+    {
+      "epoch": 0.5853497125367084,
+      "grad_norm": 837.21337890625,
+      "learning_rate": 4.619920480380127e-06,
+      "loss": 119.9007,
+      "step": 70760
+    },
+    {
+      "epoch": 0.5854324357860777,
+      "grad_norm": 10413.9169921875,
+      "learning_rate": 4.618514209195648e-06,
+      "loss": 153.6559,
+      "step": 70770
+    },
+    {
+      "epoch": 0.5855151590354469,
+      "grad_norm": 851.4038696289062,
+      "learning_rate": 4.617107968364243e-06,
+      "loss": 90.0972,
+      "step": 70780
+    },
+    {
+      "epoch": 0.5855978822848161,
+      "grad_norm": 551.212646484375,
+      "learning_rate": 4.615701757997799e-06,
+      "loss": 134.4456,
+      "step": 70790
+    },
+    {
+      "epoch": 0.5856806055341853,
+      "grad_norm": 1029.0963134765625,
+      "learning_rate": 4.614295578208202e-06,
+      "loss": 74.9261,
+      "step": 70800
+    },
+    {
+      "epoch": 0.5857633287835546,
+      "grad_norm": 1182.5975341796875,
+      "learning_rate": 4.612889429107337e-06,
+      "loss": 94.9276,
+      "step": 70810
+    },
+    {
+      "epoch": 0.5858460520329238,
+      "grad_norm": 879.9530639648438,
+      "learning_rate": 4.611483310807082e-06,
+      "loss": 109.8622,
+      "step": 70820
+    },
+    {
+      "epoch": 0.585928775282293,
+      "grad_norm": 756.7501831054688,
+      "learning_rate": 4.610077223419319e-06,
+      "loss": 111.9088,
+      "step": 70830
+    },
+    {
+      "epoch": 0.5860114985316623,
+      "grad_norm": 935.3068237304688,
+      "learning_rate": 4.608671167055922e-06,
+      "loss": 114.9015,
+      "step": 70840
+    },
+    {
+      "epoch": 0.5860942217810315,
+      "grad_norm": 764.818359375,
+      "learning_rate": 4.607265141828762e-06,
+      "loss": 95.3971,
+      "step": 70850
+    },
+    {
+      "epoch": 0.5861769450304007,
+      "grad_norm": 516.7097778320312,
+      "learning_rate": 4.605859147849713e-06,
+      "loss": 66.5408,
+      "step": 70860
+    },
+    {
+      "epoch": 0.58625966827977,
+      "grad_norm": 1200.6766357421875,
+      "learning_rate": 4.604453185230643e-06,
+      "loss": 94.5176,
+      "step": 70870
+    },
+    {
+      "epoch": 0.5863423915291393,
+      "grad_norm": 1096.214599609375,
+      "learning_rate": 4.603047254083418e-06,
+      "loss": 123.1105,
+      "step": 70880
+    },
+    {
+      "epoch": 0.5864251147785084,
+      "grad_norm": 789.3768920898438,
+      "learning_rate": 4.601641354519901e-06,
+      "loss": 97.1097,
+      "step": 70890
+    },
+    {
+      "epoch": 0.5865078380278778,
+      "grad_norm": 483.42315673828125,
+      "learning_rate": 4.6002354866519526e-06,
+      "loss": 100.7335,
+      "step": 70900
+    },
+    {
+      "epoch": 0.586590561277247,
+      "grad_norm": 978.4453125,
+      "learning_rate": 4.598829650591432e-06,
+      "loss": 91.333,
+      "step": 70910
+    },
+    {
+      "epoch": 0.5866732845266162,
+      "grad_norm": 653.1732177734375,
+      "learning_rate": 4.597423846450196e-06,
+      "loss": 86.567,
+      "step": 70920
+    },
+    {
+      "epoch": 0.5867560077759855,
+      "grad_norm": 714.1948852539062,
+      "learning_rate": 4.596018074340097e-06,
+      "loss": 86.6666,
+      "step": 70930
+    },
+    {
+      "epoch": 0.5868387310253547,
+      "grad_norm": 635.244140625,
+      "learning_rate": 4.594612334372985e-06,
+      "loss": 81.3609,
+      "step": 70940
+    },
+    {
+      "epoch": 0.5869214542747239,
+      "grad_norm": 790.4684448242188,
+      "learning_rate": 4.59320662666071e-06,
+      "loss": 91.6696,
+      "step": 70950
+    },
+    {
+      "epoch": 0.5870041775240932,
+      "grad_norm": 636.0848388671875,
+      "learning_rate": 4.591800951315116e-06,
+      "loss": 78.6052,
+      "step": 70960
+    },
+    {
+      "epoch": 0.5870869007734624,
+      "grad_norm": 1063.884765625,
+      "learning_rate": 4.590395308448046e-06,
+      "loss": 88.4161,
+      "step": 70970
+    },
+    {
+      "epoch": 0.5871696240228316,
+      "grad_norm": 1092.5953369140625,
+      "learning_rate": 4.588989698171343e-06,
+      "loss": 89.8243,
+      "step": 70980
+    },
+    {
+      "epoch": 0.5872523472722009,
+      "grad_norm": 758.7847290039062,
+      "learning_rate": 4.587584120596842e-06,
+      "loss": 95.1261,
+      "step": 70990
+    },
+    {
+      "epoch": 0.5873350705215701,
+      "grad_norm": 634.198486328125,
+      "learning_rate": 4.58617857583638e-06,
+      "loss": 72.3683,
+      "step": 71000
+    },
+    {
+      "epoch": 0.5874177937709393,
+      "grad_norm": 774.2512817382812,
+      "learning_rate": 4.5847730640017926e-06,
+      "loss": 86.8859,
+      "step": 71010
+    },
+    {
+      "epoch": 0.5875005170203086,
+      "grad_norm": 682.7206420898438,
+      "learning_rate": 4.5833675852049045e-06,
+      "loss": 95.423,
+      "step": 71020
+    },
+    {
+      "epoch": 0.5875832402696778,
+      "grad_norm": 458.109130859375,
+      "learning_rate": 4.5819621395575445e-06,
+      "loss": 90.4719,
+      "step": 71030
+    },
+    {
+      "epoch": 0.587665963519047,
+      "grad_norm": 784.8795166015625,
+      "learning_rate": 4.5805567271715395e-06,
+      "loss": 93.2228,
+      "step": 71040
+    },
+    {
+      "epoch": 0.5877486867684163,
+      "grad_norm": 802.4894409179688,
+      "learning_rate": 4.5791513481587105e-06,
+      "loss": 68.1955,
+      "step": 71050
+    },
+    {
+      "epoch": 0.5878314100177855,
+      "grad_norm": 834.863525390625,
+      "learning_rate": 4.577746002630878e-06,
+      "loss": 81.5406,
+      "step": 71060
+    },
+    {
+      "epoch": 0.5879141332671547,
+      "grad_norm": 1166.91796875,
+      "learning_rate": 4.576340690699857e-06,
+      "loss": 88.6559,
+      "step": 71070
+    },
+    {
+      "epoch": 0.587996856516524,
+      "grad_norm": 426.1348876953125,
+      "learning_rate": 4.574935412477464e-06,
+      "loss": 92.7223,
+      "step": 71080
+    },
+    {
+      "epoch": 0.5880795797658932,
+      "grad_norm": 980.8372192382812,
+      "learning_rate": 4.573530168075508e-06,
+      "loss": 126.4129,
+      "step": 71090
+    },
+    {
+      "epoch": 0.5881623030152624,
+      "grad_norm": 789.66796875,
+      "learning_rate": 4.572124957605803e-06,
+      "loss": 80.9297,
+      "step": 71100
+    },
+    {
+      "epoch": 0.5882450262646317,
+      "grad_norm": 647.4459228515625,
+      "learning_rate": 4.5707197811801484e-06,
+      "loss": 68.6454,
+      "step": 71110
+    },
+    {
+      "epoch": 0.5883277495140009,
+      "grad_norm": 1236.953369140625,
+      "learning_rate": 4.569314638910352e-06,
+      "loss": 79.3519,
+      "step": 71120
+    },
+    {
+      "epoch": 0.5884104727633701,
+      "grad_norm": 808.24462890625,
+      "learning_rate": 4.56790953090821e-06,
+      "loss": 126.6436,
+      "step": 71130
+    },
+    {
+      "epoch": 0.5884931960127394,
+      "grad_norm": 1413.169921875,
+      "learning_rate": 4.566504457285527e-06,
+      "loss": 102.8847,
+      "step": 71140
+    },
+    {
+      "epoch": 0.5885759192621086,
+      "grad_norm": 751.8162231445312,
+      "learning_rate": 4.565099418154093e-06,
+      "loss": 84.422,
+      "step": 71150
+    },
+    {
+      "epoch": 0.5886586425114778,
+      "grad_norm": 608.4118041992188,
+      "learning_rate": 4.563694413625703e-06,
+      "loss": 66.0182,
+      "step": 71160
+    },
+    {
+      "epoch": 0.5887413657608471,
+      "grad_norm": 413.452392578125,
+      "learning_rate": 4.5622894438121465e-06,
+      "loss": 75.759,
+      "step": 71170
+    },
+    {
+      "epoch": 0.5888240890102163,
+      "grad_norm": 522.5778198242188,
+      "learning_rate": 4.560884508825212e-06,
+      "loss": 99.5577,
+      "step": 71180
+    },
+    {
+      "epoch": 0.5889068122595855,
+      "grad_norm": 492.1758728027344,
+      "learning_rate": 4.559479608776679e-06,
+      "loss": 91.1286,
+      "step": 71190
+    },
+    {
+      "epoch": 0.5889895355089548,
+      "grad_norm": 694.331787109375,
+      "learning_rate": 4.558074743778333e-06,
+      "loss": 96.5787,
+      "step": 71200
+    },
+    {
+      "epoch": 0.589072258758324,
+      "grad_norm": 828.5624389648438,
+      "learning_rate": 4.556669913941951e-06,
+      "loss": 85.6967,
+      "step": 71210
+    },
+    {
+      "epoch": 0.5891549820076932,
+      "grad_norm": 963.8417358398438,
+      "learning_rate": 4.555265119379308e-06,
+      "loss": 84.3966,
+      "step": 71220
+    },
+    {
+      "epoch": 0.5892377052570625,
+      "grad_norm": 869.2294921875,
+      "learning_rate": 4.55386036020218e-06,
+      "loss": 124.0421,
+      "step": 71230
+    },
+    {
+      "epoch": 0.5893204285064317,
+      "grad_norm": 1175.8673095703125,
+      "learning_rate": 4.552455636522335e-06,
+      "loss": 127.9119,
+      "step": 71240
+    },
+    {
+      "epoch": 0.5894031517558009,
+      "grad_norm": 815.5340576171875,
+      "learning_rate": 4.551050948451542e-06,
+      "loss": 76.2241,
+      "step": 71250
+    },
+    {
+      "epoch": 0.5894858750051702,
+      "grad_norm": 769.5640869140625,
+      "learning_rate": 4.549646296101564e-06,
+      "loss": 106.5304,
+      "step": 71260
+    },
+    {
+      "epoch": 0.5895685982545394,
+      "grad_norm": 693.4418334960938,
+      "learning_rate": 4.548241679584165e-06,
+      "loss": 79.3208,
+      "step": 71270
+    },
+    {
+      "epoch": 0.5896513215039086,
+      "grad_norm": 1170.5986328125,
+      "learning_rate": 4.546837099011101e-06,
+      "loss": 126.9079,
+      "step": 71280
+    },
+    {
+      "epoch": 0.589734044753278,
+      "grad_norm": 750.7109375,
+      "learning_rate": 4.545432554494128e-06,
+      "loss": 109.4266,
+      "step": 71290
+    },
+    {
+      "epoch": 0.5898167680026472,
+      "grad_norm": 1614.88671875,
+      "learning_rate": 4.544028046145002e-06,
+      "loss": 115.2326,
+      "step": 71300
+    },
+    {
+      "epoch": 0.5898994912520164,
+      "grad_norm": 1244.8895263671875,
+      "learning_rate": 4.542623574075471e-06,
+      "loss": 115.9653,
+      "step": 71310
+    },
+    {
+      "epoch": 0.5899822145013857,
+      "grad_norm": 1223.9134521484375,
+      "learning_rate": 4.541219138397283e-06,
+      "loss": 107.9536,
+      "step": 71320
+    },
+    {
+      "epoch": 0.5900649377507549,
+      "grad_norm": 645.8054809570312,
+      "learning_rate": 4.539814739222182e-06,
+      "loss": 88.0276,
+      "step": 71330
+    },
+    {
+      "epoch": 0.5901476610001241,
+      "grad_norm": 906.608154296875,
+      "learning_rate": 4.538410376661912e-06,
+      "loss": 97.0788,
+      "step": 71340
+    },
+    {
+      "epoch": 0.5902303842494934,
+      "grad_norm": 655.9107666015625,
+      "learning_rate": 4.537006050828209e-06,
+      "loss": 102.7088,
+      "step": 71350
+    },
+    {
+      "epoch": 0.5903131074988626,
+      "grad_norm": 881.0200805664062,
+      "learning_rate": 4.535601761832811e-06,
+      "loss": 94.3685,
+      "step": 71360
+    },
+    {
+      "epoch": 0.5903958307482318,
+      "grad_norm": 593.4007568359375,
+      "learning_rate": 4.534197509787448e-06,
+      "loss": 96.9625,
+      "step": 71370
+    },
+    {
+      "epoch": 0.5904785539976011,
+      "grad_norm": 362.5093688964844,
+      "learning_rate": 4.5327932948038525e-06,
+      "loss": 76.0315,
+      "step": 71380
+    },
+    {
+      "epoch": 0.5905612772469703,
+      "grad_norm": 724.7342529296875,
+      "learning_rate": 4.5313891169937495e-06,
+      "loss": 68.4613,
+      "step": 71390
+    },
+    {
+      "epoch": 0.5906440004963395,
+      "grad_norm": 320.0503845214844,
+      "learning_rate": 4.529984976468864e-06,
+      "loss": 59.5351,
+      "step": 71400
+    },
+    {
+      "epoch": 0.5907267237457088,
+      "grad_norm": 689.7297973632812,
+      "learning_rate": 4.528580873340916e-06,
+      "loss": 71.2971,
+      "step": 71410
+    },
+    {
+      "epoch": 0.590809446995078,
+      "grad_norm": 698.4686889648438,
+      "learning_rate": 4.5271768077216245e-06,
+      "loss": 81.2217,
+      "step": 71420
+    },
+    {
+      "epoch": 0.5908921702444472,
+      "grad_norm": 778.4169311523438,
+      "learning_rate": 4.525772779722705e-06,
+      "loss": 78.8058,
+      "step": 71430
+    },
+    {
+      "epoch": 0.5909748934938165,
+      "grad_norm": 723.3839721679688,
+      "learning_rate": 4.524368789455872e-06,
+      "loss": 140.3945,
+      "step": 71440
+    },
+    {
+      "epoch": 0.5910576167431857,
+      "grad_norm": 1784.62158203125,
+      "learning_rate": 4.5229648370328276e-06,
+      "loss": 99.9209,
+      "step": 71450
+    },
+    {
+      "epoch": 0.5911403399925549,
+      "grad_norm": 688.6642456054688,
+      "learning_rate": 4.521560922565282e-06,
+      "loss": 110.5966,
+      "step": 71460
+    },
+    {
+      "epoch": 0.5912230632419242,
+      "grad_norm": 1047.2376708984375,
+      "learning_rate": 4.52015704616494e-06,
+      "loss": 67.9934,
+      "step": 71470
+    },
+    {
+      "epoch": 0.5913057864912934,
+      "grad_norm": 956.0989379882812,
+      "learning_rate": 4.518753207943498e-06,
+      "loss": 77.9636,
+      "step": 71480
+    },
+    {
+      "epoch": 0.5913885097406626,
+      "grad_norm": 832.9586181640625,
+      "learning_rate": 4.517349408012656e-06,
+      "loss": 89.3105,
+      "step": 71490
+    },
+    {
+      "epoch": 0.5914712329900319,
+      "grad_norm": 1294.6611328125,
+      "learning_rate": 4.515945646484105e-06,
+      "loss": 108.4286,
+      "step": 71500
+    },
+    {
+      "epoch": 0.5915539562394011,
+      "grad_norm": 635.0994262695312,
+      "learning_rate": 4.514541923469538e-06,
+      "loss": 74.8464,
+      "step": 71510
+    },
+    {
+      "epoch": 0.5916366794887703,
+      "grad_norm": 1128.41748046875,
+      "learning_rate": 4.513138239080641e-06,
+      "loss": 65.9964,
+      "step": 71520
+    },
+    {
+      "epoch": 0.5917194027381395,
+      "grad_norm": 722.1233520507812,
+      "learning_rate": 4.511734593429104e-06,
+      "loss": 106.5479,
+      "step": 71530
+    },
+    {
+      "epoch": 0.5918021259875088,
+      "grad_norm": 948.386474609375,
+      "learning_rate": 4.510330986626602e-06,
+      "loss": 87.1451,
+      "step": 71540
+    },
+    {
+      "epoch": 0.591884849236878,
+      "grad_norm": 1609.5699462890625,
+      "learning_rate": 4.5089274187848144e-06,
+      "loss": 74.7854,
+      "step": 71550
+    },
+    {
+      "epoch": 0.5919675724862472,
+      "grad_norm": 1418.9605712890625,
+      "learning_rate": 4.507523890015421e-06,
+      "loss": 88.9027,
+      "step": 71560
+    },
+    {
+      "epoch": 0.5920502957356165,
+      "grad_norm": 553.1968383789062,
+      "learning_rate": 4.5061204004300905e-06,
+      "loss": 79.0763,
+      "step": 71570
+    },
+    {
+      "epoch": 0.5921330189849857,
+      "grad_norm": 1095.764892578125,
+      "learning_rate": 4.504716950140492e-06,
+      "loss": 103.5799,
+      "step": 71580
+    },
+    {
+      "epoch": 0.5922157422343549,
+      "grad_norm": 527.2064819335938,
+      "learning_rate": 4.503313539258294e-06,
+      "loss": 88.4465,
+      "step": 71590
+    },
+    {
+      "epoch": 0.5922984654837242,
+      "grad_norm": 1063.0721435546875,
+      "learning_rate": 4.501910167895158e-06,
+      "loss": 98.2458,
+      "step": 71600
+    },
+    {
+      "epoch": 0.5923811887330934,
+      "grad_norm": 2767.47705078125,
+      "learning_rate": 4.500506836162746e-06,
+      "loss": 92.3343,
+      "step": 71610
+    },
+    {
+      "epoch": 0.5924639119824626,
+      "grad_norm": 455.406982421875,
+      "learning_rate": 4.499103544172711e-06,
+      "loss": 81.4368,
+      "step": 71620
+    },
+    {
+      "epoch": 0.5925466352318319,
+      "grad_norm": 850.3991088867188,
+      "learning_rate": 4.497700292036708e-06,
+      "loss": 70.8303,
+      "step": 71630
+    },
+    {
+      "epoch": 0.5926293584812011,
+      "grad_norm": 778.4962768554688,
+      "learning_rate": 4.4962970798663865e-06,
+      "loss": 66.9589,
+      "step": 71640
+    },
+    {
+      "epoch": 0.5927120817305703,
+      "grad_norm": 543.8009033203125,
+      "learning_rate": 4.494893907773394e-06,
+      "loss": 69.5775,
+      "step": 71650
+    },
+    {
+      "epoch": 0.5927948049799396,
+      "grad_norm": 799.34375,
+      "learning_rate": 4.493490775869377e-06,
+      "loss": 78.267,
+      "step": 71660
+    },
+    {
+      "epoch": 0.5928775282293088,
+      "grad_norm": 1260.3624267578125,
+      "learning_rate": 4.492087684265975e-06,
+      "loss": 91.3813,
+      "step": 71670
+    },
+    {
+      "epoch": 0.592960251478678,
+      "grad_norm": 714.5103149414062,
+      "learning_rate": 4.490684633074824e-06,
+      "loss": 87.9972,
+      "step": 71680
+    },
+    {
+      "epoch": 0.5930429747280473,
+      "grad_norm": 419.715087890625,
+      "learning_rate": 4.489281622407559e-06,
+      "loss": 85.3557,
+      "step": 71690
+    },
+    {
+      "epoch": 0.5931256979774165,
+      "grad_norm": 939.987060546875,
+      "learning_rate": 4.487878652375813e-06,
+      "loss": 132.4992,
+      "step": 71700
+    },
+    {
+      "epoch": 0.5932084212267857,
+      "grad_norm": 623.05712890625,
+      "learning_rate": 4.486475723091211e-06,
+      "loss": 97.7154,
+      "step": 71710
+    },
+    {
+      "epoch": 0.593291144476155,
+      "grad_norm": 852.7836303710938,
+      "learning_rate": 4.485072834665379e-06,
+      "loss": 133.3668,
+      "step": 71720
+    },
+    {
+      "epoch": 0.5933738677255243,
+      "grad_norm": 830.4113159179688,
+      "learning_rate": 4.483669987209938e-06,
+      "loss": 85.3274,
+      "step": 71730
+    },
+    {
+      "epoch": 0.5934565909748934,
+      "grad_norm": 1263.1221923828125,
+      "learning_rate": 4.482267180836508e-06,
+      "loss": 96.8079,
+      "step": 71740
+    },
+    {
+      "epoch": 0.5935393142242628,
+      "grad_norm": 605.0487060546875,
+      "learning_rate": 4.4808644156567e-06,
+      "loss": 86.8117,
+      "step": 71750
+    },
+    {
+      "epoch": 0.593622037473632,
+      "grad_norm": 524.0765380859375,
+      "learning_rate": 4.479461691782129e-06,
+      "loss": 74.157,
+      "step": 71760
+    },
+    {
+      "epoch": 0.5937047607230012,
+      "grad_norm": 708.2615966796875,
+      "learning_rate": 4.478059009324403e-06,
+      "loss": 106.5555,
+      "step": 71770
+    },
+    {
+      "epoch": 0.5937874839723705,
+      "grad_norm": 427.41162109375,
+      "learning_rate": 4.476656368395126e-06,
+      "loss": 91.2204,
+      "step": 71780
+    },
+    {
+      "epoch": 0.5938702072217397,
+      "grad_norm": 744.795654296875,
+      "learning_rate": 4.4752537691059e-06,
+      "loss": 78.5461,
+      "step": 71790
+    },
+    {
+      "epoch": 0.5939529304711089,
+      "grad_norm": 1040.6856689453125,
+      "learning_rate": 4.473851211568323e-06,
+      "loss": 105.2389,
+      "step": 71800
+    },
+    {
+      "epoch": 0.5940356537204782,
+      "grad_norm": 558.6827392578125,
+      "learning_rate": 4.472448695893991e-06,
+      "loss": 97.6422,
+      "step": 71810
+    },
+    {
+      "epoch": 0.5941183769698474,
+      "grad_norm": 1102.7012939453125,
+      "learning_rate": 4.471046222194494e-06,
+      "loss": 104.5667,
+      "step": 71820
+    },
+    {
+      "epoch": 0.5942011002192166,
+      "grad_norm": 1997.569091796875,
+      "learning_rate": 4.469643790581422e-06,
+      "loss": 84.4513,
+      "step": 71830
+    },
+    {
+      "epoch": 0.5942838234685859,
+      "grad_norm": 1167.2252197265625,
+      "learning_rate": 4.468241401166359e-06,
+      "loss": 102.4033,
+      "step": 71840
+    },
+    {
+      "epoch": 0.5943665467179551,
+      "grad_norm": 1216.229736328125,
+      "learning_rate": 4.466839054060888e-06,
+      "loss": 73.6999,
+      "step": 71850
+    },
+    {
+      "epoch": 0.5944492699673243,
+      "grad_norm": 371.2035217285156,
+      "learning_rate": 4.465436749376586e-06,
+      "loss": 87.3126,
+      "step": 71860
+    },
+    {
+      "epoch": 0.5945319932166936,
+      "grad_norm": 993.4617309570312,
+      "learning_rate": 4.464034487225031e-06,
+      "loss": 84.5187,
+      "step": 71870
+    },
+    {
+      "epoch": 0.5946147164660628,
+      "grad_norm": 668.97802734375,
+      "learning_rate": 4.462632267717789e-06,
+      "loss": 93.8088,
+      "step": 71880
+    },
+    {
+      "epoch": 0.594697439715432,
+      "grad_norm": 723.9701538085938,
+      "learning_rate": 4.461230090966433e-06,
+      "loss": 92.2385,
+      "step": 71890
+    },
+    {
+      "epoch": 0.5947801629648013,
+      "grad_norm": 1200.336669921875,
+      "learning_rate": 4.4598279570825244e-06,
+      "loss": 108.3329,
+      "step": 71900
+    },
+    {
+      "epoch": 0.5948628862141705,
+      "grad_norm": 749.9968872070312,
+      "learning_rate": 4.458425866177628e-06,
+      "loss": 96.5593,
+      "step": 71910
+    },
+    {
+      "epoch": 0.5949456094635397,
+      "grad_norm": 984.7999877929688,
+      "learning_rate": 4.457023818363299e-06,
+      "loss": 105.6873,
+      "step": 71920
+    },
+    {
+      "epoch": 0.595028332712909,
+      "grad_norm": 679.338134765625,
+      "learning_rate": 4.455621813751093e-06,
+      "loss": 73.7637,
+      "step": 71930
+    },
+    {
+      "epoch": 0.5951110559622782,
+      "grad_norm": 1122.4134521484375,
+      "learning_rate": 4.45421985245256e-06,
+      "loss": 98.8108,
+      "step": 71940
+    },
+    {
+      "epoch": 0.5951937792116474,
+      "grad_norm": 988.9681396484375,
+      "learning_rate": 4.452817934579249e-06,
+      "loss": 84.0397,
+      "step": 71950
+    },
+    {
+      "epoch": 0.5952765024610167,
+      "grad_norm": 918.6084594726562,
+      "learning_rate": 4.451416060242707e-06,
+      "loss": 103.6519,
+      "step": 71960
+    },
+    {
+      "epoch": 0.5953592257103859,
+      "grad_norm": 676.3186645507812,
+      "learning_rate": 4.450014229554468e-06,
+      "loss": 80.6245,
+      "step": 71970
+    },
+    {
+      "epoch": 0.5954419489597551,
+      "grad_norm": 987.3781127929688,
+      "learning_rate": 4.448612442626073e-06,
+      "loss": 89.4239,
+      "step": 71980
+    },
+    {
+      "epoch": 0.5955246722091244,
+      "grad_norm": 1492.4959716796875,
+      "learning_rate": 4.447210699569055e-06,
+      "loss": 85.7063,
+      "step": 71990
+    },
+    {
+      "epoch": 0.5956073954584936,
+      "grad_norm": 614.7178955078125,
+      "learning_rate": 4.445809000494945e-06,
+      "loss": 101.3085,
+      "step": 72000
+    },
+    {
+      "epoch": 0.5956901187078628,
+      "grad_norm": 1345.1634521484375,
+      "learning_rate": 4.4444073455152705e-06,
+      "loss": 87.2096,
+      "step": 72010
+    },
+    {
+      "epoch": 0.5957728419572321,
+      "grad_norm": 536.4259033203125,
+      "learning_rate": 4.443005734741553e-06,
+      "loss": 71.8604,
+      "step": 72020
+    },
+    {
+      "epoch": 0.5958555652066013,
+      "grad_norm": 652.9493408203125,
+      "learning_rate": 4.441604168285313e-06,
+      "loss": 121.6692,
+      "step": 72030
+    },
+    {
+      "epoch": 0.5959382884559705,
+      "grad_norm": 1071.4775390625,
+      "learning_rate": 4.440202646258067e-06,
+      "loss": 125.7619,
+      "step": 72040
+    },
+    {
+      "epoch": 0.5960210117053398,
+      "grad_norm": 540.5606689453125,
+      "learning_rate": 4.4388011687713274e-06,
+      "loss": 76.9964,
+      "step": 72050
+    },
+    {
+      "epoch": 0.596103734954709,
+      "grad_norm": 979.0853881835938,
+      "learning_rate": 4.437399735936603e-06,
+      "loss": 86.8726,
+      "step": 72060
+    },
+    {
+      "epoch": 0.5961864582040782,
+      "grad_norm": 1528.9964599609375,
+      "learning_rate": 4.435998347865399e-06,
+      "loss": 139.583,
+      "step": 72070
+    },
+    {
+      "epoch": 0.5962691814534475,
+      "grad_norm": 581.6967163085938,
+      "learning_rate": 4.4345970046692174e-06,
+      "loss": 79.9947,
+      "step": 72080
+    },
+    {
+      "epoch": 0.5963519047028167,
+      "grad_norm": 915.5377807617188,
+      "learning_rate": 4.433195706459558e-06,
+      "loss": 67.4258,
+      "step": 72090
+    },
+    {
+      "epoch": 0.5964346279521859,
+      "grad_norm": 1992.865234375,
+      "learning_rate": 4.431794453347915e-06,
+      "loss": 97.0232,
+      "step": 72100
+    },
+    {
+      "epoch": 0.5965173512015552,
+      "grad_norm": 653.1949462890625,
+      "learning_rate": 4.430393245445781e-06,
+      "loss": 99.4039,
+      "step": 72110
+    },
+    {
+      "epoch": 0.5966000744509244,
+      "grad_norm": 1475.4862060546875,
+      "learning_rate": 4.42899208286464e-06,
+      "loss": 134.1732,
+      "step": 72120
+    },
+    {
+      "epoch": 0.5966827977002936,
+      "grad_norm": 985.6160888671875,
+      "learning_rate": 4.427590965715981e-06,
+      "loss": 93.6811,
+      "step": 72130
+    },
+    {
+      "epoch": 0.596765520949663,
+      "grad_norm": 1121.754638671875,
+      "learning_rate": 4.426189894111281e-06,
+      "loss": 98.63,
+      "step": 72140
+    },
+    {
+      "epoch": 0.5968482441990322,
+      "grad_norm": 778.4381103515625,
+      "learning_rate": 4.4247888681620165e-06,
+      "loss": 95.378,
+      "step": 72150
+    },
+    {
+      "epoch": 0.5969309674484014,
+      "grad_norm": 1081.6031494140625,
+      "learning_rate": 4.423387887979663e-06,
+      "loss": 76.984,
+      "step": 72160
+    },
+    {
+      "epoch": 0.5970136906977707,
+      "grad_norm": 844.283203125,
+      "learning_rate": 4.421986953675687e-06,
+      "loss": 95.1035,
+      "step": 72170
+    },
+    {
+      "epoch": 0.5970964139471399,
+      "grad_norm": 1140.15869140625,
+      "learning_rate": 4.420586065361558e-06,
+      "loss": 98.3029,
+      "step": 72180
+    },
+    {
+      "epoch": 0.5971791371965091,
+      "grad_norm": 1047.5072021484375,
+      "learning_rate": 4.419185223148737e-06,
+      "loss": 71.558,
+      "step": 72190
+    },
+    {
+      "epoch": 0.5972618604458784,
+      "grad_norm": 512.5674438476562,
+      "learning_rate": 4.417784427148681e-06,
+      "loss": 118.2006,
+      "step": 72200
+    },
+    {
+      "epoch": 0.5973445836952476,
+      "grad_norm": 922.7410888671875,
+      "learning_rate": 4.4163836774728466e-06,
+      "loss": 88.5922,
+      "step": 72210
+    },
+    {
+      "epoch": 0.5974273069446168,
+      "grad_norm": 695.3885498046875,
+      "learning_rate": 4.414982974232686e-06,
+      "loss": 85.0964,
+      "step": 72220
+    },
+    {
+      "epoch": 0.5975100301939861,
+      "grad_norm": 756.3115844726562,
+      "learning_rate": 4.413582317539644e-06,
+      "loss": 84.8309,
+      "step": 72230
+    },
+    {
+      "epoch": 0.5975927534433553,
+      "grad_norm": 893.6836547851562,
+      "learning_rate": 4.412181707505167e-06,
+      "loss": 76.9076,
+      "step": 72240
+    },
+    {
+      "epoch": 0.5976754766927245,
+      "grad_norm": 830.41943359375,
+      "learning_rate": 4.410781144240692e-06,
+      "loss": 99.8375,
+      "step": 72250
+    },
+    {
+      "epoch": 0.5977581999420937,
+      "grad_norm": 520.194580078125,
+      "learning_rate": 4.409380627857658e-06,
+      "loss": 77.2131,
+      "step": 72260
+    },
+    {
+      "epoch": 0.597840923191463,
+      "grad_norm": 1167.1990966796875,
+      "learning_rate": 4.4079801584674955e-06,
+      "loss": 106.8954,
+      "step": 72270
+    },
+    {
+      "epoch": 0.5979236464408322,
+      "grad_norm": 987.2964477539062,
+      "learning_rate": 4.406579736181636e-06,
+      "loss": 109.5783,
+      "step": 72280
+    },
+    {
+      "epoch": 0.5980063696902014,
+      "grad_norm": 755.127197265625,
+      "learning_rate": 4.405179361111503e-06,
+      "loss": 102.5628,
+      "step": 72290
+    },
+    {
+      "epoch": 0.5980890929395707,
+      "grad_norm": 858.6559448242188,
+      "learning_rate": 4.403779033368521e-06,
+      "loss": 83.2092,
+      "step": 72300
+    },
+    {
+      "epoch": 0.5981718161889399,
+      "grad_norm": 1175.3221435546875,
+      "learning_rate": 4.402378753064102e-06,
+      "loss": 120.481,
+      "step": 72310
+    },
+    {
+      "epoch": 0.5982545394383091,
+      "grad_norm": 1265.4703369140625,
+      "learning_rate": 4.400978520309663e-06,
+      "loss": 126.6442,
+      "step": 72320
+    },
+    {
+      "epoch": 0.5983372626876784,
+      "grad_norm": 837.8228759765625,
+      "learning_rate": 4.399578335216615e-06,
+      "loss": 83.0502,
+      "step": 72330
+    },
+    {
+      "epoch": 0.5984199859370476,
+      "grad_norm": 454.5643310546875,
+      "learning_rate": 4.3981781978963625e-06,
+      "loss": 195.8178,
+      "step": 72340
+    },
+    {
+      "epoch": 0.5985027091864168,
+      "grad_norm": 1586.3955078125,
+      "learning_rate": 4.39677810846031e-06,
+      "loss": 104.6424,
+      "step": 72350
+    },
+    {
+      "epoch": 0.5985854324357861,
+      "grad_norm": 472.3536376953125,
+      "learning_rate": 4.395378067019854e-06,
+      "loss": 103.8725,
+      "step": 72360
+    },
+    {
+      "epoch": 0.5986681556851553,
+      "grad_norm": 1100.48486328125,
+      "learning_rate": 4.39397807368639e-06,
+      "loss": 95.1332,
+      "step": 72370
+    },
+    {
+      "epoch": 0.5987508789345245,
+      "grad_norm": 836.9981079101562,
+      "learning_rate": 4.39257812857131e-06,
+      "loss": 88.9254,
+      "step": 72380
+    },
+    {
+      "epoch": 0.5988336021838938,
+      "grad_norm": 861.1360473632812,
+      "learning_rate": 4.391178231786003e-06,
+      "loss": 154.0604,
+      "step": 72390
+    },
+    {
+      "epoch": 0.598916325433263,
+      "grad_norm": 801.0167846679688,
+      "learning_rate": 4.389778383441847e-06,
+      "loss": 57.2983,
+      "step": 72400
+    },
+    {
+      "epoch": 0.5989990486826322,
+      "grad_norm": 737.025390625,
+      "learning_rate": 4.388378583650225e-06,
+      "loss": 106.2906,
+      "step": 72410
+    },
+    {
+      "epoch": 0.5990817719320015,
+      "grad_norm": 770.6280517578125,
+      "learning_rate": 4.386978832522512e-06,
+      "loss": 60.7984,
+      "step": 72420
+    },
+    {
+      "epoch": 0.5991644951813707,
+      "grad_norm": 454.4928894042969,
+      "learning_rate": 4.38557913017008e-06,
+      "loss": 85.5769,
+      "step": 72430
+    },
+    {
+      "epoch": 0.5992472184307399,
+      "grad_norm": 799.1620483398438,
+      "learning_rate": 4.384179476704297e-06,
+      "loss": 69.6634,
+      "step": 72440
+    },
+    {
+      "epoch": 0.5993299416801092,
+      "grad_norm": 1317.85693359375,
+      "learning_rate": 4.382779872236527e-06,
+      "loss": 96.343,
+      "step": 72450
+    },
+    {
+      "epoch": 0.5994126649294784,
+      "grad_norm": 550.043212890625,
+      "learning_rate": 4.3813803168781295e-06,
+      "loss": 66.2317,
+      "step": 72460
+    },
+    {
+      "epoch": 0.5994953881788476,
+      "grad_norm": 359.5018005371094,
+      "learning_rate": 4.379980810740463e-06,
+      "loss": 92.4579,
+      "step": 72470
+    },
+    {
+      "epoch": 0.5995781114282169,
+      "grad_norm": 773.148193359375,
+      "learning_rate": 4.378581353934876e-06,
+      "loss": 109.3504,
+      "step": 72480
+    },
+    {
+      "epoch": 0.5996608346775861,
+      "grad_norm": 994.1666259765625,
+      "learning_rate": 4.3771819465727185e-06,
+      "loss": 106.4414,
+      "step": 72490
+    },
+    {
+      "epoch": 0.5997435579269553,
+      "grad_norm": 1295.1717529296875,
+      "learning_rate": 4.3757825887653345e-06,
+      "loss": 86.4112,
+      "step": 72500
+    },
+    {
+      "epoch": 0.5998262811763246,
+      "grad_norm": 1010.71142578125,
+      "learning_rate": 4.374383280624066e-06,
+      "loss": 85.4114,
+      "step": 72510
+    },
+    {
+      "epoch": 0.5999090044256938,
+      "grad_norm": 731.9288940429688,
+      "learning_rate": 4.372984022260249e-06,
+      "loss": 104.8326,
+      "step": 72520
+    },
+    {
+      "epoch": 0.599991727675063,
+      "grad_norm": 670.1849975585938,
+      "learning_rate": 4.371584813785216e-06,
+      "loss": 86.4142,
+      "step": 72530
+    },
+    {
+      "epoch": 0.6000744509244323,
+      "grad_norm": 669.2977905273438,
+      "learning_rate": 4.370185655310295e-06,
+      "loss": 116.1969,
+      "step": 72540
+    },
+    {
+      "epoch": 0.6001571741738015,
+      "grad_norm": 729.8284912109375,
+      "learning_rate": 4.368786546946811e-06,
+      "loss": 79.6399,
+      "step": 72550
+    },
+    {
+      "epoch": 0.6002398974231707,
+      "grad_norm": 600.2257690429688,
+      "learning_rate": 4.367387488806086e-06,
+      "loss": 77.0065,
+      "step": 72560
+    },
+    {
+      "epoch": 0.60032262067254,
+      "grad_norm": 879.3544311523438,
+      "learning_rate": 4.365988480999434e-06,
+      "loss": 83.5863,
+      "step": 72570
+    },
+    {
+      "epoch": 0.6004053439219093,
+      "grad_norm": 1029.636962890625,
+      "learning_rate": 4.364589523638168e-06,
+      "loss": 103.5862,
+      "step": 72580
+    },
+    {
+      "epoch": 0.6004880671712785,
+      "grad_norm": 1054.551513671875,
+      "learning_rate": 4.363190616833598e-06,
+      "loss": 81.133,
+      "step": 72590
+    },
+    {
+      "epoch": 0.6005707904206478,
+      "grad_norm": 1133.460693359375,
+      "learning_rate": 4.361791760697027e-06,
+      "loss": 104.2316,
+      "step": 72600
+    },
+    {
+      "epoch": 0.600653513670017,
+      "grad_norm": 795.6099243164062,
+      "learning_rate": 4.360392955339758e-06,
+      "loss": 114.3038,
+      "step": 72610
+    },
+    {
+      "epoch": 0.6007362369193862,
+      "grad_norm": 892.427001953125,
+      "learning_rate": 4.358994200873085e-06,
+      "loss": 90.4173,
+      "step": 72620
+    },
+    {
+      "epoch": 0.6008189601687555,
+      "grad_norm": 511.8495788574219,
+      "learning_rate": 4.357595497408303e-06,
+      "loss": 71.118,
+      "step": 72630
+    },
+    {
+      "epoch": 0.6009016834181247,
+      "grad_norm": 1011.441650390625,
+      "learning_rate": 4.356196845056699e-06,
+      "loss": 97.9198,
+      "step": 72640
+    },
+    {
+      "epoch": 0.6009844066674939,
+      "grad_norm": 1414.120361328125,
+      "learning_rate": 4.3547982439295576e-06,
+      "loss": 116.4794,
+      "step": 72650
+    },
+    {
+      "epoch": 0.6010671299168632,
+      "grad_norm": 1085.852294921875,
+      "learning_rate": 4.353399694138158e-06,
+      "loss": 69.8157,
+      "step": 72660
+    },
+    {
+      "epoch": 0.6011498531662324,
+      "grad_norm": 836.6179809570312,
+      "learning_rate": 4.352001195793778e-06,
+      "loss": 68.6478,
+      "step": 72670
+    },
+    {
+      "epoch": 0.6012325764156016,
+      "grad_norm": 1110.6588134765625,
+      "learning_rate": 4.350602749007688e-06,
+      "loss": 70.8156,
+      "step": 72680
+    },
+    {
+      "epoch": 0.6013152996649709,
+      "grad_norm": 648.1185302734375,
+      "learning_rate": 4.349204353891158e-06,
+      "loss": 92.2575,
+      "step": 72690
+    },
+    {
+      "epoch": 0.6013980229143401,
+      "grad_norm": 998.8807373046875,
+      "learning_rate": 4.347806010555448e-06,
+      "loss": 86.0575,
+      "step": 72700
+    },
+    {
+      "epoch": 0.6014807461637093,
+      "grad_norm": 579.22265625,
+      "learning_rate": 4.346407719111823e-06,
+      "loss": 122.89,
+      "step": 72710
+    },
+    {
+      "epoch": 0.6015634694130786,
+      "grad_norm": 1019.9585571289062,
+      "learning_rate": 4.3450094796715354e-06,
+      "loss": 119.5738,
+      "step": 72720
+    },
+    {
+      "epoch": 0.6016461926624478,
+      "grad_norm": 903.5271606445312,
+      "learning_rate": 4.343611292345839e-06,
+      "loss": 107.9593,
+      "step": 72730
+    },
+    {
+      "epoch": 0.601728915911817,
+      "grad_norm": 816.7589111328125,
+      "learning_rate": 4.342213157245979e-06,
+      "loss": 84.478,
+      "step": 72740
+    },
+    {
+      "epoch": 0.6018116391611863,
+      "grad_norm": 550.0929565429688,
+      "learning_rate": 4.340815074483199e-06,
+      "loss": 69.082,
+      "step": 72750
+    },
+    {
+      "epoch": 0.6018943624105555,
+      "grad_norm": 576.094482421875,
+      "learning_rate": 4.339417044168738e-06,
+      "loss": 81.2501,
+      "step": 72760
+    },
+    {
+      "epoch": 0.6019770856599247,
+      "grad_norm": 1026.7596435546875,
+      "learning_rate": 4.338019066413832e-06,
+      "loss": 96.2626,
+      "step": 72770
+    },
+    {
+      "epoch": 0.602059808909294,
+      "grad_norm": 739.3609008789062,
+      "learning_rate": 4.33662114132971e-06,
+      "loss": 80.703,
+      "step": 72780
+    },
+    {
+      "epoch": 0.6021425321586632,
+      "grad_norm": 1358.84716796875,
+      "learning_rate": 4.335223269027599e-06,
+      "loss": 113.0873,
+      "step": 72790
+    },
+    {
+      "epoch": 0.6022252554080324,
+      "grad_norm": 932.6506958007812,
+      "learning_rate": 4.333825449618721e-06,
+      "loss": 76.9534,
+      "step": 72800
+    },
+    {
+      "epoch": 0.6023079786574017,
+      "grad_norm": 744.7998046875,
+      "learning_rate": 4.332427683214295e-06,
+      "loss": 95.1433,
+      "step": 72810
+    },
+    {
+      "epoch": 0.6023907019067709,
+      "grad_norm": 856.265380859375,
+      "learning_rate": 4.331029969925538e-06,
+      "loss": 69.2679,
+      "step": 72820
+    },
+    {
+      "epoch": 0.6024734251561401,
+      "grad_norm": 1203.6240234375,
+      "learning_rate": 4.329632309863652e-06,
+      "loss": 75.6211,
+      "step": 72830
+    },
+    {
+      "epoch": 0.6025561484055094,
+      "grad_norm": 814.6486206054688,
+      "learning_rate": 4.328234703139847e-06,
+      "loss": 83.6226,
+      "step": 72840
+    },
+    {
+      "epoch": 0.6026388716548786,
+      "grad_norm": 1125.9552001953125,
+      "learning_rate": 4.326837149865325e-06,
+      "loss": 120.5001,
+      "step": 72850
+    },
+    {
+      "epoch": 0.6027215949042478,
+      "grad_norm": 682.9524536132812,
+      "learning_rate": 4.325439650151281e-06,
+      "loss": 59.8708,
+      "step": 72860
+    },
+    {
+      "epoch": 0.6028043181536171,
+      "grad_norm": 1073.811279296875,
+      "learning_rate": 4.324042204108908e-06,
+      "loss": 79.1899,
+      "step": 72870
+    },
+    {
+      "epoch": 0.6028870414029863,
+      "grad_norm": 764.7388305664062,
+      "learning_rate": 4.322644811849395e-06,
+      "loss": 75.46,
+      "step": 72880
+    },
+    {
+      "epoch": 0.6029697646523555,
+      "grad_norm": 827.508056640625,
+      "learning_rate": 4.321247473483924e-06,
+      "loss": 73.4172,
+      "step": 72890
+    },
+    {
+      "epoch": 0.6030524879017248,
+      "grad_norm": 989.9174194335938,
+      "learning_rate": 4.319850189123681e-06,
+      "loss": 79.5693,
+      "step": 72900
+    },
+    {
+      "epoch": 0.603135211151094,
+      "grad_norm": 1122.249267578125,
+      "learning_rate": 4.3184529588798335e-06,
+      "loss": 98.8418,
+      "step": 72910
+    },
+    {
+      "epoch": 0.6032179344004632,
+      "grad_norm": 922.3504028320312,
+      "learning_rate": 4.3170557828635565e-06,
+      "loss": 139.0302,
+      "step": 72920
+    },
+    {
+      "epoch": 0.6033006576498325,
+      "grad_norm": 818.5303955078125,
+      "learning_rate": 4.315658661186016e-06,
+      "loss": 77.4534,
+      "step": 72930
+    },
+    {
+      "epoch": 0.6033833808992017,
+      "grad_norm": 795.8893432617188,
+      "learning_rate": 4.314261593958376e-06,
+      "loss": 88.9662,
+      "step": 72940
+    },
+    {
+      "epoch": 0.6034661041485709,
+      "grad_norm": 891.454833984375,
+      "learning_rate": 4.3128645812917935e-06,
+      "loss": 83.1203,
+      "step": 72950
+    },
+    {
+      "epoch": 0.6035488273979402,
+      "grad_norm": 707.6808471679688,
+      "learning_rate": 4.311467623297423e-06,
+      "loss": 88.5659,
+      "step": 72960
+    },
+    {
+      "epoch": 0.6036315506473094,
+      "grad_norm": 584.8573608398438,
+      "learning_rate": 4.310070720086414e-06,
+      "loss": 80.7194,
+      "step": 72970
+    },
+    {
+      "epoch": 0.6037142738966786,
+      "grad_norm": 665.1079711914062,
+      "learning_rate": 4.30867387176991e-06,
+      "loss": 73.1562,
+      "step": 72980
+    },
+    {
+      "epoch": 0.6037969971460478,
+      "grad_norm": 1153.0792236328125,
+      "learning_rate": 4.307277078459057e-06,
+      "loss": 89.2579,
+      "step": 72990
+    },
+    {
+      "epoch": 0.6038797203954172,
+      "grad_norm": 922.5089721679688,
+      "learning_rate": 4.305880340264985e-06,
+      "loss": 100.4802,
+      "step": 73000
+    },
+    {
+      "epoch": 0.6039624436447864,
+      "grad_norm": 940.73681640625,
+      "learning_rate": 4.3044836572988285e-06,
+      "loss": 78.375,
+      "step": 73010
+    },
+    {
+      "epoch": 0.6040451668941555,
+      "grad_norm": 867.5308227539062,
+      "learning_rate": 4.3030870296717155e-06,
+      "loss": 87.0764,
+      "step": 73020
+    },
+    {
+      "epoch": 0.6041278901435249,
+      "grad_norm": 1026.0672607421875,
+      "learning_rate": 4.301690457494769e-06,
+      "loss": 99.893,
+      "step": 73030
+    },
+    {
+      "epoch": 0.6042106133928941,
+      "grad_norm": 835.3468017578125,
+      "learning_rate": 4.300293940879108e-06,
+      "loss": 67.1481,
+      "step": 73040
+    },
+    {
+      "epoch": 0.6042933366422633,
+      "grad_norm": 1170.596923828125,
+      "learning_rate": 4.298897479935847e-06,
+      "loss": 110.3714,
+      "step": 73050
+    },
+    {
+      "epoch": 0.6043760598916326,
+      "grad_norm": 505.8780822753906,
+      "learning_rate": 4.297501074776097e-06,
+      "loss": 94.121,
+      "step": 73060
+    },
+    {
+      "epoch": 0.6044587831410018,
+      "grad_norm": 674.9179077148438,
+      "learning_rate": 4.296104725510961e-06,
+      "loss": 79.6617,
+      "step": 73070
+    },
+    {
+      "epoch": 0.604541506390371,
+      "grad_norm": 982.7257690429688,
+      "learning_rate": 4.294708432251544e-06,
+      "loss": 82.3634,
+      "step": 73080
+    },
+    {
+      "epoch": 0.6046242296397403,
+      "grad_norm": 908.1831665039062,
+      "learning_rate": 4.293312195108938e-06,
+      "loss": 96.1006,
+      "step": 73090
+    },
+    {
+      "epoch": 0.6047069528891095,
+      "grad_norm": 594.184814453125,
+      "learning_rate": 4.291916014194238e-06,
+      "loss": 87.1127,
+      "step": 73100
+    },
+    {
+      "epoch": 0.6047896761384787,
+      "grad_norm": 1231.35107421875,
+      "learning_rate": 4.290519889618531e-06,
+      "loss": 89.3334,
+      "step": 73110
+    },
+    {
+      "epoch": 0.604872399387848,
+      "grad_norm": 1115.8880615234375,
+      "learning_rate": 4.2891238214928995e-06,
+      "loss": 65.0904,
+      "step": 73120
+    },
+    {
+      "epoch": 0.6049551226372172,
+      "grad_norm": 802.1785278320312,
+      "learning_rate": 4.287727809928423e-06,
+      "loss": 83.0461,
+      "step": 73130
+    },
+    {
+      "epoch": 0.6050378458865864,
+      "grad_norm": 779.43896484375,
+      "learning_rate": 4.286331855036177e-06,
+      "loss": 94.5422,
+      "step": 73140
+    },
+    {
+      "epoch": 0.6051205691359557,
+      "grad_norm": 808.46337890625,
+      "learning_rate": 4.284935956927229e-06,
+      "loss": 96.0789,
+      "step": 73150
+    },
+    {
+      "epoch": 0.6052032923853249,
+      "grad_norm": 1208.354736328125,
+      "learning_rate": 4.283540115712647e-06,
+      "loss": 92.2509,
+      "step": 73160
+    },
+    {
+      "epoch": 0.6052860156346941,
+      "grad_norm": 839.9061889648438,
+      "learning_rate": 4.282144331503488e-06,
+      "loss": 89.2915,
+      "step": 73170
+    },
+    {
+      "epoch": 0.6053687388840634,
+      "grad_norm": 876.1185913085938,
+      "learning_rate": 4.280748604410811e-06,
+      "loss": 70.9787,
+      "step": 73180
+    },
+    {
+      "epoch": 0.6054514621334326,
+      "grad_norm": 6405.99951171875,
+      "learning_rate": 4.279352934545666e-06,
+      "loss": 128.2961,
+      "step": 73190
+    },
+    {
+      "epoch": 0.6055341853828018,
+      "grad_norm": 900.7869873046875,
+      "learning_rate": 4.277957322019101e-06,
+      "loss": 131.7455,
+      "step": 73200
+    },
+    {
+      "epoch": 0.6056169086321711,
+      "grad_norm": 611.6780395507812,
+      "learning_rate": 4.276561766942158e-06,
+      "loss": 79.8027,
+      "step": 73210
+    },
+    {
+      "epoch": 0.6056996318815403,
+      "grad_norm": 584.4486083984375,
+      "learning_rate": 4.275166269425874e-06,
+      "loss": 94.241,
+      "step": 73220
+    },
+    {
+      "epoch": 0.6057823551309095,
+      "grad_norm": 680.3448486328125,
+      "learning_rate": 4.273770829581285e-06,
+      "loss": 82.9358,
+      "step": 73230
+    },
+    {
+      "epoch": 0.6058650783802788,
+      "grad_norm": 1168.8824462890625,
+      "learning_rate": 4.272375447519418e-06,
+      "loss": 84.9492,
+      "step": 73240
+    },
+    {
+      "epoch": 0.605947801629648,
+      "grad_norm": 1853.1300048828125,
+      "learning_rate": 4.270980123351299e-06,
+      "loss": 88.8922,
+      "step": 73250
+    },
+    {
+      "epoch": 0.6060305248790172,
+      "grad_norm": 588.9846801757812,
+      "learning_rate": 4.269584857187942e-06,
+      "loss": 86.5293,
+      "step": 73260
+    },
+    {
+      "epoch": 0.6061132481283865,
+      "grad_norm": 684.2811889648438,
+      "learning_rate": 4.268189649140369e-06,
+      "loss": 75.5877,
+      "step": 73270
+    },
+    {
+      "epoch": 0.6061959713777557,
+      "grad_norm": 751.708251953125,
+      "learning_rate": 4.266794499319585e-06,
+      "loss": 94.9124,
+      "step": 73280
+    },
+    {
+      "epoch": 0.6062786946271249,
+      "grad_norm": 709.6283569335938,
+      "learning_rate": 4.265399407836598e-06,
+      "loss": 88.3451,
+      "step": 73290
+    },
+    {
+      "epoch": 0.6063614178764942,
+      "grad_norm": 558.1085815429688,
+      "learning_rate": 4.26400437480241e-06,
+      "loss": 131.0363,
+      "step": 73300
+    },
+    {
+      "epoch": 0.6064441411258634,
+      "grad_norm": 655.4963989257812,
+      "learning_rate": 4.262609400328015e-06,
+      "loss": 95.9021,
+      "step": 73310
+    },
+    {
+      "epoch": 0.6065268643752326,
+      "grad_norm": 762.81689453125,
+      "learning_rate": 4.2612144845244044e-06,
+      "loss": 92.3398,
+      "step": 73320
+    },
+    {
+      "epoch": 0.6066095876246019,
+      "grad_norm": 1324.61962890625,
+      "learning_rate": 4.259819627502571e-06,
+      "loss": 86.3487,
+      "step": 73330
+    },
+    {
+      "epoch": 0.6066923108739711,
+      "grad_norm": 915.4453125,
+      "learning_rate": 4.258424829373491e-06,
+      "loss": 95.3454,
+      "step": 73340
+    },
+    {
+      "epoch": 0.6067750341233403,
+      "grad_norm": 1026.8973388671875,
+      "learning_rate": 4.257030090248142e-06,
+      "loss": 121.2151,
+      "step": 73350
+    },
+    {
+      "epoch": 0.6068577573727096,
+      "grad_norm": 995.72412109375,
+      "learning_rate": 4.2556354102374994e-06,
+      "loss": 93.051,
+      "step": 73360
+    },
+    {
+      "epoch": 0.6069404806220788,
+      "grad_norm": 1735.9681396484375,
+      "learning_rate": 4.254240789452532e-06,
+      "loss": 123.005,
+      "step": 73370
+    },
+    {
+      "epoch": 0.607023203871448,
+      "grad_norm": 819.5543212890625,
+      "learning_rate": 4.252846228004203e-06,
+      "loss": 89.1001,
+      "step": 73380
+    },
+    {
+      "epoch": 0.6071059271208173,
+      "grad_norm": 981.048828125,
+      "learning_rate": 4.25145172600347e-06,
+      "loss": 94.8838,
+      "step": 73390
+    },
+    {
+      "epoch": 0.6071886503701865,
+      "grad_norm": 996.7430419921875,
+      "learning_rate": 4.2500572835612876e-06,
+      "loss": 130.0341,
+      "step": 73400
+    },
+    {
+      "epoch": 0.6072713736195557,
+      "grad_norm": 890.7987060546875,
+      "learning_rate": 4.248662900788605e-06,
+      "loss": 95.0139,
+      "step": 73410
+    },
+    {
+      "epoch": 0.607354096868925,
+      "grad_norm": 766.2911376953125,
+      "learning_rate": 4.247268577796368e-06,
+      "loss": 61.7691,
+      "step": 73420
+    },
+    {
+      "epoch": 0.6074368201182943,
+      "grad_norm": 953.1488037109375,
+      "learning_rate": 4.245874314695516e-06,
+      "loss": 146.5053,
+      "step": 73430
+    },
+    {
+      "epoch": 0.6075195433676635,
+      "grad_norm": 1011.7604370117188,
+      "learning_rate": 4.244480111596984e-06,
+      "loss": 85.3021,
+      "step": 73440
+    },
+    {
+      "epoch": 0.6076022666170328,
+      "grad_norm": 575.0761108398438,
+      "learning_rate": 4.2430859686117e-06,
+      "loss": 95.1132,
+      "step": 73450
+    },
+    {
+      "epoch": 0.607684989866402,
+      "grad_norm": 753.287353515625,
+      "learning_rate": 4.241691885850593e-06,
+      "loss": 75.5522,
+      "step": 73460
+    },
+    {
+      "epoch": 0.6077677131157712,
+      "grad_norm": 2446.799072265625,
+      "learning_rate": 4.240297863424582e-06,
+      "loss": 84.9961,
+      "step": 73470
+    },
+    {
+      "epoch": 0.6078504363651405,
+      "grad_norm": 1110.462158203125,
+      "learning_rate": 4.2389039014445846e-06,
+      "loss": 92.255,
+      "step": 73480
+    },
+    {
+      "epoch": 0.6079331596145097,
+      "grad_norm": 927.8515014648438,
+      "learning_rate": 4.23751000002151e-06,
+      "loss": 101.0412,
+      "step": 73490
+    },
+    {
+      "epoch": 0.6080158828638789,
+      "grad_norm": 1213.8529052734375,
+      "learning_rate": 4.2361161592662655e-06,
+      "loss": 91.7586,
+      "step": 73500
+    },
+    {
+      "epoch": 0.6080986061132482,
+      "grad_norm": 960.1309814453125,
+      "learning_rate": 4.234722379289753e-06,
+      "loss": 129.1659,
+      "step": 73510
+    },
+    {
+      "epoch": 0.6081813293626174,
+      "grad_norm": 704.3455810546875,
+      "learning_rate": 4.233328660202869e-06,
+      "loss": 67.6366,
+      "step": 73520
+    },
+    {
+      "epoch": 0.6082640526119866,
+      "grad_norm": 843.2509765625,
+      "learning_rate": 4.231935002116504e-06,
+      "loss": 78.1124,
+      "step": 73530
+    },
+    {
+      "epoch": 0.6083467758613559,
+      "grad_norm": 598.6126708984375,
+      "learning_rate": 4.230541405141546e-06,
+      "loss": 81.4852,
+      "step": 73540
+    },
+    {
+      "epoch": 0.6084294991107251,
+      "grad_norm": 1050.9449462890625,
+      "learning_rate": 4.229147869388875e-06,
+      "loss": 94.9926,
+      "step": 73550
+    },
+    {
+      "epoch": 0.6085122223600943,
+      "grad_norm": 839.3172607421875,
+      "learning_rate": 4.227754394969373e-06,
+      "loss": 104.2477,
+      "step": 73560
+    },
+    {
+      "epoch": 0.6085949456094636,
+      "grad_norm": 916.7781372070312,
+      "learning_rate": 4.226360981993909e-06,
+      "loss": 104.4644,
+      "step": 73570
+    },
+    {
+      "epoch": 0.6086776688588328,
+      "grad_norm": 1136.1072998046875,
+      "learning_rate": 4.224967630573351e-06,
+      "loss": 97.9894,
+      "step": 73580
+    },
+    {
+      "epoch": 0.608760392108202,
+      "grad_norm": 1269.821044921875,
+      "learning_rate": 4.2235743408185635e-06,
+      "loss": 80.4433,
+      "step": 73590
+    },
+    {
+      "epoch": 0.6088431153575713,
+      "grad_norm": 922.0454711914062,
+      "learning_rate": 4.222181112840401e-06,
+      "loss": 86.2411,
+      "step": 73600
+    },
+    {
+      "epoch": 0.6089258386069405,
+      "grad_norm": 841.0752563476562,
+      "learning_rate": 4.220787946749717e-06,
+      "loss": 89.3184,
+      "step": 73610
+    },
+    {
+      "epoch": 0.6090085618563097,
+      "grad_norm": 681.7354736328125,
+      "learning_rate": 4.219394842657361e-06,
+      "loss": 86.7437,
+      "step": 73620
+    },
+    {
+      "epoch": 0.609091285105679,
+      "grad_norm": 935.4651489257812,
+      "learning_rate": 4.218001800674174e-06,
+      "loss": 88.4934,
+      "step": 73630
+    },
+    {
+      "epoch": 0.6091740083550482,
+      "grad_norm": 675.4620971679688,
+      "learning_rate": 4.216608820910995e-06,
+      "loss": 67.0971,
+      "step": 73640
+    },
+    {
+      "epoch": 0.6092567316044174,
+      "grad_norm": 710.9379272460938,
+      "learning_rate": 4.2152159034786554e-06,
+      "loss": 62.8751,
+      "step": 73650
+    },
+    {
+      "epoch": 0.6093394548537867,
+      "grad_norm": 829.56103515625,
+      "learning_rate": 4.213823048487987e-06,
+      "loss": 108.2473,
+      "step": 73660
+    },
+    {
+      "epoch": 0.6094221781031559,
+      "grad_norm": 757.4158935546875,
+      "learning_rate": 4.212430256049809e-06,
+      "loss": 99.3676,
+      "step": 73670
+    },
+    {
+      "epoch": 0.6095049013525251,
+      "grad_norm": 564.0368041992188,
+      "learning_rate": 4.2110375262749435e-06,
+      "loss": 86.9687,
+      "step": 73680
+    },
+    {
+      "epoch": 0.6095876246018944,
+      "grad_norm": 513.57470703125,
+      "learning_rate": 4.209644859274199e-06,
+      "loss": 70.3236,
+      "step": 73690
+    },
+    {
+      "epoch": 0.6096703478512636,
+      "grad_norm": 1249.1611328125,
+      "learning_rate": 4.208252255158387e-06,
+      "loss": 94.686,
+      "step": 73700
+    },
+    {
+      "epoch": 0.6097530711006328,
+      "grad_norm": 1189.990478515625,
+      "learning_rate": 4.2068597140383084e-06,
+      "loss": 114.4235,
+      "step": 73710
+    },
+    {
+      "epoch": 0.609835794350002,
+      "grad_norm": 561.5841674804688,
+      "learning_rate": 4.205467236024763e-06,
+      "loss": 97.9067,
+      "step": 73720
+    },
+    {
+      "epoch": 0.6099185175993713,
+      "grad_norm": 1291.4609375,
+      "learning_rate": 4.204074821228542e-06,
+      "loss": 97.8942,
+      "step": 73730
+    },
+    {
+      "epoch": 0.6100012408487405,
+      "grad_norm": 960.28076171875,
+      "learning_rate": 4.202682469760436e-06,
+      "loss": 82.6604,
+      "step": 73740
+    },
+    {
+      "epoch": 0.6100839640981097,
+      "grad_norm": 1440.652587890625,
+      "learning_rate": 4.2012901817312255e-06,
+      "loss": 116.6677,
+      "step": 73750
+    },
+    {
+      "epoch": 0.610166687347479,
+      "grad_norm": 665.2211303710938,
+      "learning_rate": 4.199897957251693e-06,
+      "loss": 82.3861,
+      "step": 73760
+    },
+    {
+      "epoch": 0.6102494105968482,
+      "grad_norm": 2500.0634765625,
+      "learning_rate": 4.198505796432605e-06,
+      "loss": 163.031,
+      "step": 73770
+    },
+    {
+      "epoch": 0.6103321338462174,
+      "grad_norm": 365.54327392578125,
+      "learning_rate": 4.197113699384732e-06,
+      "loss": 84.3422,
+      "step": 73780
+    },
+    {
+      "epoch": 0.6104148570955867,
+      "grad_norm": 1687.4744873046875,
+      "learning_rate": 4.1957216662188385e-06,
+      "loss": 114.6087,
+      "step": 73790
+    },
+    {
+      "epoch": 0.6104975803449559,
+      "grad_norm": 1095.3680419921875,
+      "learning_rate": 4.194329697045681e-06,
+      "loss": 79.8403,
+      "step": 73800
+    },
+    {
+      "epoch": 0.6105803035943251,
+      "grad_norm": 939.5096435546875,
+      "learning_rate": 4.19293779197601e-06,
+      "loss": 76.4956,
+      "step": 73810
+    },
+    {
+      "epoch": 0.6106630268436944,
+      "grad_norm": 1039.002197265625,
+      "learning_rate": 4.191545951120577e-06,
+      "loss": 127.6832,
+      "step": 73820
+    },
+    {
+      "epoch": 0.6107457500930636,
+      "grad_norm": 605.078857421875,
+      "learning_rate": 4.190154174590122e-06,
+      "loss": 107.0245,
+      "step": 73830
+    },
+    {
+      "epoch": 0.6108284733424328,
+      "grad_norm": 1389.932373046875,
+      "learning_rate": 4.188762462495381e-06,
+      "loss": 127.3608,
+      "step": 73840
+    },
+    {
+      "epoch": 0.6109111965918022,
+      "grad_norm": 716.9305419921875,
+      "learning_rate": 4.187370814947091e-06,
+      "loss": 74.3108,
+      "step": 73850
+    },
+    {
+      "epoch": 0.6109939198411714,
+      "grad_norm": 835.6632690429688,
+      "learning_rate": 4.185979232055975e-06,
+      "loss": 104.9701,
+      "step": 73860
+    },
+    {
+      "epoch": 0.6110766430905405,
+      "grad_norm": 960.3486938476562,
+      "learning_rate": 4.184587713932755e-06,
+      "loss": 118.767,
+      "step": 73870
+    },
+    {
+      "epoch": 0.6111593663399099,
+      "grad_norm": 811.2726440429688,
+      "learning_rate": 4.183196260688147e-06,
+      "loss": 83.4491,
+      "step": 73880
+    },
+    {
+      "epoch": 0.6112420895892791,
+      "grad_norm": 1344.4681396484375,
+      "learning_rate": 4.1818048724328646e-06,
+      "loss": 99.5997,
+      "step": 73890
+    },
+    {
+      "epoch": 0.6113248128386483,
+      "grad_norm": 1349.948486328125,
+      "learning_rate": 4.180413549277614e-06,
+      "loss": 101.2212,
+      "step": 73900
+    },
+    {
+      "epoch": 0.6114075360880176,
+      "grad_norm": 646.7472534179688,
+      "learning_rate": 4.1790222913330955e-06,
+      "loss": 89.3753,
+      "step": 73910
+    },
+    {
+      "epoch": 0.6114902593373868,
+      "grad_norm": 920.3361206054688,
+      "learning_rate": 4.1776310987100054e-06,
+      "loss": 97.7225,
+      "step": 73920
+    },
+    {
+      "epoch": 0.611572982586756,
+      "grad_norm": 1254.9158935546875,
+      "learning_rate": 4.1762399715190366e-06,
+      "loss": 104.7424,
+      "step": 73930
+    },
+    {
+      "epoch": 0.6116557058361253,
+      "grad_norm": 821.2960815429688,
+      "learning_rate": 4.1748489098708715e-06,
+      "loss": 106.9926,
+      "step": 73940
+    },
+    {
+      "epoch": 0.6117384290854945,
+      "grad_norm": 716.6043701171875,
+      "learning_rate": 4.173457913876191e-06,
+      "loss": 92.9678,
+      "step": 73950
+    },
+    {
+      "epoch": 0.6118211523348637,
+      "grad_norm": 1183.022216796875,
+      "learning_rate": 4.172066983645671e-06,
+      "loss": 117.3768,
+      "step": 73960
+    },
+    {
+      "epoch": 0.611903875584233,
+      "grad_norm": 836.3502807617188,
+      "learning_rate": 4.170676119289982e-06,
+      "loss": 81.8189,
+      "step": 73970
+    },
+    {
+      "epoch": 0.6119865988336022,
+      "grad_norm": 491.899658203125,
+      "learning_rate": 4.1692853209197865e-06,
+      "loss": 81.1659,
+      "step": 73980
+    },
+    {
+      "epoch": 0.6120693220829714,
+      "grad_norm": 1108.9029541015625,
+      "learning_rate": 4.167894588645746e-06,
+      "loss": 63.8417,
+      "step": 73990
+    },
+    {
+      "epoch": 0.6121520453323407,
+      "grad_norm": 1243.21533203125,
+      "learning_rate": 4.166503922578516e-06,
+      "loss": 71.4258,
+      "step": 74000
+    },
+    {
+      "epoch": 0.6122347685817099,
+      "grad_norm": 1081.2105712890625,
+      "learning_rate": 4.165113322828743e-06,
+      "loss": 110.2544,
+      "step": 74010
+    },
+    {
+      "epoch": 0.6123174918310791,
+      "grad_norm": 857.2208862304688,
+      "learning_rate": 4.163722789507071e-06,
+      "loss": 97.8825,
+      "step": 74020
+    },
+    {
+      "epoch": 0.6124002150804484,
+      "grad_norm": 1000.3453369140625,
+      "learning_rate": 4.162332322724139e-06,
+      "loss": 64.7686,
+      "step": 74030
+    },
+    {
+      "epoch": 0.6124829383298176,
+      "grad_norm": 536.7930297851562,
+      "learning_rate": 4.16094192259058e-06,
+      "loss": 82.5932,
+      "step": 74040
+    },
+    {
+      "epoch": 0.6125656615791868,
+      "grad_norm": 1039.748779296875,
+      "learning_rate": 4.15955158921702e-06,
+      "loss": 94.5069,
+      "step": 74050
+    },
+    {
+      "epoch": 0.6126483848285561,
+      "grad_norm": 903.4092407226562,
+      "learning_rate": 4.158161322714085e-06,
+      "loss": 93.7216,
+      "step": 74060
+    },
+    {
+      "epoch": 0.6127311080779253,
+      "grad_norm": 1123.583740234375,
+      "learning_rate": 4.1567711231923876e-06,
+      "loss": 97.1475,
+      "step": 74070
+    },
+    {
+      "epoch": 0.6128138313272945,
+      "grad_norm": 614.0092163085938,
+      "learning_rate": 4.155380990762542e-06,
+      "loss": 69.82,
+      "step": 74080
+    },
+    {
+      "epoch": 0.6128965545766638,
+      "grad_norm": 436.3868103027344,
+      "learning_rate": 4.153990925535157e-06,
+      "loss": 85.2651,
+      "step": 74090
+    },
+    {
+      "epoch": 0.612979277826033,
+      "grad_norm": 1019.7743530273438,
+      "learning_rate": 4.15260092762083e-06,
+      "loss": 68.9177,
+      "step": 74100
+    },
+    {
+      "epoch": 0.6130620010754022,
+      "grad_norm": 1233.597412109375,
+      "learning_rate": 4.151210997130159e-06,
+      "loss": 126.1689,
+      "step": 74110
+    },
+    {
+      "epoch": 0.6131447243247715,
+      "grad_norm": 1863.585205078125,
+      "learning_rate": 4.1498211341737335e-06,
+      "loss": 109.2528,
+      "step": 74120
+    },
+    {
+      "epoch": 0.6132274475741407,
+      "grad_norm": 603.75634765625,
+      "learning_rate": 4.148431338862138e-06,
+      "loss": 83.3835,
+      "step": 74130
+    },
+    {
+      "epoch": 0.6133101708235099,
+      "grad_norm": 1063.5419921875,
+      "learning_rate": 4.147041611305952e-06,
+      "loss": 81.7966,
+      "step": 74140
+    },
+    {
+      "epoch": 0.6133928940728792,
+      "grad_norm": 553.776123046875,
+      "learning_rate": 4.145651951615752e-06,
+      "loss": 103.0459,
+      "step": 74150
+    },
+    {
+      "epoch": 0.6134756173222484,
+      "grad_norm": 783.0980834960938,
+      "learning_rate": 4.144262359902104e-06,
+      "loss": 88.7107,
+      "step": 74160
+    },
+    {
+      "epoch": 0.6135583405716176,
+      "grad_norm": 1454.119873046875,
+      "learning_rate": 4.142872836275572e-06,
+      "loss": 95.9363,
+      "step": 74170
+    },
+    {
+      "epoch": 0.6136410638209869,
+      "grad_norm": 1249.62548828125,
+      "learning_rate": 4.141483380846716e-06,
+      "loss": 85.594,
+      "step": 74180
+    },
+    {
+      "epoch": 0.6137237870703561,
+      "grad_norm": 1073.4764404296875,
+      "learning_rate": 4.1400939937260894e-06,
+      "loss": 78.375,
+      "step": 74190
+    },
+    {
+      "epoch": 0.6138065103197253,
+      "grad_norm": 762.4814453125,
+      "learning_rate": 4.138704675024235e-06,
+      "loss": 87.9466,
+      "step": 74200
+    },
+    {
+      "epoch": 0.6138892335690946,
+      "grad_norm": 1005.1492919921875,
+      "learning_rate": 4.137315424851696e-06,
+      "loss": 99.7527,
+      "step": 74210
+    },
+    {
+      "epoch": 0.6139719568184638,
+      "grad_norm": 892.3167114257812,
+      "learning_rate": 4.1359262433190105e-06,
+      "loss": 88.9903,
+      "step": 74220
+    },
+    {
+      "epoch": 0.614054680067833,
+      "grad_norm": 736.7813110351562,
+      "learning_rate": 4.134537130536708e-06,
+      "loss": 72.6644,
+      "step": 74230
+    },
+    {
+      "epoch": 0.6141374033172023,
+      "grad_norm": 2169.072998046875,
+      "learning_rate": 4.133148086615314e-06,
+      "loss": 91.2334,
+      "step": 74240
+    },
+    {
+      "epoch": 0.6142201265665715,
+      "grad_norm": 918.8009643554688,
+      "learning_rate": 4.131759111665349e-06,
+      "loss": 88.9064,
+      "step": 74250
+    },
+    {
+      "epoch": 0.6143028498159407,
+      "grad_norm": 991.6587524414062,
+      "learning_rate": 4.130370205797326e-06,
+      "loss": 126.6198,
+      "step": 74260
+    },
+    {
+      "epoch": 0.61438557306531,
+      "grad_norm": 626.6166381835938,
+      "learning_rate": 4.128981369121754e-06,
+      "loss": 101.1643,
+      "step": 74270
+    },
+    {
+      "epoch": 0.6144682963146793,
+      "grad_norm": 1045.540283203125,
+      "learning_rate": 4.127592601749141e-06,
+      "loss": 100.4899,
+      "step": 74280
+    },
+    {
+      "epoch": 0.6145510195640485,
+      "grad_norm": 948.6563720703125,
+      "learning_rate": 4.1262039037899775e-06,
+      "loss": 92.1499,
+      "step": 74290
+    },
+    {
+      "epoch": 0.6146337428134178,
+      "grad_norm": 672.4278564453125,
+      "learning_rate": 4.12481527535476e-06,
+      "loss": 90.7191,
+      "step": 74300
+    },
+    {
+      "epoch": 0.614716466062787,
+      "grad_norm": 1149.4989013671875,
+      "learning_rate": 4.123426716553972e-06,
+      "loss": 97.7278,
+      "step": 74310
+    },
+    {
+      "epoch": 0.6147991893121562,
+      "grad_norm": 1198.019775390625,
+      "learning_rate": 4.122038227498101e-06,
+      "loss": 84.5663,
+      "step": 74320
+    },
+    {
+      "epoch": 0.6148819125615255,
+      "grad_norm": 1210.01611328125,
+      "learning_rate": 4.120649808297616e-06,
+      "loss": 116.1529,
+      "step": 74330
+    },
+    {
+      "epoch": 0.6149646358108947,
+      "grad_norm": 815.8023681640625,
+      "learning_rate": 4.119261459062992e-06,
+      "loss": 94.437,
+      "step": 74340
+    },
+    {
+      "epoch": 0.6150473590602639,
+      "grad_norm": 616.3191528320312,
+      "learning_rate": 4.1178731799046915e-06,
+      "loss": 110.6811,
+      "step": 74350
+    },
+    {
+      "epoch": 0.6151300823096332,
+      "grad_norm": 938.7825927734375,
+      "learning_rate": 4.116484970933174e-06,
+      "loss": 85.8687,
+      "step": 74360
+    },
+    {
+      "epoch": 0.6152128055590024,
+      "grad_norm": 754.9133911132812,
+      "learning_rate": 4.1150968322588915e-06,
+      "loss": 85.892,
+      "step": 74370
+    },
+    {
+      "epoch": 0.6152955288083716,
+      "grad_norm": 982.1900024414062,
+      "learning_rate": 4.113708763992294e-06,
+      "loss": 98.6362,
+      "step": 74380
+    },
+    {
+      "epoch": 0.6153782520577409,
+      "grad_norm": 1620.11767578125,
+      "learning_rate": 4.1123207662438216e-06,
+      "loss": 99.9468,
+      "step": 74390
+    },
+    {
+      "epoch": 0.6154609753071101,
+      "grad_norm": 843.5370483398438,
+      "learning_rate": 4.110932839123911e-06,
+      "loss": 102.8236,
+      "step": 74400
+    },
+    {
+      "epoch": 0.6155436985564793,
+      "grad_norm": 1027.692138671875,
+      "learning_rate": 4.109544982742995e-06,
+      "loss": 87.6993,
+      "step": 74410
+    },
+    {
+      "epoch": 0.6156264218058485,
+      "grad_norm": 922.636962890625,
+      "learning_rate": 4.108157197211499e-06,
+      "loss": 91.7235,
+      "step": 74420
+    },
+    {
+      "epoch": 0.6157091450552178,
+      "grad_norm": 592.3994750976562,
+      "learning_rate": 4.1067694826398405e-06,
+      "loss": 106.6859,
+      "step": 74430
+    },
+    {
+      "epoch": 0.615791868304587,
+      "grad_norm": 576.2105102539062,
+      "learning_rate": 4.105381839138436e-06,
+      "loss": 67.9388,
+      "step": 74440
+    },
+    {
+      "epoch": 0.6158745915539562,
+      "grad_norm": 1076.5267333984375,
+      "learning_rate": 4.103994266817694e-06,
+      "loss": 81.3077,
+      "step": 74450
+    },
+    {
+      "epoch": 0.6159573148033255,
+      "grad_norm": 1463.522216796875,
+      "learning_rate": 4.102606765788014e-06,
+      "loss": 121.3816,
+      "step": 74460
+    },
+    {
+      "epoch": 0.6160400380526947,
+      "grad_norm": 523.5887451171875,
+      "learning_rate": 4.101219336159795e-06,
+      "loss": 96.6892,
+      "step": 74470
+    },
+    {
+      "epoch": 0.6161227613020639,
+      "grad_norm": 801.384765625,
+      "learning_rate": 4.09983197804343e-06,
+      "loss": 90.4624,
+      "step": 74480
+    },
+    {
+      "epoch": 0.6162054845514332,
+      "grad_norm": 815.4998168945312,
+      "learning_rate": 4.098444691549302e-06,
+      "loss": 85.0159,
+      "step": 74490
+    },
+    {
+      "epoch": 0.6162882078008024,
+      "grad_norm": 1301.9669189453125,
+      "learning_rate": 4.097057476787792e-06,
+      "loss": 101.675,
+      "step": 74500
+    },
+    {
+      "epoch": 0.6163709310501716,
+      "grad_norm": 2187.890869140625,
+      "learning_rate": 4.0956703338692755e-06,
+      "loss": 93.8541,
+      "step": 74510
+    },
+    {
+      "epoch": 0.6164536542995409,
+      "grad_norm": 2067.32470703125,
+      "learning_rate": 4.09428326290412e-06,
+      "loss": 100.0631,
+      "step": 74520
+    },
+    {
+      "epoch": 0.6165363775489101,
+      "grad_norm": 1015.9877319335938,
+      "learning_rate": 4.092896264002689e-06,
+      "loss": 88.7052,
+      "step": 74530
+    },
+    {
+      "epoch": 0.6166191007982793,
+      "grad_norm": 1361.1781005859375,
+      "learning_rate": 4.09150933727534e-06,
+      "loss": 98.1841,
+      "step": 74540
+    },
+    {
+      "epoch": 0.6167018240476486,
+      "grad_norm": 1028.168701171875,
+      "learning_rate": 4.0901224828324225e-06,
+      "loss": 92.4252,
+      "step": 74550
+    },
+    {
+      "epoch": 0.6167845472970178,
+      "grad_norm": 841.4468383789062,
+      "learning_rate": 4.088735700784283e-06,
+      "loss": 96.4391,
+      "step": 74560
+    },
+    {
+      "epoch": 0.616867270546387,
+      "grad_norm": 941.0628662109375,
+      "learning_rate": 4.087348991241262e-06,
+      "loss": 79.4249,
+      "step": 74570
+    },
+    {
+      "epoch": 0.6169499937957563,
+      "grad_norm": 550.6486206054688,
+      "learning_rate": 4.0859623543136935e-06,
+      "loss": 71.2127,
+      "step": 74580
+    },
+    {
+      "epoch": 0.6170327170451255,
+      "grad_norm": 396.369140625,
+      "learning_rate": 4.084575790111905e-06,
+      "loss": 119.8758,
+      "step": 74590
+    },
+    {
+      "epoch": 0.6171154402944947,
+      "grad_norm": 714.75634765625,
+      "learning_rate": 4.08318929874622e-06,
+      "loss": 97.5772,
+      "step": 74600
+    },
+    {
+      "epoch": 0.617198163543864,
+      "grad_norm": 1025.8707275390625,
+      "learning_rate": 4.081802880326955e-06,
+      "loss": 109.0059,
+      "step": 74610
+    },
+    {
+      "epoch": 0.6172808867932332,
+      "grad_norm": 819.5011596679688,
+      "learning_rate": 4.080416534964422e-06,
+      "loss": 106.0014,
+      "step": 74620
+    },
+    {
+      "epoch": 0.6173636100426024,
+      "grad_norm": 628.18359375,
+      "learning_rate": 4.079030262768924e-06,
+      "loss": 99.0313,
+      "step": 74630
+    },
+    {
+      "epoch": 0.6174463332919717,
+      "grad_norm": 1026.5445556640625,
+      "learning_rate": 4.077644063850761e-06,
+      "loss": 106.1461,
+      "step": 74640
+    },
+    {
+      "epoch": 0.6175290565413409,
+      "grad_norm": 479.8872985839844,
+      "learning_rate": 4.076257938320226e-06,
+      "loss": 77.8964,
+      "step": 74650
+    },
+    {
+      "epoch": 0.6176117797907101,
+      "grad_norm": 782.7094116210938,
+      "learning_rate": 4.074871886287609e-06,
+      "loss": 123.7894,
+      "step": 74660
+    },
+    {
+      "epoch": 0.6176945030400794,
+      "grad_norm": 708.5579833984375,
+      "learning_rate": 4.073485907863189e-06,
+      "loss": 62.82,
+      "step": 74670
+    },
+    {
+      "epoch": 0.6177772262894486,
+      "grad_norm": 1126.7977294921875,
+      "learning_rate": 4.0721000031572445e-06,
+      "loss": 85.1428,
+      "step": 74680
+    },
+    {
+      "epoch": 0.6178599495388178,
+      "grad_norm": 664.0321655273438,
+      "learning_rate": 4.070714172280043e-06,
+      "loss": 87.29,
+      "step": 74690
+    },
+    {
+      "epoch": 0.6179426727881872,
+      "grad_norm": 716.9480590820312,
+      "learning_rate": 4.06932841534185e-06,
+      "loss": 85.6546,
+      "step": 74700
+    },
+    {
+      "epoch": 0.6180253960375564,
+      "grad_norm": 894.58935546875,
+      "learning_rate": 4.067942732452926e-06,
+      "loss": 98.9088,
+      "step": 74710
+    },
+    {
+      "epoch": 0.6181081192869256,
+      "grad_norm": 844.1010131835938,
+      "learning_rate": 4.06655712372352e-06,
+      "loss": 114.6562,
+      "step": 74720
+    },
+    {
+      "epoch": 0.6181908425362949,
+      "grad_norm": 735.4295654296875,
+      "learning_rate": 4.065171589263878e-06,
+      "loss": 87.0721,
+      "step": 74730
+    },
+    {
+      "epoch": 0.6182735657856641,
+      "grad_norm": 758.739990234375,
+      "learning_rate": 4.063786129184243e-06,
+      "loss": 108.7126,
+      "step": 74740
+    },
+    {
+      "epoch": 0.6183562890350333,
+      "grad_norm": 1705.91064453125,
+      "learning_rate": 4.06240074359485e-06,
+      "loss": 130.587,
+      "step": 74750
+    },
+    {
+      "epoch": 0.6184390122844026,
+      "grad_norm": 498.7382507324219,
+      "learning_rate": 4.061015432605927e-06,
+      "loss": 76.9041,
+      "step": 74760
+    },
+    {
+      "epoch": 0.6185217355337718,
+      "grad_norm": 610.8711547851562,
+      "learning_rate": 4.059630196327696e-06,
+      "loss": 93.8699,
+      "step": 74770
+    },
+    {
+      "epoch": 0.618604458783141,
+      "grad_norm": 850.5593872070312,
+      "learning_rate": 4.058245034870375e-06,
+      "loss": 84.9047,
+      "step": 74780
+    },
+    {
+      "epoch": 0.6186871820325103,
+      "grad_norm": 833.0304565429688,
+      "learning_rate": 4.056859948344175e-06,
+      "loss": 71.8329,
+      "step": 74790
+    },
+    {
+      "epoch": 0.6187699052818795,
+      "grad_norm": 758.42236328125,
+      "learning_rate": 4.0554749368593e-06,
+      "loss": 69.1855,
+      "step": 74800
+    },
+    {
+      "epoch": 0.6188526285312487,
+      "grad_norm": 853.135009765625,
+      "learning_rate": 4.054090000525949e-06,
+      "loss": 74.7512,
+      "step": 74810
+    },
+    {
+      "epoch": 0.618935351780618,
+      "grad_norm": 817.5387573242188,
+      "learning_rate": 4.052705139454316e-06,
+      "loss": 116.5073,
+      "step": 74820
+    },
+    {
+      "epoch": 0.6190180750299872,
+      "grad_norm": 1293.7938232421875,
+      "learning_rate": 4.051320353754586e-06,
+      "loss": 102.931,
+      "step": 74830
+    },
+    {
+      "epoch": 0.6191007982793564,
+      "grad_norm": 931.7042236328125,
+      "learning_rate": 4.049935643536943e-06,
+      "loss": 96.7613,
+      "step": 74840
+    },
+    {
+      "epoch": 0.6191835215287257,
+      "grad_norm": 495.9801940917969,
+      "learning_rate": 4.048551008911561e-06,
+      "loss": 89.8051,
+      "step": 74850
+    },
+    {
+      "epoch": 0.6192662447780949,
+      "grad_norm": 587.1924438476562,
+      "learning_rate": 4.0471664499886074e-06,
+      "loss": 116.4916,
+      "step": 74860
+    },
+    {
+      "epoch": 0.6193489680274641,
+      "grad_norm": 769.1555786132812,
+      "learning_rate": 4.045781966878247e-06,
+      "loss": 85.5647,
+      "step": 74870
+    },
+    {
+      "epoch": 0.6194316912768334,
+      "grad_norm": 1216.7843017578125,
+      "learning_rate": 4.044397559690638e-06,
+      "loss": 108.6549,
+      "step": 74880
+    },
+    {
+      "epoch": 0.6195144145262026,
+      "grad_norm": 764.6852416992188,
+      "learning_rate": 4.043013228535928e-06,
+      "loss": 113.6466,
+      "step": 74890
+    },
+    {
+      "epoch": 0.6195971377755718,
+      "grad_norm": 934.8550415039062,
+      "learning_rate": 4.041628973524264e-06,
+      "loss": 130.2357,
+      "step": 74900
+    },
+    {
+      "epoch": 0.6196798610249411,
+      "grad_norm": 377.97088623046875,
+      "learning_rate": 4.040244794765783e-06,
+      "loss": 85.8722,
+      "step": 74910
+    },
+    {
+      "epoch": 0.6197625842743103,
+      "grad_norm": 1394.6834716796875,
+      "learning_rate": 4.03886069237062e-06,
+      "loss": 91.4125,
+      "step": 74920
+    },
+    {
+      "epoch": 0.6198453075236795,
+      "grad_norm": 595.1375122070312,
+      "learning_rate": 4.037476666448899e-06,
+      "loss": 71.7473,
+      "step": 74930
+    },
+    {
+      "epoch": 0.6199280307730488,
+      "grad_norm": 595.4174194335938,
+      "learning_rate": 4.0360927171107436e-06,
+      "loss": 93.8769,
+      "step": 74940
+    },
+    {
+      "epoch": 0.620010754022418,
+      "grad_norm": 718.7639770507812,
+      "learning_rate": 4.034708844466267e-06,
+      "loss": 102.3341,
+      "step": 74950
+    },
+    {
+      "epoch": 0.6200934772717872,
+      "grad_norm": 968.3093872070312,
+      "learning_rate": 4.033325048625578e-06,
+      "loss": 90.3649,
+      "step": 74960
+    },
+    {
+      "epoch": 0.6201762005211565,
+      "grad_norm": 1063.10009765625,
+      "learning_rate": 4.031941329698778e-06,
+      "loss": 104.0913,
+      "step": 74970
+    },
+    {
+      "epoch": 0.6202589237705257,
+      "grad_norm": 1573.7166748046875,
+      "learning_rate": 4.030557687795965e-06,
+      "loss": 92.6962,
+      "step": 74980
+    },
+    {
+      "epoch": 0.6203416470198949,
+      "grad_norm": 1660.5982666015625,
+      "learning_rate": 4.029174123027226e-06,
+      "loss": 135.9649,
+      "step": 74990
+    },
+    {
+      "epoch": 0.6204243702692642,
+      "grad_norm": 1034.1180419921875,
+      "learning_rate": 4.027790635502646e-06,
+      "loss": 94.0106,
+      "step": 75000
+    },
+    {
+      "epoch": 0.6205070935186334,
+      "grad_norm": 742.1674194335938,
+      "learning_rate": 4.026407225332305e-06,
+      "loss": 109.5858,
+      "step": 75010
+    },
+    {
+      "epoch": 0.6205898167680026,
+      "grad_norm": 586.0728759765625,
+      "learning_rate": 4.025023892626272e-06,
+      "loss": 76.6323,
+      "step": 75020
+    },
+    {
+      "epoch": 0.6206725400173719,
+      "grad_norm": 1411.9517822265625,
+      "learning_rate": 4.023640637494612e-06,
+      "loss": 135.056,
+      "step": 75030
+    },
+    {
+      "epoch": 0.6207552632667411,
+      "grad_norm": 1079.8941650390625,
+      "learning_rate": 4.022257460047387e-06,
+      "loss": 97.0648,
+      "step": 75040
+    },
+    {
+      "epoch": 0.6208379865161103,
+      "grad_norm": 791.2523803710938,
+      "learning_rate": 4.0208743603946505e-06,
+      "loss": 102.2262,
+      "step": 75050
+    },
+    {
+      "epoch": 0.6209207097654796,
+      "grad_norm": 1095.387939453125,
+      "learning_rate": 4.0194913386464445e-06,
+      "loss": 95.2964,
+      "step": 75060
+    },
+    {
+      "epoch": 0.6210034330148488,
+      "grad_norm": 601.5514526367188,
+      "learning_rate": 4.018108394912814e-06,
+      "loss": 88.666,
+      "step": 75070
+    },
+    {
+      "epoch": 0.621086156264218,
+      "grad_norm": 747.2603149414062,
+      "learning_rate": 4.016725529303792e-06,
+      "loss": 84.6649,
+      "step": 75080
+    },
+    {
+      "epoch": 0.6211688795135873,
+      "grad_norm": 619.85791015625,
+      "learning_rate": 4.015342741929407e-06,
+      "loss": 73.6812,
+      "step": 75090
+    },
+    {
+      "epoch": 0.6212516027629565,
+      "grad_norm": 1243.3499755859375,
+      "learning_rate": 4.013960032899681e-06,
+      "loss": 100.2627,
+      "step": 75100
+    },
+    {
+      "epoch": 0.6213343260123257,
+      "grad_norm": 491.8591003417969,
+      "learning_rate": 4.012577402324631e-06,
+      "loss": 99.1271,
+      "step": 75110
+    },
+    {
+      "epoch": 0.621417049261695,
+      "grad_norm": 955.7748413085938,
+      "learning_rate": 4.011194850314263e-06,
+      "loss": 87.4493,
+      "step": 75120
+    },
+    {
+      "epoch": 0.6214997725110643,
+      "grad_norm": 895.6536254882812,
+      "learning_rate": 4.009812376978585e-06,
+      "loss": 111.0081,
+      "step": 75130
+    },
+    {
+      "epoch": 0.6215824957604335,
+      "grad_norm": 513.0172119140625,
+      "learning_rate": 4.0084299824275926e-06,
+      "loss": 62.0584,
+      "step": 75140
+    },
+    {
+      "epoch": 0.6216652190098026,
+      "grad_norm": 835.1646118164062,
+      "learning_rate": 4.007047666771274e-06,
+      "loss": 112.1609,
+      "step": 75150
+    },
+    {
+      "epoch": 0.621747942259172,
+      "grad_norm": 855.74169921875,
+      "learning_rate": 4.005665430119615e-06,
+      "loss": 89.9893,
+      "step": 75160
+    },
+    {
+      "epoch": 0.6218306655085412,
+      "grad_norm": 1189.672119140625,
+      "learning_rate": 4.0042832725825954e-06,
+      "loss": 78.6606,
+      "step": 75170
+    },
+    {
+      "epoch": 0.6219133887579104,
+      "grad_norm": 690.7192993164062,
+      "learning_rate": 4.002901194270186e-06,
+      "loss": 82.508,
+      "step": 75180
+    },
+    {
+      "epoch": 0.6219961120072797,
+      "grad_norm": 519.3690795898438,
+      "learning_rate": 4.001519195292352e-06,
+      "loss": 74.1029,
+      "step": 75190
+    },
+    {
+      "epoch": 0.6220788352566489,
+      "grad_norm": 664.5087890625,
+      "learning_rate": 4.000137275759053e-06,
+      "loss": 99.7511,
+      "step": 75200
+    },
+    {
+      "epoch": 0.6221615585060181,
+      "grad_norm": 966.7168579101562,
+      "learning_rate": 3.9987554357802435e-06,
+      "loss": 102.3384,
+      "step": 75210
+    },
+    {
+      "epoch": 0.6222442817553874,
+      "grad_norm": 1171.9957275390625,
+      "learning_rate": 3.997373675465869e-06,
+      "loss": 74.9776,
+      "step": 75220
+    },
+    {
+      "epoch": 0.6223270050047566,
+      "grad_norm": 534.0394897460938,
+      "learning_rate": 3.995991994925869e-06,
+      "loss": 91.2517,
+      "step": 75230
+    },
+    {
+      "epoch": 0.6224097282541258,
+      "grad_norm": 833.4016723632812,
+      "learning_rate": 3.994610394270178e-06,
+      "loss": 112.876,
+      "step": 75240
+    },
+    {
+      "epoch": 0.6224924515034951,
+      "grad_norm": 833.5524291992188,
+      "learning_rate": 3.993228873608724e-06,
+      "loss": 97.8957,
+      "step": 75250
+    },
+    {
+      "epoch": 0.6225751747528643,
+      "grad_norm": 1124.5748291015625,
+      "learning_rate": 3.991847433051427e-06,
+      "loss": 90.5661,
+      "step": 75260
+    },
+    {
+      "epoch": 0.6226578980022335,
+      "grad_norm": 701.1575927734375,
+      "learning_rate": 3.990466072708204e-06,
+      "loss": 93.4155,
+      "step": 75270
+    },
+    {
+      "epoch": 0.6227406212516028,
+      "grad_norm": 1118.724365234375,
+      "learning_rate": 3.989084792688962e-06,
+      "loss": 114.3592,
+      "step": 75280
+    },
+    {
+      "epoch": 0.622823344500972,
+      "grad_norm": 1080.4481201171875,
+      "learning_rate": 3.987703593103604e-06,
+      "loss": 91.7185,
+      "step": 75290
+    },
+    {
+      "epoch": 0.6229060677503412,
+      "grad_norm": 1988.3421630859375,
+      "learning_rate": 3.986322474062025e-06,
+      "loss": 110.2386,
+      "step": 75300
+    },
+    {
+      "epoch": 0.6229887909997105,
+      "grad_norm": 434.8983459472656,
+      "learning_rate": 3.9849414356741165e-06,
+      "loss": 66.5312,
+      "step": 75310
+    },
+    {
+      "epoch": 0.6230715142490797,
+      "grad_norm": 682.07421875,
+      "learning_rate": 3.9835604780497575e-06,
+      "loss": 85.5432,
+      "step": 75320
+    },
+    {
+      "epoch": 0.6231542374984489,
+      "grad_norm": 1016.9640502929688,
+      "learning_rate": 3.982179601298827e-06,
+      "loss": 109.5171,
+      "step": 75330
+    },
+    {
+      "epoch": 0.6232369607478182,
+      "grad_norm": 893.9984130859375,
+      "learning_rate": 3.9807988055311946e-06,
+      "loss": 78.253,
+      "step": 75340
+    },
+    {
+      "epoch": 0.6233196839971874,
+      "grad_norm": 589.87451171875,
+      "learning_rate": 3.979418090856723e-06,
+      "loss": 87.9922,
+      "step": 75350
+    },
+    {
+      "epoch": 0.6234024072465566,
+      "grad_norm": 899.80126953125,
+      "learning_rate": 3.978037457385268e-06,
+      "loss": 109.1129,
+      "step": 75360
+    },
+    {
+      "epoch": 0.6234851304959259,
+      "grad_norm": 833.7298583984375,
+      "learning_rate": 3.976656905226686e-06,
+      "loss": 104.8421,
+      "step": 75370
+    },
+    {
+      "epoch": 0.6235678537452951,
+      "grad_norm": 1044.365234375,
+      "learning_rate": 3.9752764344908155e-06,
+      "loss": 97.9259,
+      "step": 75380
+    },
+    {
+      "epoch": 0.6236505769946643,
+      "grad_norm": 1515.5921630859375,
+      "learning_rate": 3.9738960452874975e-06,
+      "loss": 102.9154,
+      "step": 75390
+    },
+    {
+      "epoch": 0.6237333002440336,
+      "grad_norm": 1511.9940185546875,
+      "learning_rate": 3.972515737726563e-06,
+      "loss": 114.3788,
+      "step": 75400
+    },
+    {
+      "epoch": 0.6238160234934028,
+      "grad_norm": 1036.8448486328125,
+      "learning_rate": 3.9711355119178345e-06,
+      "loss": 144.706,
+      "step": 75410
+    },
+    {
+      "epoch": 0.623898746742772,
+      "grad_norm": 655.7908325195312,
+      "learning_rate": 3.969755367971131e-06,
+      "loss": 90.5163,
+      "step": 75420
+    },
+    {
+      "epoch": 0.6239814699921413,
+      "grad_norm": 574.1646728515625,
+      "learning_rate": 3.9683753059962646e-06,
+      "loss": 77.3598,
+      "step": 75430
+    },
+    {
+      "epoch": 0.6240641932415105,
+      "grad_norm": 615.3191528320312,
+      "learning_rate": 3.966995326103041e-06,
+      "loss": 88.5786,
+      "step": 75440
+    },
+    {
+      "epoch": 0.6241469164908797,
+      "grad_norm": 1213.435546875,
+      "learning_rate": 3.965615428401257e-06,
+      "loss": 95.156,
+      "step": 75450
+    },
+    {
+      "epoch": 0.624229639740249,
+      "grad_norm": 960.2600708007812,
+      "learning_rate": 3.964235613000708e-06,
+      "loss": 85.2145,
+      "step": 75460
+    },
+    {
+      "epoch": 0.6243123629896182,
+      "grad_norm": 616.06689453125,
+      "learning_rate": 3.962855880011177e-06,
+      "loss": 78.2717,
+      "step": 75470
+    },
+    {
+      "epoch": 0.6243950862389874,
+      "grad_norm": 560.2363891601562,
+      "learning_rate": 3.961476229542446e-06,
+      "loss": 102.936,
+      "step": 75480
+    },
+    {
+      "epoch": 0.6244778094883567,
+      "grad_norm": 473.06793212890625,
+      "learning_rate": 3.9600966617042825e-06,
+      "loss": 72.4044,
+      "step": 75490
+    },
+    {
+      "epoch": 0.6245605327377259,
+      "grad_norm": 770.420166015625,
+      "learning_rate": 3.958717176606456e-06,
+      "loss": 99.9721,
+      "step": 75500
+    },
+    {
+      "epoch": 0.6246432559870951,
+      "grad_norm": 1100.4041748046875,
+      "learning_rate": 3.957337774358725e-06,
+      "loss": 123.5163,
+      "step": 75510
+    },
+    {
+      "epoch": 0.6247259792364644,
+      "grad_norm": 1014.705322265625,
+      "learning_rate": 3.955958455070842e-06,
+      "loss": 105.1985,
+      "step": 75520
+    },
+    {
+      "epoch": 0.6248087024858336,
+      "grad_norm": 1663.8048095703125,
+      "learning_rate": 3.954579218852553e-06,
+      "loss": 100.1772,
+      "step": 75530
+    },
+    {
+      "epoch": 0.6248914257352028,
+      "grad_norm": 701.8104248046875,
+      "learning_rate": 3.953200065813597e-06,
+      "loss": 100.0598,
+      "step": 75540
+    },
+    {
+      "epoch": 0.6249741489845722,
+      "grad_norm": 906.2036743164062,
+      "learning_rate": 3.951820996063708e-06,
+      "loss": 95.372,
+      "step": 75550
+    },
+    {
+      "epoch": 0.6250568722339414,
+      "grad_norm": 719.7368774414062,
+      "learning_rate": 3.950442009712612e-06,
+      "loss": 83.6697,
+      "step": 75560
+    },
+    {
+      "epoch": 0.6251395954833106,
+      "grad_norm": 703.7513427734375,
+      "learning_rate": 3.949063106870031e-06,
+      "loss": 85.2299,
+      "step": 75570
+    },
+    {
+      "epoch": 0.6252223187326799,
+      "grad_norm": 571.2645263671875,
+      "learning_rate": 3.9476842876456735e-06,
+      "loss": 100.489,
+      "step": 75580
+    },
+    {
+      "epoch": 0.6253050419820491,
+      "grad_norm": 829.6326293945312,
+      "learning_rate": 3.946305552149247e-06,
+      "loss": 92.2218,
+      "step": 75590
+    },
+    {
+      "epoch": 0.6253877652314183,
+      "grad_norm": 716.100341796875,
+      "learning_rate": 3.944926900490452e-06,
+      "loss": 96.999,
+      "step": 75600
+    },
+    {
+      "epoch": 0.6254704884807876,
+      "grad_norm": 731.5742797851562,
+      "learning_rate": 3.943548332778982e-06,
+      "loss": 108.1109,
+      "step": 75610
+    },
+    {
+      "epoch": 0.6255532117301568,
+      "grad_norm": 586.7816162109375,
+      "learning_rate": 3.942169849124523e-06,
+      "loss": 102.1984,
+      "step": 75620
+    },
+    {
+      "epoch": 0.625635934979526,
+      "grad_norm": 583.6131591796875,
+      "learning_rate": 3.940791449636753e-06,
+      "loss": 72.8051,
+      "step": 75630
+    },
+    {
+      "epoch": 0.6257186582288953,
+      "grad_norm": 655.7570190429688,
+      "learning_rate": 3.939413134425347e-06,
+      "loss": 92.8459,
+      "step": 75640
+    },
+    {
+      "epoch": 0.6258013814782645,
+      "grad_norm": 609.809326171875,
+      "learning_rate": 3.938034903599972e-06,
+      "loss": 67.6845,
+      "step": 75650
+    },
+    {
+      "epoch": 0.6258841047276337,
+      "grad_norm": 989.1069946289062,
+      "learning_rate": 3.9366567572702845e-06,
+      "loss": 65.7915,
+      "step": 75660
+    },
+    {
+      "epoch": 0.625966827977003,
+      "grad_norm": 746.6648559570312,
+      "learning_rate": 3.935278695545939e-06,
+      "loss": 90.3173,
+      "step": 75670
+    },
+    {
+      "epoch": 0.6260495512263722,
+      "grad_norm": 917.3569946289062,
+      "learning_rate": 3.933900718536579e-06,
+      "loss": 103.0491,
+      "step": 75680
+    },
+    {
+      "epoch": 0.6261322744757414,
+      "grad_norm": 753.1608276367188,
+      "learning_rate": 3.932522826351849e-06,
+      "loss": 111.5943,
+      "step": 75690
+    },
+    {
+      "epoch": 0.6262149977251107,
+      "grad_norm": 582.921142578125,
+      "learning_rate": 3.9311450191013774e-06,
+      "loss": 113.5137,
+      "step": 75700
+    },
+    {
+      "epoch": 0.6262977209744799,
+      "grad_norm": 944.9461669921875,
+      "learning_rate": 3.929767296894792e-06,
+      "loss": 77.7904,
+      "step": 75710
+    },
+    {
+      "epoch": 0.6263804442238491,
+      "grad_norm": 705.4637451171875,
+      "learning_rate": 3.9283896598417104e-06,
+      "loss": 79.6284,
+      "step": 75720
+    },
+    {
+      "epoch": 0.6264631674732184,
+      "grad_norm": 835.1659545898438,
+      "learning_rate": 3.927012108051746e-06,
+      "loss": 89.4407,
+      "step": 75730
+    },
+    {
+      "epoch": 0.6265458907225876,
+      "grad_norm": 1340.5850830078125,
+      "learning_rate": 3.925634641634505e-06,
+      "loss": 90.6847,
+      "step": 75740
+    },
+    {
+      "epoch": 0.6266286139719568,
+      "grad_norm": 1067.461181640625,
+      "learning_rate": 3.924257260699583e-06,
+      "loss": 90.2306,
+      "step": 75750
+    },
+    {
+      "epoch": 0.6267113372213261,
+      "grad_norm": 967.4222412109375,
+      "learning_rate": 3.922879965356574e-06,
+      "loss": 81.9187,
+      "step": 75760
+    },
+    {
+      "epoch": 0.6267940604706953,
+      "grad_norm": 1008.221923828125,
+      "learning_rate": 3.921502755715064e-06,
+      "loss": 83.478,
+      "step": 75770
+    },
+    {
+      "epoch": 0.6268767837200645,
+      "grad_norm": 506.6917724609375,
+      "learning_rate": 3.920125631884628e-06,
+      "loss": 87.9039,
+      "step": 75780
+    },
+    {
+      "epoch": 0.6269595069694338,
+      "grad_norm": 881.6430053710938,
+      "learning_rate": 3.918748593974841e-06,
+      "loss": 84.5617,
+      "step": 75790
+    },
+    {
+      "epoch": 0.627042230218803,
+      "grad_norm": 1556.2261962890625,
+      "learning_rate": 3.917371642095265e-06,
+      "loss": 108.6617,
+      "step": 75800
+    },
+    {
+      "epoch": 0.6271249534681722,
+      "grad_norm": 1210.6939697265625,
+      "learning_rate": 3.91599477635546e-06,
+      "loss": 89.0075,
+      "step": 75810
+    },
+    {
+      "epoch": 0.6272076767175415,
+      "grad_norm": 792.7057495117188,
+      "learning_rate": 3.914617996864976e-06,
+      "loss": 75.4186,
+      "step": 75820
+    },
+    {
+      "epoch": 0.6272903999669107,
+      "grad_norm": 965.24072265625,
+      "learning_rate": 3.9132413037333565e-06,
+      "loss": 85.8507,
+      "step": 75830
+    },
+    {
+      "epoch": 0.6273731232162799,
+      "grad_norm": 773.074462890625,
+      "learning_rate": 3.911864697070139e-06,
+      "loss": 97.3403,
+      "step": 75840
+    },
+    {
+      "epoch": 0.6274558464656492,
+      "grad_norm": 1299.928955078125,
+      "learning_rate": 3.910488176984853e-06,
+      "loss": 127.7312,
+      "step": 75850
+    },
+    {
+      "epoch": 0.6275385697150184,
+      "grad_norm": 662.94970703125,
+      "learning_rate": 3.909111743587023e-06,
+      "loss": 90.648,
+      "step": 75860
+    },
+    {
+      "epoch": 0.6276212929643876,
+      "grad_norm": 557.4157104492188,
+      "learning_rate": 3.907735396986166e-06,
+      "loss": 110.0861,
+      "step": 75870
+    },
+    {
+      "epoch": 0.6277040162137568,
+      "grad_norm": 1387.1092529296875,
+      "learning_rate": 3.9063591372917875e-06,
+      "loss": 83.7509,
+      "step": 75880
+    },
+    {
+      "epoch": 0.6277867394631261,
+      "grad_norm": 1644.345947265625,
+      "learning_rate": 3.904982964613395e-06,
+      "loss": 104.4969,
+      "step": 75890
+    },
+    {
+      "epoch": 0.6278694627124953,
+      "grad_norm": 663.9248657226562,
+      "learning_rate": 3.903606879060483e-06,
+      "loss": 120.2953,
+      "step": 75900
+    },
+    {
+      "epoch": 0.6279521859618645,
+      "grad_norm": 1548.13232421875,
+      "learning_rate": 3.902230880742541e-06,
+      "loss": 109.4352,
+      "step": 75910
+    },
+    {
+      "epoch": 0.6280349092112338,
+      "grad_norm": 615.7534790039062,
+      "learning_rate": 3.900854969769049e-06,
+      "loss": 82.1672,
+      "step": 75920
+    },
+    {
+      "epoch": 0.628117632460603,
+      "grad_norm": 1001.1287841796875,
+      "learning_rate": 3.899479146249482e-06,
+      "loss": 87.0741,
+      "step": 75930
+    },
+    {
+      "epoch": 0.6282003557099722,
+      "grad_norm": 1314.9921875,
+      "learning_rate": 3.898103410293309e-06,
+      "loss": 136.3624,
+      "step": 75940
+    },
+    {
+      "epoch": 0.6282830789593415,
+      "grad_norm": 1015.78759765625,
+      "learning_rate": 3.89672776200999e-06,
+      "loss": 87.5703,
+      "step": 75950
+    },
+    {
+      "epoch": 0.6283658022087107,
+      "grad_norm": 553.9141235351562,
+      "learning_rate": 3.895352201508981e-06,
+      "loss": 66.2459,
+      "step": 75960
+    },
+    {
+      "epoch": 0.6284485254580799,
+      "grad_norm": 585.1502075195312,
+      "learning_rate": 3.893976728899726e-06,
+      "loss": 119.4419,
+      "step": 75970
+    },
+    {
+      "epoch": 0.6285312487074493,
+      "grad_norm": 414.6874084472656,
+      "learning_rate": 3.892601344291667e-06,
+      "loss": 68.8029,
+      "step": 75980
+    },
+    {
+      "epoch": 0.6286139719568185,
+      "grad_norm": 912.771728515625,
+      "learning_rate": 3.891226047794237e-06,
+      "loss": 89.6607,
+      "step": 75990
+    },
+    {
+      "epoch": 0.6286966952061877,
+      "grad_norm": 777.063232421875,
+      "learning_rate": 3.8898508395168645e-06,
+      "loss": 146.2245,
+      "step": 76000
+    },
+    {
+      "epoch": 0.628779418455557,
+      "grad_norm": 884.0517578125,
+      "learning_rate": 3.888475719568961e-06,
+      "loss": 87.4548,
+      "step": 76010
+    },
+    {
+      "epoch": 0.6288621417049262,
+      "grad_norm": 768.80859375,
+      "learning_rate": 3.887100688059947e-06,
+      "loss": 83.4292,
+      "step": 76020
+    },
+    {
+      "epoch": 0.6289448649542954,
+      "grad_norm": 686.2239379882812,
+      "learning_rate": 3.885725745099222e-06,
+      "loss": 100.6639,
+      "step": 76030
+    },
+    {
+      "epoch": 0.6290275882036647,
+      "grad_norm": 1023.0579833984375,
+      "learning_rate": 3.8843508907961855e-06,
+      "loss": 81.8008,
+      "step": 76040
+    },
+    {
+      "epoch": 0.6291103114530339,
+      "grad_norm": 867.1831665039062,
+      "learning_rate": 3.882976125260229e-06,
+      "loss": 89.1627,
+      "step": 76050
+    },
+    {
+      "epoch": 0.6291930347024031,
+      "grad_norm": 1243.499755859375,
+      "learning_rate": 3.881601448600736e-06,
+      "loss": 112.3818,
+      "step": 76060
+    },
+    {
+      "epoch": 0.6292757579517724,
+      "grad_norm": 753.3609619140625,
+      "learning_rate": 3.880226860927082e-06,
+      "loss": 79.829,
+      "step": 76070
+    },
+    {
+      "epoch": 0.6293584812011416,
+      "grad_norm": 457.8770446777344,
+      "learning_rate": 3.8788523623486405e-06,
+      "loss": 115.6456,
+      "step": 76080
+    },
+    {
+      "epoch": 0.6294412044505108,
+      "grad_norm": 2279.8681640625,
+      "learning_rate": 3.877477952974768e-06,
+      "loss": 100.5304,
+      "step": 76090
+    },
+    {
+      "epoch": 0.6295239276998801,
+      "grad_norm": 1045.8399658203125,
+      "learning_rate": 3.876103632914825e-06,
+      "loss": 84.072,
+      "step": 76100
+    },
+    {
+      "epoch": 0.6296066509492493,
+      "grad_norm": 731.2930908203125,
+      "learning_rate": 3.8747294022781555e-06,
+      "loss": 91.0355,
+      "step": 76110
+    },
+    {
+      "epoch": 0.6296893741986185,
+      "grad_norm": 1290.4974365234375,
+      "learning_rate": 3.873355261174105e-06,
+      "loss": 74.0479,
+      "step": 76120
+    },
+    {
+      "epoch": 0.6297720974479878,
+      "grad_norm": 842.3202514648438,
+      "learning_rate": 3.871981209712006e-06,
+      "loss": 91.0174,
+      "step": 76130
+    },
+    {
+      "epoch": 0.629854820697357,
+      "grad_norm": 1351.25244140625,
+      "learning_rate": 3.870607248001184e-06,
+      "loss": 96.3427,
+      "step": 76140
+    },
+    {
+      "epoch": 0.6299375439467262,
+      "grad_norm": 1196.4425048828125,
+      "learning_rate": 3.869233376150961e-06,
+      "loss": 83.2907,
+      "step": 76150
+    },
+    {
+      "epoch": 0.6300202671960955,
+      "grad_norm": 630.231689453125,
+      "learning_rate": 3.867859594270649e-06,
+      "loss": 93.8257,
+      "step": 76160
+    },
+    {
+      "epoch": 0.6301029904454647,
+      "grad_norm": 636.0516357421875,
+      "learning_rate": 3.866485902469554e-06,
+      "loss": 107.9014,
+      "step": 76170
+    },
+    {
+      "epoch": 0.6301857136948339,
+      "grad_norm": 867.7330322265625,
+      "learning_rate": 3.865112300856972e-06,
+      "loss": 98.2939,
+      "step": 76180
+    },
+    {
+      "epoch": 0.6302684369442032,
+      "grad_norm": 836.81884765625,
+      "learning_rate": 3.863738789542196e-06,
+      "loss": 91.0884,
+      "step": 76190
+    },
+    {
+      "epoch": 0.6303511601935724,
+      "grad_norm": 700.2529296875,
+      "learning_rate": 3.86236536863451e-06,
+      "loss": 95.8659,
+      "step": 76200
+    },
+    {
+      "epoch": 0.6304338834429416,
+      "grad_norm": 716.2244262695312,
+      "learning_rate": 3.860992038243189e-06,
+      "loss": 61.6874,
+      "step": 76210
+    },
+    {
+      "epoch": 0.6305166066923109,
+      "grad_norm": 410.9248962402344,
+      "learning_rate": 3.8596187984775064e-06,
+      "loss": 99.224,
+      "step": 76220
+    },
+    {
+      "epoch": 0.6305993299416801,
+      "grad_norm": 907.3155517578125,
+      "learning_rate": 3.8582456494467214e-06,
+      "loss": 101.9425,
+      "step": 76230
+    },
+    {
+      "epoch": 0.6306820531910493,
+      "grad_norm": 579.9906616210938,
+      "learning_rate": 3.8568725912600904e-06,
+      "loss": 71.3855,
+      "step": 76240
+    },
+    {
+      "epoch": 0.6307647764404186,
+      "grad_norm": 1267.6279296875,
+      "learning_rate": 3.855499624026861e-06,
+      "loss": 123.397,
+      "step": 76250
+    },
+    {
+      "epoch": 0.6308474996897878,
+      "grad_norm": 720.7117309570312,
+      "learning_rate": 3.854126747856275e-06,
+      "loss": 105.1762,
+      "step": 76260
+    },
+    {
+      "epoch": 0.630930222939157,
+      "grad_norm": 1166.6748046875,
+      "learning_rate": 3.8527539628575635e-06,
+      "loss": 78.2158,
+      "step": 76270
+    },
+    {
+      "epoch": 0.6310129461885263,
+      "grad_norm": 721.6836547851562,
+      "learning_rate": 3.851381269139955e-06,
+      "loss": 75.0298,
+      "step": 76280
+    },
+    {
+      "epoch": 0.6310956694378955,
+      "grad_norm": 932.6498413085938,
+      "learning_rate": 3.8500086668126666e-06,
+      "loss": 127.4507,
+      "step": 76290
+    },
+    {
+      "epoch": 0.6311783926872647,
+      "grad_norm": 804.4441528320312,
+      "learning_rate": 3.848636155984912e-06,
+      "loss": 88.2771,
+      "step": 76300
+    },
+    {
+      "epoch": 0.631261115936634,
+      "grad_norm": 662.4979248046875,
+      "learning_rate": 3.847263736765892e-06,
+      "loss": 76.6129,
+      "step": 76310
+    },
+    {
+      "epoch": 0.6313438391860032,
+      "grad_norm": 1362.8896484375,
+      "learning_rate": 3.8458914092648074e-06,
+      "loss": 116.3484,
+      "step": 76320
+    },
+    {
+      "epoch": 0.6314265624353724,
+      "grad_norm": 1995.614013671875,
+      "learning_rate": 3.844519173590847e-06,
+      "loss": 85.0176,
+      "step": 76330
+    },
+    {
+      "epoch": 0.6315092856847417,
+      "grad_norm": 494.42132568359375,
+      "learning_rate": 3.843147029853194e-06,
+      "loss": 77.5021,
+      "step": 76340
+    },
+    {
+      "epoch": 0.6315920089341109,
+      "grad_norm": 1092.7735595703125,
+      "learning_rate": 3.841774978161022e-06,
+      "loss": 107.8559,
+      "step": 76350
+    },
+    {
+      "epoch": 0.6316747321834801,
+      "grad_norm": 746.4695434570312,
+      "learning_rate": 3.840403018623499e-06,
+      "loss": 82.9424,
+      "step": 76360
+    },
+    {
+      "epoch": 0.6317574554328494,
+      "grad_norm": 911.3170166015625,
+      "learning_rate": 3.839031151349786e-06,
+      "loss": 87.2492,
+      "step": 76370
+    },
+    {
+      "epoch": 0.6318401786822186,
+      "grad_norm": 1411.3955078125,
+      "learning_rate": 3.837659376449036e-06,
+      "loss": 103.6988,
+      "step": 76380
+    },
+    {
+      "epoch": 0.6319229019315878,
+      "grad_norm": 1356.0250244140625,
+      "learning_rate": 3.836287694030395e-06,
+      "loss": 120.5798,
+      "step": 76390
+    },
+    {
+      "epoch": 0.6320056251809572,
+      "grad_norm": 498.0589294433594,
+      "learning_rate": 3.834916104203e-06,
+      "loss": 95.7689,
+      "step": 76400
+    },
+    {
+      "epoch": 0.6320883484303264,
+      "grad_norm": 701.58740234375,
+      "learning_rate": 3.833544607075986e-06,
+      "loss": 78.422,
+      "step": 76410
+    },
+    {
+      "epoch": 0.6321710716796956,
+      "grad_norm": 596.4544677734375,
+      "learning_rate": 3.8321732027584734e-06,
+      "loss": 94.8889,
+      "step": 76420
+    },
+    {
+      "epoch": 0.6322537949290649,
+      "grad_norm": 910.3405151367188,
+      "learning_rate": 3.830801891359582e-06,
+      "loss": 81.3986,
+      "step": 76430
+    },
+    {
+      "epoch": 0.6323365181784341,
+      "grad_norm": 494.3802490234375,
+      "learning_rate": 3.829430672988414e-06,
+      "loss": 75.0982,
+      "step": 76440
+    },
+    {
+      "epoch": 0.6324192414278033,
+      "grad_norm": 773.7373046875,
+      "learning_rate": 3.828059547754078e-06,
+      "loss": 73.4452,
+      "step": 76450
+    },
+    {
+      "epoch": 0.6325019646771726,
+      "grad_norm": 597.4763793945312,
+      "learning_rate": 3.826688515765664e-06,
+      "loss": 94.815,
+      "step": 76460
+    },
+    {
+      "epoch": 0.6325846879265418,
+      "grad_norm": 1558.9375,
+      "learning_rate": 3.82531757713226e-06,
+      "loss": 89.0715,
+      "step": 76470
+    },
+    {
+      "epoch": 0.632667411175911,
+      "grad_norm": 912.0201416015625,
+      "learning_rate": 3.823946731962945e-06,
+      "loss": 89.6122,
+      "step": 76480
+    },
+    {
+      "epoch": 0.6327501344252803,
+      "grad_norm": 803.7633666992188,
+      "learning_rate": 3.8225759803667925e-06,
+      "loss": 90.7924,
+      "step": 76490
+    },
+    {
+      "epoch": 0.6328328576746495,
+      "grad_norm": 1232.05712890625,
+      "learning_rate": 3.821205322452863e-06,
+      "loss": 89.2706,
+      "step": 76500
+    },
+    {
+      "epoch": 0.6329155809240187,
+      "grad_norm": 831.2234497070312,
+      "learning_rate": 3.81983475833022e-06,
+      "loss": 75.2085,
+      "step": 76510
+    },
+    {
+      "epoch": 0.632998304173388,
+      "grad_norm": 560.6333618164062,
+      "learning_rate": 3.818464288107908e-06,
+      "loss": 80.2901,
+      "step": 76520
+    },
+    {
+      "epoch": 0.6330810274227572,
+      "grad_norm": 1093.8529052734375,
+      "learning_rate": 3.817093911894968e-06,
+      "loss": 81.6339,
+      "step": 76530
+    },
+    {
+      "epoch": 0.6331637506721264,
+      "grad_norm": 787.9254150390625,
+      "learning_rate": 3.8157236298004375e-06,
+      "loss": 98.8886,
+      "step": 76540
+    },
+    {
+      "epoch": 0.6332464739214957,
+      "grad_norm": 613.9071655273438,
+      "learning_rate": 3.814353441933343e-06,
+      "loss": 88.6119,
+      "step": 76550
+    },
+    {
+      "epoch": 0.6333291971708649,
+      "grad_norm": 1168.52685546875,
+      "learning_rate": 3.812983348402703e-06,
+      "loss": 86.0251,
+      "step": 76560
+    },
+    {
+      "epoch": 0.6334119204202341,
+      "grad_norm": 684.206298828125,
+      "learning_rate": 3.811613349317531e-06,
+      "loss": 68.7577,
+      "step": 76570
+    },
+    {
+      "epoch": 0.6334946436696034,
+      "grad_norm": 543.0419311523438,
+      "learning_rate": 3.810243444786831e-06,
+      "loss": 85.61,
+      "step": 76580
+    },
+    {
+      "epoch": 0.6335773669189726,
+      "grad_norm": 1784.14794921875,
+      "learning_rate": 3.8088736349195995e-06,
+      "loss": 119.1676,
+      "step": 76590
+    },
+    {
+      "epoch": 0.6336600901683418,
+      "grad_norm": 1234.8670654296875,
+      "learning_rate": 3.8075039198248274e-06,
+      "loss": 82.1384,
+      "step": 76600
+    },
+    {
+      "epoch": 0.633742813417711,
+      "grad_norm": 454.25372314453125,
+      "learning_rate": 3.8061342996114946e-06,
+      "loss": 81.6356,
+      "step": 76610
+    },
+    {
+      "epoch": 0.6338255366670803,
+      "grad_norm": 1700.140380859375,
+      "learning_rate": 3.8047647743885762e-06,
+      "loss": 88.7648,
+      "step": 76620
+    },
+    {
+      "epoch": 0.6339082599164495,
+      "grad_norm": 1148.6446533203125,
+      "learning_rate": 3.8033953442650382e-06,
+      "loss": 85.314,
+      "step": 76630
+    },
+    {
+      "epoch": 0.6339909831658187,
+      "grad_norm": 730.97998046875,
+      "learning_rate": 3.802026009349843e-06,
+      "loss": 98.301,
+      "step": 76640
+    },
+    {
+      "epoch": 0.634073706415188,
+      "grad_norm": 672.2242431640625,
+      "learning_rate": 3.800656769751939e-06,
+      "loss": 86.9433,
+      "step": 76650
+    },
+    {
+      "epoch": 0.6341564296645572,
+      "grad_norm": 769.8524780273438,
+      "learning_rate": 3.799287625580273e-06,
+      "loss": 93.8155,
+      "step": 76660
+    },
+    {
+      "epoch": 0.6342391529139264,
+      "grad_norm": 539.6921997070312,
+      "learning_rate": 3.7979185769437795e-06,
+      "loss": 71.8179,
+      "step": 76670
+    },
+    {
+      "epoch": 0.6343218761632957,
+      "grad_norm": 1174.0040283203125,
+      "learning_rate": 3.7965496239513875e-06,
+      "loss": 100.7191,
+      "step": 76680
+    },
+    {
+      "epoch": 0.6344045994126649,
+      "grad_norm": 654.21240234375,
+      "learning_rate": 3.79518076671202e-06,
+      "loss": 101.6481,
+      "step": 76690
+    },
+    {
+      "epoch": 0.6344873226620341,
+      "grad_norm": 1032.4493408203125,
+      "learning_rate": 3.793812005334589e-06,
+      "loss": 73.4023,
+      "step": 76700
+    },
+    {
+      "epoch": 0.6345700459114034,
+      "grad_norm": 1226.67333984375,
+      "learning_rate": 3.792443339928001e-06,
+      "loss": 94.1326,
+      "step": 76710
+    },
+    {
+      "epoch": 0.6346527691607726,
+      "grad_norm": 693.3799438476562,
+      "learning_rate": 3.7910747706011543e-06,
+      "loss": 103.0554,
+      "step": 76720
+    },
+    {
+      "epoch": 0.6347354924101418,
+      "grad_norm": 427.7120361328125,
+      "learning_rate": 3.7897062974629384e-06,
+      "loss": 101.5577,
+      "step": 76730
+    },
+    {
+      "epoch": 0.6348182156595111,
+      "grad_norm": 1844.3143310546875,
+      "learning_rate": 3.78833792062224e-06,
+      "loss": 119.7666,
+      "step": 76740
+    },
+    {
+      "epoch": 0.6349009389088803,
+      "grad_norm": 381.83380126953125,
+      "learning_rate": 3.786969640187932e-06,
+      "loss": 89.6403,
+      "step": 76750
+    },
+    {
+      "epoch": 0.6349836621582495,
+      "grad_norm": 1052.9390869140625,
+      "learning_rate": 3.785601456268882e-06,
+      "loss": 81.7052,
+      "step": 76760
+    },
+    {
+      "epoch": 0.6350663854076188,
+      "grad_norm": 1047.6627197265625,
+      "learning_rate": 3.7842333689739524e-06,
+      "loss": 87.8519,
+      "step": 76770
+    },
+    {
+      "epoch": 0.635149108656988,
+      "grad_norm": 623.3661499023438,
+      "learning_rate": 3.782865378411993e-06,
+      "loss": 104.4676,
+      "step": 76780
+    },
+    {
+      "epoch": 0.6352318319063572,
+      "grad_norm": 482.55059814453125,
+      "learning_rate": 3.7814974846918496e-06,
+      "loss": 87.8084,
+      "step": 76790
+    },
+    {
+      "epoch": 0.6353145551557265,
+      "grad_norm": 731.9932250976562,
+      "learning_rate": 3.7801296879223594e-06,
+      "loss": 64.0985,
+      "step": 76800
+    },
+    {
+      "epoch": 0.6353972784050957,
+      "grad_norm": 421.17864990234375,
+      "learning_rate": 3.7787619882123506e-06,
+      "loss": 95.0226,
+      "step": 76810
+    },
+    {
+      "epoch": 0.6354800016544649,
+      "grad_norm": 945.0263671875,
+      "learning_rate": 3.7773943856706463e-06,
+      "loss": 96.3978,
+      "step": 76820
+    },
+    {
+      "epoch": 0.6355627249038343,
+      "grad_norm": 705.6464233398438,
+      "learning_rate": 3.7760268804060583e-06,
+      "loss": 92.4526,
+      "step": 76830
+    },
+    {
+      "epoch": 0.6356454481532035,
+      "grad_norm": 624.5090942382812,
+      "learning_rate": 3.774659472527396e-06,
+      "loss": 82.1249,
+      "step": 76840
+    },
+    {
+      "epoch": 0.6357281714025727,
+      "grad_norm": 788.3277587890625,
+      "learning_rate": 3.7732921621434553e-06,
+      "loss": 66.9649,
+      "step": 76850
+    },
+    {
+      "epoch": 0.635810894651942,
+      "grad_norm": 451.6109619140625,
+      "learning_rate": 3.77192494936303e-06,
+      "loss": 113.1434,
+      "step": 76860
+    },
+    {
+      "epoch": 0.6358936179013112,
+      "grad_norm": 711.1947021484375,
+      "learning_rate": 3.7705578342948967e-06,
+      "loss": 81.1459,
+      "step": 76870
+    },
+    {
+      "epoch": 0.6359763411506804,
+      "grad_norm": 987.3251953125,
+      "learning_rate": 3.7691908170478352e-06,
+      "loss": 82.2781,
+      "step": 76880
+    },
+    {
+      "epoch": 0.6360590644000497,
+      "grad_norm": 791.8734130859375,
+      "learning_rate": 3.767823897730612e-06,
+      "loss": 87.6097,
+      "step": 76890
+    },
+    {
+      "epoch": 0.6361417876494189,
+      "grad_norm": 657.6787109375,
+      "learning_rate": 3.7664570764519865e-06,
+      "loss": 88.5876,
+      "step": 76900
+    },
+    {
+      "epoch": 0.6362245108987881,
+      "grad_norm": 958.947265625,
+      "learning_rate": 3.76509035332071e-06,
+      "loss": 94.1349,
+      "step": 76910
+    },
+    {
+      "epoch": 0.6363072341481574,
+      "grad_norm": 1417.035400390625,
+      "learning_rate": 3.7637237284455264e-06,
+      "loss": 97.1612,
+      "step": 76920
+    },
+    {
+      "epoch": 0.6363899573975266,
+      "grad_norm": 766.021240234375,
+      "learning_rate": 3.762357201935171e-06,
+      "loss": 82.1091,
+      "step": 76930
+    },
+    {
+      "epoch": 0.6364726806468958,
+      "grad_norm": 1105.8828125,
+      "learning_rate": 3.7609907738983762e-06,
+      "loss": 88.7781,
+      "step": 76940
+    },
+    {
+      "epoch": 0.6365554038962651,
+      "grad_norm": 916.3728637695312,
+      "learning_rate": 3.7596244444438577e-06,
+      "loss": 99.8711,
+      "step": 76950
+    },
+    {
+      "epoch": 0.6366381271456343,
+      "grad_norm": 1511.9302978515625,
+      "learning_rate": 3.758258213680328e-06,
+      "loss": 76.2348,
+      "step": 76960
+    },
+    {
+      "epoch": 0.6367208503950035,
+      "grad_norm": 966.1973266601562,
+      "learning_rate": 3.7568920817164945e-06,
+      "loss": 102.2466,
+      "step": 76970
+    },
+    {
+      "epoch": 0.6368035736443728,
+      "grad_norm": 1336.3597412109375,
+      "learning_rate": 3.755526048661053e-06,
+      "loss": 102.7599,
+      "step": 76980
+    },
+    {
+      "epoch": 0.636886296893742,
+      "grad_norm": 845.8065185546875,
+      "learning_rate": 3.7541601146226924e-06,
+      "loss": 92.4951,
+      "step": 76990
+    },
+    {
+      "epoch": 0.6369690201431112,
+      "grad_norm": 876.119873046875,
+      "learning_rate": 3.752794279710094e-06,
+      "loss": 90.3612,
+      "step": 77000
+    },
+    {
+      "epoch": 0.6370517433924805,
+      "grad_norm": 1019.0983276367188,
+      "learning_rate": 3.751428544031931e-06,
+      "loss": 69.2927,
+      "step": 77010
+    },
+    {
+      "epoch": 0.6371344666418497,
+      "grad_norm": 591.1396484375,
+      "learning_rate": 3.750062907696868e-06,
+      "loss": 59.3948,
+      "step": 77020
+    },
+    {
+      "epoch": 0.6372171898912189,
+      "grad_norm": 627.9861450195312,
+      "learning_rate": 3.7486973708135643e-06,
+      "loss": 89.2826,
+      "step": 77030
+    },
+    {
+      "epoch": 0.6372999131405882,
+      "grad_norm": 823.3641967773438,
+      "learning_rate": 3.7473319334906678e-06,
+      "loss": 90.4643,
+      "step": 77040
+    },
+    {
+      "epoch": 0.6373826363899574,
+      "grad_norm": 1226.17333984375,
+      "learning_rate": 3.7459665958368197e-06,
+      "loss": 88.7953,
+      "step": 77050
+    },
+    {
+      "epoch": 0.6374653596393266,
+      "grad_norm": 1117.8607177734375,
+      "learning_rate": 3.7446013579606534e-06,
+      "loss": 101.2203,
+      "step": 77060
+    },
+    {
+      "epoch": 0.6375480828886959,
+      "grad_norm": 1034.3206787109375,
+      "learning_rate": 3.743236219970796e-06,
+      "loss": 102.8942,
+      "step": 77070
+    },
+    {
+      "epoch": 0.6376308061380651,
+      "grad_norm": 929.0318603515625,
+      "learning_rate": 3.741871181975866e-06,
+      "loss": 102.5503,
+      "step": 77080
+    },
+    {
+      "epoch": 0.6377135293874343,
+      "grad_norm": 815.303466796875,
+      "learning_rate": 3.740506244084471e-06,
+      "loss": 88.2377,
+      "step": 77090
+    },
+    {
+      "epoch": 0.6377962526368036,
+      "grad_norm": 1119.909423828125,
+      "learning_rate": 3.7391414064052138e-06,
+      "loss": 85.4328,
+      "step": 77100
+    },
+    {
+      "epoch": 0.6378789758861728,
+      "grad_norm": 477.2499084472656,
+      "learning_rate": 3.737776669046689e-06,
+      "loss": 85.3801,
+      "step": 77110
+    },
+    {
+      "epoch": 0.637961699135542,
+      "grad_norm": 1207.775634765625,
+      "learning_rate": 3.7364120321174826e-06,
+      "loss": 105.7464,
+      "step": 77120
+    },
+    {
+      "epoch": 0.6380444223849113,
+      "grad_norm": 664.5465698242188,
+      "learning_rate": 3.7350474957261705e-06,
+      "loss": 119.5242,
+      "step": 77130
+    },
+    {
+      "epoch": 0.6381271456342805,
+      "grad_norm": 686.4176025390625,
+      "learning_rate": 3.7336830599813245e-06,
+      "loss": 96.0158,
+      "step": 77140
+    },
+    {
+      "epoch": 0.6382098688836497,
+      "grad_norm": 537.5717163085938,
+      "learning_rate": 3.732318724991505e-06,
+      "loss": 75.7251,
+      "step": 77150
+    },
+    {
+      "epoch": 0.638292592133019,
+      "grad_norm": 1141.5447998046875,
+      "learning_rate": 3.730954490865266e-06,
+      "loss": 88.8078,
+      "step": 77160
+    },
+    {
+      "epoch": 0.6383753153823882,
+      "grad_norm": 990.2410888671875,
+      "learning_rate": 3.7295903577111548e-06,
+      "loss": 86.1368,
+      "step": 77170
+    },
+    {
+      "epoch": 0.6384580386317574,
+      "grad_norm": 613.973876953125,
+      "learning_rate": 3.728226325637709e-06,
+      "loss": 57.9084,
+      "step": 77180
+    },
+    {
+      "epoch": 0.6385407618811267,
+      "grad_norm": 959.3004150390625,
+      "learning_rate": 3.726862394753457e-06,
+      "loss": 88.2764,
+      "step": 77190
+    },
+    {
+      "epoch": 0.6386234851304959,
+      "grad_norm": 1009.2028198242188,
+      "learning_rate": 3.725498565166923e-06,
+      "loss": 60.0482,
+      "step": 77200
+    },
+    {
+      "epoch": 0.6387062083798651,
+      "grad_norm": 731.6983032226562,
+      "learning_rate": 3.7241348369866183e-06,
+      "loss": 90.616,
+      "step": 77210
+    },
+    {
+      "epoch": 0.6387889316292344,
+      "grad_norm": 1137.9534912109375,
+      "learning_rate": 3.7227712103210485e-06,
+      "loss": 77.4084,
+      "step": 77220
+    },
+    {
+      "epoch": 0.6388716548786036,
+      "grad_norm": 740.2508544921875,
+      "learning_rate": 3.721407685278712e-06,
+      "loss": 99.5695,
+      "step": 77230
+    },
+    {
+      "epoch": 0.6389543781279728,
+      "grad_norm": 921.7864379882812,
+      "learning_rate": 3.7200442619680976e-06,
+      "loss": 62.2452,
+      "step": 77240
+    },
+    {
+      "epoch": 0.6390371013773422,
+      "grad_norm": 1076.6083984375,
+      "learning_rate": 3.7186809404976877e-06,
+      "loss": 95.747,
+      "step": 77250
+    },
+    {
+      "epoch": 0.6391198246267114,
+      "grad_norm": 971.0541381835938,
+      "learning_rate": 3.7173177209759538e-06,
+      "loss": 99.368,
+      "step": 77260
+    },
+    {
+      "epoch": 0.6392025478760806,
+      "grad_norm": 1410.000732421875,
+      "learning_rate": 3.715954603511363e-06,
+      "loss": 106.5322,
+      "step": 77270
+    },
+    {
+      "epoch": 0.6392852711254499,
+      "grad_norm": 587.7864990234375,
+      "learning_rate": 3.714591588212372e-06,
+      "loss": 74.5204,
+      "step": 77280
+    },
+    {
+      "epoch": 0.6393679943748191,
+      "grad_norm": 632.3125,
+      "learning_rate": 3.713228675187429e-06,
+      "loss": 105.5088,
+      "step": 77290
+    },
+    {
+      "epoch": 0.6394507176241883,
+      "grad_norm": 768.724609375,
+      "learning_rate": 3.7118658645449745e-06,
+      "loss": 87.9743,
+      "step": 77300
+    },
+    {
+      "epoch": 0.6395334408735576,
+      "grad_norm": 721.2608642578125,
+      "learning_rate": 3.710503156393441e-06,
+      "loss": 79.1286,
+      "step": 77310
+    },
+    {
+      "epoch": 0.6396161641229268,
+      "grad_norm": 1327.6044921875,
+      "learning_rate": 3.7091405508412538e-06,
+      "loss": 95.4713,
+      "step": 77320
+    },
+    {
+      "epoch": 0.639698887372296,
+      "grad_norm": 1236.894287109375,
+      "learning_rate": 3.7077780479968286e-06,
+      "loss": 91.4254,
+      "step": 77330
+    },
+    {
+      "epoch": 0.6397816106216652,
+      "grad_norm": 883.6204833984375,
+      "learning_rate": 3.7064156479685736e-06,
+      "loss": 119.5111,
+      "step": 77340
+    },
+    {
+      "epoch": 0.6398643338710345,
+      "grad_norm": 589.4072265625,
+      "learning_rate": 3.705053350864888e-06,
+      "loss": 71.212,
+      "step": 77350
+    },
+    {
+      "epoch": 0.6399470571204037,
+      "grad_norm": 1030.01904296875,
+      "learning_rate": 3.703691156794165e-06,
+      "loss": 68.7815,
+      "step": 77360
+    },
+    {
+      "epoch": 0.6400297803697729,
+      "grad_norm": 1243.1724853515625,
+      "learning_rate": 3.7023290658647893e-06,
+      "loss": 81.1806,
+      "step": 77370
+    },
+    {
+      "epoch": 0.6401125036191422,
+      "grad_norm": 1076.9923095703125,
+      "learning_rate": 3.7009670781851326e-06,
+      "loss": 92.6175,
+      "step": 77380
+    },
+    {
+      "epoch": 0.6401952268685114,
+      "grad_norm": 811.0065307617188,
+      "learning_rate": 3.6996051938635626e-06,
+      "loss": 94.767,
+      "step": 77390
+    },
+    {
+      "epoch": 0.6402779501178806,
+      "grad_norm": 707.3673095703125,
+      "learning_rate": 3.69824341300844e-06,
+      "loss": 82.464,
+      "step": 77400
+    },
+    {
+      "epoch": 0.6403606733672499,
+      "grad_norm": 801.9874267578125,
+      "learning_rate": 3.696881735728115e-06,
+      "loss": 89.9301,
+      "step": 77410
+    },
+    {
+      "epoch": 0.6404433966166191,
+      "grad_norm": 782.5747680664062,
+      "learning_rate": 3.6955201621309302e-06,
+      "loss": 91.4034,
+      "step": 77420
+    },
+    {
+      "epoch": 0.6405261198659883,
+      "grad_norm": 1794.2557373046875,
+      "learning_rate": 3.6941586923252194e-06,
+      "loss": 89.6232,
+      "step": 77430
+    },
+    {
+      "epoch": 0.6406088431153576,
+      "grad_norm": 1282.8133544921875,
+      "learning_rate": 3.6927973264193074e-06,
+      "loss": 84.1603,
+      "step": 77440
+    },
+    {
+      "epoch": 0.6406915663647268,
+      "grad_norm": 615.4580078125,
+      "learning_rate": 3.691436064521513e-06,
+      "loss": 106.5725,
+      "step": 77450
+    },
+    {
+      "epoch": 0.640774289614096,
+      "grad_norm": 885.3602905273438,
+      "learning_rate": 3.6900749067401474e-06,
+      "loss": 100.404,
+      "step": 77460
+    },
+    {
+      "epoch": 0.6408570128634653,
+      "grad_norm": 883.349853515625,
+      "learning_rate": 3.6887138531835085e-06,
+      "loss": 120.5579,
+      "step": 77470
+    },
+    {
+      "epoch": 0.6409397361128345,
+      "grad_norm": 510.637939453125,
+      "learning_rate": 3.6873529039598903e-06,
+      "loss": 72.0067,
+      "step": 77480
+    },
+    {
+      "epoch": 0.6410224593622037,
+      "grad_norm": 1126.9063720703125,
+      "learning_rate": 3.6859920591775763e-06,
+      "loss": 89.0631,
+      "step": 77490
+    },
+    {
+      "epoch": 0.641105182611573,
+      "grad_norm": 897.1529541015625,
+      "learning_rate": 3.6846313189448447e-06,
+      "loss": 95.6343,
+      "step": 77500
+    },
+    {
+      "epoch": 0.6411879058609422,
+      "grad_norm": 1228.466796875,
+      "learning_rate": 3.6832706833699616e-06,
+      "loss": 107.3611,
+      "step": 77510
+    },
+    {
+      "epoch": 0.6412706291103114,
+      "grad_norm": 386.9453125,
+      "learning_rate": 3.681910152561188e-06,
+      "loss": 56.1387,
+      "step": 77520
+    },
+    {
+      "epoch": 0.6413533523596807,
+      "grad_norm": 651.796142578125,
+      "learning_rate": 3.6805497266267742e-06,
+      "loss": 80.0673,
+      "step": 77530
+    },
+    {
+      "epoch": 0.6414360756090499,
+      "grad_norm": 970.6996459960938,
+      "learning_rate": 3.679189405674963e-06,
+      "loss": 86.8163,
+      "step": 77540
+    },
+    {
+      "epoch": 0.6415187988584191,
+      "grad_norm": 921.419189453125,
+      "learning_rate": 3.6778291898139907e-06,
+      "loss": 99.753,
+      "step": 77550
+    },
+    {
+      "epoch": 0.6416015221077884,
+      "grad_norm": 940.6666259765625,
+      "learning_rate": 3.6764690791520797e-06,
+      "loss": 72.5782,
+      "step": 77560
+    },
+    {
+      "epoch": 0.6416842453571576,
+      "grad_norm": 927.505126953125,
+      "learning_rate": 3.6751090737974506e-06,
+      "loss": 80.6144,
+      "step": 77570
+    },
+    {
+      "epoch": 0.6417669686065268,
+      "grad_norm": 1062.0577392578125,
+      "learning_rate": 3.673749173858312e-06,
+      "loss": 80.1158,
+      "step": 77580
+    },
+    {
+      "epoch": 0.6418496918558961,
+      "grad_norm": 1165.107177734375,
+      "learning_rate": 3.672389379442864e-06,
+      "loss": 109.7003,
+      "step": 77590
+    },
+    {
+      "epoch": 0.6419324151052653,
+      "grad_norm": 929.5150756835938,
+      "learning_rate": 3.6710296906593012e-06,
+      "loss": 76.4433,
+      "step": 77600
+    },
+    {
+      "epoch": 0.6420151383546345,
+      "grad_norm": 709.593017578125,
+      "learning_rate": 3.6696701076158064e-06,
+      "loss": 78.3428,
+      "step": 77610
+    },
+    {
+      "epoch": 0.6420978616040038,
+      "grad_norm": 995.3925170898438,
+      "learning_rate": 3.6683106304205564e-06,
+      "loss": 92.2917,
+      "step": 77620
+    },
+    {
+      "epoch": 0.642180584853373,
+      "grad_norm": 426.57952880859375,
+      "learning_rate": 3.666951259181718e-06,
+      "loss": 80.3045,
+      "step": 77630
+    },
+    {
+      "epoch": 0.6422633081027422,
+      "grad_norm": 1803.5283203125,
+      "learning_rate": 3.6655919940074497e-06,
+      "loss": 85.4238,
+      "step": 77640
+    },
+    {
+      "epoch": 0.6423460313521115,
+      "grad_norm": 863.35302734375,
+      "learning_rate": 3.664232835005902e-06,
+      "loss": 112.9134,
+      "step": 77650
+    },
+    {
+      "epoch": 0.6424287546014807,
+      "grad_norm": 444.5658264160156,
+      "learning_rate": 3.6628737822852177e-06,
+      "loss": 73.9999,
+      "step": 77660
+    },
+    {
+      "epoch": 0.6425114778508499,
+      "grad_norm": 895.299560546875,
+      "learning_rate": 3.66151483595353e-06,
+      "loss": 98.808,
+      "step": 77670
+    },
+    {
+      "epoch": 0.6425942011002193,
+      "grad_norm": 622.2603149414062,
+      "learning_rate": 3.6601559961189626e-06,
+      "loss": 72.3235,
+      "step": 77680
+    },
+    {
+      "epoch": 0.6426769243495885,
+      "grad_norm": 751.9857788085938,
+      "learning_rate": 3.6587972628896345e-06,
+      "loss": 65.3022,
+      "step": 77690
+    },
+    {
+      "epoch": 0.6427596475989577,
+      "grad_norm": 696.5802001953125,
+      "learning_rate": 3.6574386363736532e-06,
+      "loss": 111.5915,
+      "step": 77700
+    },
+    {
+      "epoch": 0.642842370848327,
+      "grad_norm": 1348.688232421875,
+      "learning_rate": 3.6560801166791183e-06,
+      "loss": 111.6634,
+      "step": 77710
+    },
+    {
+      "epoch": 0.6429250940976962,
+      "grad_norm": 933.01953125,
+      "learning_rate": 3.654721703914121e-06,
+      "loss": 101.2537,
+      "step": 77720
+    },
+    {
+      "epoch": 0.6430078173470654,
+      "grad_norm": 555.1231689453125,
+      "learning_rate": 3.6533633981867433e-06,
+      "loss": 71.2294,
+      "step": 77730
+    },
+    {
+      "epoch": 0.6430905405964347,
+      "grad_norm": 681.1109008789062,
+      "learning_rate": 3.652005199605059e-06,
+      "loss": 114.1313,
+      "step": 77740
+    },
+    {
+      "epoch": 0.6431732638458039,
+      "grad_norm": 1005.3911743164062,
+      "learning_rate": 3.6506471082771357e-06,
+      "loss": 98.2529,
+      "step": 77750
+    },
+    {
+      "epoch": 0.6432559870951731,
+      "grad_norm": 1053.3704833984375,
+      "learning_rate": 3.6492891243110283e-06,
+      "loss": 72.2338,
+      "step": 77760
+    },
+    {
+      "epoch": 0.6433387103445424,
+      "grad_norm": 664.69482421875,
+      "learning_rate": 3.6479312478147866e-06,
+      "loss": 93.1901,
+      "step": 77770
+    },
+    {
+      "epoch": 0.6434214335939116,
+      "grad_norm": 517.3734741210938,
+      "learning_rate": 3.64657347889645e-06,
+      "loss": 81.5787,
+      "step": 77780
+    },
+    {
+      "epoch": 0.6435041568432808,
+      "grad_norm": 778.450927734375,
+      "learning_rate": 3.6452158176640505e-06,
+      "loss": 111.1286,
+      "step": 77790
+    },
+    {
+      "epoch": 0.6435868800926501,
+      "grad_norm": 795.8098754882812,
+      "learning_rate": 3.6438582642256138e-06,
+      "loss": 80.7202,
+      "step": 77800
+    },
+    {
+      "epoch": 0.6436696033420193,
+      "grad_norm": 1202.6724853515625,
+      "learning_rate": 3.642500818689148e-06,
+      "loss": 89.1335,
+      "step": 77810
+    },
+    {
+      "epoch": 0.6437523265913885,
+      "grad_norm": 1074.335205078125,
+      "learning_rate": 3.641143481162661e-06,
+      "loss": 71.3191,
+      "step": 77820
+    },
+    {
+      "epoch": 0.6438350498407578,
+      "grad_norm": 1161.094482421875,
+      "learning_rate": 3.639786251754153e-06,
+      "loss": 85.943,
+      "step": 77830
+    },
+    {
+      "epoch": 0.643917773090127,
+      "grad_norm": 3801.1142578125,
+      "learning_rate": 3.638429130571609e-06,
+      "loss": 89.4513,
+      "step": 77840
+    },
+    {
+      "epoch": 0.6440004963394962,
+      "grad_norm": 818.4242553710938,
+      "learning_rate": 3.637072117723012e-06,
+      "loss": 108.8363,
+      "step": 77850
+    },
+    {
+      "epoch": 0.6440832195888655,
+      "grad_norm": 541.69921875,
+      "learning_rate": 3.6357152133163297e-06,
+      "loss": 83.608,
+      "step": 77860
+    },
+    {
+      "epoch": 0.6441659428382347,
+      "grad_norm": 540.3868408203125,
+      "learning_rate": 3.634358417459528e-06,
+      "loss": 109.2081,
+      "step": 77870
+    },
+    {
+      "epoch": 0.6442486660876039,
+      "grad_norm": 2112.091796875,
+      "learning_rate": 3.633001730260558e-06,
+      "loss": 145.9334,
+      "step": 77880
+    },
+    {
+      "epoch": 0.6443313893369732,
+      "grad_norm": 1037.028076171875,
+      "learning_rate": 3.63164515182737e-06,
+      "loss": 128.6462,
+      "step": 77890
+    },
+    {
+      "epoch": 0.6444141125863424,
+      "grad_norm": 710.5433959960938,
+      "learning_rate": 3.630288682267895e-06,
+      "loss": 95.2309,
+      "step": 77900
+    },
+    {
+      "epoch": 0.6444968358357116,
+      "grad_norm": 542.4974365234375,
+      "learning_rate": 3.628932321690063e-06,
+      "loss": 89.335,
+      "step": 77910
+    },
+    {
+      "epoch": 0.6445795590850809,
+      "grad_norm": 910.8280639648438,
+      "learning_rate": 3.6275760702017938e-06,
+      "loss": 123.32,
+      "step": 77920
+    },
+    {
+      "epoch": 0.6446622823344501,
+      "grad_norm": 474.7346496582031,
+      "learning_rate": 3.626219927910999e-06,
+      "loss": 144.9698,
+      "step": 77930
+    },
+    {
+      "epoch": 0.6447450055838193,
+      "grad_norm": 729.6742553710938,
+      "learning_rate": 3.6248638949255795e-06,
+      "loss": 97.5584,
+      "step": 77940
+    },
+    {
+      "epoch": 0.6448277288331886,
+      "grad_norm": 692.0260009765625,
+      "learning_rate": 3.6235079713534287e-06,
+      "loss": 96.659,
+      "step": 77950
+    },
+    {
+      "epoch": 0.6449104520825578,
+      "grad_norm": 1216.1015625,
+      "learning_rate": 3.6221521573024316e-06,
+      "loss": 119.2402,
+      "step": 77960
+    },
+    {
+      "epoch": 0.644993175331927,
+      "grad_norm": 1217.374267578125,
+      "learning_rate": 3.620796452880464e-06,
+      "loss": 110.5306,
+      "step": 77970
+    },
+    {
+      "epoch": 0.6450758985812963,
+      "grad_norm": 606.9527587890625,
+      "learning_rate": 3.6194408581953934e-06,
+      "loss": 99.9628,
+      "step": 77980
+    },
+    {
+      "epoch": 0.6451586218306655,
+      "grad_norm": 649.1746826171875,
+      "learning_rate": 3.618085373355077e-06,
+      "loss": 95.9513,
+      "step": 77990
+    },
+    {
+      "epoch": 0.6452413450800347,
+      "grad_norm": 3910.442626953125,
+      "learning_rate": 3.6167299984673655e-06,
+      "loss": 100.3708,
+      "step": 78000
+    },
+    {
+      "epoch": 0.645324068329404,
+      "grad_norm": 991.4779663085938,
+      "learning_rate": 3.615374733640099e-06,
+      "loss": 89.7993,
+      "step": 78010
+    },
+    {
+      "epoch": 0.6454067915787732,
+      "grad_norm": 814.968017578125,
+      "learning_rate": 3.6140195789811108e-06,
+      "loss": 95.6553,
+      "step": 78020
+    },
+    {
+      "epoch": 0.6454895148281424,
+      "grad_norm": 974.3428344726562,
+      "learning_rate": 3.6126645345982243e-06,
+      "loss": 88.6417,
+      "step": 78030
+    },
+    {
+      "epoch": 0.6455722380775117,
+      "grad_norm": 700.50390625,
+      "learning_rate": 3.611309600599253e-06,
+      "loss": 97.1519,
+      "step": 78040
+    },
+    {
+      "epoch": 0.6456549613268809,
+      "grad_norm": 1007.422607421875,
+      "learning_rate": 3.6099547770920046e-06,
+      "loss": 96.4095,
+      "step": 78050
+    },
+    {
+      "epoch": 0.6457376845762501,
+      "grad_norm": 833.5045776367188,
+      "learning_rate": 3.6086000641842757e-06,
+      "loss": 90.0102,
+      "step": 78060
+    },
+    {
+      "epoch": 0.6458204078256193,
+      "grad_norm": 593.3013916015625,
+      "learning_rate": 3.6072454619838525e-06,
+      "loss": 89.5628,
+      "step": 78070
+    },
+    {
+      "epoch": 0.6459031310749886,
+      "grad_norm": 937.5786743164062,
+      "learning_rate": 3.6058909705985166e-06,
+      "loss": 93.1092,
+      "step": 78080
+    },
+    {
+      "epoch": 0.6459858543243578,
+      "grad_norm": 1777.1658935546875,
+      "learning_rate": 3.6045365901360385e-06,
+      "loss": 75.8696,
+      "step": 78090
+    },
+    {
+      "epoch": 0.646068577573727,
+      "grad_norm": 819.875732421875,
+      "learning_rate": 3.603182320704179e-06,
+      "loss": 77.7278,
+      "step": 78100
+    },
+    {
+      "epoch": 0.6461513008230964,
+      "grad_norm": 661.6198120117188,
+      "learning_rate": 3.601828162410691e-06,
+      "loss": 108.2391,
+      "step": 78110
+    },
+    {
+      "epoch": 0.6462340240724656,
+      "grad_norm": 1499.3050537109375,
+      "learning_rate": 3.6004741153633194e-06,
+      "loss": 92.8364,
+      "step": 78120
+    },
+    {
+      "epoch": 0.6463167473218348,
+      "grad_norm": 854.0673828125,
+      "learning_rate": 3.5991201796698006e-06,
+      "loss": 108.2219,
+      "step": 78130
+    },
+    {
+      "epoch": 0.6463994705712041,
+      "grad_norm": 1086.9384765625,
+      "learning_rate": 3.5977663554378594e-06,
+      "loss": 84.7707,
+      "step": 78140
+    },
+    {
+      "epoch": 0.6464821938205733,
+      "grad_norm": 1125.5399169921875,
+      "learning_rate": 3.5964126427752155e-06,
+      "loss": 109.9038,
+      "step": 78150
+    },
+    {
+      "epoch": 0.6465649170699425,
+      "grad_norm": 1124.92919921875,
+      "learning_rate": 3.595059041789575e-06,
+      "loss": 101.9598,
+      "step": 78160
+    },
+    {
+      "epoch": 0.6466476403193118,
+      "grad_norm": 868.62255859375,
+      "learning_rate": 3.5937055525886377e-06,
+      "loss": 80.6028,
+      "step": 78170
+    },
+    {
+      "epoch": 0.646730363568681,
+      "grad_norm": 605.7478637695312,
+      "learning_rate": 3.592352175280096e-06,
+      "loss": 91.8966,
+      "step": 78180
+    },
+    {
+      "epoch": 0.6468130868180502,
+      "grad_norm": 630.140380859375,
+      "learning_rate": 3.5909989099716325e-06,
+      "loss": 91.8721,
+      "step": 78190
+    },
+    {
+      "epoch": 0.6468958100674195,
+      "grad_norm": 599.5414428710938,
+      "learning_rate": 3.589645756770918e-06,
+      "loss": 73.2456,
+      "step": 78200
+    },
+    {
+      "epoch": 0.6469785333167887,
+      "grad_norm": 549.9411010742188,
+      "learning_rate": 3.5882927157856175e-06,
+      "loss": 76.2016,
+      "step": 78210
+    },
+    {
+      "epoch": 0.6470612565661579,
+      "grad_norm": 965.70947265625,
+      "learning_rate": 3.586939787123388e-06,
+      "loss": 89.8131,
+      "step": 78220
+    },
+    {
+      "epoch": 0.6471439798155272,
+      "grad_norm": 705.2805786132812,
+      "learning_rate": 3.585586970891876e-06,
+      "loss": 85.9309,
+      "step": 78230
+    },
+    {
+      "epoch": 0.6472267030648964,
+      "grad_norm": 539.7587890625,
+      "learning_rate": 3.584234267198715e-06,
+      "loss": 83.7423,
+      "step": 78240
+    },
+    {
+      "epoch": 0.6473094263142656,
+      "grad_norm": 904.0587768554688,
+      "learning_rate": 3.582881676151536e-06,
+      "loss": 102.2972,
+      "step": 78250
+    },
+    {
+      "epoch": 0.6473921495636349,
+      "grad_norm": 1084.20263671875,
+      "learning_rate": 3.581529197857959e-06,
+      "loss": 72.5396,
+      "step": 78260
+    },
+    {
+      "epoch": 0.6474748728130041,
+      "grad_norm": 1199.1031494140625,
+      "learning_rate": 3.580176832425594e-06,
+      "loss": 96.1683,
+      "step": 78270
+    },
+    {
+      "epoch": 0.6475575960623733,
+      "grad_norm": 1210.6788330078125,
+      "learning_rate": 3.5788245799620425e-06,
+      "loss": 103.4515,
+      "step": 78280
+    },
+    {
+      "epoch": 0.6476403193117426,
+      "grad_norm": 670.8250122070312,
+      "learning_rate": 3.577472440574896e-06,
+      "loss": 80.9642,
+      "step": 78290
+    },
+    {
+      "epoch": 0.6477230425611118,
+      "grad_norm": 965.8837890625,
+      "learning_rate": 3.5761204143717387e-06,
+      "loss": 61.6829,
+      "step": 78300
+    },
+    {
+      "epoch": 0.647805765810481,
+      "grad_norm": 566.632568359375,
+      "learning_rate": 3.5747685014601456e-06,
+      "loss": 104.7201,
+      "step": 78310
+    },
+    {
+      "epoch": 0.6478884890598503,
+      "grad_norm": 464.7363586425781,
+      "learning_rate": 3.5734167019476845e-06,
+      "loss": 90.921,
+      "step": 78320
+    },
+    {
+      "epoch": 0.6479712123092195,
+      "grad_norm": 498.57763671875,
+      "learning_rate": 3.572065015941907e-06,
+      "loss": 85.318,
+      "step": 78330
+    },
+    {
+      "epoch": 0.6480539355585887,
+      "grad_norm": 826.5978393554688,
+      "learning_rate": 3.570713443550362e-06,
+      "loss": 94.5103,
+      "step": 78340
+    },
+    {
+      "epoch": 0.648136658807958,
+      "grad_norm": 807.204345703125,
+      "learning_rate": 3.5693619848805892e-06,
+      "loss": 90.0785,
+      "step": 78350
+    },
+    {
+      "epoch": 0.6482193820573272,
+      "grad_norm": 774.7725219726562,
+      "learning_rate": 3.568010640040118e-06,
+      "loss": 70.4705,
+      "step": 78360
+    },
+    {
+      "epoch": 0.6483021053066964,
+      "grad_norm": 504.8932189941406,
+      "learning_rate": 3.566659409136468e-06,
+      "loss": 90.6975,
+      "step": 78370
+    },
+    {
+      "epoch": 0.6483848285560657,
+      "grad_norm": 746.927978515625,
+      "learning_rate": 3.565308292277151e-06,
+      "loss": 116.9276,
+      "step": 78380
+    },
+    {
+      "epoch": 0.6484675518054349,
+      "grad_norm": 912.6074829101562,
+      "learning_rate": 3.563957289569669e-06,
+      "loss": 78.5412,
+      "step": 78390
+    },
+    {
+      "epoch": 0.6485502750548041,
+      "grad_norm": 737.116455078125,
+      "learning_rate": 3.5626064011215135e-06,
+      "loss": 76.1263,
+      "step": 78400
+    },
+    {
+      "epoch": 0.6486329983041734,
+      "grad_norm": 932.8018188476562,
+      "learning_rate": 3.5612556270401733e-06,
+      "loss": 84.2955,
+      "step": 78410
+    },
+    {
+      "epoch": 0.6487157215535426,
+      "grad_norm": 862.2860717773438,
+      "learning_rate": 3.5599049674331175e-06,
+      "loss": 102.538,
+      "step": 78420
+    },
+    {
+      "epoch": 0.6487984448029118,
+      "grad_norm": 1257.6517333984375,
+      "learning_rate": 3.5585544224078143e-06,
+      "loss": 101.6207,
+      "step": 78430
+    },
+    {
+      "epoch": 0.6488811680522811,
+      "grad_norm": 948.5725708007812,
+      "learning_rate": 3.5572039920717192e-06,
+      "loss": 79.6746,
+      "step": 78440
+    },
+    {
+      "epoch": 0.6489638913016503,
+      "grad_norm": 1046.1217041015625,
+      "learning_rate": 3.5558536765322825e-06,
+      "loss": 113.3602,
+      "step": 78450
+    },
+    {
+      "epoch": 0.6490466145510195,
+      "grad_norm": 915.197509765625,
+      "learning_rate": 3.554503475896941e-06,
+      "loss": 97.8924,
+      "step": 78460
+    },
+    {
+      "epoch": 0.6491293378003888,
+      "grad_norm": 1135.54248046875,
+      "learning_rate": 3.553153390273124e-06,
+      "loss": 69.3663,
+      "step": 78470
+    },
+    {
+      "epoch": 0.649212061049758,
+      "grad_norm": 998.7444458007812,
+      "learning_rate": 3.551803419768251e-06,
+      "loss": 73.8026,
+      "step": 78480
+    },
+    {
+      "epoch": 0.6492947842991272,
+      "grad_norm": 949.7462158203125,
+      "learning_rate": 3.5504535644897352e-06,
+      "loss": 105.2546,
+      "step": 78490
+    },
+    {
+      "epoch": 0.6493775075484965,
+      "grad_norm": 992.8348388671875,
+      "learning_rate": 3.549103824544975e-06,
+      "loss": 106.6997,
+      "step": 78500
+    },
+    {
+      "epoch": 0.6494602307978657,
+      "grad_norm": 748.7544555664062,
+      "learning_rate": 3.5477542000413657e-06,
+      "loss": 85.0411,
+      "step": 78510
+    },
+    {
+      "epoch": 0.6495429540472349,
+      "grad_norm": 1108.8211669921875,
+      "learning_rate": 3.546404691086289e-06,
+      "loss": 102.6315,
+      "step": 78520
+    },
+    {
+      "epoch": 0.6496256772966043,
+      "grad_norm": 552.9169311523438,
+      "learning_rate": 3.5450552977871207e-06,
+      "loss": 83.2827,
+      "step": 78530
+    },
+    {
+      "epoch": 0.6497084005459735,
+      "grad_norm": 615.8543090820312,
+      "learning_rate": 3.543706020251223e-06,
+      "loss": 93.9779,
+      "step": 78540
+    },
+    {
+      "epoch": 0.6497911237953427,
+      "grad_norm": 740.286376953125,
+      "learning_rate": 3.542356858585956e-06,
+      "loss": 85.2819,
+      "step": 78550
+    },
+    {
+      "epoch": 0.649873847044712,
+      "grad_norm": 898.7952270507812,
+      "learning_rate": 3.541007812898663e-06,
+      "loss": 105.5582,
+      "step": 78560
+    },
+    {
+      "epoch": 0.6499565702940812,
+      "grad_norm": 719.4003295898438,
+      "learning_rate": 3.539658883296683e-06,
+      "loss": 73.1953,
+      "step": 78570
+    },
+    {
+      "epoch": 0.6500392935434504,
+      "grad_norm": 1007.62109375,
+      "learning_rate": 3.5383100698873446e-06,
+      "loss": 103.7223,
+      "step": 78580
+    },
+    {
+      "epoch": 0.6501220167928197,
+      "grad_norm": 1619.2178955078125,
+      "learning_rate": 3.536961372777965e-06,
+      "loss": 125.7312,
+      "step": 78590
+    },
+    {
+      "epoch": 0.6502047400421889,
+      "grad_norm": 796.3026123046875,
+      "learning_rate": 3.535612792075854e-06,
+      "loss": 86.965,
+      "step": 78600
+    },
+    {
+      "epoch": 0.6502874632915581,
+      "grad_norm": 686.6871337890625,
+      "learning_rate": 3.5342643278883127e-06,
+      "loss": 67.3191,
+      "step": 78610
+    },
+    {
+      "epoch": 0.6503701865409274,
+      "grad_norm": 950.4487915039062,
+      "learning_rate": 3.532915980322632e-06,
+      "loss": 76.5455,
+      "step": 78620
+    },
+    {
+      "epoch": 0.6504529097902966,
+      "grad_norm": 750.7484741210938,
+      "learning_rate": 3.5315677494860923e-06,
+      "loss": 94.5387,
+      "step": 78630
+    },
+    {
+      "epoch": 0.6505356330396658,
+      "grad_norm": 599.2245483398438,
+      "learning_rate": 3.5302196354859693e-06,
+      "loss": 97.6871,
+      "step": 78640
+    },
+    {
+      "epoch": 0.6506183562890351,
+      "grad_norm": 915.01123046875,
+      "learning_rate": 3.528871638429524e-06,
+      "loss": 84.8775,
+      "step": 78650
+    },
+    {
+      "epoch": 0.6507010795384043,
+      "grad_norm": 1431.2159423828125,
+      "learning_rate": 3.527523758424013e-06,
+      "loss": 101.8159,
+      "step": 78660
+    },
+    {
+      "epoch": 0.6507838027877735,
+      "grad_norm": 907.7205810546875,
+      "learning_rate": 3.526175995576676e-06,
+      "loss": 92.8583,
+      "step": 78670
+    },
+    {
+      "epoch": 0.6508665260371428,
+      "grad_norm": 685.8528442382812,
+      "learning_rate": 3.524828349994752e-06,
+      "loss": 108.4734,
+      "step": 78680
+    },
+    {
+      "epoch": 0.650949249286512,
+      "grad_norm": 1018.92919921875,
+      "learning_rate": 3.523480821785466e-06,
+      "loss": 116.4635,
+      "step": 78690
+    },
+    {
+      "epoch": 0.6510319725358812,
+      "grad_norm": 616.6637573242188,
+      "learning_rate": 3.5221334110560345e-06,
+      "loss": 70.0087,
+      "step": 78700
+    },
+    {
+      "epoch": 0.6511146957852505,
+      "grad_norm": 1268.9815673828125,
+      "learning_rate": 3.5207861179136654e-06,
+      "loss": 87.3583,
+      "step": 78710
+    },
+    {
+      "epoch": 0.6511974190346197,
+      "grad_norm": 787.7352905273438,
+      "learning_rate": 3.519438942465556e-06,
+      "loss": 79.1112,
+      "step": 78720
+    },
+    {
+      "epoch": 0.6512801422839889,
+      "grad_norm": 756.2986450195312,
+      "learning_rate": 3.5180918848188937e-06,
+      "loss": 89.4581,
+      "step": 78730
+    },
+    {
+      "epoch": 0.6513628655333582,
+      "grad_norm": 439.9861145019531,
+      "learning_rate": 3.516744945080861e-06,
+      "loss": 82.7982,
+      "step": 78740
+    },
+    {
+      "epoch": 0.6514455887827274,
+      "grad_norm": 702.6160278320312,
+      "learning_rate": 3.5153981233586277e-06,
+      "loss": 87.4703,
+      "step": 78750
+    },
+    {
+      "epoch": 0.6515283120320966,
+      "grad_norm": 973.2195434570312,
+      "learning_rate": 3.5140514197593494e-06,
+      "loss": 105.3042,
+      "step": 78760
+    },
+    {
+      "epoch": 0.6516110352814659,
+      "grad_norm": 547.9530639648438,
+      "learning_rate": 3.512704834390179e-06,
+      "loss": 90.783,
+      "step": 78770
+    },
+    {
+      "epoch": 0.6516937585308351,
+      "grad_norm": 734.762939453125,
+      "learning_rate": 3.5113583673582613e-06,
+      "loss": 93.9857,
+      "step": 78780
+    },
+    {
+      "epoch": 0.6517764817802043,
+      "grad_norm": 730.4273681640625,
+      "learning_rate": 3.510012018770726e-06,
+      "loss": 96.3544,
+      "step": 78790
+    },
+    {
+      "epoch": 0.6518592050295735,
+      "grad_norm": 1371.4984130859375,
+      "learning_rate": 3.508665788734696e-06,
+      "loss": 97.4626,
+      "step": 78800
+    },
+    {
+      "epoch": 0.6519419282789428,
+      "grad_norm": 974.1882934570312,
+      "learning_rate": 3.507319677357285e-06,
+      "loss": 106.0504,
+      "step": 78810
+    },
+    {
+      "epoch": 0.652024651528312,
+      "grad_norm": 973.0641479492188,
+      "learning_rate": 3.5059736847455967e-06,
+      "loss": 85.558,
+      "step": 78820
+    },
+    {
+      "epoch": 0.6521073747776812,
+      "grad_norm": 1314.4976806640625,
+      "learning_rate": 3.5046278110067242e-06,
+      "loss": 122.3348,
+      "step": 78830
+    },
+    {
+      "epoch": 0.6521900980270505,
+      "grad_norm": 687.9398803710938,
+      "learning_rate": 3.5032820562477577e-06,
+      "loss": 150.5666,
+      "step": 78840
+    },
+    {
+      "epoch": 0.6522728212764197,
+      "grad_norm": 1185.294189453125,
+      "learning_rate": 3.5019364205757667e-06,
+      "loss": 89.5953,
+      "step": 78850
+    },
+    {
+      "epoch": 0.6523555445257889,
+      "grad_norm": 736.0501708984375,
+      "learning_rate": 3.5005909040978188e-06,
+      "loss": 88.9674,
+      "step": 78860
+    },
+    {
+      "epoch": 0.6524382677751582,
+      "grad_norm": 644.609619140625,
+      "learning_rate": 3.4992455069209717e-06,
+      "loss": 84.699,
+      "step": 78870
+    },
+    {
+      "epoch": 0.6525209910245274,
+      "grad_norm": 954.2164306640625,
+      "learning_rate": 3.4979002291522723e-06,
+      "loss": 128.6724,
+      "step": 78880
+    },
+    {
+      "epoch": 0.6526037142738966,
+      "grad_norm": 750.8978881835938,
+      "learning_rate": 3.4965550708987583e-06,
+      "loss": 71.4564,
+      "step": 78890
+    },
+    {
+      "epoch": 0.6526864375232659,
+      "grad_norm": 927.7744140625,
+      "learning_rate": 3.4952100322674574e-06,
+      "loss": 93.2375,
+      "step": 78900
+    },
+    {
+      "epoch": 0.6527691607726351,
+      "grad_norm": 882.0057983398438,
+      "learning_rate": 3.4938651133653877e-06,
+      "loss": 89.3872,
+      "step": 78910
+    },
+    {
+      "epoch": 0.6528518840220043,
+      "grad_norm": 909.5263671875,
+      "learning_rate": 3.49252031429956e-06,
+      "loss": 90.456,
+      "step": 78920
+    },
+    {
+      "epoch": 0.6529346072713736,
+      "grad_norm": 810.2213134765625,
+      "learning_rate": 3.4911756351769722e-06,
+      "loss": 93.9017,
+      "step": 78930
+    },
+    {
+      "epoch": 0.6530173305207428,
+      "grad_norm": 510.4373474121094,
+      "learning_rate": 3.4898310761046133e-06,
+      "loss": 75.6976,
+      "step": 78940
+    },
+    {
+      "epoch": 0.653100053770112,
+      "grad_norm": 876.5424194335938,
+      "learning_rate": 3.4884866371894654e-06,
+      "loss": 82.5301,
+      "step": 78950
+    },
+    {
+      "epoch": 0.6531827770194814,
+      "grad_norm": 496.5128173828125,
+      "learning_rate": 3.487142318538498e-06,
+      "loss": 124.7649,
+      "step": 78960
+    },
+    {
+      "epoch": 0.6532655002688506,
+      "grad_norm": 574.9053344726562,
+      "learning_rate": 3.4857981202586742e-06,
+      "loss": 117.4653,
+      "step": 78970
+    },
+    {
+      "epoch": 0.6533482235182198,
+      "grad_norm": 619.2501831054688,
+      "learning_rate": 3.4844540424569453e-06,
+      "loss": 75.0704,
+      "step": 78980
+    },
+    {
+      "epoch": 0.6534309467675891,
+      "grad_norm": 502.5335693359375,
+      "learning_rate": 3.483110085240252e-06,
+      "loss": 117.5442,
+      "step": 78990
+    },
+    {
+      "epoch": 0.6535136700169583,
+      "grad_norm": 802.0072631835938,
+      "learning_rate": 3.481766248715528e-06,
+      "loss": 100.3333,
+      "step": 79000
+    },
+    {
+      "epoch": 0.6535963932663275,
+      "grad_norm": 762.8905639648438,
+      "learning_rate": 3.4804225329896963e-06,
+      "loss": 100.5814,
+      "step": 79010
+    },
+    {
+      "epoch": 0.6536791165156968,
+      "grad_norm": 732.5484619140625,
+      "learning_rate": 3.4790789381696686e-06,
+      "loss": 110.0594,
+      "step": 79020
+    },
+    {
+      "epoch": 0.653761839765066,
+      "grad_norm": 865.77734375,
+      "learning_rate": 3.4777354643623506e-06,
+      "loss": 80.9146,
+      "step": 79030
+    },
+    {
+      "epoch": 0.6538445630144352,
+      "grad_norm": 957.7134399414062,
+      "learning_rate": 3.4763921116746352e-06,
+      "loss": 93.2645,
+      "step": 79040
+    },
+    {
+      "epoch": 0.6539272862638045,
+      "grad_norm": 900.92041015625,
+      "learning_rate": 3.475048880213407e-06,
+      "loss": 62.1217,
+      "step": 79050
+    },
+    {
+      "epoch": 0.6540100095131737,
+      "grad_norm": 544.1870727539062,
+      "learning_rate": 3.473705770085539e-06,
+      "loss": 93.6251,
+      "step": 79060
+    },
+    {
+      "epoch": 0.6540927327625429,
+      "grad_norm": 1076.5374755859375,
+      "learning_rate": 3.4723627813979005e-06,
+      "loss": 113.8579,
+      "step": 79070
+    },
+    {
+      "epoch": 0.6541754560119122,
+      "grad_norm": 862.5486450195312,
+      "learning_rate": 3.471019914257344e-06,
+      "loss": 94.458,
+      "step": 79080
+    },
+    {
+      "epoch": 0.6542581792612814,
+      "grad_norm": 637.738037109375,
+      "learning_rate": 3.4696771687707176e-06,
+      "loss": 86.9744,
+      "step": 79090
+    },
+    {
+      "epoch": 0.6543409025106506,
+      "grad_norm": 979.8980712890625,
+      "learning_rate": 3.468334545044853e-06,
+      "loss": 89.8826,
+      "step": 79100
+    },
+    {
+      "epoch": 0.6544236257600199,
+      "grad_norm": 1138.6566162109375,
+      "learning_rate": 3.46699204318658e-06,
+      "loss": 93.8205,
+      "step": 79110
+    },
+    {
+      "epoch": 0.6545063490093891,
+      "grad_norm": 680.276611328125,
+      "learning_rate": 3.465649663302715e-06,
+      "loss": 87.8101,
+      "step": 79120
+    },
+    {
+      "epoch": 0.6545890722587583,
+      "grad_norm": 803.8458251953125,
+      "learning_rate": 3.464307405500064e-06,
+      "loss": 99.0789,
+      "step": 79130
+    },
+    {
+      "epoch": 0.6546717955081276,
+      "grad_norm": 1364.3984375,
+      "learning_rate": 3.4629652698854254e-06,
+      "loss": 92.8128,
+      "step": 79140
+    },
+    {
+      "epoch": 0.6547545187574968,
+      "grad_norm": 751.8421630859375,
+      "learning_rate": 3.461623256565586e-06,
+      "loss": 78.3212,
+      "step": 79150
+    },
+    {
+      "epoch": 0.654837242006866,
+      "grad_norm": 862.8961181640625,
+      "learning_rate": 3.4602813656473223e-06,
+      "loss": 94.2935,
+      "step": 79160
+    },
+    {
+      "epoch": 0.6549199652562353,
+      "grad_norm": 1315.6492919921875,
+      "learning_rate": 3.4589395972374055e-06,
+      "loss": 94.4292,
+      "step": 79170
+    },
+    {
+      "epoch": 0.6550026885056045,
+      "grad_norm": 714.10546875,
+      "learning_rate": 3.457597951442595e-06,
+      "loss": 61.3309,
+      "step": 79180
+    },
+    {
+      "epoch": 0.6550854117549737,
+      "grad_norm": 2307.056640625,
+      "learning_rate": 3.456256428369633e-06,
+      "loss": 95.6213,
+      "step": 79190
+    },
+    {
+      "epoch": 0.655168135004343,
+      "grad_norm": 1832.7984619140625,
+      "learning_rate": 3.4549150281252635e-06,
+      "loss": 94.4903,
+      "step": 79200
+    },
+    {
+      "epoch": 0.6552508582537122,
+      "grad_norm": 0.0,
+      "learning_rate": 3.453573750816214e-06,
+      "loss": 69.4073,
+      "step": 79210
+    },
+    {
+      "epoch": 0.6553335815030814,
+      "grad_norm": 550.095947265625,
+      "learning_rate": 3.452232596549204e-06,
+      "loss": 83.0661,
+      "step": 79220
+    },
+    {
+      "epoch": 0.6554163047524507,
+      "grad_norm": 833.4959716796875,
+      "learning_rate": 3.4508915654309438e-06,
+      "loss": 93.9101,
+      "step": 79230
+    },
+    {
+      "epoch": 0.6554990280018199,
+      "grad_norm": 584.6357421875,
+      "learning_rate": 3.4495506575681313e-06,
+      "loss": 89.3119,
+      "step": 79240
+    },
+    {
+      "epoch": 0.6555817512511891,
+      "grad_norm": 557.8964233398438,
+      "learning_rate": 3.4482098730674577e-06,
+      "loss": 78.8267,
+      "step": 79250
+    },
+    {
+      "epoch": 0.6556644745005584,
+      "grad_norm": 1147.6644287109375,
+      "learning_rate": 3.4468692120356017e-06,
+      "loss": 115.0388,
+      "step": 79260
+    },
+    {
+      "epoch": 0.6557471977499276,
+      "grad_norm": 1800.31494140625,
+      "learning_rate": 3.4455286745792383e-06,
+      "loss": 90.1449,
+      "step": 79270
+    },
+    {
+      "epoch": 0.6558299209992968,
+      "grad_norm": 1146.2457275390625,
+      "learning_rate": 3.4441882608050216e-06,
+      "loss": 99.0781,
+      "step": 79280
+    },
+    {
+      "epoch": 0.6559126442486661,
+      "grad_norm": 784.1929931640625,
+      "learning_rate": 3.442847970819604e-06,
+      "loss": 86.3861,
+      "step": 79290
+    },
+    {
+      "epoch": 0.6559953674980353,
+      "grad_norm": 1415.5955810546875,
+      "learning_rate": 3.441507804729627e-06,
+      "loss": 112.1447,
+      "step": 79300
+    },
+    {
+      "epoch": 0.6560780907474045,
+      "grad_norm": 643.2744140625,
+      "learning_rate": 3.440167762641722e-06,
+      "loss": 97.26,
+      "step": 79310
+    },
+    {
+      "epoch": 0.6561608139967738,
+      "grad_norm": 990.8764038085938,
+      "learning_rate": 3.43882784466251e-06,
+      "loss": 70.3671,
+      "step": 79320
+    },
+    {
+      "epoch": 0.656243537246143,
+      "grad_norm": 1008.8643188476562,
+      "learning_rate": 3.4374880508986013e-06,
+      "loss": 106.8238,
+      "step": 79330
+    },
+    {
+      "epoch": 0.6563262604955122,
+      "grad_norm": 616.8389892578125,
+      "learning_rate": 3.436148381456598e-06,
+      "loss": 88.8686,
+      "step": 79340
+    },
+    {
+      "epoch": 0.6564089837448815,
+      "grad_norm": 1004.9009399414062,
+      "learning_rate": 3.434808836443091e-06,
+      "loss": 79.8238,
+      "step": 79350
+    },
+    {
+      "epoch": 0.6564917069942507,
+      "grad_norm": 1268.065673828125,
+      "learning_rate": 3.4334694159646608e-06,
+      "loss": 86.7182,
+      "step": 79360
+    },
+    {
+      "epoch": 0.6565744302436199,
+      "grad_norm": 1004.0042724609375,
+      "learning_rate": 3.43213012012788e-06,
+      "loss": 108.0468,
+      "step": 79370
+    },
+    {
+      "epoch": 0.6566571534929893,
+      "grad_norm": 563.8744506835938,
+      "learning_rate": 3.43079094903931e-06,
+      "loss": 74.4405,
+      "step": 79380
+    },
+    {
+      "epoch": 0.6567398767423585,
+      "grad_norm": 410.0007019042969,
+      "learning_rate": 3.4294519028055014e-06,
+      "loss": 89.0153,
+      "step": 79390
+    },
+    {
+      "epoch": 0.6568225999917277,
+      "grad_norm": 816.875,
+      "learning_rate": 3.428112981532998e-06,
+      "loss": 66.4603,
+      "step": 79400
+    },
+    {
+      "epoch": 0.656905323241097,
+      "grad_norm": 665.1447143554688,
+      "learning_rate": 3.4267741853283305e-06,
+      "loss": 123.0643,
+      "step": 79410
+    },
+    {
+      "epoch": 0.6569880464904662,
+      "grad_norm": 1130.827392578125,
+      "learning_rate": 3.425435514298021e-06,
+      "loss": 113.6049,
+      "step": 79420
+    },
+    {
+      "epoch": 0.6570707697398354,
+      "grad_norm": 933.1907348632812,
+      "learning_rate": 3.4240969685485813e-06,
+      "loss": 85.3566,
+      "step": 79430
+    },
+    {
+      "epoch": 0.6571534929892047,
+      "grad_norm": 930.4938354492188,
+      "learning_rate": 3.422758548186515e-06,
+      "loss": 110.3764,
+      "step": 79440
+    },
+    {
+      "epoch": 0.6572362162385739,
+      "grad_norm": 698.0180053710938,
+      "learning_rate": 3.4214202533183104e-06,
+      "loss": 89.8429,
+      "step": 79450
+    },
+    {
+      "epoch": 0.6573189394879431,
+      "grad_norm": 949.0037841796875,
+      "learning_rate": 3.420082084050453e-06,
+      "loss": 125.2701,
+      "step": 79460
+    },
+    {
+      "epoch": 0.6574016627373124,
+      "grad_norm": 1515.08349609375,
+      "learning_rate": 3.4187440404894123e-06,
+      "loss": 111.6007,
+      "step": 79470
+    },
+    {
+      "epoch": 0.6574843859866816,
+      "grad_norm": 531.5831298828125,
+      "learning_rate": 3.417406122741651e-06,
+      "loss": 115.491,
+      "step": 79480
+    },
+    {
+      "epoch": 0.6575671092360508,
+      "grad_norm": 951.4327392578125,
+      "learning_rate": 3.416068330913621e-06,
+      "loss": 77.1653,
+      "step": 79490
+    },
+    {
+      "epoch": 0.6576498324854201,
+      "grad_norm": 1479.0516357421875,
+      "learning_rate": 3.4147306651117663e-06,
+      "loss": 102.2539,
+      "step": 79500
+    },
+    {
+      "epoch": 0.6577325557347893,
+      "grad_norm": 1768.2347412109375,
+      "learning_rate": 3.4133931254425156e-06,
+      "loss": 139.0158,
+      "step": 79510
+    },
+    {
+      "epoch": 0.6578152789841585,
+      "grad_norm": 519.06103515625,
+      "learning_rate": 3.4120557120122944e-06,
+      "loss": 103.6259,
+      "step": 79520
+    },
+    {
+      "epoch": 0.6578980022335277,
+      "grad_norm": 923.902587890625,
+      "learning_rate": 3.4107184249275114e-06,
+      "loss": 69.9052,
+      "step": 79530
+    },
+    {
+      "epoch": 0.657980725482897,
+      "grad_norm": 812.4449462890625,
+      "learning_rate": 3.4093812642945694e-06,
+      "loss": 82.5353,
+      "step": 79540
+    },
+    {
+      "epoch": 0.6580634487322662,
+      "grad_norm": 669.1686401367188,
+      "learning_rate": 3.40804423021986e-06,
+      "loss": 75.8008,
+      "step": 79550
+    },
+    {
+      "epoch": 0.6581461719816354,
+      "grad_norm": 556.7330322265625,
+      "learning_rate": 3.4067073228097656e-06,
+      "loss": 79.9305,
+      "step": 79560
+    },
+    {
+      "epoch": 0.6582288952310047,
+      "grad_norm": 828.5440673828125,
+      "learning_rate": 3.4053705421706574e-06,
+      "loss": 119.8367,
+      "step": 79570
+    },
+    {
+      "epoch": 0.6583116184803739,
+      "grad_norm": 982.3250732421875,
+      "learning_rate": 3.4040338884088955e-06,
+      "loss": 103.1818,
+      "step": 79580
+    },
+    {
+      "epoch": 0.6583943417297431,
+      "grad_norm": 556.98193359375,
+      "learning_rate": 3.4026973616308334e-06,
+      "loss": 127.7505,
+      "step": 79590
+    },
+    {
+      "epoch": 0.6584770649791124,
+      "grad_norm": 930.4917602539062,
+      "learning_rate": 3.401360961942812e-06,
+      "loss": 85.2934,
+      "step": 79600
+    },
+    {
+      "epoch": 0.6585597882284816,
+      "grad_norm": 1134.147705078125,
+      "learning_rate": 3.4000246894511634e-06,
+      "loss": 70.3682,
+      "step": 79610
+    },
+    {
+      "epoch": 0.6586425114778508,
+      "grad_norm": 524.334228515625,
+      "learning_rate": 3.398688544262205e-06,
+      "loss": 74.7115,
+      "step": 79620
+    },
+    {
+      "epoch": 0.6587252347272201,
+      "grad_norm": 1209.0157470703125,
+      "learning_rate": 3.397352526482251e-06,
+      "loss": 86.3904,
+      "step": 79630
+    },
+    {
+      "epoch": 0.6588079579765893,
+      "grad_norm": 833.2462768554688,
+      "learning_rate": 3.396016636217601e-06,
+      "loss": 103.8662,
+      "step": 79640
+    },
+    {
+      "epoch": 0.6588906812259585,
+      "grad_norm": 1902.74951171875,
+      "learning_rate": 3.394680873574546e-06,
+      "loss": 89.4633,
+      "step": 79650
+    },
+    {
+      "epoch": 0.6589734044753278,
+      "grad_norm": 1193.58740234375,
+      "learning_rate": 3.3933452386593666e-06,
+      "loss": 83.5736,
+      "step": 79660
+    },
+    {
+      "epoch": 0.659056127724697,
+      "grad_norm": 602.8118896484375,
+      "learning_rate": 3.392009731578334e-06,
+      "loss": 115.8359,
+      "step": 79670
+    },
+    {
+      "epoch": 0.6591388509740662,
+      "grad_norm": 744.4619140625,
+      "learning_rate": 3.3906743524377053e-06,
+      "loss": 78.6802,
+      "step": 79680
+    },
+    {
+      "epoch": 0.6592215742234355,
+      "grad_norm": 1028.5284423828125,
+      "learning_rate": 3.3893391013437338e-06,
+      "loss": 82.8485,
+      "step": 79690
+    },
+    {
+      "epoch": 0.6593042974728047,
+      "grad_norm": 1039.501708984375,
+      "learning_rate": 3.38800397840266e-06,
+      "loss": 97.3438,
+      "step": 79700
+    },
+    {
+      "epoch": 0.6593870207221739,
+      "grad_norm": 1091.6658935546875,
+      "learning_rate": 3.3866689837207094e-06,
+      "loss": 103.66,
+      "step": 79710
+    },
+    {
+      "epoch": 0.6594697439715432,
+      "grad_norm": 1788.04345703125,
+      "learning_rate": 3.3853341174041025e-06,
+      "loss": 97.2377,
+      "step": 79720
+    },
+    {
+      "epoch": 0.6595524672209124,
+      "grad_norm": 842.3888549804688,
+      "learning_rate": 3.3839993795590507e-06,
+      "loss": 101.914,
+      "step": 79730
+    },
+    {
+      "epoch": 0.6596351904702816,
+      "grad_norm": 1032.7957763671875,
+      "learning_rate": 3.3826647702917526e-06,
+      "loss": 83.7671,
+      "step": 79740
+    },
+    {
+      "epoch": 0.6597179137196509,
+      "grad_norm": 910.4108276367188,
+      "learning_rate": 3.3813302897083955e-06,
+      "loss": 114.1942,
+      "step": 79750
+    },
+    {
+      "epoch": 0.6598006369690201,
+      "grad_norm": 710.4639282226562,
+      "learning_rate": 3.379995937915158e-06,
+      "loss": 84.6564,
+      "step": 79760
+    },
+    {
+      "epoch": 0.6598833602183893,
+      "grad_norm": 667.979248046875,
+      "learning_rate": 3.37866171501821e-06,
+      "loss": 77.0747,
+      "step": 79770
+    },
+    {
+      "epoch": 0.6599660834677586,
+      "grad_norm": 1670.4046630859375,
+      "learning_rate": 3.3773276211237087e-06,
+      "loss": 120.1049,
+      "step": 79780
+    },
+    {
+      "epoch": 0.6600488067171278,
+      "grad_norm": 613.4122314453125,
+      "learning_rate": 3.3759936563378004e-06,
+      "loss": 83.0683,
+      "step": 79790
+    },
+    {
+      "epoch": 0.660131529966497,
+      "grad_norm": 902.4213256835938,
+      "learning_rate": 3.374659820766625e-06,
+      "loss": 99.4036,
+      "step": 79800
+    },
+    {
+      "epoch": 0.6602142532158664,
+      "grad_norm": 841.2720336914062,
+      "learning_rate": 3.3733261145163064e-06,
+      "loss": 112.4562,
+      "step": 79810
+    },
+    {
+      "epoch": 0.6602969764652356,
+      "grad_norm": 609.0618286132812,
+      "learning_rate": 3.371992537692964e-06,
+      "loss": 95.7773,
+      "step": 79820
+    },
+    {
+      "epoch": 0.6603796997146048,
+      "grad_norm": 1067.176025390625,
+      "learning_rate": 3.370659090402704e-06,
+      "loss": 115.8151,
+      "step": 79830
+    },
+    {
+      "epoch": 0.6604624229639741,
+      "grad_norm": 1055.511474609375,
+      "learning_rate": 3.3693257727516227e-06,
+      "loss": 96.7919,
+      "step": 79840
+    },
+    {
+      "epoch": 0.6605451462133433,
+      "grad_norm": 510.99395751953125,
+      "learning_rate": 3.367992584845806e-06,
+      "loss": 76.2172,
+      "step": 79850
+    },
+    {
+      "epoch": 0.6606278694627125,
+      "grad_norm": 407.9087219238281,
+      "learning_rate": 3.3666595267913293e-06,
+      "loss": 86.5948,
+      "step": 79860
+    },
+    {
+      "epoch": 0.6607105927120818,
+      "grad_norm": 1010.9825439453125,
+      "learning_rate": 3.365326598694259e-06,
+      "loss": 80.669,
+      "step": 79870
+    },
+    {
+      "epoch": 0.660793315961451,
+      "grad_norm": 707.4743041992188,
+      "learning_rate": 3.3639938006606483e-06,
+      "loss": 89.7645,
+      "step": 79880
+    },
+    {
+      "epoch": 0.6608760392108202,
+      "grad_norm": 1092.1805419921875,
+      "learning_rate": 3.3626611327965418e-06,
+      "loss": 103.0097,
+      "step": 79890
+    },
+    {
+      "epoch": 0.6609587624601895,
+      "grad_norm": 530.4946899414062,
+      "learning_rate": 3.3613285952079754e-06,
+      "loss": 87.2853,
+      "step": 79900
+    },
+    {
+      "epoch": 0.6610414857095587,
+      "grad_norm": 750.7198486328125,
+      "learning_rate": 3.3599961880009713e-06,
+      "loss": 80.3749,
+      "step": 79910
+    },
+    {
+      "epoch": 0.6611242089589279,
+      "grad_norm": 728.5155029296875,
+      "learning_rate": 3.3586639112815446e-06,
+      "loss": 104.2869,
+      "step": 79920
+    },
+    {
+      "epoch": 0.6612069322082972,
+      "grad_norm": 1172.182373046875,
+      "learning_rate": 3.357331765155698e-06,
+      "loss": 76.1472,
+      "step": 79930
+    },
+    {
+      "epoch": 0.6612896554576664,
+      "grad_norm": 1009.7491455078125,
+      "learning_rate": 3.355999749729424e-06,
+      "loss": 83.3518,
+      "step": 79940
+    },
+    {
+      "epoch": 0.6613723787070356,
+      "grad_norm": 865.0538940429688,
+      "learning_rate": 3.354667865108706e-06,
+      "loss": 121.849,
+      "step": 79950
+    },
+    {
+      "epoch": 0.6614551019564049,
+      "grad_norm": 635.9669189453125,
+      "learning_rate": 3.353336111399513e-06,
+      "loss": 103.0901,
+      "step": 79960
+    },
+    {
+      "epoch": 0.6615378252057741,
+      "grad_norm": 3667.30615234375,
+      "learning_rate": 3.3520044887078096e-06,
+      "loss": 148.1079,
+      "step": 79970
+    },
+    {
+      "epoch": 0.6616205484551433,
+      "grad_norm": 1349.1805419921875,
+      "learning_rate": 3.350672997139546e-06,
+      "loss": 102.0127,
+      "step": 79980
+    },
+    {
+      "epoch": 0.6617032717045126,
+      "grad_norm": 795.7191162109375,
+      "learning_rate": 3.3493416368006614e-06,
+      "loss": 75.6753,
+      "step": 79990
+    },
+    {
+      "epoch": 0.6617859949538818,
+      "grad_norm": 692.658935546875,
+      "learning_rate": 3.348010407797088e-06,
+      "loss": 82.7135,
+      "step": 80000
+    },
+    {
+      "epoch": 0.661868718203251,
+      "grad_norm": 1096.899169921875,
+      "learning_rate": 3.346679310234744e-06,
+      "loss": 79.7514,
+      "step": 80010
+    },
+    {
+      "epoch": 0.6619514414526203,
+      "grad_norm": 1015.4619140625,
+      "learning_rate": 3.34534834421954e-06,
+      "loss": 59.5193,
+      "step": 80020
+    },
+    {
+      "epoch": 0.6620341647019895,
+      "grad_norm": 537.7772216796875,
+      "learning_rate": 3.3440175098573748e-06,
+      "loss": 87.1953,
+      "step": 80030
+    },
+    {
+      "epoch": 0.6621168879513587,
+      "grad_norm": 721.2218627929688,
+      "learning_rate": 3.3426868072541386e-06,
+      "loss": 81.0099,
+      "step": 80040
+    },
+    {
+      "epoch": 0.662199611200728,
+      "grad_norm": 862.6383666992188,
+      "learning_rate": 3.3413562365157037e-06,
+      "loss": 82.9833,
+      "step": 80050
+    },
+    {
+      "epoch": 0.6622823344500972,
+      "grad_norm": 879.6834106445312,
+      "learning_rate": 3.340025797747942e-06,
+      "loss": 72.3241,
+      "step": 80060
+    },
+    {
+      "epoch": 0.6623650576994664,
+      "grad_norm": 831.2039184570312,
+      "learning_rate": 3.3386954910567094e-06,
+      "loss": 66.0481,
+      "step": 80070
+    },
+    {
+      "epoch": 0.6624477809488357,
+      "grad_norm": 699.9284057617188,
+      "learning_rate": 3.337365316547852e-06,
+      "loss": 98.672,
+      "step": 80080
+    },
+    {
+      "epoch": 0.6625305041982049,
+      "grad_norm": 831.2557983398438,
+      "learning_rate": 3.336035274327206e-06,
+      "loss": 99.8643,
+      "step": 80090
+    },
+    {
+      "epoch": 0.6626132274475741,
+      "grad_norm": 1425.111328125,
+      "learning_rate": 3.3347053645005965e-06,
+      "loss": 82.7985,
+      "step": 80100
+    },
+    {
+      "epoch": 0.6626959506969434,
+      "grad_norm": 794.6343994140625,
+      "learning_rate": 3.333375587173838e-06,
+      "loss": 107.7234,
+      "step": 80110
+    },
+    {
+      "epoch": 0.6627786739463126,
+      "grad_norm": 481.92596435546875,
+      "learning_rate": 3.332045942452738e-06,
+      "loss": 62.8261,
+      "step": 80120
+    },
+    {
+      "epoch": 0.6628613971956818,
+      "grad_norm": 651.9033813476562,
+      "learning_rate": 3.330716430443085e-06,
+      "loss": 76.272,
+      "step": 80130
+    },
+    {
+      "epoch": 0.6629441204450511,
+      "grad_norm": 692.5248413085938,
+      "learning_rate": 3.329387051250664e-06,
+      "loss": 92.6327,
+      "step": 80140
+    },
+    {
+      "epoch": 0.6630268436944203,
+      "grad_norm": 537.0725708007812,
+      "learning_rate": 3.3280578049812493e-06,
+      "loss": 73.6244,
+      "step": 80150
+    },
+    {
+      "epoch": 0.6631095669437895,
+      "grad_norm": 891.6332397460938,
+      "learning_rate": 3.3267286917406027e-06,
+      "loss": 67.5265,
+      "step": 80160
+    },
+    {
+      "epoch": 0.6631922901931588,
+      "grad_norm": 662.6693115234375,
+      "learning_rate": 3.3253997116344737e-06,
+      "loss": 102.1911,
+      "step": 80170
+    },
+    {
+      "epoch": 0.663275013442528,
+      "grad_norm": 1005.4918823242188,
+      "learning_rate": 3.3240708647686047e-06,
+      "loss": 76.6256,
+      "step": 80180
+    },
+    {
+      "epoch": 0.6633577366918972,
+      "grad_norm": 629.626953125,
+      "learning_rate": 3.322742151248726e-06,
+      "loss": 87.3736,
+      "step": 80190
+    },
+    {
+      "epoch": 0.6634404599412665,
+      "grad_norm": 760.7479858398438,
+      "learning_rate": 3.3214135711805555e-06,
+      "loss": 96.766,
+      "step": 80200
+    },
+    {
+      "epoch": 0.6635231831906357,
+      "grad_norm": 556.74365234375,
+      "learning_rate": 3.3200851246698053e-06,
+      "loss": 77.8871,
+      "step": 80210
+    },
+    {
+      "epoch": 0.663605906440005,
+      "grad_norm": 645.0292358398438,
+      "learning_rate": 3.318756811822171e-06,
+      "loss": 74.2956,
+      "step": 80220
+    },
+    {
+      "epoch": 0.6636886296893741,
+      "grad_norm": 1107.0675048828125,
+      "learning_rate": 3.3174286327433408e-06,
+      "loss": 87.9465,
+      "step": 80230
+    },
+    {
+      "epoch": 0.6637713529387435,
+      "grad_norm": 770.3927001953125,
+      "learning_rate": 3.3161005875389916e-06,
+      "loss": 91.9657,
+      "step": 80240
+    },
+    {
+      "epoch": 0.6638540761881127,
+      "grad_norm": 1054.7138671875,
+      "learning_rate": 3.3147726763147913e-06,
+      "loss": 84.3892,
+      "step": 80250
+    },
+    {
+      "epoch": 0.6639367994374819,
+      "grad_norm": 1000.5108032226562,
+      "learning_rate": 3.3134448991763957e-06,
+      "loss": 101.0391,
+      "step": 80260
+    },
+    {
+      "epoch": 0.6640195226868512,
+      "grad_norm": 588.8154296875,
+      "learning_rate": 3.312117256229449e-06,
+      "loss": 91.1146,
+      "step": 80270
+    },
+    {
+      "epoch": 0.6641022459362204,
+      "grad_norm": 722.9697875976562,
+      "learning_rate": 3.310789747579586e-06,
+      "loss": 84.0725,
+      "step": 80280
+    },
+    {
+      "epoch": 0.6641849691855896,
+      "grad_norm": 645.8081665039062,
+      "learning_rate": 3.30946237333243e-06,
+      "loss": 92.8005,
+      "step": 80290
+    },
+    {
+      "epoch": 0.6642676924349589,
+      "grad_norm": 1031.3046875,
+      "learning_rate": 3.308135133593595e-06,
+      "loss": 108.4536,
+      "step": 80300
+    },
+    {
+      "epoch": 0.6643504156843281,
+      "grad_norm": 596.3705444335938,
+      "learning_rate": 3.3068080284686825e-06,
+      "loss": 104.7899,
+      "step": 80310
+    },
+    {
+      "epoch": 0.6644331389336973,
+      "grad_norm": 792.6683349609375,
+      "learning_rate": 3.3054810580632844e-06,
+      "loss": 100.9892,
+      "step": 80320
+    },
+    {
+      "epoch": 0.6645158621830666,
+      "grad_norm": 803.8338623046875,
+      "learning_rate": 3.304154222482982e-06,
+      "loss": 72.7554,
+      "step": 80330
+    },
+    {
+      "epoch": 0.6645985854324358,
+      "grad_norm": 1118.839111328125,
+      "learning_rate": 3.3028275218333438e-06,
+      "loss": 67.1966,
+      "step": 80340
+    },
+    {
+      "epoch": 0.664681308681805,
+      "grad_norm": 929.019287109375,
+      "learning_rate": 3.301500956219932e-06,
+      "loss": 107.5598,
+      "step": 80350
+    },
+    {
+      "epoch": 0.6647640319311743,
+      "grad_norm": 1357.971923828125,
+      "learning_rate": 3.3001745257482935e-06,
+      "loss": 130.7297,
+      "step": 80360
+    },
+    {
+      "epoch": 0.6648467551805435,
+      "grad_norm": 299.3155212402344,
+      "learning_rate": 3.2988482305239673e-06,
+      "loss": 80.1713,
+      "step": 80370
+    },
+    {
+      "epoch": 0.6649294784299127,
+      "grad_norm": 658.03125,
+      "learning_rate": 3.2975220706524813e-06,
+      "loss": 120.3528,
+      "step": 80380
+    },
+    {
+      "epoch": 0.665012201679282,
+      "grad_norm": 1064.147216796875,
+      "learning_rate": 3.2961960462393492e-06,
+      "loss": 102.2007,
+      "step": 80390
+    },
+    {
+      "epoch": 0.6650949249286512,
+      "grad_norm": 783.42431640625,
+      "learning_rate": 3.2948701573900786e-06,
+      "loss": 104.9912,
+      "step": 80400
+    },
+    {
+      "epoch": 0.6651776481780204,
+      "grad_norm": 729.3176879882812,
+      "learning_rate": 3.2935444042101646e-06,
+      "loss": 95.9438,
+      "step": 80410
+    },
+    {
+      "epoch": 0.6652603714273897,
+      "grad_norm": 819.0823974609375,
+      "learning_rate": 3.29221878680509e-06,
+      "loss": 118.4774,
+      "step": 80420
+    },
+    {
+      "epoch": 0.6653430946767589,
+      "grad_norm": 698.5296020507812,
+      "learning_rate": 3.2908933052803292e-06,
+      "loss": 79.8682,
+      "step": 80430
+    },
+    {
+      "epoch": 0.6654258179261281,
+      "grad_norm": 1220.9241943359375,
+      "learning_rate": 3.2895679597413433e-06,
+      "loss": 97.124,
+      "step": 80440
+    },
+    {
+      "epoch": 0.6655085411754974,
+      "grad_norm": 1185.635009765625,
+      "learning_rate": 3.2882427502935867e-06,
+      "loss": 80.9898,
+      "step": 80450
+    },
+    {
+      "epoch": 0.6655912644248666,
+      "grad_norm": 1041.6197509765625,
+      "learning_rate": 3.2869176770424976e-06,
+      "loss": 91.9751,
+      "step": 80460
+    },
+    {
+      "epoch": 0.6656739876742358,
+      "grad_norm": 909.6505737304688,
+      "learning_rate": 3.2855927400935085e-06,
+      "loss": 83.0227,
+      "step": 80470
+    },
+    {
+      "epoch": 0.6657567109236051,
+      "grad_norm": 978.8333740234375,
+      "learning_rate": 3.2842679395520363e-06,
+      "loss": 87.613,
+      "step": 80480
+    },
+    {
+      "epoch": 0.6658394341729743,
+      "grad_norm": 1578.8958740234375,
+      "learning_rate": 3.282943275523489e-06,
+      "loss": 86.7915,
+      "step": 80490
+    },
+    {
+      "epoch": 0.6659221574223435,
+      "grad_norm": 1016.648681640625,
+      "learning_rate": 3.2816187481132655e-06,
+      "loss": 89.614,
+      "step": 80500
+    },
+    {
+      "epoch": 0.6660048806717128,
+      "grad_norm": 795.9924926757812,
+      "learning_rate": 3.280294357426752e-06,
+      "loss": 92.8654,
+      "step": 80510
+    },
+    {
+      "epoch": 0.666087603921082,
+      "grad_norm": 486.5448913574219,
+      "learning_rate": 3.2789701035693242e-06,
+      "loss": 68.977,
+      "step": 80520
+    },
+    {
+      "epoch": 0.6661703271704512,
+      "grad_norm": 1071.9464111328125,
+      "learning_rate": 3.277645986646346e-06,
+      "loss": 97.0568,
+      "step": 80530
+    },
+    {
+      "epoch": 0.6662530504198205,
+      "grad_norm": 1014.2429809570312,
+      "learning_rate": 3.276322006763172e-06,
+      "loss": 111.9005,
+      "step": 80540
+    },
+    {
+      "epoch": 0.6663357736691897,
+      "grad_norm": 571.8171997070312,
+      "learning_rate": 3.274998164025148e-06,
+      "loss": 73.2075,
+      "step": 80550
+    },
+    {
+      "epoch": 0.6664184969185589,
+      "grad_norm": 1098.88720703125,
+      "learning_rate": 3.2736744585376016e-06,
+      "loss": 73.6912,
+      "step": 80560
+    },
+    {
+      "epoch": 0.6665012201679282,
+      "grad_norm": 1065.5684814453125,
+      "learning_rate": 3.2723508904058547e-06,
+      "loss": 86.9936,
+      "step": 80570
+    },
+    {
+      "epoch": 0.6665839434172974,
+      "grad_norm": 1688.9266357421875,
+      "learning_rate": 3.27102745973522e-06,
+      "loss": 95.265,
+      "step": 80580
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 1830.2410888671875,
+      "learning_rate": 3.269704166630995e-06,
+      "loss": 107.0393,
+      "step": 80590
+    },
+    {
+      "epoch": 0.6667493899160359,
+      "grad_norm": 662.8126831054688,
+      "learning_rate": 3.268381011198468e-06,
+      "loss": 81.1234,
+      "step": 80600
+    },
+    {
+      "epoch": 0.6668321131654051,
+      "grad_norm": 780.1619873046875,
+      "learning_rate": 3.2670579935429176e-06,
+      "loss": 72.5676,
+      "step": 80610
+    },
+    {
+      "epoch": 0.6669148364147743,
+      "grad_norm": 900.4564819335938,
+      "learning_rate": 3.265735113769609e-06,
+      "loss": 72.4489,
+      "step": 80620
+    },
+    {
+      "epoch": 0.6669975596641436,
+      "grad_norm": 737.3795166015625,
+      "learning_rate": 3.264412371983797e-06,
+      "loss": 104.431,
+      "step": 80630
+    },
+    {
+      "epoch": 0.6670802829135128,
+      "grad_norm": 705.3508911132812,
+      "learning_rate": 3.2630897682907312e-06,
+      "loss": 82.3283,
+      "step": 80640
+    },
+    {
+      "epoch": 0.667163006162882,
+      "grad_norm": 1338.135498046875,
+      "learning_rate": 3.261767302795639e-06,
+      "loss": 102.0957,
+      "step": 80650
+    },
+    {
+      "epoch": 0.6672457294122514,
+      "grad_norm": 1417.84375,
+      "learning_rate": 3.2604449756037447e-06,
+      "loss": 140.2835,
+      "step": 80660
+    },
+    {
+      "epoch": 0.6673284526616206,
+      "grad_norm": 899.5838012695312,
+      "learning_rate": 3.2591227868202592e-06,
+      "loss": 106.4462,
+      "step": 80670
+    },
+    {
+      "epoch": 0.6674111759109898,
+      "grad_norm": 1201.1510009765625,
+      "learning_rate": 3.257800736550385e-06,
+      "loss": 88.8766,
+      "step": 80680
+    },
+    {
+      "epoch": 0.6674938991603591,
+      "grad_norm": 1101.5433349609375,
+      "learning_rate": 3.2564788248993105e-06,
+      "loss": 91.0437,
+      "step": 80690
+    },
+    {
+      "epoch": 0.6675766224097283,
+      "grad_norm": 609.85400390625,
+      "learning_rate": 3.2551570519722155e-06,
+      "loss": 104.3016,
+      "step": 80700
+    },
+    {
+      "epoch": 0.6676593456590975,
+      "grad_norm": 1198.7747802734375,
+      "learning_rate": 3.2538354178742648e-06,
+      "loss": 102.0451,
+      "step": 80710
+    },
+    {
+      "epoch": 0.6677420689084668,
+      "grad_norm": 494.413818359375,
+      "learning_rate": 3.2525139227106163e-06,
+      "loss": 70.2176,
+      "step": 80720
+    },
+    {
+      "epoch": 0.667824792157836,
+      "grad_norm": 614.208251953125,
+      "learning_rate": 3.2511925665864164e-06,
+      "loss": 67.1571,
+      "step": 80730
+    },
+    {
+      "epoch": 0.6679075154072052,
+      "grad_norm": 730.7908935546875,
+      "learning_rate": 3.2498713496067963e-06,
+      "loss": 73.7568,
+      "step": 80740
+    },
+    {
+      "epoch": 0.6679902386565745,
+      "grad_norm": 1220.75341796875,
+      "learning_rate": 3.2485502718768814e-06,
+      "loss": 103.8804,
+      "step": 80750
+    },
+    {
+      "epoch": 0.6680729619059437,
+      "grad_norm": 909.8517456054688,
+      "learning_rate": 3.2472293335017836e-06,
+      "loss": 90.6027,
+      "step": 80760
+    },
+    {
+      "epoch": 0.6681556851553129,
+      "grad_norm": 1850.9075927734375,
+      "learning_rate": 3.245908534586602e-06,
+      "loss": 87.6312,
+      "step": 80770
+    },
+    {
+      "epoch": 0.6682384084046822,
+      "grad_norm": 984.0964965820312,
+      "learning_rate": 3.2445878752364298e-06,
+      "loss": 94.8259,
+      "step": 80780
+    },
+    {
+      "epoch": 0.6683211316540514,
+      "grad_norm": 605.5479736328125,
+      "learning_rate": 3.2432673555563433e-06,
+      "loss": 113.5487,
+      "step": 80790
+    },
+    {
+      "epoch": 0.6684038549034206,
+      "grad_norm": 800.7566528320312,
+      "learning_rate": 3.2419469756514116e-06,
+      "loss": 99.107,
+      "step": 80800
+    },
+    {
+      "epoch": 0.6684865781527899,
+      "grad_norm": 1075.5169677734375,
+      "learning_rate": 3.2406267356266918e-06,
+      "loss": 82.8865,
+      "step": 80810
+    },
+    {
+      "epoch": 0.6685693014021591,
+      "grad_norm": 1107.46240234375,
+      "learning_rate": 3.2393066355872264e-06,
+      "loss": 75.885,
+      "step": 80820
+    },
+    {
+      "epoch": 0.6686520246515283,
+      "grad_norm": 1359.361083984375,
+      "learning_rate": 3.237986675638052e-06,
+      "loss": 115.8543,
+      "step": 80830
+    },
+    {
+      "epoch": 0.6687347479008976,
+      "grad_norm": 1057.6058349609375,
+      "learning_rate": 3.236666855884192e-06,
+      "loss": 75.9444,
+      "step": 80840
+    },
+    {
+      "epoch": 0.6688174711502668,
+      "grad_norm": 1190.32421875,
+      "learning_rate": 3.2353471764306567e-06,
+      "loss": 97.1294,
+      "step": 80850
+    },
+    {
+      "epoch": 0.668900194399636,
+      "grad_norm": 431.67431640625,
+      "learning_rate": 3.234027637382447e-06,
+      "loss": 78.7077,
+      "step": 80860
+    },
+    {
+      "epoch": 0.6689829176490053,
+      "grad_norm": 1030.784912109375,
+      "learning_rate": 3.2327082388445545e-06,
+      "loss": 123.1514,
+      "step": 80870
+    },
+    {
+      "epoch": 0.6690656408983745,
+      "grad_norm": 712.6047973632812,
+      "learning_rate": 3.2313889809219568e-06,
+      "loss": 93.2329,
+      "step": 80880
+    },
+    {
+      "epoch": 0.6691483641477437,
+      "grad_norm": 793.4395141601562,
+      "learning_rate": 3.2300698637196217e-06,
+      "loss": 88.246,
+      "step": 80890
+    },
+    {
+      "epoch": 0.669231087397113,
+      "grad_norm": 594.5629272460938,
+      "learning_rate": 3.2287508873425043e-06,
+      "loss": 66.3094,
+      "step": 80900
+    },
+    {
+      "epoch": 0.6693138106464822,
+      "grad_norm": 589.426513671875,
+      "learning_rate": 3.22743205189555e-06,
+      "loss": 72.8444,
+      "step": 80910
+    },
+    {
+      "epoch": 0.6693965338958514,
+      "grad_norm": 545.673095703125,
+      "learning_rate": 3.2261133574836918e-06,
+      "loss": 89.7926,
+      "step": 80920
+    },
+    {
+      "epoch": 0.6694792571452207,
+      "grad_norm": 809.666748046875,
+      "learning_rate": 3.2247948042118525e-06,
+      "loss": 81.8926,
+      "step": 80930
+    },
+    {
+      "epoch": 0.6695619803945899,
+      "grad_norm": 1300.5181884765625,
+      "learning_rate": 3.223476392184944e-06,
+      "loss": 93.8708,
+      "step": 80940
+    },
+    {
+      "epoch": 0.6696447036439591,
+      "grad_norm": 767.7609252929688,
+      "learning_rate": 3.2221581215078656e-06,
+      "loss": 86.6903,
+      "step": 80950
+    },
+    {
+      "epoch": 0.6697274268933283,
+      "grad_norm": 568.2315063476562,
+      "learning_rate": 3.2208399922855055e-06,
+      "loss": 88.5472,
+      "step": 80960
+    },
+    {
+      "epoch": 0.6698101501426976,
+      "grad_norm": 971.0084228515625,
+      "learning_rate": 3.2195220046227425e-06,
+      "loss": 104.4163,
+      "step": 80970
+    },
+    {
+      "epoch": 0.6698928733920668,
+      "grad_norm": 955.1143188476562,
+      "learning_rate": 3.218204158624445e-06,
+      "loss": 71.1244,
+      "step": 80980
+    },
+    {
+      "epoch": 0.669975596641436,
+      "grad_norm": 953.3546142578125,
+      "learning_rate": 3.216886454395463e-06,
+      "loss": 89.6237,
+      "step": 80990
+    },
+    {
+      "epoch": 0.6700583198908053,
+      "grad_norm": 684.8885498046875,
+      "learning_rate": 3.2155688920406415e-06,
+      "loss": 99.5519,
+      "step": 81000
+    },
+    {
+      "epoch": 0.6701410431401745,
+      "grad_norm": 1077.44677734375,
+      "learning_rate": 3.2142514716648143e-06,
+      "loss": 82.6027,
+      "step": 81010
+    },
+    {
+      "epoch": 0.6702237663895437,
+      "grad_norm": 455.7202453613281,
+      "learning_rate": 3.212934193372803e-06,
+      "loss": 94.583,
+      "step": 81020
+    },
+    {
+      "epoch": 0.670306489638913,
+      "grad_norm": 445.9930419921875,
+      "learning_rate": 3.2116170572694156e-06,
+      "loss": 94.4956,
+      "step": 81030
+    },
+    {
+      "epoch": 0.6703892128882822,
+      "grad_norm": 668.4183349609375,
+      "learning_rate": 3.2103000634594518e-06,
+      "loss": 96.0746,
+      "step": 81040
+    },
+    {
+      "epoch": 0.6704719361376514,
+      "grad_norm": 1390.123291015625,
+      "learning_rate": 3.2089832120476983e-06,
+      "loss": 89.5325,
+      "step": 81050
+    },
+    {
+      "epoch": 0.6705546593870207,
+      "grad_norm": 847.5530395507812,
+      "learning_rate": 3.2076665031389294e-06,
+      "loss": 92.6629,
+      "step": 81060
+    },
+    {
+      "epoch": 0.67063738263639,
+      "grad_norm": 1006.8488159179688,
+      "learning_rate": 3.2063499368379146e-06,
+      "loss": 84.7229,
+      "step": 81070
+    },
+    {
+      "epoch": 0.6707201058857591,
+      "grad_norm": 802.8297119140625,
+      "learning_rate": 3.2050335132494014e-06,
+      "loss": 91.7173,
+      "step": 81080
+    },
+    {
+      "epoch": 0.6708028291351285,
+      "grad_norm": 806.7569580078125,
+      "learning_rate": 3.203717232478133e-06,
+      "loss": 95.0556,
+      "step": 81090
+    },
+    {
+      "epoch": 0.6708855523844977,
+      "grad_norm": 1519.580810546875,
+      "learning_rate": 3.2024010946288415e-06,
+      "loss": 93.1881,
+      "step": 81100
+    },
+    {
+      "epoch": 0.6709682756338669,
+      "grad_norm": 539.17919921875,
+      "learning_rate": 3.201085099806245e-06,
+      "loss": 98.1286,
+      "step": 81110
+    },
+    {
+      "epoch": 0.6710509988832362,
+      "grad_norm": 1128.193359375,
+      "learning_rate": 3.199769248115051e-06,
+      "loss": 89.5094,
+      "step": 81120
+    },
+    {
+      "epoch": 0.6711337221326054,
+      "grad_norm": 1057.2535400390625,
+      "learning_rate": 3.1984535396599565e-06,
+      "loss": 103.7013,
+      "step": 81130
+    },
+    {
+      "epoch": 0.6712164453819746,
+      "grad_norm": 699.2178955078125,
+      "learning_rate": 3.1971379745456452e-06,
+      "loss": 88.7541,
+      "step": 81140
+    },
+    {
+      "epoch": 0.6712991686313439,
+      "grad_norm": 515.211669921875,
+      "learning_rate": 3.1958225528767918e-06,
+      "loss": 85.0595,
+      "step": 81150
+    },
+    {
+      "epoch": 0.6713818918807131,
+      "grad_norm": 1344.10595703125,
+      "learning_rate": 3.1945072747580585e-06,
+      "loss": 89.7821,
+      "step": 81160
+    },
+    {
+      "epoch": 0.6714646151300823,
+      "grad_norm": 1213.0770263671875,
+      "learning_rate": 3.1931921402940946e-06,
+      "loss": 73.3116,
+      "step": 81170
+    },
+    {
+      "epoch": 0.6715473383794516,
+      "grad_norm": 654.7890625,
+      "learning_rate": 3.1918771495895395e-06,
+      "loss": 78.7428,
+      "step": 81180
+    },
+    {
+      "epoch": 0.6716300616288208,
+      "grad_norm": 1164.929443359375,
+      "learning_rate": 3.1905623027490205e-06,
+      "loss": 84.9769,
+      "step": 81190
+    },
+    {
+      "epoch": 0.67171278487819,
+      "grad_norm": 1309.23779296875,
+      "learning_rate": 3.1892475998771567e-06,
+      "loss": 81.4625,
+      "step": 81200
+    },
+    {
+      "epoch": 0.6717955081275593,
+      "grad_norm": 654.0279541015625,
+      "learning_rate": 3.1879330410785503e-06,
+      "loss": 71.7905,
+      "step": 81210
+    },
+    {
+      "epoch": 0.6718782313769285,
+      "grad_norm": 503.28863525390625,
+      "learning_rate": 3.186618626457796e-06,
+      "loss": 82.309,
+      "step": 81220
+    },
+    {
+      "epoch": 0.6719609546262977,
+      "grad_norm": 951.95654296875,
+      "learning_rate": 3.1853043561194748e-06,
+      "loss": 88.6264,
+      "step": 81230
+    },
+    {
+      "epoch": 0.672043677875667,
+      "grad_norm": 778.96484375,
+      "learning_rate": 3.183990230168159e-06,
+      "loss": 87.3521,
+      "step": 81240
+    },
+    {
+      "epoch": 0.6721264011250362,
+      "grad_norm": 1273.863525390625,
+      "learning_rate": 3.1826762487084053e-06,
+      "loss": 96.637,
+      "step": 81250
+    },
+    {
+      "epoch": 0.6722091243744054,
+      "grad_norm": 951.4760131835938,
+      "learning_rate": 3.1813624118447615e-06,
+      "loss": 77.3764,
+      "step": 81260
+    },
+    {
+      "epoch": 0.6722918476237747,
+      "grad_norm": 613.3677978515625,
+      "learning_rate": 3.180048719681765e-06,
+      "loss": 81.409,
+      "step": 81270
+    },
+    {
+      "epoch": 0.6723745708731439,
+      "grad_norm": 1063.3297119140625,
+      "learning_rate": 3.178735172323939e-06,
+      "loss": 68.0863,
+      "step": 81280
+    },
+    {
+      "epoch": 0.6724572941225131,
+      "grad_norm": 799.1066284179688,
+      "learning_rate": 3.177421769875796e-06,
+      "loss": 90.396,
+      "step": 81290
+    },
+    {
+      "epoch": 0.6725400173718824,
+      "grad_norm": 1305.4615478515625,
+      "learning_rate": 3.176108512441839e-06,
+      "loss": 110.8276,
+      "step": 81300
+    },
+    {
+      "epoch": 0.6726227406212516,
+      "grad_norm": 630.9846801757812,
+      "learning_rate": 3.174795400126557e-06,
+      "loss": 84.7067,
+      "step": 81310
+    },
+    {
+      "epoch": 0.6727054638706208,
+      "grad_norm": 2631.676025390625,
+      "learning_rate": 3.173482433034429e-06,
+      "loss": 118.5525,
+      "step": 81320
+    },
+    {
+      "epoch": 0.6727881871199901,
+      "grad_norm": 1203.627197265625,
+      "learning_rate": 3.1721696112699217e-06,
+      "loss": 102.9443,
+      "step": 81330
+    },
+    {
+      "epoch": 0.6728709103693593,
+      "grad_norm": 1666.17236328125,
+      "learning_rate": 3.1708569349374896e-06,
+      "loss": 100.6521,
+      "step": 81340
+    },
+    {
+      "epoch": 0.6729536336187285,
+      "grad_norm": 969.8807983398438,
+      "learning_rate": 3.1695444041415757e-06,
+      "loss": 88.7701,
+      "step": 81350
+    },
+    {
+      "epoch": 0.6730363568680978,
+      "grad_norm": 1491.0135498046875,
+      "learning_rate": 3.1682320189866133e-06,
+      "loss": 87.3877,
+      "step": 81360
+    },
+    {
+      "epoch": 0.673119080117467,
+      "grad_norm": 506.7419128417969,
+      "learning_rate": 3.1669197795770225e-06,
+      "loss": 107.3529,
+      "step": 81370
+    },
+    {
+      "epoch": 0.6732018033668362,
+      "grad_norm": 839.7028198242188,
+      "learning_rate": 3.165607686017212e-06,
+      "loss": 62.6055,
+      "step": 81380
+    },
+    {
+      "epoch": 0.6732845266162055,
+      "grad_norm": 372.7325134277344,
+      "learning_rate": 3.164295738411578e-06,
+      "loss": 79.6807,
+      "step": 81390
+    },
+    {
+      "epoch": 0.6733672498655747,
+      "grad_norm": 940.877685546875,
+      "learning_rate": 3.1629839368645087e-06,
+      "loss": 109.5253,
+      "step": 81400
+    },
+    {
+      "epoch": 0.6734499731149439,
+      "grad_norm": 571.5814208984375,
+      "learning_rate": 3.161672281480379e-06,
+      "loss": 101.1825,
+      "step": 81410
+    },
+    {
+      "epoch": 0.6735326963643132,
+      "grad_norm": 1666.25634765625,
+      "learning_rate": 3.1603607723635455e-06,
+      "loss": 92.3682,
+      "step": 81420
+    },
+    {
+      "epoch": 0.6736154196136824,
+      "grad_norm": 394.33026123046875,
+      "learning_rate": 3.1590494096183643e-06,
+      "loss": 68.1128,
+      "step": 81430
+    },
+    {
+      "epoch": 0.6736981428630516,
+      "grad_norm": 943.1664428710938,
+      "learning_rate": 3.1577381933491718e-06,
+      "loss": 92.4026,
+      "step": 81440
+    },
+    {
+      "epoch": 0.6737808661124209,
+      "grad_norm": 479.8966064453125,
+      "learning_rate": 3.156427123660297e-06,
+      "loss": 101.9956,
+      "step": 81450
+    },
+    {
+      "epoch": 0.6738635893617901,
+      "grad_norm": 1042.3021240234375,
+      "learning_rate": 3.1551162006560554e-06,
+      "loss": 92.2319,
+      "step": 81460
+    },
+    {
+      "epoch": 0.6739463126111593,
+      "grad_norm": 640.0420532226562,
+      "learning_rate": 3.15380542444075e-06,
+      "loss": 115.9655,
+      "step": 81470
+    },
+    {
+      "epoch": 0.6740290358605286,
+      "grad_norm": 636.8284912109375,
+      "learning_rate": 3.1524947951186746e-06,
+      "loss": 80.7586,
+      "step": 81480
+    },
+    {
+      "epoch": 0.6741117591098978,
+      "grad_norm": 886.6581420898438,
+      "learning_rate": 3.1511843127941085e-06,
+      "loss": 116.3692,
+      "step": 81490
+    },
+    {
+      "epoch": 0.674194482359267,
+      "grad_norm": 881.8652954101562,
+      "learning_rate": 3.149873977571324e-06,
+      "loss": 81.8739,
+      "step": 81500
+    },
+    {
+      "epoch": 0.6742772056086364,
+      "grad_norm": 663.4722900390625,
+      "learning_rate": 3.148563789554575e-06,
+      "loss": 74.8076,
+      "step": 81510
+    },
+    {
+      "epoch": 0.6743599288580056,
+      "grad_norm": 1380.9359130859375,
+      "learning_rate": 3.147253748848107e-06,
+      "loss": 138.0078,
+      "step": 81520
+    },
+    {
+      "epoch": 0.6744426521073748,
+      "grad_norm": 402.91241455078125,
+      "learning_rate": 3.1459438555561565e-06,
+      "loss": 111.0916,
+      "step": 81530
+    },
+    {
+      "epoch": 0.6745253753567441,
+      "grad_norm": 256.97442626953125,
+      "learning_rate": 3.1446341097829446e-06,
+      "loss": 81.1193,
+      "step": 81540
+    },
+    {
+      "epoch": 0.6746080986061133,
+      "grad_norm": 847.0762329101562,
+      "learning_rate": 3.1433245116326812e-06,
+      "loss": 80.5571,
+      "step": 81550
+    },
+    {
+      "epoch": 0.6746908218554825,
+      "grad_norm": 387.09881591796875,
+      "learning_rate": 3.1420150612095653e-06,
+      "loss": 85.5534,
+      "step": 81560
+    },
+    {
+      "epoch": 0.6747735451048518,
+      "grad_norm": 1107.9180908203125,
+      "learning_rate": 3.140705758617784e-06,
+      "loss": 92.0838,
+      "step": 81570
+    },
+    {
+      "epoch": 0.674856268354221,
+      "grad_norm": 940.799560546875,
+      "learning_rate": 3.139396603961512e-06,
+      "loss": 95.1766,
+      "step": 81580
+    },
+    {
+      "epoch": 0.6749389916035902,
+      "grad_norm": 753.7284545898438,
+      "learning_rate": 3.1380875973449155e-06,
+      "loss": 79.7191,
+      "step": 81590
+    },
+    {
+      "epoch": 0.6750217148529595,
+      "grad_norm": 1136.5980224609375,
+      "learning_rate": 3.1367787388721427e-06,
+      "loss": 69.0357,
+      "step": 81600
+    },
+    {
+      "epoch": 0.6751044381023287,
+      "grad_norm": 724.4802856445312,
+      "learning_rate": 3.135470028647334e-06,
+      "loss": 87.2878,
+      "step": 81610
+    },
+    {
+      "epoch": 0.6751871613516979,
+      "grad_norm": 742.6774291992188,
+      "learning_rate": 3.134161466774617e-06,
+      "loss": 72.0161,
+      "step": 81620
+    },
+    {
+      "epoch": 0.6752698846010672,
+      "grad_norm": 855.7781372070312,
+      "learning_rate": 3.1328530533581102e-06,
+      "loss": 76.9828,
+      "step": 81630
+    },
+    {
+      "epoch": 0.6753526078504364,
+      "grad_norm": 848.9146118164062,
+      "learning_rate": 3.131544788501917e-06,
+      "loss": 57.4658,
+      "step": 81640
+    },
+    {
+      "epoch": 0.6754353310998056,
+      "grad_norm": 800.960205078125,
+      "learning_rate": 3.1302366723101294e-06,
+      "loss": 94.8087,
+      "step": 81650
+    },
+    {
+      "epoch": 0.6755180543491749,
+      "grad_norm": 1037.375,
+      "learning_rate": 3.12892870488683e-06,
+      "loss": 102.3649,
+      "step": 81660
+    },
+    {
+      "epoch": 0.6756007775985441,
+      "grad_norm": 530.1427001953125,
+      "learning_rate": 3.1276208863360862e-06,
+      "loss": 79.7667,
+      "step": 81670
+    },
+    {
+      "epoch": 0.6756835008479133,
+      "grad_norm": 1380.489501953125,
+      "learning_rate": 3.126313216761955e-06,
+      "loss": 76.0104,
+      "step": 81680
+    },
+    {
+      "epoch": 0.6757662240972825,
+      "grad_norm": 503.63250732421875,
+      "learning_rate": 3.125005696268482e-06,
+      "loss": 117.2083,
+      "step": 81690
+    },
+    {
+      "epoch": 0.6758489473466518,
+      "grad_norm": 867.2805786132812,
+      "learning_rate": 3.1236983249597007e-06,
+      "loss": 107.8847,
+      "step": 81700
+    },
+    {
+      "epoch": 0.675931670596021,
+      "grad_norm": 753.2324829101562,
+      "learning_rate": 3.1223911029396324e-06,
+      "loss": 78.1021,
+      "step": 81710
+    },
+    {
+      "epoch": 0.6760143938453902,
+      "grad_norm": 1229.90673828125,
+      "learning_rate": 3.121084030312286e-06,
+      "loss": 83.4004,
+      "step": 81720
+    },
+    {
+      "epoch": 0.6760971170947595,
+      "grad_norm": 641.0833740234375,
+      "learning_rate": 3.1197771071816617e-06,
+      "loss": 97.8247,
+      "step": 81730
+    },
+    {
+      "epoch": 0.6761798403441287,
+      "grad_norm": 761.6908569335938,
+      "learning_rate": 3.118470333651744e-06,
+      "loss": 86.3788,
+      "step": 81740
+    },
+    {
+      "epoch": 0.6762625635934979,
+      "grad_norm": 763.9179077148438,
+      "learning_rate": 3.1171637098265063e-06,
+      "loss": 107.4264,
+      "step": 81750
+    },
+    {
+      "epoch": 0.6763452868428672,
+      "grad_norm": 1150.12890625,
+      "learning_rate": 3.1158572358099127e-06,
+      "loss": 118.2742,
+      "step": 81760
+    },
+    {
+      "epoch": 0.6764280100922364,
+      "grad_norm": 1080.7073974609375,
+      "learning_rate": 3.11455091170591e-06,
+      "loss": 106.7814,
+      "step": 81770
+    },
+    {
+      "epoch": 0.6765107333416056,
+      "grad_norm": 642.6464233398438,
+      "learning_rate": 3.1132447376184383e-06,
+      "loss": 99.3273,
+      "step": 81780
+    },
+    {
+      "epoch": 0.6765934565909749,
+      "grad_norm": 808.6451416015625,
+      "learning_rate": 3.1119387136514246e-06,
+      "loss": 75.5224,
+      "step": 81790
+    },
+    {
+      "epoch": 0.6766761798403441,
+      "grad_norm": 1050.21533203125,
+      "learning_rate": 3.1106328399087814e-06,
+      "loss": 79.97,
+      "step": 81800
+    },
+    {
+      "epoch": 0.6767589030897133,
+      "grad_norm": 1466.090087890625,
+      "learning_rate": 3.1093271164944116e-06,
+      "loss": 89.5285,
+      "step": 81810
+    },
+    {
+      "epoch": 0.6768416263390826,
+      "grad_norm": 1522.2515869140625,
+      "learning_rate": 3.1080215435122072e-06,
+      "loss": 116.0093,
+      "step": 81820
+    },
+    {
+      "epoch": 0.6769243495884518,
+      "grad_norm": 1189.734375,
+      "learning_rate": 3.106716121066046e-06,
+      "loss": 100.7569,
+      "step": 81830
+    },
+    {
+      "epoch": 0.677007072837821,
+      "grad_norm": 1023.820556640625,
+      "learning_rate": 3.105410849259796e-06,
+      "loss": 89.4581,
+      "step": 81840
+    },
+    {
+      "epoch": 0.6770897960871903,
+      "grad_norm": 938.9853515625,
+      "learning_rate": 3.104105728197306e-06,
+      "loss": 85.5331,
+      "step": 81850
+    },
+    {
+      "epoch": 0.6771725193365595,
+      "grad_norm": 752.8026123046875,
+      "learning_rate": 3.1028007579824234e-06,
+      "loss": 93.6751,
+      "step": 81860
+    },
+    {
+      "epoch": 0.6772552425859287,
+      "grad_norm": 913.8092041015625,
+      "learning_rate": 3.1014959387189774e-06,
+      "loss": 68.4734,
+      "step": 81870
+    },
+    {
+      "epoch": 0.677337965835298,
+      "grad_norm": 593.3901977539062,
+      "learning_rate": 3.1001912705107874e-06,
+      "loss": 82.684,
+      "step": 81880
+    },
+    {
+      "epoch": 0.6774206890846672,
+      "grad_norm": 824.601318359375,
+      "learning_rate": 3.0988867534616586e-06,
+      "loss": 70.1524,
+      "step": 81890
+    },
+    {
+      "epoch": 0.6775034123340364,
+      "grad_norm": 495.2556457519531,
+      "learning_rate": 3.097582387675385e-06,
+      "loss": 101.8257,
+      "step": 81900
+    },
+    {
+      "epoch": 0.6775861355834057,
+      "grad_norm": 515.7427978515625,
+      "learning_rate": 3.09627817325575e-06,
+      "loss": 109.3507,
+      "step": 81910
+    },
+    {
+      "epoch": 0.677668858832775,
+      "grad_norm": 457.17315673828125,
+      "learning_rate": 3.0949741103065246e-06,
+      "loss": 102.0512,
+      "step": 81920
+    },
+    {
+      "epoch": 0.6777515820821441,
+      "grad_norm": 652.50537109375,
+      "learning_rate": 3.093670198931469e-06,
+      "loss": 70.5868,
+      "step": 81930
+    },
+    {
+      "epoch": 0.6778343053315135,
+      "grad_norm": 1835.66748046875,
+      "learning_rate": 3.0923664392343233e-06,
+      "loss": 87.3482,
+      "step": 81940
+    },
+    {
+      "epoch": 0.6779170285808827,
+      "grad_norm": 632.263671875,
+      "learning_rate": 3.091062831318825e-06,
+      "loss": 104.5366,
+      "step": 81950
+    },
+    {
+      "epoch": 0.6779997518302519,
+      "grad_norm": 766.2409057617188,
+      "learning_rate": 3.089759375288698e-06,
+      "loss": 81.9699,
+      "step": 81960
+    },
+    {
+      "epoch": 0.6780824750796212,
+      "grad_norm": 743.1511840820312,
+      "learning_rate": 3.0884560712476497e-06,
+      "loss": 110.5213,
+      "step": 81970
+    },
+    {
+      "epoch": 0.6781651983289904,
+      "grad_norm": 979.716064453125,
+      "learning_rate": 3.0871529192993794e-06,
+      "loss": 85.1489,
+      "step": 81980
+    },
+    {
+      "epoch": 0.6782479215783596,
+      "grad_norm": 713.2177124023438,
+      "learning_rate": 3.085849919547572e-06,
+      "loss": 92.1346,
+      "step": 81990
+    },
+    {
+      "epoch": 0.6783306448277289,
+      "grad_norm": 1064.33837890625,
+      "learning_rate": 3.0845470720959027e-06,
+      "loss": 92.5992,
+      "step": 82000
+    },
+    {
+      "epoch": 0.6784133680770981,
+      "grad_norm": 541.3108520507812,
+      "learning_rate": 3.08324437704803e-06,
+      "loss": 81.4435,
+      "step": 82010
+    },
+    {
+      "epoch": 0.6784960913264673,
+      "grad_norm": 867.9805908203125,
+      "learning_rate": 3.0819418345076095e-06,
+      "loss": 86.1453,
+      "step": 82020
+    },
+    {
+      "epoch": 0.6785788145758366,
+      "grad_norm": 1205.291259765625,
+      "learning_rate": 3.080639444578272e-06,
+      "loss": 77.0935,
+      "step": 82030
+    },
+    {
+      "epoch": 0.6786615378252058,
+      "grad_norm": 821.0764770507812,
+      "learning_rate": 3.0793372073636455e-06,
+      "loss": 76.2802,
+      "step": 82040
+    },
+    {
+      "epoch": 0.678744261074575,
+      "grad_norm": 1098.4530029296875,
+      "learning_rate": 3.0780351229673423e-06,
+      "loss": 99.049,
+      "step": 82050
+    },
+    {
+      "epoch": 0.6788269843239443,
+      "grad_norm": 1023.1441650390625,
+      "learning_rate": 3.0767331914929638e-06,
+      "loss": 86.1433,
+      "step": 82060
+    },
+    {
+      "epoch": 0.6789097075733135,
+      "grad_norm": 648.2615966796875,
+      "learning_rate": 3.075431413044099e-06,
+      "loss": 67.8278,
+      "step": 82070
+    },
+    {
+      "epoch": 0.6789924308226827,
+      "grad_norm": 800.4606323242188,
+      "learning_rate": 3.074129787724324e-06,
+      "loss": 87.548,
+      "step": 82080
+    },
+    {
+      "epoch": 0.679075154072052,
+      "grad_norm": 1863.6453857421875,
+      "learning_rate": 3.072828315637203e-06,
+      "loss": 109.8302,
+      "step": 82090
+    },
+    {
+      "epoch": 0.6791578773214212,
+      "grad_norm": 1212.3822021484375,
+      "learning_rate": 3.0715269968862898e-06,
+      "loss": 122.7618,
+      "step": 82100
+    },
+    {
+      "epoch": 0.6792406005707904,
+      "grad_norm": 599.944580078125,
+      "learning_rate": 3.0702258315751223e-06,
+      "loss": 69.2954,
+      "step": 82110
+    },
+    {
+      "epoch": 0.6793233238201597,
+      "grad_norm": 1033.164794921875,
+      "learning_rate": 3.0689248198072282e-06,
+      "loss": 95.2678,
+      "step": 82120
+    },
+    {
+      "epoch": 0.6794060470695289,
+      "grad_norm": 803.4859008789062,
+      "learning_rate": 3.0676239616861234e-06,
+      "loss": 93.7857,
+      "step": 82130
+    },
+    {
+      "epoch": 0.6794887703188981,
+      "grad_norm": 790.937744140625,
+      "learning_rate": 3.066323257315311e-06,
+      "loss": 79.9522,
+      "step": 82140
+    },
+    {
+      "epoch": 0.6795714935682674,
+      "grad_norm": 756.3161010742188,
+      "learning_rate": 3.065022706798284e-06,
+      "loss": 62.7251,
+      "step": 82150
+    },
+    {
+      "epoch": 0.6796542168176366,
+      "grad_norm": 976.355224609375,
+      "learning_rate": 3.06372231023852e-06,
+      "loss": 115.6686,
+      "step": 82160
+    },
+    {
+      "epoch": 0.6797369400670058,
+      "grad_norm": 841.8123168945312,
+      "learning_rate": 3.0624220677394854e-06,
+      "loss": 87.1789,
+      "step": 82170
+    },
+    {
+      "epoch": 0.6798196633163751,
+      "grad_norm": 945.3234252929688,
+      "learning_rate": 3.0611219794046344e-06,
+      "loss": 93.9989,
+      "step": 82180
+    },
+    {
+      "epoch": 0.6799023865657443,
+      "grad_norm": 1082.9246826171875,
+      "learning_rate": 3.05982204533741e-06,
+      "loss": 66.153,
+      "step": 82190
+    },
+    {
+      "epoch": 0.6799851098151135,
+      "grad_norm": 707.8014526367188,
+      "learning_rate": 3.0585222656412406e-06,
+      "loss": 114.1069,
+      "step": 82200
+    },
+    {
+      "epoch": 0.6800678330644828,
+      "grad_norm": 942.4901123046875,
+      "learning_rate": 3.0572226404195436e-06,
+      "loss": 63.9763,
+      "step": 82210
+    },
+    {
+      "epoch": 0.680150556313852,
+      "grad_norm": 668.8705444335938,
+      "learning_rate": 3.055923169775726e-06,
+      "loss": 71.8056,
+      "step": 82220
+    },
+    {
+      "epoch": 0.6802332795632212,
+      "grad_norm": 1010.2730712890625,
+      "learning_rate": 3.054623853813179e-06,
+      "loss": 116.1978,
+      "step": 82230
+    },
+    {
+      "epoch": 0.6803160028125905,
+      "grad_norm": 636.508544921875,
+      "learning_rate": 3.0533246926352834e-06,
+      "loss": 88.7034,
+      "step": 82240
+    },
+    {
+      "epoch": 0.6803987260619597,
+      "grad_norm": 1168.9595947265625,
+      "learning_rate": 3.0520256863454077e-06,
+      "loss": 94.9172,
+      "step": 82250
+    },
+    {
+      "epoch": 0.6804814493113289,
+      "grad_norm": 727.0571899414062,
+      "learning_rate": 3.05072683504691e-06,
+      "loss": 85.0885,
+      "step": 82260
+    },
+    {
+      "epoch": 0.6805641725606982,
+      "grad_norm": 1059.238037109375,
+      "learning_rate": 3.049428138843133e-06,
+      "loss": 87.404,
+      "step": 82270
+    },
+    {
+      "epoch": 0.6806468958100674,
+      "grad_norm": 720.4179077148438,
+      "learning_rate": 3.0481295978374037e-06,
+      "loss": 84.2605,
+      "step": 82280
+    },
+    {
+      "epoch": 0.6807296190594366,
+      "grad_norm": 1156.1129150390625,
+      "learning_rate": 3.0468312121330464e-06,
+      "loss": 99.4175,
+      "step": 82290
+    },
+    {
+      "epoch": 0.6808123423088059,
+      "grad_norm": 987.896240234375,
+      "learning_rate": 3.0455329818333652e-06,
+      "loss": 66.1081,
+      "step": 82300
+    },
+    {
+      "epoch": 0.6808950655581751,
+      "grad_norm": 878.2027587890625,
+      "learning_rate": 3.044234907041655e-06,
+      "loss": 98.8819,
+      "step": 82310
+    },
+    {
+      "epoch": 0.6809777888075443,
+      "grad_norm": 1048.861083984375,
+      "learning_rate": 3.0429369878611968e-06,
+      "loss": 67.5868,
+      "step": 82320
+    },
+    {
+      "epoch": 0.6810605120569136,
+      "grad_norm": 1312.079833984375,
+      "learning_rate": 3.041639224395262e-06,
+      "loss": 120.5818,
+      "step": 82330
+    },
+    {
+      "epoch": 0.6811432353062828,
+      "grad_norm": 1217.2681884765625,
+      "learning_rate": 3.0403416167471044e-06,
+      "loss": 84.3008,
+      "step": 82340
+    },
+    {
+      "epoch": 0.681225958555652,
+      "grad_norm": 1105.7882080078125,
+      "learning_rate": 3.0390441650199727e-06,
+      "loss": 80.4502,
+      "step": 82350
+    },
+    {
+      "epoch": 0.6813086818050214,
+      "grad_norm": 722.1676635742188,
+      "learning_rate": 3.0377468693170985e-06,
+      "loss": 109.3831,
+      "step": 82360
+    },
+    {
+      "epoch": 0.6813914050543906,
+      "grad_norm": 734.0096435546875,
+      "learning_rate": 3.0364497297416973e-06,
+      "loss": 80.3656,
+      "step": 82370
+    },
+    {
+      "epoch": 0.6814741283037598,
+      "grad_norm": 392.9991760253906,
+      "learning_rate": 3.035152746396981e-06,
+      "loss": 93.2643,
+      "step": 82380
+    },
+    {
+      "epoch": 0.6815568515531291,
+      "grad_norm": 724.57177734375,
+      "learning_rate": 3.0338559193861434e-06,
+      "loss": 106.8783,
+      "step": 82390
+    },
+    {
+      "epoch": 0.6816395748024983,
+      "grad_norm": 511.555419921875,
+      "learning_rate": 3.032559248812367e-06,
+      "loss": 118.9703,
+      "step": 82400
+    },
+    {
+      "epoch": 0.6817222980518675,
+      "grad_norm": 378.6773376464844,
+      "learning_rate": 3.0312627347788208e-06,
+      "loss": 82.5903,
+      "step": 82410
+    },
+    {
+      "epoch": 0.6818050213012367,
+      "grad_norm": 438.7552490234375,
+      "learning_rate": 3.0299663773886646e-06,
+      "loss": 73.4946,
+      "step": 82420
+    },
+    {
+      "epoch": 0.681887744550606,
+      "grad_norm": 1227.794921875,
+      "learning_rate": 3.0286701767450423e-06,
+      "loss": 114.8965,
+      "step": 82430
+    },
+    {
+      "epoch": 0.6819704677999752,
+      "grad_norm": 591.5338745117188,
+      "learning_rate": 3.0273741329510852e-06,
+      "loss": 116.4364,
+      "step": 82440
+    },
+    {
+      "epoch": 0.6820531910493444,
+      "grad_norm": 852.6351928710938,
+      "learning_rate": 3.0260782461099192e-06,
+      "loss": 71.9436,
+      "step": 82450
+    },
+    {
+      "epoch": 0.6821359142987137,
+      "grad_norm": 758.0744018554688,
+      "learning_rate": 3.024782516324645e-06,
+      "loss": 82.7659,
+      "step": 82460
+    },
+    {
+      "epoch": 0.6822186375480829,
+      "grad_norm": 1753.00048828125,
+      "learning_rate": 3.0234869436983606e-06,
+      "loss": 109.4363,
+      "step": 82470
+    },
+    {
+      "epoch": 0.6823013607974521,
+      "grad_norm": 1189.2587890625,
+      "learning_rate": 3.02219152833415e-06,
+      "loss": 96.6973,
+      "step": 82480
+    },
+    {
+      "epoch": 0.6823840840468214,
+      "grad_norm": 894.1843872070312,
+      "learning_rate": 3.0208962703350832e-06,
+      "loss": 86.2049,
+      "step": 82490
+    },
+    {
+      "epoch": 0.6824668072961906,
+      "grad_norm": 675.1846313476562,
+      "learning_rate": 3.019601169804216e-06,
+      "loss": 73.5452,
+      "step": 82500
+    },
+    {
+      "epoch": 0.6825495305455598,
+      "grad_norm": 733.6863403320312,
+      "learning_rate": 3.0183062268445964e-06,
+      "loss": 96.7175,
+      "step": 82510
+    },
+    {
+      "epoch": 0.6826322537949291,
+      "grad_norm": 953.9824829101562,
+      "learning_rate": 3.0170114415592543e-06,
+      "loss": 95.8179,
+      "step": 82520
+    },
+    {
+      "epoch": 0.6827149770442983,
+      "grad_norm": 776.6597290039062,
+      "learning_rate": 3.015716814051213e-06,
+      "loss": 75.2235,
+      "step": 82530
+    },
+    {
+      "epoch": 0.6827977002936675,
+      "grad_norm": 741.5625610351562,
+      "learning_rate": 3.0144223444234767e-06,
+      "loss": 90.4091,
+      "step": 82540
+    },
+    {
+      "epoch": 0.6828804235430368,
+      "grad_norm": 1144.602783203125,
+      "learning_rate": 3.0131280327790412e-06,
+      "loss": 74.8416,
+      "step": 82550
+    },
+    {
+      "epoch": 0.682963146792406,
+      "grad_norm": 1054.5020751953125,
+      "learning_rate": 3.0118338792208912e-06,
+      "loss": 93.5738,
+      "step": 82560
+    },
+    {
+      "epoch": 0.6830458700417752,
+      "grad_norm": 1366.202392578125,
+      "learning_rate": 3.010539883851993e-06,
+      "loss": 99.257,
+      "step": 82570
+    },
+    {
+      "epoch": 0.6831285932911445,
+      "grad_norm": 303.204345703125,
+      "learning_rate": 3.009246046775307e-06,
+      "loss": 79.5888,
+      "step": 82580
+    },
+    {
+      "epoch": 0.6832113165405137,
+      "grad_norm": 1691.6275634765625,
+      "learning_rate": 3.0079523680937766e-06,
+      "loss": 130.3326,
+      "step": 82590
+    },
+    {
+      "epoch": 0.6832940397898829,
+      "grad_norm": 1158.856689453125,
+      "learning_rate": 3.006658847910334e-06,
+      "loss": 113.3093,
+      "step": 82600
+    },
+    {
+      "epoch": 0.6833767630392522,
+      "grad_norm": 804.7373046875,
+      "learning_rate": 3.005365486327899e-06,
+      "loss": 85.9386,
+      "step": 82610
+    },
+    {
+      "epoch": 0.6834594862886214,
+      "grad_norm": 961.7406005859375,
+      "learning_rate": 3.004072283449379e-06,
+      "loss": 88.6196,
+      "step": 82620
+    },
+    {
+      "epoch": 0.6835422095379906,
+      "grad_norm": 764.2142333984375,
+      "learning_rate": 3.0027792393776666e-06,
+      "loss": 89.8795,
+      "step": 82630
+    },
+    {
+      "epoch": 0.6836249327873599,
+      "grad_norm": 751.17822265625,
+      "learning_rate": 3.001486354215644e-06,
+      "loss": 70.7893,
+      "step": 82640
+    },
+    {
+      "epoch": 0.6837076560367291,
+      "grad_norm": 1221.4046630859375,
+      "learning_rate": 3.0001936280661794e-06,
+      "loss": 80.4374,
+      "step": 82650
+    },
+    {
+      "epoch": 0.6837903792860983,
+      "grad_norm": 748.9243774414062,
+      "learning_rate": 2.998901061032131e-06,
+      "loss": 115.65,
+      "step": 82660
+    },
+    {
+      "epoch": 0.6838731025354676,
+      "grad_norm": 847.3457641601562,
+      "learning_rate": 2.9976086532163397e-06,
+      "loss": 88.8179,
+      "step": 82670
+    },
+    {
+      "epoch": 0.6839558257848368,
+      "grad_norm": 363.82830810546875,
+      "learning_rate": 2.9963164047216397e-06,
+      "loss": 113.4569,
+      "step": 82680
+    },
+    {
+      "epoch": 0.684038549034206,
+      "grad_norm": 855.93212890625,
+      "learning_rate": 2.9950243156508473e-06,
+      "loss": 99.6168,
+      "step": 82690
+    },
+    {
+      "epoch": 0.6841212722835753,
+      "grad_norm": 877.383544921875,
+      "learning_rate": 2.9937323861067695e-06,
+      "loss": 107.2685,
+      "step": 82700
+    },
+    {
+      "epoch": 0.6842039955329445,
+      "grad_norm": 737.8973388671875,
+      "learning_rate": 2.992440616192197e-06,
+      "loss": 73.6583,
+      "step": 82710
+    },
+    {
+      "epoch": 0.6842867187823137,
+      "grad_norm": 878.8938598632812,
+      "learning_rate": 2.9911490060099117e-06,
+      "loss": 110.4591,
+      "step": 82720
+    },
+    {
+      "epoch": 0.684369442031683,
+      "grad_norm": 1008.2698364257812,
+      "learning_rate": 2.9898575556626807e-06,
+      "loss": 84.5192,
+      "step": 82730
+    },
+    {
+      "epoch": 0.6844521652810522,
+      "grad_norm": 942.9556884765625,
+      "learning_rate": 2.9885662652532586e-06,
+      "loss": 101.3532,
+      "step": 82740
+    },
+    {
+      "epoch": 0.6845348885304214,
+      "grad_norm": 903.5498657226562,
+      "learning_rate": 2.9872751348843875e-06,
+      "loss": 87.6757,
+      "step": 82750
+    },
+    {
+      "epoch": 0.6846176117797907,
+      "grad_norm": 1304.6390380859375,
+      "learning_rate": 2.985984164658796e-06,
+      "loss": 78.2163,
+      "step": 82760
+    },
+    {
+      "epoch": 0.68470033502916,
+      "grad_norm": 691.3413696289062,
+      "learning_rate": 2.9846933546792012e-06,
+      "loss": 79.0234,
+      "step": 82770
+    },
+    {
+      "epoch": 0.6847830582785291,
+      "grad_norm": 488.5516662597656,
+      "learning_rate": 2.9834027050483085e-06,
+      "loss": 106.5184,
+      "step": 82780
+    },
+    {
+      "epoch": 0.6848657815278985,
+      "grad_norm": 1274.3031005859375,
+      "learning_rate": 2.9821122158688086e-06,
+      "loss": 75.8053,
+      "step": 82790
+    },
+    {
+      "epoch": 0.6849485047772677,
+      "grad_norm": 719.878173828125,
+      "learning_rate": 2.980821887243377e-06,
+      "loss": 88.5654,
+      "step": 82800
+    },
+    {
+      "epoch": 0.6850312280266369,
+      "grad_norm": 974.0200805664062,
+      "learning_rate": 2.979531719274681e-06,
+      "loss": 79.3588,
+      "step": 82810
+    },
+    {
+      "epoch": 0.6851139512760062,
+      "grad_norm": 875.3156127929688,
+      "learning_rate": 2.978241712065374e-06,
+      "loss": 97.1635,
+      "step": 82820
+    },
+    {
+      "epoch": 0.6851966745253754,
+      "grad_norm": 1160.015869140625,
+      "learning_rate": 2.9769518657180953e-06,
+      "loss": 93.5529,
+      "step": 82830
+    },
+    {
+      "epoch": 0.6852793977747446,
+      "grad_norm": 419.8360290527344,
+      "learning_rate": 2.9756621803354722e-06,
+      "loss": 74.9933,
+      "step": 82840
+    },
+    {
+      "epoch": 0.6853621210241139,
+      "grad_norm": 822.18505859375,
+      "learning_rate": 2.9743726560201185e-06,
+      "loss": 147.5206,
+      "step": 82850
+    },
+    {
+      "epoch": 0.6854448442734831,
+      "grad_norm": 1076.56787109375,
+      "learning_rate": 2.9730832928746355e-06,
+      "loss": 89.3311,
+      "step": 82860
+    },
+    {
+      "epoch": 0.6855275675228523,
+      "grad_norm": 563.077392578125,
+      "learning_rate": 2.9717940910016135e-06,
+      "loss": 94.1015,
+      "step": 82870
+    },
+    {
+      "epoch": 0.6856102907722216,
+      "grad_norm": 1070.950927734375,
+      "learning_rate": 2.9705050505036294e-06,
+      "loss": 75.7543,
+      "step": 82880
+    },
+    {
+      "epoch": 0.6856930140215908,
+      "grad_norm": 563.2598876953125,
+      "learning_rate": 2.9692161714832422e-06,
+      "loss": 81.8349,
+      "step": 82890
+    },
+    {
+      "epoch": 0.68577573727096,
+      "grad_norm": 779.0994262695312,
+      "learning_rate": 2.9679274540430037e-06,
+      "loss": 77.1619,
+      "step": 82900
+    },
+    {
+      "epoch": 0.6858584605203293,
+      "grad_norm": 1350.3271484375,
+      "learning_rate": 2.966638898285452e-06,
+      "loss": 76.3261,
+      "step": 82910
+    },
+    {
+      "epoch": 0.6859411837696985,
+      "grad_norm": 790.4927368164062,
+      "learning_rate": 2.9653505043131125e-06,
+      "loss": 89.7235,
+      "step": 82920
+    },
+    {
+      "epoch": 0.6860239070190677,
+      "grad_norm": 549.5680541992188,
+      "learning_rate": 2.9640622722284944e-06,
+      "loss": 84.1619,
+      "step": 82930
+    },
+    {
+      "epoch": 0.686106630268437,
+      "grad_norm": 602.5223388671875,
+      "learning_rate": 2.962774202134098e-06,
+      "loss": 115.692,
+      "step": 82940
+    },
+    {
+      "epoch": 0.6861893535178062,
+      "grad_norm": 574.7587890625,
+      "learning_rate": 2.961486294132409e-06,
+      "loss": 103.2164,
+      "step": 82950
+    },
+    {
+      "epoch": 0.6862720767671754,
+      "grad_norm": 1748.5323486328125,
+      "learning_rate": 2.960198548325901e-06,
+      "loss": 89.1274,
+      "step": 82960
+    },
+    {
+      "epoch": 0.6863548000165447,
+      "grad_norm": 626.42041015625,
+      "learning_rate": 2.958910964817032e-06,
+      "loss": 85.8427,
+      "step": 82970
+    },
+    {
+      "epoch": 0.6864375232659139,
+      "grad_norm": 692.4397583007812,
+      "learning_rate": 2.9576235437082502e-06,
+      "loss": 100.6521,
+      "step": 82980
+    },
+    {
+      "epoch": 0.6865202465152831,
+      "grad_norm": 587.5997314453125,
+      "learning_rate": 2.9563362851019893e-06,
+      "loss": 65.4427,
+      "step": 82990
+    },
+    {
+      "epoch": 0.6866029697646524,
+      "grad_norm": 582.0680541992188,
+      "learning_rate": 2.9550491891006704e-06,
+      "loss": 82.0156,
+      "step": 83000
+    },
+    {
+      "epoch": 0.6866856930140216,
+      "grad_norm": 1114.3927001953125,
+      "learning_rate": 2.9537622558067036e-06,
+      "loss": 87.1522,
+      "step": 83010
+    },
+    {
+      "epoch": 0.6867684162633908,
+      "grad_norm": 487.21307373046875,
+      "learning_rate": 2.9524754853224837e-06,
+      "loss": 94.6792,
+      "step": 83020
+    },
+    {
+      "epoch": 0.6868511395127601,
+      "grad_norm": 713.6537475585938,
+      "learning_rate": 2.9511888777503916e-06,
+      "loss": 67.1539,
+      "step": 83030
+    },
+    {
+      "epoch": 0.6869338627621293,
+      "grad_norm": 840.3438110351562,
+      "learning_rate": 2.949902433192798e-06,
+      "loss": 84.9945,
+      "step": 83040
+    },
+    {
+      "epoch": 0.6870165860114985,
+      "grad_norm": 1528.6619873046875,
+      "learning_rate": 2.94861615175206e-06,
+      "loss": 77.7083,
+      "step": 83050
+    },
+    {
+      "epoch": 0.6870993092608678,
+      "grad_norm": 1220.947998046875,
+      "learning_rate": 2.9473300335305193e-06,
+      "loss": 89.6249,
+      "step": 83060
+    },
+    {
+      "epoch": 0.687182032510237,
+      "grad_norm": 588.5947875976562,
+      "learning_rate": 2.946044078630508e-06,
+      "loss": 85.9502,
+      "step": 83070
+    },
+    {
+      "epoch": 0.6872647557596062,
+      "grad_norm": 1005.1710815429688,
+      "learning_rate": 2.9447582871543423e-06,
+      "loss": 85.558,
+      "step": 83080
+    },
+    {
+      "epoch": 0.6873474790089755,
+      "grad_norm": 658.07177734375,
+      "learning_rate": 2.9434726592043263e-06,
+      "loss": 92.5939,
+      "step": 83090
+    },
+    {
+      "epoch": 0.6874302022583447,
+      "grad_norm": 996.118896484375,
+      "learning_rate": 2.942187194882754e-06,
+      "loss": 109.5573,
+      "step": 83100
+    },
+    {
+      "epoch": 0.6875129255077139,
+      "grad_norm": 852.5332641601562,
+      "learning_rate": 2.940901894291902e-06,
+      "loss": 113.4957,
+      "step": 83110
+    },
+    {
+      "epoch": 0.6875956487570832,
+      "grad_norm": 6915.46484375,
+      "learning_rate": 2.939616757534037e-06,
+      "loss": 119.719,
+      "step": 83120
+    },
+    {
+      "epoch": 0.6876783720064524,
+      "grad_norm": 852.2628173828125,
+      "learning_rate": 2.938331784711411e-06,
+      "loss": 89.6527,
+      "step": 83130
+    },
+    {
+      "epoch": 0.6877610952558216,
+      "grad_norm": 790.9977416992188,
+      "learning_rate": 2.937046975926262e-06,
+      "loss": 83.5397,
+      "step": 83140
+    },
+    {
+      "epoch": 0.6878438185051908,
+      "grad_norm": 721.8024291992188,
+      "learning_rate": 2.9357623312808183e-06,
+      "loss": 67.4693,
+      "step": 83150
+    },
+    {
+      "epoch": 0.6879265417545601,
+      "grad_norm": 525.3212280273438,
+      "learning_rate": 2.934477850877292e-06,
+      "loss": 100.1755,
+      "step": 83160
+    },
+    {
+      "epoch": 0.6880092650039293,
+      "grad_norm": 536.9306640625,
+      "learning_rate": 2.9331935348178838e-06,
+      "loss": 91.7662,
+      "step": 83170
+    },
+    {
+      "epoch": 0.6880919882532985,
+      "grad_norm": 924.3001098632812,
+      "learning_rate": 2.931909383204781e-06,
+      "loss": 77.1973,
+      "step": 83180
+    },
+    {
+      "epoch": 0.6881747115026678,
+      "grad_norm": 688.9435424804688,
+      "learning_rate": 2.9306253961401553e-06,
+      "loss": 79.3377,
+      "step": 83190
+    },
+    {
+      "epoch": 0.688257434752037,
+      "grad_norm": 1393.37451171875,
+      "learning_rate": 2.929341573726171e-06,
+      "loss": 111.1658,
+      "step": 83200
+    },
+    {
+      "epoch": 0.6883401580014062,
+      "grad_norm": 524.9993896484375,
+      "learning_rate": 2.928057916064975e-06,
+      "loss": 68.5311,
+      "step": 83210
+    },
+    {
+      "epoch": 0.6884228812507756,
+      "grad_norm": 1237.0621337890625,
+      "learning_rate": 2.9267744232587035e-06,
+      "loss": 102.3083,
+      "step": 83220
+    },
+    {
+      "epoch": 0.6885056045001448,
+      "grad_norm": 620.7877807617188,
+      "learning_rate": 2.925491095409473e-06,
+      "loss": 107.3997,
+      "step": 83230
+    },
+    {
+      "epoch": 0.688588327749514,
+      "grad_norm": 1383.7750244140625,
+      "learning_rate": 2.924207932619397e-06,
+      "loss": 91.8689,
+      "step": 83240
+    },
+    {
+      "epoch": 0.6886710509988833,
+      "grad_norm": 775.3110961914062,
+      "learning_rate": 2.9229249349905686e-06,
+      "loss": 87.8382,
+      "step": 83250
+    },
+    {
+      "epoch": 0.6887537742482525,
+      "grad_norm": 681.6326904296875,
+      "learning_rate": 2.9216421026250707e-06,
+      "loss": 70.0503,
+      "step": 83260
+    },
+    {
+      "epoch": 0.6888364974976217,
+      "grad_norm": 786.73046875,
+      "learning_rate": 2.9203594356249726e-06,
+      "loss": 117.5731,
+      "step": 83270
+    },
+    {
+      "epoch": 0.688919220746991,
+      "grad_norm": 575.7755737304688,
+      "learning_rate": 2.919076934092329e-06,
+      "loss": 64.0,
+      "step": 83280
+    },
+    {
+      "epoch": 0.6890019439963602,
+      "grad_norm": 677.3495483398438,
+      "learning_rate": 2.9177945981291843e-06,
+      "loss": 69.7188,
+      "step": 83290
+    },
+    {
+      "epoch": 0.6890846672457294,
+      "grad_norm": 704.7107543945312,
+      "learning_rate": 2.916512427837568e-06,
+      "loss": 112.342,
+      "step": 83300
+    },
+    {
+      "epoch": 0.6891673904950987,
+      "grad_norm": 861.6221923828125,
+      "learning_rate": 2.9152304233194974e-06,
+      "loss": 63.6381,
+      "step": 83310
+    },
+    {
+      "epoch": 0.6892501137444679,
+      "grad_norm": 722.7597045898438,
+      "learning_rate": 2.9139485846769723e-06,
+      "loss": 98.6105,
+      "step": 83320
+    },
+    {
+      "epoch": 0.6893328369938371,
+      "grad_norm": 586.8096313476562,
+      "learning_rate": 2.9126669120119846e-06,
+      "loss": 112.3304,
+      "step": 83330
+    },
+    {
+      "epoch": 0.6894155602432064,
+      "grad_norm": 1111.9354248046875,
+      "learning_rate": 2.9113854054265112e-06,
+      "loss": 104.5657,
+      "step": 83340
+    },
+    {
+      "epoch": 0.6894982834925756,
+      "grad_norm": 1186.3795166015625,
+      "learning_rate": 2.9101040650225155e-06,
+      "loss": 112.6777,
+      "step": 83350
+    },
+    {
+      "epoch": 0.6895810067419448,
+      "grad_norm": 627.06298828125,
+      "learning_rate": 2.9088228909019455e-06,
+      "loss": 91.1352,
+      "step": 83360
+    },
+    {
+      "epoch": 0.6896637299913141,
+      "grad_norm": 530.3230590820312,
+      "learning_rate": 2.9075418831667436e-06,
+      "loss": 74.27,
+      "step": 83370
+    },
+    {
+      "epoch": 0.6897464532406833,
+      "grad_norm": 1148.4014892578125,
+      "learning_rate": 2.906261041918831e-06,
+      "loss": 91.6355,
+      "step": 83380
+    },
+    {
+      "epoch": 0.6898291764900525,
+      "grad_norm": 896.2343139648438,
+      "learning_rate": 2.90498036726012e-06,
+      "loss": 101.3476,
+      "step": 83390
+    },
+    {
+      "epoch": 0.6899118997394218,
+      "grad_norm": 748.8731689453125,
+      "learning_rate": 2.903699859292505e-06,
+      "loss": 81.6133,
+      "step": 83400
+    },
+    {
+      "epoch": 0.689994622988791,
+      "grad_norm": 1056.1656494140625,
+      "learning_rate": 2.9024195181178704e-06,
+      "loss": 90.9416,
+      "step": 83410
+    },
+    {
+      "epoch": 0.6900773462381602,
+      "grad_norm": 1155.9085693359375,
+      "learning_rate": 2.9011393438380884e-06,
+      "loss": 80.1188,
+      "step": 83420
+    },
+    {
+      "epoch": 0.6901600694875295,
+      "grad_norm": 807.7344970703125,
+      "learning_rate": 2.8998593365550178e-06,
+      "loss": 121.4385,
+      "step": 83430
+    },
+    {
+      "epoch": 0.6902427927368987,
+      "grad_norm": 552.521484375,
+      "learning_rate": 2.8985794963704992e-06,
+      "loss": 92.7039,
+      "step": 83440
+    },
+    {
+      "epoch": 0.6903255159862679,
+      "grad_norm": 1697.0496826171875,
+      "learning_rate": 2.8972998233863657e-06,
+      "loss": 101.2329,
+      "step": 83450
+    },
+    {
+      "epoch": 0.6904082392356372,
+      "grad_norm": 824.8051147460938,
+      "learning_rate": 2.8960203177044364e-06,
+      "loss": 94.2846,
+      "step": 83460
+    },
+    {
+      "epoch": 0.6904909624850064,
+      "grad_norm": 1690.2552490234375,
+      "learning_rate": 2.8947409794265146e-06,
+      "loss": 85.7355,
+      "step": 83470
+    },
+    {
+      "epoch": 0.6905736857343756,
+      "grad_norm": 961.0975341796875,
+      "learning_rate": 2.893461808654393e-06,
+      "loss": 99.6032,
+      "step": 83480
+    },
+    {
+      "epoch": 0.6906564089837449,
+      "grad_norm": 622.6866455078125,
+      "learning_rate": 2.892182805489846e-06,
+      "loss": 78.2167,
+      "step": 83490
+    },
+    {
+      "epoch": 0.6907391322331141,
+      "grad_norm": 1131.5731201171875,
+      "learning_rate": 2.8909039700346385e-06,
+      "loss": 95.0546,
+      "step": 83500
+    },
+    {
+      "epoch": 0.6908218554824833,
+      "grad_norm": 324.06011962890625,
+      "learning_rate": 2.889625302390524e-06,
+      "loss": 72.3154,
+      "step": 83510
+    },
+    {
+      "epoch": 0.6909045787318526,
+      "grad_norm": 1070.0745849609375,
+      "learning_rate": 2.8883468026592382e-06,
+      "loss": 91.6812,
+      "step": 83520
+    },
+    {
+      "epoch": 0.6909873019812218,
+      "grad_norm": 876.22607421875,
+      "learning_rate": 2.8870684709425063e-06,
+      "loss": 86.1727,
+      "step": 83530
+    },
+    {
+      "epoch": 0.691070025230591,
+      "grad_norm": 1324.11376953125,
+      "learning_rate": 2.885790307342039e-06,
+      "loss": 97.0591,
+      "step": 83540
+    },
+    {
+      "epoch": 0.6911527484799603,
+      "grad_norm": 1690.0306396484375,
+      "learning_rate": 2.884512311959532e-06,
+      "loss": 101.5719,
+      "step": 83550
+    },
+    {
+      "epoch": 0.6912354717293295,
+      "grad_norm": 436.6270446777344,
+      "learning_rate": 2.8832344848966758e-06,
+      "loss": 64.3056,
+      "step": 83560
+    },
+    {
+      "epoch": 0.6913181949786987,
+      "grad_norm": 1113.216552734375,
+      "learning_rate": 2.8819568262551344e-06,
+      "loss": 134.5069,
+      "step": 83570
+    },
+    {
+      "epoch": 0.691400918228068,
+      "grad_norm": 676.7778930664062,
+      "learning_rate": 2.8806793361365686e-06,
+      "loss": 128.0765,
+      "step": 83580
+    },
+    {
+      "epoch": 0.6914836414774372,
+      "grad_norm": 1021.91259765625,
+      "learning_rate": 2.8794020146426217e-06,
+      "loss": 95.5648,
+      "step": 83590
+    },
+    {
+      "epoch": 0.6915663647268064,
+      "grad_norm": 963.874755859375,
+      "learning_rate": 2.8781248618749235e-06,
+      "loss": 85.2975,
+      "step": 83600
+    },
+    {
+      "epoch": 0.6916490879761757,
+      "grad_norm": 819.314453125,
+      "learning_rate": 2.8768478779350927e-06,
+      "loss": 76.2316,
+      "step": 83610
+    },
+    {
+      "epoch": 0.691731811225545,
+      "grad_norm": 1420.544921875,
+      "learning_rate": 2.875571062924732e-06,
+      "loss": 102.6076,
+      "step": 83620
+    },
+    {
+      "epoch": 0.6918145344749141,
+      "grad_norm": 1988.164794921875,
+      "learning_rate": 2.874294416945432e-06,
+      "loss": 109.0461,
+      "step": 83630
+    },
+    {
+      "epoch": 0.6918972577242835,
+      "grad_norm": 808.8279418945312,
+      "learning_rate": 2.8730179400987697e-06,
+      "loss": 102.7774,
+      "step": 83640
+    },
+    {
+      "epoch": 0.6919799809736527,
+      "grad_norm": 1196.42626953125,
+      "learning_rate": 2.871741632486308e-06,
+      "loss": 86.4877,
+      "step": 83650
+    },
+    {
+      "epoch": 0.6920627042230219,
+      "grad_norm": 940.0408935546875,
+      "learning_rate": 2.8704654942095977e-06,
+      "loss": 103.2302,
+      "step": 83660
+    },
+    {
+      "epoch": 0.6921454274723912,
+      "grad_norm": 955.0111083984375,
+      "learning_rate": 2.869189525370174e-06,
+      "loss": 86.1693,
+      "step": 83670
+    },
+    {
+      "epoch": 0.6922281507217604,
+      "grad_norm": 411.0567932128906,
+      "learning_rate": 2.8679137260695614e-06,
+      "loss": 82.702,
+      "step": 83680
+    },
+    {
+      "epoch": 0.6923108739711296,
+      "grad_norm": 586.8140258789062,
+      "learning_rate": 2.866638096409269e-06,
+      "loss": 97.6693,
+      "step": 83690
+    },
+    {
+      "epoch": 0.6923935972204989,
+      "grad_norm": 781.5844116210938,
+      "learning_rate": 2.8653626364907918e-06,
+      "loss": 52.9038,
+      "step": 83700
+    },
+    {
+      "epoch": 0.6924763204698681,
+      "grad_norm": 1218.3243408203125,
+      "learning_rate": 2.8640873464156127e-06,
+      "loss": 77.371,
+      "step": 83710
+    },
+    {
+      "epoch": 0.6925590437192373,
+      "grad_norm": 860.3048706054688,
+      "learning_rate": 2.8628122262852015e-06,
+      "loss": 78.9779,
+      "step": 83720
+    },
+    {
+      "epoch": 0.6926417669686066,
+      "grad_norm": 655.4887084960938,
+      "learning_rate": 2.861537276201013e-06,
+      "loss": 95.3996,
+      "step": 83730
+    },
+    {
+      "epoch": 0.6927244902179758,
+      "grad_norm": 611.69384765625,
+      "learning_rate": 2.860262496264489e-06,
+      "loss": 93.9206,
+      "step": 83740
+    },
+    {
+      "epoch": 0.692807213467345,
+      "grad_norm": 1187.9385986328125,
+      "learning_rate": 2.858987886577058e-06,
+      "loss": 85.537,
+      "step": 83750
+    },
+    {
+      "epoch": 0.6928899367167143,
+      "grad_norm": 594.444580078125,
+      "learning_rate": 2.857713447240135e-06,
+      "loss": 96.7337,
+      "step": 83760
+    },
+    {
+      "epoch": 0.6929726599660835,
+      "grad_norm": 702.6067504882812,
+      "learning_rate": 2.8564391783551214e-06,
+      "loss": 96.4278,
+      "step": 83770
+    },
+    {
+      "epoch": 0.6930553832154527,
+      "grad_norm": 799.9725341796875,
+      "learning_rate": 2.855165080023405e-06,
+      "loss": 85.3117,
+      "step": 83780
+    },
+    {
+      "epoch": 0.693138106464822,
+      "grad_norm": 865.3021240234375,
+      "learning_rate": 2.85389115234636e-06,
+      "loss": 97.0047,
+      "step": 83790
+    },
+    {
+      "epoch": 0.6932208297141912,
+      "grad_norm": 595.7553100585938,
+      "learning_rate": 2.8526173954253458e-06,
+      "loss": 87.2789,
+      "step": 83800
+    },
+    {
+      "epoch": 0.6933035529635604,
+      "grad_norm": 1058.0035400390625,
+      "learning_rate": 2.8513438093617107e-06,
+      "loss": 119.3666,
+      "step": 83810
+    },
+    {
+      "epoch": 0.6933862762129297,
+      "grad_norm": 1175.2041015625,
+      "learning_rate": 2.8500703942567874e-06,
+      "loss": 82.3834,
+      "step": 83820
+    },
+    {
+      "epoch": 0.6934689994622989,
+      "grad_norm": 966.0508422851562,
+      "learning_rate": 2.848797150211896e-06,
+      "loss": 89.0491,
+      "step": 83830
+    },
+    {
+      "epoch": 0.6935517227116681,
+      "grad_norm": 732.8645629882812,
+      "learning_rate": 2.847524077328343e-06,
+      "loss": 90.359,
+      "step": 83840
+    },
+    {
+      "epoch": 0.6936344459610374,
+      "grad_norm": 575.0491943359375,
+      "learning_rate": 2.8462511757074205e-06,
+      "loss": 73.8018,
+      "step": 83850
+    },
+    {
+      "epoch": 0.6937171692104066,
+      "grad_norm": 1305.1068115234375,
+      "learning_rate": 2.844978445450408e-06,
+      "loss": 90.5077,
+      "step": 83860
+    },
+    {
+      "epoch": 0.6937998924597758,
+      "grad_norm": 1045.33935546875,
+      "learning_rate": 2.8437058866585698e-06,
+      "loss": 114.761,
+      "step": 83870
+    },
+    {
+      "epoch": 0.693882615709145,
+      "grad_norm": 688.1589965820312,
+      "learning_rate": 2.842433499433158e-06,
+      "loss": 91.853,
+      "step": 83880
+    },
+    {
+      "epoch": 0.6939653389585143,
+      "grad_norm": 436.8230285644531,
+      "learning_rate": 2.841161283875411e-06,
+      "loss": 106.294,
+      "step": 83890
+    },
+    {
+      "epoch": 0.6940480622078835,
+      "grad_norm": 906.2509765625,
+      "learning_rate": 2.8398892400865537e-06,
+      "loss": 74.6637,
+      "step": 83900
+    },
+    {
+      "epoch": 0.6941307854572527,
+      "grad_norm": 450.2344665527344,
+      "learning_rate": 2.838617368167797e-06,
+      "loss": 143.2664,
+      "step": 83910
+    },
+    {
+      "epoch": 0.694213508706622,
+      "grad_norm": 952.6104125976562,
+      "learning_rate": 2.837345668220333e-06,
+      "loss": 69.9202,
+      "step": 83920
+    },
+    {
+      "epoch": 0.6942962319559912,
+      "grad_norm": 755.5713500976562,
+      "learning_rate": 2.836074140345352e-06,
+      "loss": 82.3544,
+      "step": 83930
+    },
+    {
+      "epoch": 0.6943789552053604,
+      "grad_norm": 659.2360229492188,
+      "learning_rate": 2.834802784644019e-06,
+      "loss": 92.3773,
+      "step": 83940
+    },
+    {
+      "epoch": 0.6944616784547297,
+      "grad_norm": 567.8734130859375,
+      "learning_rate": 2.8335316012174925e-06,
+      "loss": 102.9247,
+      "step": 83950
+    },
+    {
+      "epoch": 0.6945444017040989,
+      "grad_norm": 434.0081787109375,
+      "learning_rate": 2.8322605901669133e-06,
+      "loss": 73.0931,
+      "step": 83960
+    },
+    {
+      "epoch": 0.6946271249534681,
+      "grad_norm": 453.88153076171875,
+      "learning_rate": 2.8309897515934104e-06,
+      "loss": 98.706,
+      "step": 83970
+    },
+    {
+      "epoch": 0.6947098482028374,
+      "grad_norm": 726.1422119140625,
+      "learning_rate": 2.8297190855980987e-06,
+      "loss": 87.1649,
+      "step": 83980
+    },
+    {
+      "epoch": 0.6947925714522066,
+      "grad_norm": 784.1115112304688,
+      "learning_rate": 2.8284485922820814e-06,
+      "loss": 74.6825,
+      "step": 83990
+    },
+    {
+      "epoch": 0.6948752947015758,
+      "grad_norm": 674.6708984375,
+      "learning_rate": 2.8271782717464413e-06,
+      "loss": 59.7278,
+      "step": 84000
+    },
+    {
+      "epoch": 0.6949580179509451,
+      "grad_norm": 1085.0732421875,
+      "learning_rate": 2.8259081240922522e-06,
+      "loss": 110.7847,
+      "step": 84010
+    },
+    {
+      "epoch": 0.6950407412003143,
+      "grad_norm": 902.2394409179688,
+      "learning_rate": 2.8246381494205775e-06,
+      "loss": 72.2629,
+      "step": 84020
+    },
+    {
+      "epoch": 0.6951234644496835,
+      "grad_norm": 1027.4735107421875,
+      "learning_rate": 2.8233683478324627e-06,
+      "loss": 99.9426,
+      "step": 84030
+    },
+    {
+      "epoch": 0.6952061876990528,
+      "grad_norm": 709.7374267578125,
+      "learning_rate": 2.822098719428938e-06,
+      "loss": 100.229,
+      "step": 84040
+    },
+    {
+      "epoch": 0.695288910948422,
+      "grad_norm": 1114.0150146484375,
+      "learning_rate": 2.8208292643110237e-06,
+      "loss": 94.4212,
+      "step": 84050
+    },
+    {
+      "epoch": 0.6953716341977912,
+      "grad_norm": 679.0440063476562,
+      "learning_rate": 2.8195599825797233e-06,
+      "loss": 76.9554,
+      "step": 84060
+    },
+    {
+      "epoch": 0.6954543574471606,
+      "grad_norm": 947.027099609375,
+      "learning_rate": 2.818290874336028e-06,
+      "loss": 100.9757,
+      "step": 84070
+    },
+    {
+      "epoch": 0.6955370806965298,
+      "grad_norm": 2001.5252685546875,
+      "learning_rate": 2.817021939680918e-06,
+      "loss": 95.2409,
+      "step": 84080
+    },
+    {
+      "epoch": 0.695619803945899,
+      "grad_norm": 871.5374145507812,
+      "learning_rate": 2.8157531787153515e-06,
+      "loss": 94.9407,
+      "step": 84090
+    },
+    {
+      "epoch": 0.6957025271952683,
+      "grad_norm": 1146.988037109375,
+      "learning_rate": 2.8144845915402796e-06,
+      "loss": 102.4395,
+      "step": 84100
+    },
+    {
+      "epoch": 0.6957852504446375,
+      "grad_norm": 855.0643920898438,
+      "learning_rate": 2.813216178256637e-06,
+      "loss": 105.7368,
+      "step": 84110
+    },
+    {
+      "epoch": 0.6958679736940067,
+      "grad_norm": 662.9407958984375,
+      "learning_rate": 2.8119479389653492e-06,
+      "loss": 76.1168,
+      "step": 84120
+    },
+    {
+      "epoch": 0.695950696943376,
+      "grad_norm": 1097.759765625,
+      "learning_rate": 2.8106798737673223e-06,
+      "loss": 86.9396,
+      "step": 84130
+    },
+    {
+      "epoch": 0.6960334201927452,
+      "grad_norm": 740.43896484375,
+      "learning_rate": 2.8094119827634496e-06,
+      "loss": 84.4411,
+      "step": 84140
+    },
+    {
+      "epoch": 0.6961161434421144,
+      "grad_norm": 1305.01171875,
+      "learning_rate": 2.8081442660546126e-06,
+      "loss": 100.1945,
+      "step": 84150
+    },
+    {
+      "epoch": 0.6961988666914837,
+      "grad_norm": 1327.839599609375,
+      "learning_rate": 2.806876723741677e-06,
+      "loss": 115.4466,
+      "step": 84160
+    },
+    {
+      "epoch": 0.6962815899408529,
+      "grad_norm": 795.6221923828125,
+      "learning_rate": 2.805609355925497e-06,
+      "loss": 75.2351,
+      "step": 84170
+    },
+    {
+      "epoch": 0.6963643131902221,
+      "grad_norm": 712.4464721679688,
+      "learning_rate": 2.8043421627069077e-06,
+      "loss": 70.8538,
+      "step": 84180
+    },
+    {
+      "epoch": 0.6964470364395914,
+      "grad_norm": 838.209716796875,
+      "learning_rate": 2.8030751441867364e-06,
+      "loss": 86.4729,
+      "step": 84190
+    },
+    {
+      "epoch": 0.6965297596889606,
+      "grad_norm": 584.37646484375,
+      "learning_rate": 2.8018083004657924e-06,
+      "loss": 112.4415,
+      "step": 84200
+    },
+    {
+      "epoch": 0.6966124829383298,
+      "grad_norm": 881.0332641601562,
+      "learning_rate": 2.800541631644873e-06,
+      "loss": 145.887,
+      "step": 84210
+    },
+    {
+      "epoch": 0.6966952061876991,
+      "grad_norm": 664.2578735351562,
+      "learning_rate": 2.7992751378247627e-06,
+      "loss": 124.8047,
+      "step": 84220
+    },
+    {
+      "epoch": 0.6967779294370683,
+      "grad_norm": 929.6114501953125,
+      "learning_rate": 2.79800881910623e-06,
+      "loss": 95.0422,
+      "step": 84230
+    },
+    {
+      "epoch": 0.6968606526864375,
+      "grad_norm": 777.854248046875,
+      "learning_rate": 2.7967426755900293e-06,
+      "loss": 84.8647,
+      "step": 84240
+    },
+    {
+      "epoch": 0.6969433759358068,
+      "grad_norm": 749.425048828125,
+      "learning_rate": 2.795476707376905e-06,
+      "loss": 89.1832,
+      "step": 84250
+    },
+    {
+      "epoch": 0.697026099185176,
+      "grad_norm": 844.8867797851562,
+      "learning_rate": 2.79421091456758e-06,
+      "loss": 75.1808,
+      "step": 84260
+    },
+    {
+      "epoch": 0.6971088224345452,
+      "grad_norm": 1129.9508056640625,
+      "learning_rate": 2.7929452972627685e-06,
+      "loss": 108.7002,
+      "step": 84270
+    },
+    {
+      "epoch": 0.6971915456839145,
+      "grad_norm": 868.1597900390625,
+      "learning_rate": 2.791679855563171e-06,
+      "loss": 70.2446,
+      "step": 84280
+    },
+    {
+      "epoch": 0.6972742689332837,
+      "grad_norm": 1031.17578125,
+      "learning_rate": 2.790414589569473e-06,
+      "loss": 93.3332,
+      "step": 84290
+    },
+    {
+      "epoch": 0.6973569921826529,
+      "grad_norm": 544.4410400390625,
+      "learning_rate": 2.789149499382345e-06,
+      "loss": 78.5159,
+      "step": 84300
+    },
+    {
+      "epoch": 0.6974397154320222,
+      "grad_norm": 861.5923461914062,
+      "learning_rate": 2.7878845851024426e-06,
+      "loss": 84.6842,
+      "step": 84310
+    },
+    {
+      "epoch": 0.6975224386813914,
+      "grad_norm": 761.0271606445312,
+      "learning_rate": 2.786619846830414e-06,
+      "loss": 91.5659,
+      "step": 84320
+    },
+    {
+      "epoch": 0.6976051619307606,
+      "grad_norm": 1823.0885009765625,
+      "learning_rate": 2.7853552846668865e-06,
+      "loss": 96.6269,
+      "step": 84330
+    },
+    {
+      "epoch": 0.6976878851801299,
+      "grad_norm": 1062.7515869140625,
+      "learning_rate": 2.784090898712476e-06,
+      "loss": 87.3296,
+      "step": 84340
+    },
+    {
+      "epoch": 0.6977706084294991,
+      "grad_norm": 582.9891967773438,
+      "learning_rate": 2.7828266890677825e-06,
+      "loss": 89.4428,
+      "step": 84350
+    },
+    {
+      "epoch": 0.6978533316788683,
+      "grad_norm": 1207.8365478515625,
+      "learning_rate": 2.781562655833393e-06,
+      "loss": 109.0874,
+      "step": 84360
+    },
+    {
+      "epoch": 0.6979360549282376,
+      "grad_norm": 583.7366943359375,
+      "learning_rate": 2.7802987991098816e-06,
+      "loss": 77.5733,
+      "step": 84370
+    },
+    {
+      "epoch": 0.6980187781776068,
+      "grad_norm": 479.65802001953125,
+      "learning_rate": 2.7790351189978083e-06,
+      "loss": 83.5992,
+      "step": 84380
+    },
+    {
+      "epoch": 0.698101501426976,
+      "grad_norm": 1373.9141845703125,
+      "learning_rate": 2.777771615597717e-06,
+      "loss": 124.8873,
+      "step": 84390
+    },
+    {
+      "epoch": 0.6981842246763453,
+      "grad_norm": 1068.3160400390625,
+      "learning_rate": 2.776508289010138e-06,
+      "loss": 102.6386,
+      "step": 84400
+    },
+    {
+      "epoch": 0.6982669479257145,
+      "grad_norm": 761.4998779296875,
+      "learning_rate": 2.7752451393355916e-06,
+      "loss": 103.4375,
+      "step": 84410
+    },
+    {
+      "epoch": 0.6983496711750837,
+      "grad_norm": 594.727294921875,
+      "learning_rate": 2.773982166674582e-06,
+      "loss": 93.2319,
+      "step": 84420
+    },
+    {
+      "epoch": 0.698432394424453,
+      "grad_norm": 516.1226806640625,
+      "learning_rate": 2.772719371127593e-06,
+      "loss": 117.761,
+      "step": 84430
+    },
+    {
+      "epoch": 0.6985151176738222,
+      "grad_norm": 1146.329833984375,
+      "learning_rate": 2.771456752795102e-06,
+      "loss": 98.0758,
+      "step": 84440
+    },
+    {
+      "epoch": 0.6985978409231914,
+      "grad_norm": 1272.3282470703125,
+      "learning_rate": 2.7701943117775686e-06,
+      "loss": 71.8762,
+      "step": 84450
+    },
+    {
+      "epoch": 0.6986805641725607,
+      "grad_norm": 938.8295288085938,
+      "learning_rate": 2.7689320481754414e-06,
+      "loss": 78.6577,
+      "step": 84460
+    },
+    {
+      "epoch": 0.69876328742193,
+      "grad_norm": 1171.89306640625,
+      "learning_rate": 2.7676699620891514e-06,
+      "loss": 65.5108,
+      "step": 84470
+    },
+    {
+      "epoch": 0.6988460106712991,
+      "grad_norm": 1358.8653564453125,
+      "learning_rate": 2.7664080536191178e-06,
+      "loss": 85.8439,
+      "step": 84480
+    },
+    {
+      "epoch": 0.6989287339206685,
+      "grad_norm": 1054.154541015625,
+      "learning_rate": 2.7651463228657444e-06,
+      "loss": 93.1995,
+      "step": 84490
+    },
+    {
+      "epoch": 0.6990114571700377,
+      "grad_norm": 1459.2327880859375,
+      "learning_rate": 2.7638847699294196e-06,
+      "loss": 96.8761,
+      "step": 84500
+    },
+    {
+      "epoch": 0.6990941804194069,
+      "grad_norm": 607.018310546875,
+      "learning_rate": 2.7626233949105252e-06,
+      "loss": 69.8457,
+      "step": 84510
+    },
+    {
+      "epoch": 0.6991769036687762,
+      "grad_norm": 713.6427612304688,
+      "learning_rate": 2.7613621979094173e-06,
+      "loss": 61.2357,
+      "step": 84520
+    },
+    {
+      "epoch": 0.6992596269181454,
+      "grad_norm": 1142.56201171875,
+      "learning_rate": 2.7601011790264454e-06,
+      "loss": 81.4103,
+      "step": 84530
+    },
+    {
+      "epoch": 0.6993423501675146,
+      "grad_norm": 354.2678527832031,
+      "learning_rate": 2.758840338361942e-06,
+      "loss": 79.8373,
+      "step": 84540
+    },
+    {
+      "epoch": 0.6994250734168839,
+      "grad_norm": 730.341552734375,
+      "learning_rate": 2.7575796760162288e-06,
+      "loss": 84.2706,
+      "step": 84550
+    },
+    {
+      "epoch": 0.6995077966662531,
+      "grad_norm": 687.3909301757812,
+      "learning_rate": 2.7563191920896084e-06,
+      "loss": 79.1268,
+      "step": 84560
+    },
+    {
+      "epoch": 0.6995905199156223,
+      "grad_norm": 771.8450927734375,
+      "learning_rate": 2.755058886682373e-06,
+      "loss": 86.2774,
+      "step": 84570
+    },
+    {
+      "epoch": 0.6996732431649916,
+      "grad_norm": 826.2738037109375,
+      "learning_rate": 2.753798759894799e-06,
+      "loss": 96.8274,
+      "step": 84580
+    },
+    {
+      "epoch": 0.6997559664143608,
+      "grad_norm": 755.7567749023438,
+      "learning_rate": 2.7525388118271495e-06,
+      "loss": 90.1103,
+      "step": 84590
+    },
+    {
+      "epoch": 0.69983868966373,
+      "grad_norm": 1258.190185546875,
+      "learning_rate": 2.751279042579672e-06,
+      "loss": 95.2349,
+      "step": 84600
+    },
+    {
+      "epoch": 0.6999214129130992,
+      "grad_norm": 915.1702270507812,
+      "learning_rate": 2.7500194522526007e-06,
+      "loss": 97.5897,
+      "step": 84610
+    },
+    {
+      "epoch": 0.7000041361624685,
+      "grad_norm": 395.96044921875,
+      "learning_rate": 2.748760040946156e-06,
+      "loss": 71.4906,
+      "step": 84620
+    },
+    {
+      "epoch": 0.7000868594118377,
+      "grad_norm": 1328.4857177734375,
+      "learning_rate": 2.7475008087605428e-06,
+      "loss": 88.9978,
+      "step": 84630
+    },
+    {
+      "epoch": 0.7001695826612069,
+      "grad_norm": 1446.4951171875,
+      "learning_rate": 2.746241755795952e-06,
+      "loss": 129.5656,
+      "step": 84640
+    },
+    {
+      "epoch": 0.7002523059105762,
+      "grad_norm": 316.505615234375,
+      "learning_rate": 2.7449828821525624e-06,
+      "loss": 87.3932,
+      "step": 84650
+    },
+    {
+      "epoch": 0.7003350291599454,
+      "grad_norm": 738.2881469726562,
+      "learning_rate": 2.7437241879305354e-06,
+      "loss": 81.33,
+      "step": 84660
+    },
+    {
+      "epoch": 0.7004177524093146,
+      "grad_norm": 449.9028625488281,
+      "learning_rate": 2.7424656732300193e-06,
+      "loss": 71.7711,
+      "step": 84670
+    },
+    {
+      "epoch": 0.7005004756586839,
+      "grad_norm": 856.5056762695312,
+      "learning_rate": 2.7412073381511495e-06,
+      "loss": 66.1047,
+      "step": 84680
+    },
+    {
+      "epoch": 0.7005831989080531,
+      "grad_norm": 1258.2718505859375,
+      "learning_rate": 2.739949182794045e-06,
+      "loss": 119.7207,
+      "step": 84690
+    },
+    {
+      "epoch": 0.7006659221574223,
+      "grad_norm": 1023.8765869140625,
+      "learning_rate": 2.7386912072588123e-06,
+      "loss": 84.5958,
+      "step": 84700
+    },
+    {
+      "epoch": 0.7007486454067916,
+      "grad_norm": 1277.8900146484375,
+      "learning_rate": 2.737433411645542e-06,
+      "loss": 85.9086,
+      "step": 84710
+    },
+    {
+      "epoch": 0.7008313686561608,
+      "grad_norm": 739.3309326171875,
+      "learning_rate": 2.7361757960543114e-06,
+      "loss": 81.2444,
+      "step": 84720
+    },
+    {
+      "epoch": 0.70091409190553,
+      "grad_norm": 902.0711669921875,
+      "learning_rate": 2.7349183605851824e-06,
+      "loss": 72.4432,
+      "step": 84730
+    },
+    {
+      "epoch": 0.7009968151548993,
+      "grad_norm": 1022.815185546875,
+      "learning_rate": 2.733661105338205e-06,
+      "loss": 131.5625,
+      "step": 84740
+    },
+    {
+      "epoch": 0.7010795384042685,
+      "grad_norm": 1463.1163330078125,
+      "learning_rate": 2.7324040304134125e-06,
+      "loss": 156.1384,
+      "step": 84750
+    },
+    {
+      "epoch": 0.7011622616536377,
+      "grad_norm": 762.2933959960938,
+      "learning_rate": 2.731147135910824e-06,
+      "loss": 100.2397,
+      "step": 84760
+    },
+    {
+      "epoch": 0.701244984903007,
+      "grad_norm": 1155.1771240234375,
+      "learning_rate": 2.729890421930445e-06,
+      "loss": 89.7689,
+      "step": 84770
+    },
+    {
+      "epoch": 0.7013277081523762,
+      "grad_norm": 649.2015991210938,
+      "learning_rate": 2.7286338885722674e-06,
+      "loss": 105.2359,
+      "step": 84780
+    },
+    {
+      "epoch": 0.7014104314017454,
+      "grad_norm": 776.761962890625,
+      "learning_rate": 2.7273775359362665e-06,
+      "loss": 78.5909,
+      "step": 84790
+    },
+    {
+      "epoch": 0.7014931546511147,
+      "grad_norm": 1017.0020141601562,
+      "learning_rate": 2.7261213641224056e-06,
+      "loss": 87.9357,
+      "step": 84800
+    },
+    {
+      "epoch": 0.7015758779004839,
+      "grad_norm": 754.7698974609375,
+      "learning_rate": 2.724865373230632e-06,
+      "loss": 105.561,
+      "step": 84810
+    },
+    {
+      "epoch": 0.7016586011498531,
+      "grad_norm": 519.2734375,
+      "learning_rate": 2.723609563360879e-06,
+      "loss": 87.3953,
+      "step": 84820
+    },
+    {
+      "epoch": 0.7017413243992224,
+      "grad_norm": 740.0153198242188,
+      "learning_rate": 2.7223539346130655e-06,
+      "loss": 105.0598,
+      "step": 84830
+    },
+    {
+      "epoch": 0.7018240476485916,
+      "grad_norm": 761.6763305664062,
+      "learning_rate": 2.7210984870870972e-06,
+      "loss": 110.6421,
+      "step": 84840
+    },
+    {
+      "epoch": 0.7019067708979608,
+      "grad_norm": 1044.949951171875,
+      "learning_rate": 2.7198432208828653e-06,
+      "loss": 116.031,
+      "step": 84850
+    },
+    {
+      "epoch": 0.7019894941473301,
+      "grad_norm": 934.532958984375,
+      "learning_rate": 2.7185881361002415e-06,
+      "loss": 75.7842,
+      "step": 84860
+    },
+    {
+      "epoch": 0.7020722173966993,
+      "grad_norm": 714.5556030273438,
+      "learning_rate": 2.717333232839088e-06,
+      "loss": 105.3236,
+      "step": 84870
+    },
+    {
+      "epoch": 0.7021549406460685,
+      "grad_norm": 589.3107299804688,
+      "learning_rate": 2.7160785111992546e-06,
+      "loss": 82.1901,
+      "step": 84880
+    },
+    {
+      "epoch": 0.7022376638954378,
+      "grad_norm": 1139.3077392578125,
+      "learning_rate": 2.7148239712805725e-06,
+      "loss": 90.0878,
+      "step": 84890
+    },
+    {
+      "epoch": 0.702320387144807,
+      "grad_norm": 951.9703979492188,
+      "learning_rate": 2.713569613182859e-06,
+      "loss": 78.3111,
+      "step": 84900
+    },
+    {
+      "epoch": 0.7024031103941762,
+      "grad_norm": 1005.8824462890625,
+      "learning_rate": 2.7123154370059185e-06,
+      "loss": 107.2209,
+      "step": 84910
+    },
+    {
+      "epoch": 0.7024858336435456,
+      "grad_norm": 1006.9483642578125,
+      "learning_rate": 2.7110614428495396e-06,
+      "loss": 116.84,
+      "step": 84920
+    },
+    {
+      "epoch": 0.7025685568929148,
+      "grad_norm": 607.9546508789062,
+      "learning_rate": 2.709807630813497e-06,
+      "loss": 106.0352,
+      "step": 84930
+    },
+    {
+      "epoch": 0.702651280142284,
+      "grad_norm": 458.4266662597656,
+      "learning_rate": 2.7085540009975526e-06,
+      "loss": 78.4597,
+      "step": 84940
+    },
+    {
+      "epoch": 0.7027340033916533,
+      "grad_norm": 1103.226318359375,
+      "learning_rate": 2.707300553501448e-06,
+      "loss": 70.445,
+      "step": 84950
+    },
+    {
+      "epoch": 0.7028167266410225,
+      "grad_norm": 1002.1087036132812,
+      "learning_rate": 2.7060472884249145e-06,
+      "loss": 87.4876,
+      "step": 84960
+    },
+    {
+      "epoch": 0.7028994498903917,
+      "grad_norm": 993.1524047851562,
+      "learning_rate": 2.7047942058676717e-06,
+      "loss": 82.637,
+      "step": 84970
+    },
+    {
+      "epoch": 0.702982173139761,
+      "grad_norm": 1153.7645263671875,
+      "learning_rate": 2.703541305929421e-06,
+      "loss": 105.1839,
+      "step": 84980
+    },
+    {
+      "epoch": 0.7030648963891302,
+      "grad_norm": 765.0949096679688,
+      "learning_rate": 2.7022885887098492e-06,
+      "loss": 69.9202,
+      "step": 84990
+    },
+    {
+      "epoch": 0.7031476196384994,
+      "grad_norm": 930.6118774414062,
+      "learning_rate": 2.701036054308629e-06,
+      "loss": 78.9953,
+      "step": 85000
+    },
+    {
+      "epoch": 0.7032303428878687,
+      "grad_norm": 728.7538452148438,
+      "learning_rate": 2.699783702825419e-06,
+      "loss": 86.7408,
+      "step": 85010
+    },
+    {
+      "epoch": 0.7033130661372379,
+      "grad_norm": 731.0704345703125,
+      "learning_rate": 2.698531534359864e-06,
+      "loss": 94.9619,
+      "step": 85020
+    },
+    {
+      "epoch": 0.7033957893866071,
+      "grad_norm": 795.151611328125,
+      "learning_rate": 2.6972795490115944e-06,
+      "loss": 78.0553,
+      "step": 85030
+    },
+    {
+      "epoch": 0.7034785126359764,
+      "grad_norm": 806.3612060546875,
+      "learning_rate": 2.6960277468802203e-06,
+      "loss": 75.0848,
+      "step": 85040
+    },
+    {
+      "epoch": 0.7035612358853456,
+      "grad_norm": 1016.3757934570312,
+      "learning_rate": 2.694776128065345e-06,
+      "loss": 70.555,
+      "step": 85050
+    },
+    {
+      "epoch": 0.7036439591347148,
+      "grad_norm": 759.8416137695312,
+      "learning_rate": 2.6935246926665513e-06,
+      "loss": 75.158,
+      "step": 85060
+    },
+    {
+      "epoch": 0.7037266823840841,
+      "grad_norm": 671.8663940429688,
+      "learning_rate": 2.692273440783415e-06,
+      "loss": 107.0158,
+      "step": 85070
+    },
+    {
+      "epoch": 0.7038094056334533,
+      "grad_norm": 978.9486694335938,
+      "learning_rate": 2.6910223725154903e-06,
+      "loss": 105.8965,
+      "step": 85080
+    },
+    {
+      "epoch": 0.7038921288828225,
+      "grad_norm": 1325.02294921875,
+      "learning_rate": 2.6897714879623184e-06,
+      "loss": 97.7983,
+      "step": 85090
+    },
+    {
+      "epoch": 0.7039748521321918,
+      "grad_norm": 592.0951538085938,
+      "learning_rate": 2.688520787223426e-06,
+      "loss": 86.1239,
+      "step": 85100
+    },
+    {
+      "epoch": 0.704057575381561,
+      "grad_norm": 648.145263671875,
+      "learning_rate": 2.6872702703983287e-06,
+      "loss": 81.3647,
+      "step": 85110
+    },
+    {
+      "epoch": 0.7041402986309302,
+      "grad_norm": 702.9576416015625,
+      "learning_rate": 2.6860199375865203e-06,
+      "loss": 125.3935,
+      "step": 85120
+    },
+    {
+      "epoch": 0.7042230218802995,
+      "grad_norm": 605.6353759765625,
+      "learning_rate": 2.6847697888874853e-06,
+      "loss": 98.5384,
+      "step": 85130
+    },
+    {
+      "epoch": 0.7043057451296687,
+      "grad_norm": 843.0704956054688,
+      "learning_rate": 2.683519824400693e-06,
+      "loss": 100.6389,
+      "step": 85140
+    },
+    {
+      "epoch": 0.7043884683790379,
+      "grad_norm": 1235.5345458984375,
+      "learning_rate": 2.6822700442255965e-06,
+      "loss": 93.4948,
+      "step": 85150
+    },
+    {
+      "epoch": 0.7044711916284072,
+      "grad_norm": 1300.0670166015625,
+      "learning_rate": 2.681020448461634e-06,
+      "loss": 77.3637,
+      "step": 85160
+    },
+    {
+      "epoch": 0.7045539148777764,
+      "grad_norm": 2526.052734375,
+      "learning_rate": 2.679771037208234e-06,
+      "loss": 179.5842,
+      "step": 85170
+    },
+    {
+      "epoch": 0.7046366381271456,
+      "grad_norm": 994.899169921875,
+      "learning_rate": 2.678521810564804e-06,
+      "loss": 68.4581,
+      "step": 85180
+    },
+    {
+      "epoch": 0.7047193613765149,
+      "grad_norm": 894.9552612304688,
+      "learning_rate": 2.6772727686307398e-06,
+      "loss": 98.303,
+      "step": 85190
+    },
+    {
+      "epoch": 0.7048020846258841,
+      "grad_norm": 1177.982421875,
+      "learning_rate": 2.676023911505423e-06,
+      "loss": 82.6399,
+      "step": 85200
+    },
+    {
+      "epoch": 0.7048848078752533,
+      "grad_norm": 1218.446533203125,
+      "learning_rate": 2.674775239288216e-06,
+      "loss": 101.0908,
+      "step": 85210
+    },
+    {
+      "epoch": 0.7049675311246226,
+      "grad_norm": 1381.9801025390625,
+      "learning_rate": 2.673526752078472e-06,
+      "loss": 102.3884,
+      "step": 85220
+    },
+    {
+      "epoch": 0.7050502543739918,
+      "grad_norm": 900.25634765625,
+      "learning_rate": 2.6722784499755273e-06,
+      "loss": 70.0581,
+      "step": 85230
+    },
+    {
+      "epoch": 0.705132977623361,
+      "grad_norm": 407.5909729003906,
+      "learning_rate": 2.6710303330787035e-06,
+      "loss": 83.3081,
+      "step": 85240
+    },
+    {
+      "epoch": 0.7052157008727303,
+      "grad_norm": 914.4041137695312,
+      "learning_rate": 2.6697824014873076e-06,
+      "loss": 82.6285,
+      "step": 85250
+    },
+    {
+      "epoch": 0.7052984241220995,
+      "grad_norm": 731.021240234375,
+      "learning_rate": 2.6685346553006293e-06,
+      "loss": 78.5455,
+      "step": 85260
+    },
+    {
+      "epoch": 0.7053811473714687,
+      "grad_norm": 1125.2713623046875,
+      "learning_rate": 2.6672870946179506e-06,
+      "loss": 105.0443,
+      "step": 85270
+    },
+    {
+      "epoch": 0.705463870620838,
+      "grad_norm": 688.19140625,
+      "learning_rate": 2.6660397195385344e-06,
+      "loss": 114.3066,
+      "step": 85280
+    },
+    {
+      "epoch": 0.7055465938702072,
+      "grad_norm": 1272.36083984375,
+      "learning_rate": 2.664792530161624e-06,
+      "loss": 86.4369,
+      "step": 85290
+    },
+    {
+      "epoch": 0.7056293171195764,
+      "grad_norm": 1196.80224609375,
+      "learning_rate": 2.6635455265864553e-06,
+      "loss": 107.9678,
+      "step": 85300
+    },
+    {
+      "epoch": 0.7057120403689456,
+      "grad_norm": 773.6207275390625,
+      "learning_rate": 2.662298708912246e-06,
+      "loss": 97.1165,
+      "step": 85310
+    },
+    {
+      "epoch": 0.705794763618315,
+      "grad_norm": 636.7437744140625,
+      "learning_rate": 2.6610520772382e-06,
+      "loss": 81.5361,
+      "step": 85320
+    },
+    {
+      "epoch": 0.7058774868676841,
+      "grad_norm": 875.7274780273438,
+      "learning_rate": 2.659805631663505e-06,
+      "loss": 78.7864,
+      "step": 85330
+    },
+    {
+      "epoch": 0.7059602101170533,
+      "grad_norm": 1031.9066162109375,
+      "learning_rate": 2.658559372287337e-06,
+      "loss": 92.2199,
+      "step": 85340
+    },
+    {
+      "epoch": 0.7060429333664227,
+      "grad_norm": 583.7274780273438,
+      "learning_rate": 2.6573132992088534e-06,
+      "loss": 63.5923,
+      "step": 85350
+    },
+    {
+      "epoch": 0.7061256566157919,
+      "grad_norm": 890.5264282226562,
+      "learning_rate": 2.656067412527197e-06,
+      "loss": 84.5146,
+      "step": 85360
+    },
+    {
+      "epoch": 0.706208379865161,
+      "grad_norm": 744.9577026367188,
+      "learning_rate": 2.6548217123415033e-06,
+      "loss": 90.214,
+      "step": 85370
+    },
+    {
+      "epoch": 0.7062911031145304,
+      "grad_norm": 792.3532104492188,
+      "learning_rate": 2.6535761987508813e-06,
+      "loss": 79.4676,
+      "step": 85380
+    },
+    {
+      "epoch": 0.7063738263638996,
+      "grad_norm": 739.8575439453125,
+      "learning_rate": 2.652330871854433e-06,
+      "loss": 90.5693,
+      "step": 85390
+    },
+    {
+      "epoch": 0.7064565496132688,
+      "grad_norm": 450.8844909667969,
+      "learning_rate": 2.651085731751242e-06,
+      "loss": 90.0703,
+      "step": 85400
+    },
+    {
+      "epoch": 0.7065392728626381,
+      "grad_norm": 1677.9515380859375,
+      "learning_rate": 2.6498407785403794e-06,
+      "loss": 93.2693,
+      "step": 85410
+    },
+    {
+      "epoch": 0.7066219961120073,
+      "grad_norm": 899.0578002929688,
+      "learning_rate": 2.648596012320901e-06,
+      "loss": 79.1233,
+      "step": 85420
+    },
+    {
+      "epoch": 0.7067047193613765,
+      "grad_norm": 735.9163208007812,
+      "learning_rate": 2.647351433191846e-06,
+      "loss": 90.7922,
+      "step": 85430
+    },
+    {
+      "epoch": 0.7067874426107458,
+      "grad_norm": 951.7572631835938,
+      "learning_rate": 2.64610704125224e-06,
+      "loss": 88.4487,
+      "step": 85440
+    },
+    {
+      "epoch": 0.706870165860115,
+      "grad_norm": 1408.339111328125,
+      "learning_rate": 2.644862836601092e-06,
+      "loss": 95.58,
+      "step": 85450
+    },
+    {
+      "epoch": 0.7069528891094842,
+      "grad_norm": 850.1331787109375,
+      "learning_rate": 2.6436188193374035e-06,
+      "loss": 96.218,
+      "step": 85460
+    },
+    {
+      "epoch": 0.7070356123588535,
+      "grad_norm": 864.1105346679688,
+      "learning_rate": 2.6423749895601494e-06,
+      "loss": 113.5922,
+      "step": 85470
+    },
+    {
+      "epoch": 0.7071183356082227,
+      "grad_norm": 969.8984985351562,
+      "learning_rate": 2.6411313473682966e-06,
+      "loss": 98.6914,
+      "step": 85480
+    },
+    {
+      "epoch": 0.7072010588575919,
+      "grad_norm": 965.0258178710938,
+      "learning_rate": 2.6398878928607973e-06,
+      "loss": 111.6938,
+      "step": 85490
+    },
+    {
+      "epoch": 0.7072837821069612,
+      "grad_norm": 2024.0552978515625,
+      "learning_rate": 2.6386446261365874e-06,
+      "loss": 122.5149,
+      "step": 85500
+    },
+    {
+      "epoch": 0.7073665053563304,
+      "grad_norm": 342.0011291503906,
+      "learning_rate": 2.6374015472945868e-06,
+      "loss": 109.7086,
+      "step": 85510
+    },
+    {
+      "epoch": 0.7074492286056996,
+      "grad_norm": 848.0896606445312,
+      "learning_rate": 2.6361586564337023e-06,
+      "loss": 91.5193,
+      "step": 85520
+    },
+    {
+      "epoch": 0.7075319518550689,
+      "grad_norm": 559.4988403320312,
+      "learning_rate": 2.6349159536528245e-06,
+      "loss": 93.1284,
+      "step": 85530
+    },
+    {
+      "epoch": 0.7076146751044381,
+      "grad_norm": 870.3743896484375,
+      "learning_rate": 2.633673439050831e-06,
+      "loss": 80.7735,
+      "step": 85540
+    },
+    {
+      "epoch": 0.7076973983538073,
+      "grad_norm": 547.6781616210938,
+      "learning_rate": 2.6324311127265812e-06,
+      "loss": 106.9985,
+      "step": 85550
+    },
+    {
+      "epoch": 0.7077801216031766,
+      "grad_norm": 691.0943603515625,
+      "learning_rate": 2.6311889747789225e-06,
+      "loss": 102.5531,
+      "step": 85560
+    },
+    {
+      "epoch": 0.7078628448525458,
+      "grad_norm": 845.0540771484375,
+      "learning_rate": 2.6299470253066863e-06,
+      "loss": 86.6591,
+      "step": 85570
+    },
+    {
+      "epoch": 0.707945568101915,
+      "grad_norm": 847.3442993164062,
+      "learning_rate": 2.628705264408687e-06,
+      "loss": 106.6626,
+      "step": 85580
+    },
+    {
+      "epoch": 0.7080282913512843,
+      "grad_norm": 1003.5369262695312,
+      "learning_rate": 2.6274636921837272e-06,
+      "loss": 106.0019,
+      "step": 85590
+    },
+    {
+      "epoch": 0.7081110146006535,
+      "grad_norm": 774.9839477539062,
+      "learning_rate": 2.626222308730594e-06,
+      "loss": 88.3581,
+      "step": 85600
+    },
+    {
+      "epoch": 0.7081937378500227,
+      "grad_norm": 854.4312133789062,
+      "learning_rate": 2.6249811141480564e-06,
+      "loss": 89.8101,
+      "step": 85610
+    },
+    {
+      "epoch": 0.708276461099392,
+      "grad_norm": 445.865234375,
+      "learning_rate": 2.6237401085348723e-06,
+      "loss": 62.2433,
+      "step": 85620
+    },
+    {
+      "epoch": 0.7083591843487612,
+      "grad_norm": 720.7432250976562,
+      "learning_rate": 2.6224992919897817e-06,
+      "loss": 99.8638,
+      "step": 85630
+    },
+    {
+      "epoch": 0.7084419075981304,
+      "grad_norm": 997.8077392578125,
+      "learning_rate": 2.6212586646115114e-06,
+      "loss": 77.9413,
+      "step": 85640
+    },
+    {
+      "epoch": 0.7085246308474997,
+      "grad_norm": 1052.6624755859375,
+      "learning_rate": 2.620018226498772e-06,
+      "loss": 104.5083,
+      "step": 85650
+    },
+    {
+      "epoch": 0.7086073540968689,
+      "grad_norm": 999.076171875,
+      "learning_rate": 2.61877797775026e-06,
+      "loss": 89.7667,
+      "step": 85660
+    },
+    {
+      "epoch": 0.7086900773462381,
+      "grad_norm": 1541.536376953125,
+      "learning_rate": 2.6175379184646565e-06,
+      "loss": 92.7317,
+      "step": 85670
+    },
+    {
+      "epoch": 0.7087728005956074,
+      "grad_norm": 876.532470703125,
+      "learning_rate": 2.616298048740626e-06,
+      "loss": 85.8813,
+      "step": 85680
+    },
+    {
+      "epoch": 0.7088555238449766,
+      "grad_norm": 896.5637817382812,
+      "learning_rate": 2.6150583686768203e-06,
+      "loss": 120.6416,
+      "step": 85690
+    },
+    {
+      "epoch": 0.7089382470943458,
+      "grad_norm": 913.8203125,
+      "learning_rate": 2.6138188783718745e-06,
+      "loss": 100.0601,
+      "step": 85700
+    },
+    {
+      "epoch": 0.7090209703437151,
+      "grad_norm": 662.0992431640625,
+      "learning_rate": 2.6125795779244125e-06,
+      "loss": 90.1497,
+      "step": 85710
+    },
+    {
+      "epoch": 0.7091036935930843,
+      "grad_norm": 738.2545776367188,
+      "learning_rate": 2.611340467433031e-06,
+      "loss": 94.7059,
+      "step": 85720
+    },
+    {
+      "epoch": 0.7091864168424535,
+      "grad_norm": 925.9259643554688,
+      "learning_rate": 2.61010154699633e-06,
+      "loss": 64.9561,
+      "step": 85730
+    },
+    {
+      "epoch": 0.7092691400918228,
+      "grad_norm": 657.1807861328125,
+      "learning_rate": 2.6088628167128794e-06,
+      "loss": 81.276,
+      "step": 85740
+    },
+    {
+      "epoch": 0.709351863341192,
+      "grad_norm": 703.9519653320312,
+      "learning_rate": 2.607624276681241e-06,
+      "loss": 73.3136,
+      "step": 85750
+    },
+    {
+      "epoch": 0.7094345865905612,
+      "grad_norm": 596.41064453125,
+      "learning_rate": 2.6063859269999594e-06,
+      "loss": 77.5182,
+      "step": 85760
+    },
+    {
+      "epoch": 0.7095173098399306,
+      "grad_norm": 327.9614562988281,
+      "learning_rate": 2.605147767767564e-06,
+      "loss": 84.4486,
+      "step": 85770
+    },
+    {
+      "epoch": 0.7096000330892998,
+      "grad_norm": 914.2791137695312,
+      "learning_rate": 2.6039097990825703e-06,
+      "loss": 99.9321,
+      "step": 85780
+    },
+    {
+      "epoch": 0.709682756338669,
+      "grad_norm": 758.622802734375,
+      "learning_rate": 2.602672021043477e-06,
+      "loss": 83.2458,
+      "step": 85790
+    },
+    {
+      "epoch": 0.7097654795880383,
+      "grad_norm": 801.3850708007812,
+      "learning_rate": 2.601434433748771e-06,
+      "loss": 99.5949,
+      "step": 85800
+    },
+    {
+      "epoch": 0.7098482028374075,
+      "grad_norm": 531.544189453125,
+      "learning_rate": 2.600197037296917e-06,
+      "loss": 85.0115,
+      "step": 85810
+    },
+    {
+      "epoch": 0.7099309260867767,
+      "grad_norm": 977.3123779296875,
+      "learning_rate": 2.5989598317863694e-06,
+      "loss": 93.289,
+      "step": 85820
+    },
+    {
+      "epoch": 0.710013649336146,
+      "grad_norm": 613.4664916992188,
+      "learning_rate": 2.59772281731557e-06,
+      "loss": 80.9545,
+      "step": 85830
+    },
+    {
+      "epoch": 0.7100963725855152,
+      "grad_norm": 646.8933715820312,
+      "learning_rate": 2.5964859939829423e-06,
+      "loss": 107.6766,
+      "step": 85840
+    },
+    {
+      "epoch": 0.7101790958348844,
+      "grad_norm": 650.1854858398438,
+      "learning_rate": 2.595249361886892e-06,
+      "loss": 75.5614,
+      "step": 85850
+    },
+    {
+      "epoch": 0.7102618190842537,
+      "grad_norm": 816.5127563476562,
+      "learning_rate": 2.5940129211258147e-06,
+      "loss": 94.7183,
+      "step": 85860
+    },
+    {
+      "epoch": 0.7103445423336229,
+      "grad_norm": 719.228271484375,
+      "learning_rate": 2.5927766717980873e-06,
+      "loss": 90.4481,
+      "step": 85870
+    },
+    {
+      "epoch": 0.7104272655829921,
+      "grad_norm": 1630.4447021484375,
+      "learning_rate": 2.5915406140020738e-06,
+      "loss": 99.5369,
+      "step": 85880
+    },
+    {
+      "epoch": 0.7105099888323614,
+      "grad_norm": 830.96044921875,
+      "learning_rate": 2.590304747836119e-06,
+      "loss": 106.8879,
+      "step": 85890
+    },
+    {
+      "epoch": 0.7105927120817306,
+      "grad_norm": 1144.8453369140625,
+      "learning_rate": 2.5890690733985555e-06,
+      "loss": 93.4202,
+      "step": 85900
+    },
+    {
+      "epoch": 0.7106754353310998,
+      "grad_norm": 1315.0010986328125,
+      "learning_rate": 2.5878335907876997e-06,
+      "loss": 75.4638,
+      "step": 85910
+    },
+    {
+      "epoch": 0.7107581585804691,
+      "grad_norm": 748.9456176757812,
+      "learning_rate": 2.5865983001018567e-06,
+      "loss": 79.2242,
+      "step": 85920
+    },
+    {
+      "epoch": 0.7108408818298383,
+      "grad_norm": 456.5758972167969,
+      "learning_rate": 2.5853632014393108e-06,
+      "loss": 61.4048,
+      "step": 85930
+    },
+    {
+      "epoch": 0.7109236050792075,
+      "grad_norm": 565.6151733398438,
+      "learning_rate": 2.584128294898334e-06,
+      "loss": 79.6808,
+      "step": 85940
+    },
+    {
+      "epoch": 0.7110063283285768,
+      "grad_norm": 922.67529296875,
+      "learning_rate": 2.5828935805771804e-06,
+      "loss": 141.0256,
+      "step": 85950
+    },
+    {
+      "epoch": 0.711089051577946,
+      "grad_norm": 889.4005737304688,
+      "learning_rate": 2.581659058574092e-06,
+      "loss": 67.4175,
+      "step": 85960
+    },
+    {
+      "epoch": 0.7111717748273152,
+      "grad_norm": 669.8568725585938,
+      "learning_rate": 2.580424728987296e-06,
+      "loss": 88.6796,
+      "step": 85970
+    },
+    {
+      "epoch": 0.7112544980766845,
+      "grad_norm": 1054.4835205078125,
+      "learning_rate": 2.5791905919149973e-06,
+      "loss": 108.1136,
+      "step": 85980
+    },
+    {
+      "epoch": 0.7113372213260537,
+      "grad_norm": 1388.4984130859375,
+      "learning_rate": 2.5779566474553934e-06,
+      "loss": 109.7086,
+      "step": 85990
+    },
+    {
+      "epoch": 0.7114199445754229,
+      "grad_norm": 624.65673828125,
+      "learning_rate": 2.5767228957066635e-06,
+      "loss": 94.2603,
+      "step": 86000
+    },
+    {
+      "epoch": 0.7115026678247922,
+      "grad_norm": 607.4688110351562,
+      "learning_rate": 2.5754893367669697e-06,
+      "loss": 75.7593,
+      "step": 86010
+    },
+    {
+      "epoch": 0.7115853910741614,
+      "grad_norm": 1318.2784423828125,
+      "learning_rate": 2.5742559707344638e-06,
+      "loss": 98.0989,
+      "step": 86020
+    },
+    {
+      "epoch": 0.7116681143235306,
+      "grad_norm": 647.1195068359375,
+      "learning_rate": 2.573022797707278e-06,
+      "loss": 79.2859,
+      "step": 86030
+    },
+    {
+      "epoch": 0.7117508375728998,
+      "grad_norm": 461.97894287109375,
+      "learning_rate": 2.57178981778353e-06,
+      "loss": 101.9,
+      "step": 86040
+    },
+    {
+      "epoch": 0.7118335608222691,
+      "grad_norm": 616.1342163085938,
+      "learning_rate": 2.5705570310613215e-06,
+      "loss": 101.6059,
+      "step": 86050
+    },
+    {
+      "epoch": 0.7119162840716383,
+      "grad_norm": 997.8457641601562,
+      "learning_rate": 2.5693244376387435e-06,
+      "loss": 83.1273,
+      "step": 86060
+    },
+    {
+      "epoch": 0.7119990073210075,
+      "grad_norm": 954.9032592773438,
+      "learning_rate": 2.568092037613862e-06,
+      "loss": 109.8029,
+      "step": 86070
+    },
+    {
+      "epoch": 0.7120817305703768,
+      "grad_norm": 704.2922973632812,
+      "learning_rate": 2.566859831084736e-06,
+      "loss": 93.9186,
+      "step": 86080
+    },
+    {
+      "epoch": 0.712164453819746,
+      "grad_norm": 826.8761596679688,
+      "learning_rate": 2.5656278181494072e-06,
+      "loss": 85.7292,
+      "step": 86090
+    },
+    {
+      "epoch": 0.7122471770691152,
+      "grad_norm": 1042.4456787109375,
+      "learning_rate": 2.5643959989058997e-06,
+      "loss": 103.6534,
+      "step": 86100
+    },
+    {
+      "epoch": 0.7123299003184845,
+      "grad_norm": 1412.099365234375,
+      "learning_rate": 2.563164373452224e-06,
+      "loss": 96.1712,
+      "step": 86110
+    },
+    {
+      "epoch": 0.7124126235678537,
+      "grad_norm": 846.20947265625,
+      "learning_rate": 2.561932941886377e-06,
+      "loss": 93.9482,
+      "step": 86120
+    },
+    {
+      "epoch": 0.7124953468172229,
+      "grad_norm": 650.3298950195312,
+      "learning_rate": 2.560701704306336e-06,
+      "loss": 94.7339,
+      "step": 86130
+    },
+    {
+      "epoch": 0.7125780700665922,
+      "grad_norm": 903.107666015625,
+      "learning_rate": 2.5594706608100677e-06,
+      "loss": 68.6888,
+      "step": 86140
+    },
+    {
+      "epoch": 0.7126607933159614,
+      "grad_norm": 2324.131591796875,
+      "learning_rate": 2.5582398114955164e-06,
+      "loss": 88.7436,
+      "step": 86150
+    },
+    {
+      "epoch": 0.7127435165653306,
+      "grad_norm": 676.3530883789062,
+      "learning_rate": 2.5570091564606182e-06,
+      "loss": 95.5154,
+      "step": 86160
+    },
+    {
+      "epoch": 0.7128262398147,
+      "grad_norm": 713.0216064453125,
+      "learning_rate": 2.555778695803288e-06,
+      "loss": 48.4695,
+      "step": 86170
+    },
+    {
+      "epoch": 0.7129089630640691,
+      "grad_norm": 877.8251342773438,
+      "learning_rate": 2.554548429621431e-06,
+      "loss": 99.8856,
+      "step": 86180
+    },
+    {
+      "epoch": 0.7129916863134383,
+      "grad_norm": 679.6746215820312,
+      "learning_rate": 2.5533183580129317e-06,
+      "loss": 109.55,
+      "step": 86190
+    },
+    {
+      "epoch": 0.7130744095628077,
+      "grad_norm": 1079.2734375,
+      "learning_rate": 2.5520884810756614e-06,
+      "loss": 92.1818,
+      "step": 86200
+    },
+    {
+      "epoch": 0.7131571328121769,
+      "grad_norm": 1415.2052001953125,
+      "learning_rate": 2.550858798907475e-06,
+      "loss": 98.666,
+      "step": 86210
+    },
+    {
+      "epoch": 0.713239856061546,
+      "grad_norm": 498.3135986328125,
+      "learning_rate": 2.5496293116062154e-06,
+      "loss": 100.0309,
+      "step": 86220
+    },
+    {
+      "epoch": 0.7133225793109154,
+      "grad_norm": 556.9967651367188,
+      "learning_rate": 2.5484000192697078e-06,
+      "loss": 128.9959,
+      "step": 86230
+    },
+    {
+      "epoch": 0.7134053025602846,
+      "grad_norm": 1304.005126953125,
+      "learning_rate": 2.5471709219957573e-06,
+      "loss": 92.9146,
+      "step": 86240
+    },
+    {
+      "epoch": 0.7134880258096538,
+      "grad_norm": 667.9055786132812,
+      "learning_rate": 2.5459420198821604e-06,
+      "loss": 96.6319,
+      "step": 86250
+    },
+    {
+      "epoch": 0.7135707490590231,
+      "grad_norm": 969.6152954101562,
+      "learning_rate": 2.5447133130266937e-06,
+      "loss": 112.7888,
+      "step": 86260
+    },
+    {
+      "epoch": 0.7136534723083923,
+      "grad_norm": 1060.4730224609375,
+      "learning_rate": 2.5434848015271206e-06,
+      "loss": 96.5464,
+      "step": 86270
+    },
+    {
+      "epoch": 0.7137361955577615,
+      "grad_norm": 1293.969970703125,
+      "learning_rate": 2.542256485481188e-06,
+      "loss": 81.4303,
+      "step": 86280
+    },
+    {
+      "epoch": 0.7138189188071308,
+      "grad_norm": 737.7543334960938,
+      "learning_rate": 2.5410283649866272e-06,
+      "loss": 86.3164,
+      "step": 86290
+    },
+    {
+      "epoch": 0.7139016420565,
+      "grad_norm": 494.8547058105469,
+      "learning_rate": 2.539800440141154e-06,
+      "loss": 97.7299,
+      "step": 86300
+    },
+    {
+      "epoch": 0.7139843653058692,
+      "grad_norm": 751.9924926757812,
+      "learning_rate": 2.5385727110424697e-06,
+      "loss": 82.0479,
+      "step": 86310
+    },
+    {
+      "epoch": 0.7140670885552385,
+      "grad_norm": 854.956298828125,
+      "learning_rate": 2.5373451777882575e-06,
+      "loss": 98.6645,
+      "step": 86320
+    },
+    {
+      "epoch": 0.7141498118046077,
+      "grad_norm": 744.7928466796875,
+      "learning_rate": 2.5361178404761876e-06,
+      "loss": 81.0549,
+      "step": 86330
+    },
+    {
+      "epoch": 0.7142325350539769,
+      "grad_norm": 546.837646484375,
+      "learning_rate": 2.534890699203914e-06,
+      "loss": 85.5998,
+      "step": 86340
+    },
+    {
+      "epoch": 0.7143152583033462,
+      "grad_norm": 846.6171875,
+      "learning_rate": 2.533663754069074e-06,
+      "loss": 82.133,
+      "step": 86350
+    },
+    {
+      "epoch": 0.7143979815527154,
+      "grad_norm": 787.9846801757812,
+      "learning_rate": 2.5324370051692905e-06,
+      "loss": 96.9177,
+      "step": 86360
+    },
+    {
+      "epoch": 0.7144807048020846,
+      "grad_norm": 494.6238708496094,
+      "learning_rate": 2.5312104526021687e-06,
+      "loss": 90.6376,
+      "step": 86370
+    },
+    {
+      "epoch": 0.7145634280514539,
+      "grad_norm": 868.8283081054688,
+      "learning_rate": 2.529984096465302e-06,
+      "loss": 93.8985,
+      "step": 86380
+    },
+    {
+      "epoch": 0.7146461513008231,
+      "grad_norm": 971.2010498046875,
+      "learning_rate": 2.528757936856264e-06,
+      "loss": 89.9199,
+      "step": 86390
+    },
+    {
+      "epoch": 0.7147288745501923,
+      "grad_norm": 757.8761596679688,
+      "learning_rate": 2.527531973872617e-06,
+      "loss": 74.3093,
+      "step": 86400
+    },
+    {
+      "epoch": 0.7148115977995616,
+      "grad_norm": 610.7327270507812,
+      "learning_rate": 2.5263062076119026e-06,
+      "loss": 65.9494,
+      "step": 86410
+    },
+    {
+      "epoch": 0.7148943210489308,
+      "grad_norm": 3729.9736328125,
+      "learning_rate": 2.525080638171651e-06,
+      "loss": 91.3989,
+      "step": 86420
+    },
+    {
+      "epoch": 0.7149770442983,
+      "grad_norm": 1108.1942138671875,
+      "learning_rate": 2.5238552656493743e-06,
+      "loss": 89.1207,
+      "step": 86430
+    },
+    {
+      "epoch": 0.7150597675476693,
+      "grad_norm": 811.8157958984375,
+      "learning_rate": 2.52263009014257e-06,
+      "loss": 111.9879,
+      "step": 86440
+    },
+    {
+      "epoch": 0.7151424907970385,
+      "grad_norm": 570.6813354492188,
+      "learning_rate": 2.5214051117487205e-06,
+      "loss": 85.2475,
+      "step": 86450
+    },
+    {
+      "epoch": 0.7152252140464077,
+      "grad_norm": 888.626953125,
+      "learning_rate": 2.52018033056529e-06,
+      "loss": 82.2838,
+      "step": 86460
+    },
+    {
+      "epoch": 0.715307937295777,
+      "grad_norm": 272.4425964355469,
+      "learning_rate": 2.5189557466897306e-06,
+      "loss": 62.3443,
+      "step": 86470
+    },
+    {
+      "epoch": 0.7153906605451462,
+      "grad_norm": 869.8495483398438,
+      "learning_rate": 2.517731360219476e-06,
+      "loss": 92.5027,
+      "step": 86480
+    },
+    {
+      "epoch": 0.7154733837945154,
+      "grad_norm": 892.153564453125,
+      "learning_rate": 2.5165071712519447e-06,
+      "loss": 101.1932,
+      "step": 86490
+    },
+    {
+      "epoch": 0.7155561070438847,
+      "grad_norm": 1358.9947509765625,
+      "learning_rate": 2.51528317988454e-06,
+      "loss": 110.4798,
+      "step": 86500
+    },
+    {
+      "epoch": 0.7156388302932539,
+      "grad_norm": 894.7091674804688,
+      "learning_rate": 2.5140593862146496e-06,
+      "loss": 90.021,
+      "step": 86510
+    },
+    {
+      "epoch": 0.7157215535426231,
+      "grad_norm": 682.6785888671875,
+      "learning_rate": 2.512835790339645e-06,
+      "loss": 95.0613,
+      "step": 86520
+    },
+    {
+      "epoch": 0.7158042767919924,
+      "grad_norm": 1361.6385498046875,
+      "learning_rate": 2.5116123923568815e-06,
+      "loss": 111.4644,
+      "step": 86530
+    },
+    {
+      "epoch": 0.7158870000413616,
+      "grad_norm": 625.406982421875,
+      "learning_rate": 2.5103891923637e-06,
+      "loss": 98.9246,
+      "step": 86540
+    },
+    {
+      "epoch": 0.7159697232907308,
+      "grad_norm": 509.01702880859375,
+      "learning_rate": 2.509166190457425e-06,
+      "loss": 85.337,
+      "step": 86550
+    },
+    {
+      "epoch": 0.7160524465401001,
+      "grad_norm": 746.6972045898438,
+      "learning_rate": 2.5079433867353646e-06,
+      "loss": 92.2448,
+      "step": 86560
+    },
+    {
+      "epoch": 0.7161351697894693,
+      "grad_norm": 1138.3980712890625,
+      "learning_rate": 2.5067207812948123e-06,
+      "loss": 92.1766,
+      "step": 86570
+    },
+    {
+      "epoch": 0.7162178930388385,
+      "grad_norm": 1579.0123291015625,
+      "learning_rate": 2.505498374233044e-06,
+      "loss": 92.1637,
+      "step": 86580
+    },
+    {
+      "epoch": 0.7163006162882078,
+      "grad_norm": 728.6637573242188,
+      "learning_rate": 2.5042761656473226e-06,
+      "loss": 103.8948,
+      "step": 86590
+    },
+    {
+      "epoch": 0.716383339537577,
+      "grad_norm": 946.8383178710938,
+      "learning_rate": 2.503054155634893e-06,
+      "loss": 106.1606,
+      "step": 86600
+    },
+    {
+      "epoch": 0.7164660627869462,
+      "grad_norm": 1170.6138916015625,
+      "learning_rate": 2.5018323442929844e-06,
+      "loss": 66.9248,
+      "step": 86610
+    },
+    {
+      "epoch": 0.7165487860363156,
+      "grad_norm": 403.2840881347656,
+      "learning_rate": 2.500610731718811e-06,
+      "loss": 96.3434,
+      "step": 86620
+    },
+    {
+      "epoch": 0.7166315092856848,
+      "grad_norm": 940.978271484375,
+      "learning_rate": 2.499389318009571e-06,
+      "loss": 102.5182,
+      "step": 86630
+    },
+    {
+      "epoch": 0.716714232535054,
+      "grad_norm": 513.7085571289062,
+      "learning_rate": 2.4981681032624473e-06,
+      "loss": 78.1403,
+      "step": 86640
+    },
+    {
+      "epoch": 0.7167969557844233,
+      "grad_norm": 1063.4349365234375,
+      "learning_rate": 2.4969470875746055e-06,
+      "loss": 110.9906,
+      "step": 86650
+    },
+    {
+      "epoch": 0.7168796790337925,
+      "grad_norm": 577.010986328125,
+      "learning_rate": 2.495726271043198e-06,
+      "loss": 77.6551,
+      "step": 86660
+    },
+    {
+      "epoch": 0.7169624022831617,
+      "grad_norm": 457.9773864746094,
+      "learning_rate": 2.4945056537653545e-06,
+      "loss": 65.822,
+      "step": 86670
+    },
+    {
+      "epoch": 0.717045125532531,
+      "grad_norm": 985.6693115234375,
+      "learning_rate": 2.493285235838199e-06,
+      "loss": 76.2427,
+      "step": 86680
+    },
+    {
+      "epoch": 0.7171278487819002,
+      "grad_norm": 1075.8038330078125,
+      "learning_rate": 2.492065017358834e-06,
+      "loss": 108.8681,
+      "step": 86690
+    },
+    {
+      "epoch": 0.7172105720312694,
+      "grad_norm": 618.4483642578125,
+      "learning_rate": 2.4908449984243448e-06,
+      "loss": 81.0635,
+      "step": 86700
+    },
+    {
+      "epoch": 0.7172932952806387,
+      "grad_norm": 1000.34521484375,
+      "learning_rate": 2.4896251791318036e-06,
+      "loss": 88.2856,
+      "step": 86710
+    },
+    {
+      "epoch": 0.7173760185300079,
+      "grad_norm": 840.3250732421875,
+      "learning_rate": 2.4884055595782666e-06,
+      "loss": 76.9793,
+      "step": 86720
+    },
+    {
+      "epoch": 0.7174587417793771,
+      "grad_norm": 1622.5933837890625,
+      "learning_rate": 2.487186139860772e-06,
+      "loss": 99.8122,
+      "step": 86730
+    },
+    {
+      "epoch": 0.7175414650287464,
+      "grad_norm": 805.3701171875,
+      "learning_rate": 2.485966920076346e-06,
+      "loss": 88.3674,
+      "step": 86740
+    },
+    {
+      "epoch": 0.7176241882781156,
+      "grad_norm": 995.889892578125,
+      "learning_rate": 2.4847479003219926e-06,
+      "loss": 91.9693,
+      "step": 86750
+    },
+    {
+      "epoch": 0.7177069115274848,
+      "grad_norm": 657.5465087890625,
+      "learning_rate": 2.4835290806947047e-06,
+      "loss": 78.4104,
+      "step": 86760
+    },
+    {
+      "epoch": 0.717789634776854,
+      "grad_norm": 632.3702392578125,
+      "learning_rate": 2.4823104612914578e-06,
+      "loss": 109.4438,
+      "step": 86770
+    },
+    {
+      "epoch": 0.7178723580262233,
+      "grad_norm": 484.9309997558594,
+      "learning_rate": 2.4810920422092137e-06,
+      "loss": 92.3368,
+      "step": 86780
+    },
+    {
+      "epoch": 0.7179550812755925,
+      "grad_norm": 791.4384155273438,
+      "learning_rate": 2.4798738235449164e-06,
+      "loss": 90.1838,
+      "step": 86790
+    },
+    {
+      "epoch": 0.7180378045249617,
+      "grad_norm": 768.782470703125,
+      "learning_rate": 2.478655805395493e-06,
+      "loss": 93.8179,
+      "step": 86800
+    },
+    {
+      "epoch": 0.718120527774331,
+      "grad_norm": 603.9031982421875,
+      "learning_rate": 2.477437987857856e-06,
+      "loss": 66.2067,
+      "step": 86810
+    },
+    {
+      "epoch": 0.7182032510237002,
+      "grad_norm": 695.658447265625,
+      "learning_rate": 2.4762203710289008e-06,
+      "loss": 90.285,
+      "step": 86820
+    },
+    {
+      "epoch": 0.7182859742730694,
+      "grad_norm": 688.7564697265625,
+      "learning_rate": 2.4750029550055098e-06,
+      "loss": 104.6844,
+      "step": 86830
+    },
+    {
+      "epoch": 0.7183686975224387,
+      "grad_norm": 594.9547729492188,
+      "learning_rate": 2.473785739884544e-06,
+      "loss": 79.156,
+      "step": 86840
+    },
+    {
+      "epoch": 0.7184514207718079,
+      "grad_norm": 776.5545043945312,
+      "learning_rate": 2.4725687257628533e-06,
+      "loss": 110.7488,
+      "step": 86850
+    },
+    {
+      "epoch": 0.7185341440211771,
+      "grad_norm": 1479.708740234375,
+      "learning_rate": 2.47135191273727e-06,
+      "loss": 157.2341,
+      "step": 86860
+    },
+    {
+      "epoch": 0.7186168672705464,
+      "grad_norm": 855.310302734375,
+      "learning_rate": 2.4701353009046075e-06,
+      "loss": 98.2323,
+      "step": 86870
+    },
+    {
+      "epoch": 0.7186995905199156,
+      "grad_norm": 1696.113525390625,
+      "learning_rate": 2.4689188903616707e-06,
+      "loss": 87.6517,
+      "step": 86880
+    },
+    {
+      "epoch": 0.7187823137692848,
+      "grad_norm": 622.81787109375,
+      "learning_rate": 2.467702681205241e-06,
+      "loss": 83.5084,
+      "step": 86890
+    },
+    {
+      "epoch": 0.7188650370186541,
+      "grad_norm": 815.8912353515625,
+      "learning_rate": 2.4664866735320886e-06,
+      "loss": 93.1341,
+      "step": 86900
+    },
+    {
+      "epoch": 0.7189477602680233,
+      "grad_norm": 599.5869140625,
+      "learning_rate": 2.4652708674389636e-06,
+      "loss": 106.3732,
+      "step": 86910
+    },
+    {
+      "epoch": 0.7190304835173925,
+      "grad_norm": 917.40380859375,
+      "learning_rate": 2.464055263022605e-06,
+      "loss": 90.2012,
+      "step": 86920
+    },
+    {
+      "epoch": 0.7191132067667618,
+      "grad_norm": 765.8475952148438,
+      "learning_rate": 2.462839860379729e-06,
+      "loss": 76.7954,
+      "step": 86930
+    },
+    {
+      "epoch": 0.719195930016131,
+      "grad_norm": 1118.771240234375,
+      "learning_rate": 2.46162465960704e-06,
+      "loss": 88.3979,
+      "step": 86940
+    },
+    {
+      "epoch": 0.7192786532655002,
+      "grad_norm": 841.2666625976562,
+      "learning_rate": 2.460409660801229e-06,
+      "loss": 77.0938,
+      "step": 86950
+    },
+    {
+      "epoch": 0.7193613765148695,
+      "grad_norm": 579.8666381835938,
+      "learning_rate": 2.459194864058963e-06,
+      "loss": 81.2424,
+      "step": 86960
+    },
+    {
+      "epoch": 0.7194440997642387,
+      "grad_norm": 2329.31884765625,
+      "learning_rate": 2.457980269476903e-06,
+      "loss": 117.6933,
+      "step": 86970
+    },
+    {
+      "epoch": 0.7195268230136079,
+      "grad_norm": 662.1182250976562,
+      "learning_rate": 2.4567658771516876e-06,
+      "loss": 80.9285,
+      "step": 86980
+    },
+    {
+      "epoch": 0.7196095462629772,
+      "grad_norm": 798.1659545898438,
+      "learning_rate": 2.455551687179939e-06,
+      "loss": 75.5553,
+      "step": 86990
+    },
+    {
+      "epoch": 0.7196922695123464,
+      "grad_norm": 685.1201171875,
+      "learning_rate": 2.454337699658267e-06,
+      "loss": 100.3396,
+      "step": 87000
+    },
+    {
+      "epoch": 0.7197749927617156,
+      "grad_norm": 951.0033569335938,
+      "learning_rate": 2.453123914683259e-06,
+      "loss": 96.5214,
+      "step": 87010
+    },
+    {
+      "epoch": 0.719857716011085,
+      "grad_norm": 571.2054443359375,
+      "learning_rate": 2.4519103323514932e-06,
+      "loss": 74.3453,
+      "step": 87020
+    },
+    {
+      "epoch": 0.7199404392604541,
+      "grad_norm": 1552.3475341796875,
+      "learning_rate": 2.4506969527595277e-06,
+      "loss": 86.0868,
+      "step": 87030
+    },
+    {
+      "epoch": 0.7200231625098233,
+      "grad_norm": 716.04931640625,
+      "learning_rate": 2.4494837760039057e-06,
+      "loss": 91.2232,
+      "step": 87040
+    },
+    {
+      "epoch": 0.7201058857591927,
+      "grad_norm": 1257.593994140625,
+      "learning_rate": 2.4482708021811546e-06,
+      "loss": 86.8641,
+      "step": 87050
+    },
+    {
+      "epoch": 0.7201886090085619,
+      "grad_norm": 950.4786376953125,
+      "learning_rate": 2.4470580313877833e-06,
+      "loss": 97.9864,
+      "step": 87060
+    },
+    {
+      "epoch": 0.720271332257931,
+      "grad_norm": 1004.8903198242188,
+      "learning_rate": 2.44584546372029e-06,
+      "loss": 88.7277,
+      "step": 87070
+    },
+    {
+      "epoch": 0.7203540555073004,
+      "grad_norm": 937.1966552734375,
+      "learning_rate": 2.4446330992751504e-06,
+      "loss": 93.3525,
+      "step": 87080
+    },
+    {
+      "epoch": 0.7204367787566696,
+      "grad_norm": 994.6897583007812,
+      "learning_rate": 2.44342093814883e-06,
+      "loss": 79.5034,
+      "step": 87090
+    },
+    {
+      "epoch": 0.7205195020060388,
+      "grad_norm": 663.3712768554688,
+      "learning_rate": 2.442208980437771e-06,
+      "loss": 119.1683,
+      "step": 87100
+    },
+    {
+      "epoch": 0.7206022252554081,
+      "grad_norm": 734.4451904296875,
+      "learning_rate": 2.4409972262384037e-06,
+      "loss": 82.272,
+      "step": 87110
+    },
+    {
+      "epoch": 0.7206849485047773,
+      "grad_norm": 511.676513671875,
+      "learning_rate": 2.4397856756471435e-06,
+      "loss": 75.2757,
+      "step": 87120
+    },
+    {
+      "epoch": 0.7207676717541465,
+      "grad_norm": 1072.48046875,
+      "learning_rate": 2.438574328760387e-06,
+      "loss": 118.7081,
+      "step": 87130
+    },
+    {
+      "epoch": 0.7208503950035158,
+      "grad_norm": 1304.2197265625,
+      "learning_rate": 2.437363185674516e-06,
+      "loss": 65.1885,
+      "step": 87140
+    },
+    {
+      "epoch": 0.720933118252885,
+      "grad_norm": 734.7349243164062,
+      "learning_rate": 2.4361522464858956e-06,
+      "loss": 75.8047,
+      "step": 87150
+    },
+    {
+      "epoch": 0.7210158415022542,
+      "grad_norm": 775.7440795898438,
+      "learning_rate": 2.434941511290872e-06,
+      "loss": 84.4206,
+      "step": 87160
+    },
+    {
+      "epoch": 0.7210985647516235,
+      "grad_norm": 846.2425537109375,
+      "learning_rate": 2.4337309801857846e-06,
+      "loss": 75.5298,
+      "step": 87170
+    },
+    {
+      "epoch": 0.7211812880009927,
+      "grad_norm": 950.1725463867188,
+      "learning_rate": 2.432520653266943e-06,
+      "loss": 137.4889,
+      "step": 87180
+    },
+    {
+      "epoch": 0.7212640112503619,
+      "grad_norm": 1078.6646728515625,
+      "learning_rate": 2.4313105306306505e-06,
+      "loss": 104.9158,
+      "step": 87190
+    },
+    {
+      "epoch": 0.7213467344997312,
+      "grad_norm": 807.844970703125,
+      "learning_rate": 2.4301006123731908e-06,
+      "loss": 86.7134,
+      "step": 87200
+    },
+    {
+      "epoch": 0.7214294577491004,
+      "grad_norm": 1981.1280517578125,
+      "learning_rate": 2.4288908985908304e-06,
+      "loss": 90.906,
+      "step": 87210
+    },
+    {
+      "epoch": 0.7215121809984696,
+      "grad_norm": 573.0137939453125,
+      "learning_rate": 2.4276813893798212e-06,
+      "loss": 86.2283,
+      "step": 87220
+    },
+    {
+      "epoch": 0.7215949042478389,
+      "grad_norm": 1042.208984375,
+      "learning_rate": 2.4264720848363992e-06,
+      "loss": 79.0382,
+      "step": 87230
+    },
+    {
+      "epoch": 0.7216776274972081,
+      "grad_norm": 667.932861328125,
+      "learning_rate": 2.4252629850567823e-06,
+      "loss": 80.4661,
+      "step": 87240
+    },
+    {
+      "epoch": 0.7217603507465773,
+      "grad_norm": 869.255615234375,
+      "learning_rate": 2.4240540901371727e-06,
+      "loss": 93.8904,
+      "step": 87250
+    },
+    {
+      "epoch": 0.7218430739959466,
+      "grad_norm": 1087.067626953125,
+      "learning_rate": 2.4228454001737576e-06,
+      "loss": 70.2305,
+      "step": 87260
+    },
+    {
+      "epoch": 0.7219257972453158,
+      "grad_norm": 853.7138061523438,
+      "learning_rate": 2.421636915262707e-06,
+      "loss": 82.3333,
+      "step": 87270
+    },
+    {
+      "epoch": 0.722008520494685,
+      "grad_norm": 575.6585083007812,
+      "learning_rate": 2.420428635500173e-06,
+      "loss": 71.9151,
+      "step": 87280
+    },
+    {
+      "epoch": 0.7220912437440543,
+      "grad_norm": 846.3626708984375,
+      "learning_rate": 2.419220560982294e-06,
+      "loss": 111.3039,
+      "step": 87290
+    },
+    {
+      "epoch": 0.7221739669934235,
+      "grad_norm": 1180.4281005859375,
+      "learning_rate": 2.418012691805191e-06,
+      "loss": 96.9584,
+      "step": 87300
+    },
+    {
+      "epoch": 0.7222566902427927,
+      "grad_norm": 992.0172119140625,
+      "learning_rate": 2.4168050280649686e-06,
+      "loss": 136.7477,
+      "step": 87310
+    },
+    {
+      "epoch": 0.722339413492162,
+      "grad_norm": 1166.8099365234375,
+      "learning_rate": 2.4155975698577146e-06,
+      "loss": 88.4807,
+      "step": 87320
+    },
+    {
+      "epoch": 0.7224221367415312,
+      "grad_norm": 758.3084106445312,
+      "learning_rate": 2.4143903172795014e-06,
+      "loss": 73.7974,
+      "step": 87330
+    },
+    {
+      "epoch": 0.7225048599909004,
+      "grad_norm": 805.8015747070312,
+      "learning_rate": 2.4131832704263842e-06,
+      "loss": 91.9121,
+      "step": 87340
+    },
+    {
+      "epoch": 0.7225875832402697,
+      "grad_norm": 889.5006103515625,
+      "learning_rate": 2.411976429394402e-06,
+      "loss": 82.7861,
+      "step": 87350
+    },
+    {
+      "epoch": 0.7226703064896389,
+      "grad_norm": 859.7445068359375,
+      "learning_rate": 2.4107697942795782e-06,
+      "loss": 86.8242,
+      "step": 87360
+    },
+    {
+      "epoch": 0.7227530297390081,
+      "grad_norm": 1024.7366943359375,
+      "learning_rate": 2.4095633651779186e-06,
+      "loss": 75.2,
+      "step": 87370
+    },
+    {
+      "epoch": 0.7228357529883774,
+      "grad_norm": 732.5393676757812,
+      "learning_rate": 2.4083571421854137e-06,
+      "loss": 83.1405,
+      "step": 87380
+    },
+    {
+      "epoch": 0.7229184762377466,
+      "grad_norm": 754.288818359375,
+      "learning_rate": 2.407151125398037e-06,
+      "loss": 87.5245,
+      "step": 87390
+    },
+    {
+      "epoch": 0.7230011994871158,
+      "grad_norm": 859.1074829101562,
+      "learning_rate": 2.405945314911746e-06,
+      "loss": 90.8254,
+      "step": 87400
+    },
+    {
+      "epoch": 0.7230839227364851,
+      "grad_norm": 841.62744140625,
+      "learning_rate": 2.4047397108224807e-06,
+      "loss": 84.0488,
+      "step": 87410
+    },
+    {
+      "epoch": 0.7231666459858543,
+      "grad_norm": 1061.734375,
+      "learning_rate": 2.403534313226166e-06,
+      "loss": 95.578,
+      "step": 87420
+    },
+    {
+      "epoch": 0.7232493692352235,
+      "grad_norm": 678.4471435546875,
+      "learning_rate": 2.40232912221871e-06,
+      "loss": 119.1306,
+      "step": 87430
+    },
+    {
+      "epoch": 0.7233320924845928,
+      "grad_norm": 1203.9510498046875,
+      "learning_rate": 2.4011241378960037e-06,
+      "loss": 87.4557,
+      "step": 87440
+    },
+    {
+      "epoch": 0.723414815733962,
+      "grad_norm": 475.9383239746094,
+      "learning_rate": 2.3999193603539234e-06,
+      "loss": 83.3428,
+      "step": 87450
+    },
+    {
+      "epoch": 0.7234975389833312,
+      "grad_norm": 1285.626708984375,
+      "learning_rate": 2.3987147896883263e-06,
+      "loss": 116.1732,
+      "step": 87460
+    },
+    {
+      "epoch": 0.7235802622327006,
+      "grad_norm": 1077.05078125,
+      "learning_rate": 2.397510425995055e-06,
+      "loss": 82.0933,
+      "step": 87470
+    },
+    {
+      "epoch": 0.7236629854820698,
+      "grad_norm": 661.8814697265625,
+      "learning_rate": 2.3963062693699353e-06,
+      "loss": 78.9931,
+      "step": 87480
+    },
+    {
+      "epoch": 0.723745708731439,
+      "grad_norm": 866.9628295898438,
+      "learning_rate": 2.3951023199087763e-06,
+      "loss": 92.3042,
+      "step": 87490
+    },
+    {
+      "epoch": 0.7238284319808082,
+      "grad_norm": 463.5540771484375,
+      "learning_rate": 2.393898577707371e-06,
+      "loss": 90.7406,
+      "step": 87500
+    },
+    {
+      "epoch": 0.7239111552301775,
+      "grad_norm": 667.8403930664062,
+      "learning_rate": 2.392695042861495e-06,
+      "loss": 99.6054,
+      "step": 87510
+    },
+    {
+      "epoch": 0.7239938784795467,
+      "grad_norm": 685.7771606445312,
+      "learning_rate": 2.391491715466909e-06,
+      "loss": 75.8269,
+      "step": 87520
+    },
+    {
+      "epoch": 0.7240766017289159,
+      "grad_norm": 666.2769775390625,
+      "learning_rate": 2.390288595619356e-06,
+      "loss": 74.8925,
+      "step": 87530
+    },
+    {
+      "epoch": 0.7241593249782852,
+      "grad_norm": 1205.9483642578125,
+      "learning_rate": 2.3890856834145625e-06,
+      "loss": 104.4413,
+      "step": 87540
+    },
+    {
+      "epoch": 0.7242420482276544,
+      "grad_norm": 823.16748046875,
+      "learning_rate": 2.3878829789482385e-06,
+      "loss": 112.2172,
+      "step": 87550
+    },
+    {
+      "epoch": 0.7243247714770236,
+      "grad_norm": 675.4729614257812,
+      "learning_rate": 2.3866804823160776e-06,
+      "loss": 71.6798,
+      "step": 87560
+    },
+    {
+      "epoch": 0.7244074947263929,
+      "grad_norm": 931.8217163085938,
+      "learning_rate": 2.385478193613758e-06,
+      "loss": 102.3784,
+      "step": 87570
+    },
+    {
+      "epoch": 0.7244902179757621,
+      "grad_norm": 906.8097534179688,
+      "learning_rate": 2.3842761129369387e-06,
+      "loss": 97.9349,
+      "step": 87580
+    },
+    {
+      "epoch": 0.7245729412251313,
+      "grad_norm": 899.681884765625,
+      "learning_rate": 2.3830742403812646e-06,
+      "loss": 72.5381,
+      "step": 87590
+    },
+    {
+      "epoch": 0.7246556644745006,
+      "grad_norm": 828.892822265625,
+      "learning_rate": 2.381872576042365e-06,
+      "loss": 87.6623,
+      "step": 87600
+    },
+    {
+      "epoch": 0.7247383877238698,
+      "grad_norm": 1114.9024658203125,
+      "learning_rate": 2.3806711200158473e-06,
+      "loss": 94.0342,
+      "step": 87610
+    },
+    {
+      "epoch": 0.724821110973239,
+      "grad_norm": 899.71630859375,
+      "learning_rate": 2.3794698723973057e-06,
+      "loss": 63.5924,
+      "step": 87620
+    },
+    {
+      "epoch": 0.7249038342226083,
+      "grad_norm": 1092.0023193359375,
+      "learning_rate": 2.3782688332823212e-06,
+      "loss": 83.4462,
+      "step": 87630
+    },
+    {
+      "epoch": 0.7249865574719775,
+      "grad_norm": 982.8212280273438,
+      "learning_rate": 2.3770680027664537e-06,
+      "loss": 82.2134,
+      "step": 87640
+    },
+    {
+      "epoch": 0.7250692807213467,
+      "grad_norm": 1915.584228515625,
+      "learning_rate": 2.3758673809452484e-06,
+      "loss": 76.1468,
+      "step": 87650
+    },
+    {
+      "epoch": 0.725152003970716,
+      "grad_norm": 647.2635498046875,
+      "learning_rate": 2.3746669679142315e-06,
+      "loss": 87.5242,
+      "step": 87660
+    },
+    {
+      "epoch": 0.7252347272200852,
+      "grad_norm": 834.3574829101562,
+      "learning_rate": 2.373466763768915e-06,
+      "loss": 96.0043,
+      "step": 87670
+    },
+    {
+      "epoch": 0.7253174504694544,
+      "grad_norm": 553.946044921875,
+      "learning_rate": 2.3722667686047945e-06,
+      "loss": 73.195,
+      "step": 87680
+    },
+    {
+      "epoch": 0.7254001737188237,
+      "grad_norm": 719.322998046875,
+      "learning_rate": 2.37106698251735e-06,
+      "loss": 87.8059,
+      "step": 87690
+    },
+    {
+      "epoch": 0.7254828969681929,
+      "grad_norm": 1009.9353637695312,
+      "learning_rate": 2.3698674056020378e-06,
+      "loss": 86.6256,
+      "step": 87700
+    },
+    {
+      "epoch": 0.7255656202175621,
+      "grad_norm": 732.2600708007812,
+      "learning_rate": 2.3686680379543057e-06,
+      "loss": 105.3399,
+      "step": 87710
+    },
+    {
+      "epoch": 0.7256483434669314,
+      "grad_norm": 1122.77783203125,
+      "learning_rate": 2.36746887966958e-06,
+      "loss": 75.9752,
+      "step": 87720
+    },
+    {
+      "epoch": 0.7257310667163006,
+      "grad_norm": 834.73388671875,
+      "learning_rate": 2.366269930843275e-06,
+      "loss": 101.7426,
+      "step": 87730
+    },
+    {
+      "epoch": 0.7258137899656698,
+      "grad_norm": 1176.048583984375,
+      "learning_rate": 2.3650711915707852e-06,
+      "loss": 89.6517,
+      "step": 87740
+    },
+    {
+      "epoch": 0.7258965132150391,
+      "grad_norm": 902.6798706054688,
+      "learning_rate": 2.363872661947488e-06,
+      "loss": 94.8531,
+      "step": 87750
+    },
+    {
+      "epoch": 0.7259792364644083,
+      "grad_norm": 875.5723876953125,
+      "learning_rate": 2.362674342068744e-06,
+      "loss": 98.4001,
+      "step": 87760
+    },
+    {
+      "epoch": 0.7260619597137775,
+      "grad_norm": 1041.4979248046875,
+      "learning_rate": 2.3614762320299e-06,
+      "loss": 86.5132,
+      "step": 87770
+    },
+    {
+      "epoch": 0.7261446829631468,
+      "grad_norm": 634.3117065429688,
+      "learning_rate": 2.3602783319262847e-06,
+      "loss": 88.5851,
+      "step": 87780
+    },
+    {
+      "epoch": 0.726227406212516,
+      "grad_norm": 468.2691955566406,
+      "learning_rate": 2.3590806418532052e-06,
+      "loss": 107.173,
+      "step": 87790
+    },
+    {
+      "epoch": 0.7263101294618852,
+      "grad_norm": 743.6190185546875,
+      "learning_rate": 2.3578831619059595e-06,
+      "loss": 75.1913,
+      "step": 87800
+    },
+    {
+      "epoch": 0.7263928527112545,
+      "grad_norm": 485.16510009765625,
+      "learning_rate": 2.3566858921798246e-06,
+      "loss": 69.3648,
+      "step": 87810
+    },
+    {
+      "epoch": 0.7264755759606237,
+      "grad_norm": 559.470703125,
+      "learning_rate": 2.3554888327700604e-06,
+      "loss": 79.9045,
+      "step": 87820
+    },
+    {
+      "epoch": 0.7265582992099929,
+      "grad_norm": 720.0167236328125,
+      "learning_rate": 2.3542919837719154e-06,
+      "loss": 82.7708,
+      "step": 87830
+    },
+    {
+      "epoch": 0.7266410224593622,
+      "grad_norm": 1213.031005859375,
+      "learning_rate": 2.3530953452806143e-06,
+      "loss": 86.5757,
+      "step": 87840
+    },
+    {
+      "epoch": 0.7267237457087314,
+      "grad_norm": 729.4437866210938,
+      "learning_rate": 2.351898917391369e-06,
+      "loss": 78.6642,
+      "step": 87850
+    },
+    {
+      "epoch": 0.7268064689581006,
+      "grad_norm": 938.8148193359375,
+      "learning_rate": 2.350702700199376e-06,
+      "loss": 78.4311,
+      "step": 87860
+    },
+    {
+      "epoch": 0.72688919220747,
+      "grad_norm": 770.9730834960938,
+      "learning_rate": 2.3495066937998085e-06,
+      "loss": 83.714,
+      "step": 87870
+    },
+    {
+      "epoch": 0.7269719154568391,
+      "grad_norm": 1002.52490234375,
+      "learning_rate": 2.3483108982878294e-06,
+      "loss": 80.0576,
+      "step": 87880
+    },
+    {
+      "epoch": 0.7270546387062083,
+      "grad_norm": 594.104248046875,
+      "learning_rate": 2.3471153137585823e-06,
+      "loss": 66.6908,
+      "step": 87890
+    },
+    {
+      "epoch": 0.7271373619555777,
+      "grad_norm": 819.463134765625,
+      "learning_rate": 2.345919940307195e-06,
+      "loss": 78.4313,
+      "step": 87900
+    },
+    {
+      "epoch": 0.7272200852049469,
+      "grad_norm": 1145.914306640625,
+      "learning_rate": 2.3447247780287746e-06,
+      "loss": 127.2008,
+      "step": 87910
+    },
+    {
+      "epoch": 0.727302808454316,
+      "grad_norm": 698.8175048828125,
+      "learning_rate": 2.3435298270184204e-06,
+      "loss": 99.664,
+      "step": 87920
+    },
+    {
+      "epoch": 0.7273855317036854,
+      "grad_norm": 763.3477172851562,
+      "learning_rate": 2.3423350873712057e-06,
+      "loss": 84.8625,
+      "step": 87930
+    },
+    {
+      "epoch": 0.7274682549530546,
+      "grad_norm": 1032.08349609375,
+      "learning_rate": 2.341140559182192e-06,
+      "loss": 83.9295,
+      "step": 87940
+    },
+    {
+      "epoch": 0.7275509782024238,
+      "grad_norm": 821.6600341796875,
+      "learning_rate": 2.339946242546422e-06,
+      "loss": 84.5949,
+      "step": 87950
+    },
+    {
+      "epoch": 0.7276337014517931,
+      "grad_norm": 1196.482421875,
+      "learning_rate": 2.3387521375589205e-06,
+      "loss": 103.7103,
+      "step": 87960
+    },
+    {
+      "epoch": 0.7277164247011623,
+      "grad_norm": 939.4242553710938,
+      "learning_rate": 2.3375582443146977e-06,
+      "loss": 97.3099,
+      "step": 87970
+    },
+    {
+      "epoch": 0.7277991479505315,
+      "grad_norm": 656.9828491210938,
+      "learning_rate": 2.3363645629087467e-06,
+      "loss": 77.6787,
+      "step": 87980
+    },
+    {
+      "epoch": 0.7278818711999008,
+      "grad_norm": 467.5138854980469,
+      "learning_rate": 2.3351710934360426e-06,
+      "loss": 96.5642,
+      "step": 87990
+    },
+    {
+      "epoch": 0.72796459444927,
+      "grad_norm": 705.623046875,
+      "learning_rate": 2.333977835991545e-06,
+      "loss": 90.1867,
+      "step": 88000
+    },
+    {
+      "epoch": 0.7280473176986392,
+      "grad_norm": 1090.185302734375,
+      "learning_rate": 2.3327847906701932e-06,
+      "loss": 85.2803,
+      "step": 88010
+    },
+    {
+      "epoch": 0.7281300409480085,
+      "grad_norm": 793.4896240234375,
+      "learning_rate": 2.3315919575669172e-06,
+      "loss": 94.3953,
+      "step": 88020
+    },
+    {
+      "epoch": 0.7282127641973777,
+      "grad_norm": 584.8360595703125,
+      "learning_rate": 2.330399336776625e-06,
+      "loss": 111.1713,
+      "step": 88030
+    },
+    {
+      "epoch": 0.7282954874467469,
+      "grad_norm": 953.3636474609375,
+      "learning_rate": 2.329206928394203e-06,
+      "loss": 98.6218,
+      "step": 88040
+    },
+    {
+      "epoch": 0.7283782106961162,
+      "grad_norm": 915.4071655273438,
+      "learning_rate": 2.3280147325145285e-06,
+      "loss": 115.2945,
+      "step": 88050
+    },
+    {
+      "epoch": 0.7284609339454854,
+      "grad_norm": 500.0581970214844,
+      "learning_rate": 2.3268227492324594e-06,
+      "loss": 102.7827,
+      "step": 88060
+    },
+    {
+      "epoch": 0.7285436571948546,
+      "grad_norm": 864.3734130859375,
+      "learning_rate": 2.325630978642836e-06,
+      "loss": 87.891,
+      "step": 88070
+    },
+    {
+      "epoch": 0.7286263804442239,
+      "grad_norm": 819.2616577148438,
+      "learning_rate": 2.3244394208404816e-06,
+      "loss": 80.8149,
+      "step": 88080
+    },
+    {
+      "epoch": 0.7287091036935931,
+      "grad_norm": 1025.54443359375,
+      "learning_rate": 2.3232480759202035e-06,
+      "loss": 112.5396,
+      "step": 88090
+    },
+    {
+      "epoch": 0.7287918269429623,
+      "grad_norm": 786.0123291015625,
+      "learning_rate": 2.3220569439767907e-06,
+      "loss": 95.3773,
+      "step": 88100
+    },
+    {
+      "epoch": 0.7288745501923316,
+      "grad_norm": 1449.097900390625,
+      "learning_rate": 2.320866025105016e-06,
+      "loss": 100.4124,
+      "step": 88110
+    },
+    {
+      "epoch": 0.7289572734417008,
+      "grad_norm": 668.387939453125,
+      "learning_rate": 2.319675319399639e-06,
+      "loss": 90.4953,
+      "step": 88120
+    },
+    {
+      "epoch": 0.72903999669107,
+      "grad_norm": 753.0546264648438,
+      "learning_rate": 2.3184848269553944e-06,
+      "loss": 94.4148,
+      "step": 88130
+    },
+    {
+      "epoch": 0.7291227199404393,
+      "grad_norm": 1106.7774658203125,
+      "learning_rate": 2.3172945478670056e-06,
+      "loss": 91.0953,
+      "step": 88140
+    },
+    {
+      "epoch": 0.7292054431898085,
+      "grad_norm": 582.337646484375,
+      "learning_rate": 2.316104482229178e-06,
+      "loss": 84.3528,
+      "step": 88150
+    },
+    {
+      "epoch": 0.7292881664391777,
+      "grad_norm": 1405.619873046875,
+      "learning_rate": 2.314914630136599e-06,
+      "loss": 76.6839,
+      "step": 88160
+    },
+    {
+      "epoch": 0.729370889688547,
+      "grad_norm": 870.0715942382812,
+      "learning_rate": 2.3137249916839394e-06,
+      "loss": 72.3327,
+      "step": 88170
+    },
+    {
+      "epoch": 0.7294536129379162,
+      "grad_norm": 719.5054931640625,
+      "learning_rate": 2.3125355669658547e-06,
+      "loss": 90.0703,
+      "step": 88180
+    },
+    {
+      "epoch": 0.7295363361872854,
+      "grad_norm": 780.942138671875,
+      "learning_rate": 2.3113463560769807e-06,
+      "loss": 109.0758,
+      "step": 88190
+    },
+    {
+      "epoch": 0.7296190594366547,
+      "grad_norm": 927.2631225585938,
+      "learning_rate": 2.310157359111938e-06,
+      "loss": 97.629,
+      "step": 88200
+    },
+    {
+      "epoch": 0.7297017826860239,
+      "grad_norm": 558.8131103515625,
+      "learning_rate": 2.3089685761653296e-06,
+      "loss": 102.3254,
+      "step": 88210
+    },
+    {
+      "epoch": 0.7297845059353931,
+      "grad_norm": 982.3890380859375,
+      "learning_rate": 2.3077800073317415e-06,
+      "loss": 102.417,
+      "step": 88220
+    },
+    {
+      "epoch": 0.7298672291847623,
+      "grad_norm": 754.8167724609375,
+      "learning_rate": 2.3065916527057426e-06,
+      "loss": 88.5576,
+      "step": 88230
+    },
+    {
+      "epoch": 0.7299499524341316,
+      "grad_norm": 659.7408447265625,
+      "learning_rate": 2.305403512381884e-06,
+      "loss": 85.5432,
+      "step": 88240
+    },
+    {
+      "epoch": 0.7300326756835008,
+      "grad_norm": 838.64697265625,
+      "learning_rate": 2.3042155864547024e-06,
+      "loss": 78.8474,
+      "step": 88250
+    },
+    {
+      "epoch": 0.73011539893287,
+      "grad_norm": 750.2058715820312,
+      "learning_rate": 2.303027875018714e-06,
+      "loss": 88.2092,
+      "step": 88260
+    },
+    {
+      "epoch": 0.7301981221822393,
+      "grad_norm": 926.1990966796875,
+      "learning_rate": 2.3018403781684205e-06,
+      "loss": 79.4068,
+      "step": 88270
+    },
+    {
+      "epoch": 0.7302808454316085,
+      "grad_norm": 645.7316284179688,
+      "learning_rate": 2.3006530959983055e-06,
+      "loss": 67.1499,
+      "step": 88280
+    },
+    {
+      "epoch": 0.7303635686809777,
+      "grad_norm": 733.4593505859375,
+      "learning_rate": 2.299466028602835e-06,
+      "loss": 58.1151,
+      "step": 88290
+    },
+    {
+      "epoch": 0.730446291930347,
+      "grad_norm": 1105.638671875,
+      "learning_rate": 2.298279176076459e-06,
+      "loss": 87.0662,
+      "step": 88300
+    },
+    {
+      "epoch": 0.7305290151797162,
+      "grad_norm": 1261.8275146484375,
+      "learning_rate": 2.2970925385136093e-06,
+      "loss": 91.2717,
+      "step": 88310
+    },
+    {
+      "epoch": 0.7306117384290854,
+      "grad_norm": 857.1454467773438,
+      "learning_rate": 2.295906116008702e-06,
+      "loss": 77.8175,
+      "step": 88320
+    },
+    {
+      "epoch": 0.7306944616784548,
+      "grad_norm": 852.4544067382812,
+      "learning_rate": 2.2947199086561346e-06,
+      "loss": 92.5309,
+      "step": 88330
+    },
+    {
+      "epoch": 0.730777184927824,
+      "grad_norm": 1291.9940185546875,
+      "learning_rate": 2.293533916550289e-06,
+      "loss": 93.7867,
+      "step": 88340
+    },
+    {
+      "epoch": 0.7308599081771932,
+      "grad_norm": 805.106201171875,
+      "learning_rate": 2.292348139785528e-06,
+      "loss": 91.1033,
+      "step": 88350
+    },
+    {
+      "epoch": 0.7309426314265625,
+      "grad_norm": 1165.837890625,
+      "learning_rate": 2.2911625784562e-06,
+      "loss": 102.6606,
+      "step": 88360
+    },
+    {
+      "epoch": 0.7310253546759317,
+      "grad_norm": 1081.798095703125,
+      "learning_rate": 2.2899772326566327e-06,
+      "loss": 115.492,
+      "step": 88370
+    },
+    {
+      "epoch": 0.7311080779253009,
+      "grad_norm": 2020.0845947265625,
+      "learning_rate": 2.2887921024811405e-06,
+      "loss": 91.6865,
+      "step": 88380
+    },
+    {
+      "epoch": 0.7311908011746702,
+      "grad_norm": 739.7776489257812,
+      "learning_rate": 2.2876071880240174e-06,
+      "loss": 87.2011,
+      "step": 88390
+    },
+    {
+      "epoch": 0.7312735244240394,
+      "grad_norm": 1269.20751953125,
+      "learning_rate": 2.2864224893795423e-06,
+      "loss": 90.7345,
+      "step": 88400
+    },
+    {
+      "epoch": 0.7313562476734086,
+      "grad_norm": 859.2783203125,
+      "learning_rate": 2.285238006641976e-06,
+      "loss": 100.769,
+      "step": 88410
+    },
+    {
+      "epoch": 0.7314389709227779,
+      "grad_norm": 957.8426513671875,
+      "learning_rate": 2.284053739905563e-06,
+      "loss": 106.6209,
+      "step": 88420
+    },
+    {
+      "epoch": 0.7315216941721471,
+      "grad_norm": 1027.73095703125,
+      "learning_rate": 2.282869689264529e-06,
+      "loss": 90.6714,
+      "step": 88430
+    },
+    {
+      "epoch": 0.7316044174215163,
+      "grad_norm": 711.40576171875,
+      "learning_rate": 2.2816858548130837e-06,
+      "loss": 69.7044,
+      "step": 88440
+    },
+    {
+      "epoch": 0.7316871406708856,
+      "grad_norm": 721.7193603515625,
+      "learning_rate": 2.28050223664542e-06,
+      "loss": 75.2567,
+      "step": 88450
+    },
+    {
+      "epoch": 0.7317698639202548,
+      "grad_norm": 1569.3946533203125,
+      "learning_rate": 2.2793188348557136e-06,
+      "loss": 88.7201,
+      "step": 88460
+    },
+    {
+      "epoch": 0.731852587169624,
+      "grad_norm": 997.7343139648438,
+      "learning_rate": 2.2781356495381186e-06,
+      "loss": 142.2485,
+      "step": 88470
+    },
+    {
+      "epoch": 0.7319353104189933,
+      "grad_norm": 973.0025634765625,
+      "learning_rate": 2.276952680786779e-06,
+      "loss": 96.4867,
+      "step": 88480
+    },
+    {
+      "epoch": 0.7320180336683625,
+      "grad_norm": 626.2188110351562,
+      "learning_rate": 2.2757699286958186e-06,
+      "loss": 77.2272,
+      "step": 88490
+    },
+    {
+      "epoch": 0.7321007569177317,
+      "grad_norm": 411.275146484375,
+      "learning_rate": 2.274587393359342e-06,
+      "loss": 80.0432,
+      "step": 88500
+    },
+    {
+      "epoch": 0.732183480167101,
+      "grad_norm": 534.1046142578125,
+      "learning_rate": 2.273405074871438e-06,
+      "loss": 84.1023,
+      "step": 88510
+    },
+    {
+      "epoch": 0.7322662034164702,
+      "grad_norm": 793.4088745117188,
+      "learning_rate": 2.2722229733261795e-06,
+      "loss": 70.7006,
+      "step": 88520
+    },
+    {
+      "epoch": 0.7323489266658394,
+      "grad_norm": 710.7265014648438,
+      "learning_rate": 2.2710410888176205e-06,
+      "loss": 67.0532,
+      "step": 88530
+    },
+    {
+      "epoch": 0.7324316499152087,
+      "grad_norm": 667.3509521484375,
+      "learning_rate": 2.2698594214397966e-06,
+      "loss": 90.6114,
+      "step": 88540
+    },
+    {
+      "epoch": 0.7325143731645779,
+      "grad_norm": 981.4866943359375,
+      "learning_rate": 2.268677971286732e-06,
+      "loss": 102.2464,
+      "step": 88550
+    },
+    {
+      "epoch": 0.7325970964139471,
+      "grad_norm": 807.9349975585938,
+      "learning_rate": 2.2674967384524237e-06,
+      "loss": 76.0214,
+      "step": 88560
+    },
+    {
+      "epoch": 0.7326798196633164,
+      "grad_norm": 629.3372192382812,
+      "learning_rate": 2.2663157230308576e-06,
+      "loss": 127.0412,
+      "step": 88570
+    },
+    {
+      "epoch": 0.7327625429126856,
+      "grad_norm": 1088.6207275390625,
+      "learning_rate": 2.2651349251160055e-06,
+      "loss": 87.4096,
+      "step": 88580
+    },
+    {
+      "epoch": 0.7328452661620548,
+      "grad_norm": 692.8621215820312,
+      "learning_rate": 2.263954344801816e-06,
+      "loss": 73.1671,
+      "step": 88590
+    },
+    {
+      "epoch": 0.7329279894114241,
+      "grad_norm": 654.3468017578125,
+      "learning_rate": 2.2627739821822226e-06,
+      "loss": 78.6952,
+      "step": 88600
+    },
+    {
+      "epoch": 0.7330107126607933,
+      "grad_norm": 467.1300964355469,
+      "learning_rate": 2.261593837351141e-06,
+      "loss": 64.8719,
+      "step": 88610
+    },
+    {
+      "epoch": 0.7330934359101625,
+      "grad_norm": 724.4471435546875,
+      "learning_rate": 2.26041391040247e-06,
+      "loss": 114.528,
+      "step": 88620
+    },
+    {
+      "epoch": 0.7331761591595318,
+      "grad_norm": 960.2774658203125,
+      "learning_rate": 2.259234201430092e-06,
+      "loss": 106.1695,
+      "step": 88630
+    },
+    {
+      "epoch": 0.733258882408901,
+      "grad_norm": 810.5166625976562,
+      "learning_rate": 2.2580547105278716e-06,
+      "loss": 80.9764,
+      "step": 88640
+    },
+    {
+      "epoch": 0.7333416056582702,
+      "grad_norm": 919.3231811523438,
+      "learning_rate": 2.2568754377896516e-06,
+      "loss": 100.5872,
+      "step": 88650
+    },
+    {
+      "epoch": 0.7334243289076395,
+      "grad_norm": 684.9686889648438,
+      "learning_rate": 2.255696383309265e-06,
+      "loss": 85.3732,
+      "step": 88660
+    },
+    {
+      "epoch": 0.7335070521570087,
+      "grad_norm": 1018.3485107421875,
+      "learning_rate": 2.2545175471805197e-06,
+      "loss": 82.5631,
+      "step": 88670
+    },
+    {
+      "epoch": 0.7335897754063779,
+      "grad_norm": 658.5642700195312,
+      "learning_rate": 2.2533389294972153e-06,
+      "loss": 127.2706,
+      "step": 88680
+    },
+    {
+      "epoch": 0.7336724986557472,
+      "grad_norm": 582.38134765625,
+      "learning_rate": 2.2521605303531267e-06,
+      "loss": 81.0251,
+      "step": 88690
+    },
+    {
+      "epoch": 0.7337552219051164,
+      "grad_norm": 940.820556640625,
+      "learning_rate": 2.2509823498420142e-06,
+      "loss": 96.1019,
+      "step": 88700
+    },
+    {
+      "epoch": 0.7338379451544856,
+      "grad_norm": 1125.545654296875,
+      "learning_rate": 2.2498043880576193e-06,
+      "loss": 93.1604,
+      "step": 88710
+    },
+    {
+      "epoch": 0.733920668403855,
+      "grad_norm": 742.275634765625,
+      "learning_rate": 2.2486266450936695e-06,
+      "loss": 69.7256,
+      "step": 88720
+    },
+    {
+      "epoch": 0.7340033916532241,
+      "grad_norm": 985.7758178710938,
+      "learning_rate": 2.2474491210438687e-06,
+      "loss": 80.2849,
+      "step": 88730
+    },
+    {
+      "epoch": 0.7340861149025933,
+      "grad_norm": 607.6522216796875,
+      "learning_rate": 2.2462718160019086e-06,
+      "loss": 90.2372,
+      "step": 88740
+    },
+    {
+      "epoch": 0.7341688381519627,
+      "grad_norm": 690.2867431640625,
+      "learning_rate": 2.245094730061463e-06,
+      "loss": 67.3423,
+      "step": 88750
+    },
+    {
+      "epoch": 0.7342515614013319,
+      "grad_norm": 643.0440063476562,
+      "learning_rate": 2.2439178633161855e-06,
+      "loss": 68.1398,
+      "step": 88760
+    },
+    {
+      "epoch": 0.734334284650701,
+      "grad_norm": 737.9159545898438,
+      "learning_rate": 2.2427412158597133e-06,
+      "loss": 111.8004,
+      "step": 88770
+    },
+    {
+      "epoch": 0.7344170079000704,
+      "grad_norm": 791.5084228515625,
+      "learning_rate": 2.2415647877856706e-06,
+      "loss": 97.5052,
+      "step": 88780
+    },
+    {
+      "epoch": 0.7344997311494396,
+      "grad_norm": 746.7666015625,
+      "learning_rate": 2.240388579187658e-06,
+      "loss": 96.2792,
+      "step": 88790
+    },
+    {
+      "epoch": 0.7345824543988088,
+      "grad_norm": 920.7720947265625,
+      "learning_rate": 2.2392125901592615e-06,
+      "loss": 74.4708,
+      "step": 88800
+    },
+    {
+      "epoch": 0.7346651776481781,
+      "grad_norm": 1364.5985107421875,
+      "learning_rate": 2.23803682079405e-06,
+      "loss": 94.9622,
+      "step": 88810
+    },
+    {
+      "epoch": 0.7347479008975473,
+      "grad_norm": 917.6015625,
+      "learning_rate": 2.236861271185572e-06,
+      "loss": 80.0564,
+      "step": 88820
+    },
+    {
+      "epoch": 0.7348306241469165,
+      "grad_norm": 1057.025634765625,
+      "learning_rate": 2.2356859414273613e-06,
+      "loss": 73.4881,
+      "step": 88830
+    },
+    {
+      "epoch": 0.7349133473962858,
+      "grad_norm": 886.4762573242188,
+      "learning_rate": 2.2345108316129333e-06,
+      "loss": 103.8498,
+      "step": 88840
+    },
+    {
+      "epoch": 0.734996070645655,
+      "grad_norm": 528.6309814453125,
+      "learning_rate": 2.233335941835787e-06,
+      "loss": 77.8374,
+      "step": 88850
+    },
+    {
+      "epoch": 0.7350787938950242,
+      "grad_norm": 901.2240600585938,
+      "learning_rate": 2.232161272189401e-06,
+      "loss": 81.7305,
+      "step": 88860
+    },
+    {
+      "epoch": 0.7351615171443935,
+      "grad_norm": 1238.568359375,
+      "learning_rate": 2.230986822767241e-06,
+      "loss": 78.068,
+      "step": 88870
+    },
+    {
+      "epoch": 0.7352442403937627,
+      "grad_norm": 1001.0896606445312,
+      "learning_rate": 2.2298125936627517e-06,
+      "loss": 70.9309,
+      "step": 88880
+    },
+    {
+      "epoch": 0.7353269636431319,
+      "grad_norm": 1090.4779052734375,
+      "learning_rate": 2.228638584969363e-06,
+      "loss": 85.4851,
+      "step": 88890
+    },
+    {
+      "epoch": 0.7354096868925012,
+      "grad_norm": 472.36602783203125,
+      "learning_rate": 2.227464796780481e-06,
+      "loss": 83.7774,
+      "step": 88900
+    },
+    {
+      "epoch": 0.7354924101418704,
+      "grad_norm": 890.2085571289062,
+      "learning_rate": 2.226291229189501e-06,
+      "loss": 103.158,
+      "step": 88910
+    },
+    {
+      "epoch": 0.7355751333912396,
+      "grad_norm": 735.7196044921875,
+      "learning_rate": 2.225117882289799e-06,
+      "loss": 105.7497,
+      "step": 88920
+    },
+    {
+      "epoch": 0.7356578566406089,
+      "grad_norm": 647.6386108398438,
+      "learning_rate": 2.223944756174731e-06,
+      "loss": 79.5156,
+      "step": 88930
+    },
+    {
+      "epoch": 0.7357405798899781,
+      "grad_norm": 1861.40869140625,
+      "learning_rate": 2.2227718509376395e-06,
+      "loss": 133.0161,
+      "step": 88940
+    },
+    {
+      "epoch": 0.7358233031393473,
+      "grad_norm": 916.1025390625,
+      "learning_rate": 2.221599166671845e-06,
+      "loss": 104.4202,
+      "step": 88950
+    },
+    {
+      "epoch": 0.7359060263887165,
+      "grad_norm": 1034.66015625,
+      "learning_rate": 2.220426703470653e-06,
+      "loss": 90.1667,
+      "step": 88960
+    },
+    {
+      "epoch": 0.7359887496380858,
+      "grad_norm": 748.0284423828125,
+      "learning_rate": 2.2192544614273526e-06,
+      "loss": 58.7081,
+      "step": 88970
+    },
+    {
+      "epoch": 0.736071472887455,
+      "grad_norm": 1415.480224609375,
+      "learning_rate": 2.218082440635215e-06,
+      "loss": 77.1677,
+      "step": 88980
+    },
+    {
+      "epoch": 0.7361541961368242,
+      "grad_norm": 431.0531921386719,
+      "learning_rate": 2.216910641187488e-06,
+      "loss": 132.108,
+      "step": 88990
+    },
+    {
+      "epoch": 0.7362369193861935,
+      "grad_norm": 609.1949462890625,
+      "learning_rate": 2.215739063177409e-06,
+      "loss": 85.791,
+      "step": 89000
+    },
+    {
+      "epoch": 0.7363196426355627,
+      "grad_norm": 3827.787841796875,
+      "learning_rate": 2.2145677066981948e-06,
+      "loss": 111.6699,
+      "step": 89010
+    },
+    {
+      "epoch": 0.7364023658849319,
+      "grad_norm": 393.4620361328125,
+      "learning_rate": 2.213396571843045e-06,
+      "loss": 80.0387,
+      "step": 89020
+    },
+    {
+      "epoch": 0.7364850891343012,
+      "grad_norm": 636.8660278320312,
+      "learning_rate": 2.2122256587051404e-06,
+      "loss": 74.4118,
+      "step": 89030
+    },
+    {
+      "epoch": 0.7365678123836704,
+      "grad_norm": 783.8397216796875,
+      "learning_rate": 2.211054967377647e-06,
+      "loss": 88.0103,
+      "step": 89040
+    },
+    {
+      "epoch": 0.7366505356330396,
+      "grad_norm": 449.6473388671875,
+      "learning_rate": 2.2098844979537093e-06,
+      "loss": 105.3095,
+      "step": 89050
+    },
+    {
+      "epoch": 0.7367332588824089,
+      "grad_norm": 1187.8668212890625,
+      "learning_rate": 2.208714250526456e-06,
+      "loss": 82.4545,
+      "step": 89060
+    },
+    {
+      "epoch": 0.7368159821317781,
+      "grad_norm": 714.7520751953125,
+      "learning_rate": 2.207544225189003e-06,
+      "loss": 105.3385,
+      "step": 89070
+    },
+    {
+      "epoch": 0.7368987053811473,
+      "grad_norm": 1098.642578125,
+      "learning_rate": 2.2063744220344386e-06,
+      "loss": 87.2493,
+      "step": 89080
+    },
+    {
+      "epoch": 0.7369814286305166,
+      "grad_norm": 833.3667602539062,
+      "learning_rate": 2.2052048411558403e-06,
+      "loss": 89.0781,
+      "step": 89090
+    },
+    {
+      "epoch": 0.7370641518798858,
+      "grad_norm": 1110.5281982421875,
+      "learning_rate": 2.204035482646267e-06,
+      "loss": 79.2814,
+      "step": 89100
+    },
+    {
+      "epoch": 0.737146875129255,
+      "grad_norm": 981.9432983398438,
+      "learning_rate": 2.2028663465987576e-06,
+      "loss": 73.8698,
+      "step": 89110
+    },
+    {
+      "epoch": 0.7372295983786243,
+      "grad_norm": 894.4478149414062,
+      "learning_rate": 2.201697433106336e-06,
+      "loss": 101.7881,
+      "step": 89120
+    },
+    {
+      "epoch": 0.7373123216279935,
+      "grad_norm": 516.5750732421875,
+      "learning_rate": 2.2005287422620083e-06,
+      "loss": 82.2681,
+      "step": 89130
+    },
+    {
+      "epoch": 0.7373950448773627,
+      "grad_norm": 549.2446899414062,
+      "learning_rate": 2.19936027415876e-06,
+      "loss": 72.0674,
+      "step": 89140
+    },
+    {
+      "epoch": 0.737477768126732,
+      "grad_norm": 639.7827758789062,
+      "learning_rate": 2.1981920288895615e-06,
+      "loss": 86.3662,
+      "step": 89150
+    },
+    {
+      "epoch": 0.7375604913761012,
+      "grad_norm": 978.5205078125,
+      "learning_rate": 2.197024006547364e-06,
+      "loss": 94.2316,
+      "step": 89160
+    },
+    {
+      "epoch": 0.7376432146254704,
+      "grad_norm": 625.5220947265625,
+      "learning_rate": 2.1958562072251033e-06,
+      "loss": 97.0904,
+      "step": 89170
+    },
+    {
+      "epoch": 0.7377259378748398,
+      "grad_norm": 1031.7874755859375,
+      "learning_rate": 2.1946886310156945e-06,
+      "loss": 97.9485,
+      "step": 89180
+    },
+    {
+      "epoch": 0.737808661124209,
+      "grad_norm": 531.7047729492188,
+      "learning_rate": 2.193521278012037e-06,
+      "loss": 99.0671,
+      "step": 89190
+    },
+    {
+      "epoch": 0.7378913843735782,
+      "grad_norm": 391.3578796386719,
+      "learning_rate": 2.1923541483070114e-06,
+      "loss": 91.6598,
+      "step": 89200
+    },
+    {
+      "epoch": 0.7379741076229475,
+      "grad_norm": 606.8892211914062,
+      "learning_rate": 2.1911872419934804e-06,
+      "loss": 65.9362,
+      "step": 89210
+    },
+    {
+      "epoch": 0.7380568308723167,
+      "grad_norm": 890.9434814453125,
+      "learning_rate": 2.1900205591642904e-06,
+      "loss": 82.1377,
+      "step": 89220
+    },
+    {
+      "epoch": 0.7381395541216859,
+      "grad_norm": 1102.2276611328125,
+      "learning_rate": 2.188854099912268e-06,
+      "loss": 81.4802,
+      "step": 89230
+    },
+    {
+      "epoch": 0.7382222773710552,
+      "grad_norm": 614.1528930664062,
+      "learning_rate": 2.187687864330224e-06,
+      "loss": 88.8945,
+      "step": 89240
+    },
+    {
+      "epoch": 0.7383050006204244,
+      "grad_norm": 611.352294921875,
+      "learning_rate": 2.1865218525109496e-06,
+      "loss": 78.2255,
+      "step": 89250
+    },
+    {
+      "epoch": 0.7383877238697936,
+      "grad_norm": 934.4541015625,
+      "learning_rate": 2.185356064547219e-06,
+      "loss": 87.7939,
+      "step": 89260
+    },
+    {
+      "epoch": 0.7384704471191629,
+      "grad_norm": 1047.531494140625,
+      "learning_rate": 2.1841905005317893e-06,
+      "loss": 76.6948,
+      "step": 89270
+    },
+    {
+      "epoch": 0.7385531703685321,
+      "grad_norm": 455.1351623535156,
+      "learning_rate": 2.1830251605573984e-06,
+      "loss": 57.0161,
+      "step": 89280
+    },
+    {
+      "epoch": 0.7386358936179013,
+      "grad_norm": 902.6798095703125,
+      "learning_rate": 2.181860044716767e-06,
+      "loss": 77.3232,
+      "step": 89290
+    },
+    {
+      "epoch": 0.7387186168672706,
+      "grad_norm": 1158.1624755859375,
+      "learning_rate": 2.180695153102599e-06,
+      "loss": 94.8786,
+      "step": 89300
+    },
+    {
+      "epoch": 0.7388013401166398,
+      "grad_norm": 644.2076416015625,
+      "learning_rate": 2.179530485807578e-06,
+      "loss": 104.4438,
+      "step": 89310
+    },
+    {
+      "epoch": 0.738884063366009,
+      "grad_norm": 919.8838500976562,
+      "learning_rate": 2.1783660429243747e-06,
+      "loss": 127.025,
+      "step": 89320
+    },
+    {
+      "epoch": 0.7389667866153783,
+      "grad_norm": 1036.4383544921875,
+      "learning_rate": 2.177201824545632e-06,
+      "loss": 100.3359,
+      "step": 89330
+    },
+    {
+      "epoch": 0.7390495098647475,
+      "grad_norm": 775.053955078125,
+      "learning_rate": 2.1760378307639867e-06,
+      "loss": 101.0328,
+      "step": 89340
+    },
+    {
+      "epoch": 0.7391322331141167,
+      "grad_norm": 891.0201416015625,
+      "learning_rate": 2.174874061672051e-06,
+      "loss": 80.9924,
+      "step": 89350
+    },
+    {
+      "epoch": 0.739214956363486,
+      "grad_norm": 919.8565063476562,
+      "learning_rate": 2.173710517362421e-06,
+      "loss": 103.8065,
+      "step": 89360
+    },
+    {
+      "epoch": 0.7392976796128552,
+      "grad_norm": 565.3506469726562,
+      "learning_rate": 2.1725471979276734e-06,
+      "loss": 72.2678,
+      "step": 89370
+    },
+    {
+      "epoch": 0.7393804028622244,
+      "grad_norm": 666.4319458007812,
+      "learning_rate": 2.17138410346037e-06,
+      "loss": 75.9782,
+      "step": 89380
+    },
+    {
+      "epoch": 0.7394631261115937,
+      "grad_norm": 745.9927368164062,
+      "learning_rate": 2.1702212340530515e-06,
+      "loss": 111.4429,
+      "step": 89390
+    },
+    {
+      "epoch": 0.7395458493609629,
+      "grad_norm": 968.5650024414062,
+      "learning_rate": 2.1690585897982423e-06,
+      "loss": 100.4713,
+      "step": 89400
+    },
+    {
+      "epoch": 0.7396285726103321,
+      "grad_norm": 831.1185302734375,
+      "learning_rate": 2.167896170788451e-06,
+      "loss": 62.0955,
+      "step": 89410
+    },
+    {
+      "epoch": 0.7397112958597014,
+      "grad_norm": 1192.0394287109375,
+      "learning_rate": 2.16673397711616e-06,
+      "loss": 110.7184,
+      "step": 89420
+    },
+    {
+      "epoch": 0.7397940191090706,
+      "grad_norm": 985.3240966796875,
+      "learning_rate": 2.165572008873845e-06,
+      "loss": 95.5798,
+      "step": 89430
+    },
+    {
+      "epoch": 0.7398767423584398,
+      "grad_norm": 285.2931213378906,
+      "learning_rate": 2.1644102661539573e-06,
+      "loss": 76.2906,
+      "step": 89440
+    },
+    {
+      "epoch": 0.7399594656078091,
+      "grad_norm": 1738.4462890625,
+      "learning_rate": 2.1632487490489314e-06,
+      "loss": 90.5874,
+      "step": 89450
+    },
+    {
+      "epoch": 0.7400421888571783,
+      "grad_norm": 746.9173583984375,
+      "learning_rate": 2.162087457651183e-06,
+      "loss": 92.9749,
+      "step": 89460
+    },
+    {
+      "epoch": 0.7401249121065475,
+      "grad_norm": 530.8255004882812,
+      "learning_rate": 2.1609263920531115e-06,
+      "loss": 77.7234,
+      "step": 89470
+    },
+    {
+      "epoch": 0.7402076353559168,
+      "grad_norm": 0.0,
+      "learning_rate": 2.159765552347098e-06,
+      "loss": 63.3263,
+      "step": 89480
+    },
+    {
+      "epoch": 0.740290358605286,
+      "grad_norm": 911.8055419921875,
+      "learning_rate": 2.1586049386255036e-06,
+      "loss": 127.8942,
+      "step": 89490
+    },
+    {
+      "epoch": 0.7403730818546552,
+      "grad_norm": 723.2269897460938,
+      "learning_rate": 2.1574445509806764e-06,
+      "loss": 89.7021,
+      "step": 89500
+    },
+    {
+      "epoch": 0.7404558051040245,
+      "grad_norm": 584.854736328125,
+      "learning_rate": 2.156284389504939e-06,
+      "loss": 96.6398,
+      "step": 89510
+    },
+    {
+      "epoch": 0.7405385283533937,
+      "grad_norm": 399.6296691894531,
+      "learning_rate": 2.1551244542905995e-06,
+      "loss": 63.739,
+      "step": 89520
+    },
+    {
+      "epoch": 0.7406212516027629,
+      "grad_norm": 729.1290893554688,
+      "learning_rate": 2.1539647454299535e-06,
+      "loss": 64.4943,
+      "step": 89530
+    },
+    {
+      "epoch": 0.7407039748521322,
+      "grad_norm": 1142.3839111328125,
+      "learning_rate": 2.152805263015271e-06,
+      "loss": 98.2541,
+      "step": 89540
+    },
+    {
+      "epoch": 0.7407866981015014,
+      "grad_norm": 761.3294067382812,
+      "learning_rate": 2.1516460071388062e-06,
+      "loss": 78.2914,
+      "step": 89550
+    },
+    {
+      "epoch": 0.7408694213508706,
+      "grad_norm": 777.5975341796875,
+      "learning_rate": 2.1504869778927965e-06,
+      "loss": 86.9927,
+      "step": 89560
+    },
+    {
+      "epoch": 0.74095214460024,
+      "grad_norm": 767.6286010742188,
+      "learning_rate": 2.149328175369461e-06,
+      "loss": 74.6266,
+      "step": 89570
+    },
+    {
+      "epoch": 0.7410348678496091,
+      "grad_norm": 1057.6436767578125,
+      "learning_rate": 2.148169599661001e-06,
+      "loss": 79.5687,
+      "step": 89580
+    },
+    {
+      "epoch": 0.7411175910989783,
+      "grad_norm": 512.66015625,
+      "learning_rate": 2.147011250859597e-06,
+      "loss": 92.7999,
+      "step": 89590
+    },
+    {
+      "epoch": 0.7412003143483477,
+      "grad_norm": 878.7671508789062,
+      "learning_rate": 2.1458531290574138e-06,
+      "loss": 114.662,
+      "step": 89600
+    },
+    {
+      "epoch": 0.7412830375977169,
+      "grad_norm": 781.1104125976562,
+      "learning_rate": 2.144695234346598e-06,
+      "loss": 89.1826,
+      "step": 89610
+    },
+    {
+      "epoch": 0.741365760847086,
+      "grad_norm": 868.6851196289062,
+      "learning_rate": 2.143537566819277e-06,
+      "loss": 94.632,
+      "step": 89620
+    },
+    {
+      "epoch": 0.7414484840964554,
+      "grad_norm": 1113.6846923828125,
+      "learning_rate": 2.1423801265675643e-06,
+      "loss": 114.5532,
+      "step": 89630
+    },
+    {
+      "epoch": 0.7415312073458246,
+      "grad_norm": 987.2115478515625,
+      "learning_rate": 2.14122291368355e-06,
+      "loss": 101.5493,
+      "step": 89640
+    },
+    {
+      "epoch": 0.7416139305951938,
+      "grad_norm": 961.3772583007812,
+      "learning_rate": 2.1400659282593083e-06,
+      "loss": 84.6896,
+      "step": 89650
+    },
+    {
+      "epoch": 0.7416966538445631,
+      "grad_norm": 910.09033203125,
+      "learning_rate": 2.1389091703868954e-06,
+      "loss": 99.8653,
+      "step": 89660
+    },
+    {
+      "epoch": 0.7417793770939323,
+      "grad_norm": 1119.726318359375,
+      "learning_rate": 2.137752640158351e-06,
+      "loss": 99.0549,
+      "step": 89670
+    },
+    {
+      "epoch": 0.7418621003433015,
+      "grad_norm": 716.283447265625,
+      "learning_rate": 2.136596337665691e-06,
+      "loss": 94.7824,
+      "step": 89680
+    },
+    {
+      "epoch": 0.7419448235926707,
+      "grad_norm": 1223.6080322265625,
+      "learning_rate": 2.13544026300092e-06,
+      "loss": 125.8443,
+      "step": 89690
+    },
+    {
+      "epoch": 0.74202754684204,
+      "grad_norm": 1124.6939697265625,
+      "learning_rate": 2.13428441625602e-06,
+      "loss": 81.6936,
+      "step": 89700
+    },
+    {
+      "epoch": 0.7421102700914092,
+      "grad_norm": 617.048095703125,
+      "learning_rate": 2.1331287975229574e-06,
+      "loss": 88.8294,
+      "step": 89710
+    },
+    {
+      "epoch": 0.7421929933407784,
+      "grad_norm": 373.8196105957031,
+      "learning_rate": 2.131973406893677e-06,
+      "loss": 72.712,
+      "step": 89720
+    },
+    {
+      "epoch": 0.7422757165901477,
+      "grad_norm": 483.8746643066406,
+      "learning_rate": 2.1308182444601126e-06,
+      "loss": 92.3466,
+      "step": 89730
+    },
+    {
+      "epoch": 0.7423584398395169,
+      "grad_norm": 708.9064331054688,
+      "learning_rate": 2.1296633103141724e-06,
+      "loss": 105.1006,
+      "step": 89740
+    },
+    {
+      "epoch": 0.7424411630888861,
+      "grad_norm": 1248.171142578125,
+      "learning_rate": 2.1285086045477515e-06,
+      "loss": 82.8471,
+      "step": 89750
+    },
+    {
+      "epoch": 0.7425238863382554,
+      "grad_norm": 773.7487182617188,
+      "learning_rate": 2.12735412725272e-06,
+      "loss": 86.4886,
+      "step": 89760
+    },
+    {
+      "epoch": 0.7426066095876246,
+      "grad_norm": 810.4151000976562,
+      "learning_rate": 2.1261998785209382e-06,
+      "loss": 83.5308,
+      "step": 89770
+    },
+    {
+      "epoch": 0.7426893328369938,
+      "grad_norm": 606.5125732421875,
+      "learning_rate": 2.125045858444242e-06,
+      "loss": 60.5102,
+      "step": 89780
+    },
+    {
+      "epoch": 0.7427720560863631,
+      "grad_norm": 420.3428955078125,
+      "learning_rate": 2.1238920671144534e-06,
+      "loss": 87.7662,
+      "step": 89790
+    },
+    {
+      "epoch": 0.7428547793357323,
+      "grad_norm": 703.5516357421875,
+      "learning_rate": 2.122738504623373e-06,
+      "loss": 106.8324,
+      "step": 89800
+    },
+    {
+      "epoch": 0.7429375025851015,
+      "grad_norm": 1249.0703125,
+      "learning_rate": 2.121585171062785e-06,
+      "loss": 86.0322,
+      "step": 89810
+    },
+    {
+      "epoch": 0.7430202258344708,
+      "grad_norm": 1007.9696655273438,
+      "learning_rate": 2.1204320665244533e-06,
+      "loss": 83.4045,
+      "step": 89820
+    },
+    {
+      "epoch": 0.74310294908384,
+      "grad_norm": 926.9172973632812,
+      "learning_rate": 2.119279191100128e-06,
+      "loss": 96.8335,
+      "step": 89830
+    },
+    {
+      "epoch": 0.7431856723332092,
+      "grad_norm": 743.4456176757812,
+      "learning_rate": 2.1181265448815388e-06,
+      "loss": 71.0544,
+      "step": 89840
+    },
+    {
+      "epoch": 0.7432683955825785,
+      "grad_norm": 1535.4232177734375,
+      "learning_rate": 2.1169741279603927e-06,
+      "loss": 123.924,
+      "step": 89850
+    },
+    {
+      "epoch": 0.7433511188319477,
+      "grad_norm": 1149.1324462890625,
+      "learning_rate": 2.1158219404283836e-06,
+      "loss": 84.1763,
+      "step": 89860
+    },
+    {
+      "epoch": 0.7434338420813169,
+      "grad_norm": 632.4330444335938,
+      "learning_rate": 2.1146699823771867e-06,
+      "loss": 66.9591,
+      "step": 89870
+    },
+    {
+      "epoch": 0.7435165653306862,
+      "grad_norm": 1532.528076171875,
+      "learning_rate": 2.1135182538984565e-06,
+      "loss": 107.8088,
+      "step": 89880
+    },
+    {
+      "epoch": 0.7435992885800554,
+      "grad_norm": 1111.4703369140625,
+      "learning_rate": 2.1123667550838322e-06,
+      "loss": 78.3602,
+      "step": 89890
+    },
+    {
+      "epoch": 0.7436820118294246,
+      "grad_norm": 775.93994140625,
+      "learning_rate": 2.1112154860249327e-06,
+      "loss": 82.3048,
+      "step": 89900
+    },
+    {
+      "epoch": 0.7437647350787939,
+      "grad_norm": 840.6805419921875,
+      "learning_rate": 2.1100644468133574e-06,
+      "loss": 63.061,
+      "step": 89910
+    },
+    {
+      "epoch": 0.7438474583281631,
+      "grad_norm": 707.5786743164062,
+      "learning_rate": 2.1089136375406934e-06,
+      "loss": 85.1319,
+      "step": 89920
+    },
+    {
+      "epoch": 0.7439301815775323,
+      "grad_norm": 895.3738403320312,
+      "learning_rate": 2.107763058298504e-06,
+      "loss": 109.1324,
+      "step": 89930
+    },
+    {
+      "epoch": 0.7440129048269016,
+      "grad_norm": 349.6246643066406,
+      "learning_rate": 2.106612709178333e-06,
+      "loss": 71.1513,
+      "step": 89940
+    },
+    {
+      "epoch": 0.7440956280762708,
+      "grad_norm": 1578.1759033203125,
+      "learning_rate": 2.10546259027171e-06,
+      "loss": 89.1613,
+      "step": 89950
+    },
+    {
+      "epoch": 0.74417835132564,
+      "grad_norm": 891.1797485351562,
+      "learning_rate": 2.1043127016701442e-06,
+      "loss": 84.265,
+      "step": 89960
+    },
+    {
+      "epoch": 0.7442610745750093,
+      "grad_norm": 481.6813659667969,
+      "learning_rate": 2.1031630434651277e-06,
+      "loss": 81.8503,
+      "step": 89970
+    },
+    {
+      "epoch": 0.7443437978243785,
+      "grad_norm": 1180.74169921875,
+      "learning_rate": 2.102013615748133e-06,
+      "loss": 103.7244,
+      "step": 89980
+    },
+    {
+      "epoch": 0.7444265210737477,
+      "grad_norm": 867.1542358398438,
+      "learning_rate": 2.1008644186106146e-06,
+      "loss": 77.129,
+      "step": 89990
+    },
+    {
+      "epoch": 0.744509244323117,
+      "grad_norm": 0.0,
+      "learning_rate": 2.09971545214401e-06,
+      "loss": 64.4092,
+      "step": 90000
+    },
+    {
+      "epoch": 0.7445919675724862,
+      "grad_norm": 453.9908142089844,
+      "learning_rate": 2.0985667164397355e-06,
+      "loss": 56.7564,
+      "step": 90010
+    },
+    {
+      "epoch": 0.7446746908218554,
+      "grad_norm": 363.618896484375,
+      "learning_rate": 2.0974182115891924e-06,
+      "loss": 91.9298,
+      "step": 90020
+    },
+    {
+      "epoch": 0.7447574140712248,
+      "grad_norm": 873.480224609375,
+      "learning_rate": 2.0962699376837604e-06,
+      "loss": 74.911,
+      "step": 90030
+    },
+    {
+      "epoch": 0.744840137320594,
+      "grad_norm": 507.69085693359375,
+      "learning_rate": 2.0951218948148034e-06,
+      "loss": 99.115,
+      "step": 90040
+    },
+    {
+      "epoch": 0.7449228605699632,
+      "grad_norm": 1094.7142333984375,
+      "learning_rate": 2.093974083073666e-06,
+      "loss": 78.3589,
+      "step": 90050
+    },
+    {
+      "epoch": 0.7450055838193325,
+      "grad_norm": 914.5409545898438,
+      "learning_rate": 2.0928265025516737e-06,
+      "loss": 77.5211,
+      "step": 90060
+    },
+    {
+      "epoch": 0.7450883070687017,
+      "grad_norm": 433.937744140625,
+      "learning_rate": 2.0916791533401344e-06,
+      "loss": 117.6988,
+      "step": 90070
+    },
+    {
+      "epoch": 0.7451710303180709,
+      "grad_norm": 891.2142944335938,
+      "learning_rate": 2.090532035530337e-06,
+      "loss": 70.6285,
+      "step": 90080
+    },
+    {
+      "epoch": 0.7452537535674402,
+      "grad_norm": 684.631103515625,
+      "learning_rate": 2.0893851492135536e-06,
+      "loss": 66.5324,
+      "step": 90090
+    },
+    {
+      "epoch": 0.7453364768168094,
+      "grad_norm": 881.5846557617188,
+      "learning_rate": 2.0882384944810358e-06,
+      "loss": 112.9255,
+      "step": 90100
+    },
+    {
+      "epoch": 0.7454192000661786,
+      "grad_norm": 773.873291015625,
+      "learning_rate": 2.087092071424017e-06,
+      "loss": 105.3122,
+      "step": 90110
+    },
+    {
+      "epoch": 0.7455019233155479,
+      "grad_norm": 1046.751708984375,
+      "learning_rate": 2.085945880133715e-06,
+      "loss": 73.7329,
+      "step": 90120
+    },
+    {
+      "epoch": 0.7455846465649171,
+      "grad_norm": 880.5443725585938,
+      "learning_rate": 2.0847999207013247e-06,
+      "loss": 77.2783,
+      "step": 90130
+    },
+    {
+      "epoch": 0.7456673698142863,
+      "grad_norm": 625.60107421875,
+      "learning_rate": 2.083654193218026e-06,
+      "loss": 91.1625,
+      "step": 90140
+    },
+    {
+      "epoch": 0.7457500930636556,
+      "grad_norm": 1652.64697265625,
+      "learning_rate": 2.0825086977749793e-06,
+      "loss": 76.7295,
+      "step": 90150
+    },
+    {
+      "epoch": 0.7458328163130248,
+      "grad_norm": 902.3091430664062,
+      "learning_rate": 2.0813634344633256e-06,
+      "loss": 83.2874,
+      "step": 90160
+    },
+    {
+      "epoch": 0.745915539562394,
+      "grad_norm": 697.8639526367188,
+      "learning_rate": 2.0802184033741886e-06,
+      "loss": 94.8818,
+      "step": 90170
+    },
+    {
+      "epoch": 0.7459982628117633,
+      "grad_norm": 1140.19287109375,
+      "learning_rate": 2.0790736045986737e-06,
+      "loss": 84.9298,
+      "step": 90180
+    },
+    {
+      "epoch": 0.7460809860611325,
+      "grad_norm": 509.9717712402344,
+      "learning_rate": 2.077929038227867e-06,
+      "loss": 78.4466,
+      "step": 90190
+    },
+    {
+      "epoch": 0.7461637093105017,
+      "grad_norm": 620.2896118164062,
+      "learning_rate": 2.076784704352835e-06,
+      "loss": 67.6305,
+      "step": 90200
+    },
+    {
+      "epoch": 0.746246432559871,
+      "grad_norm": 776.4671020507812,
+      "learning_rate": 2.075640603064629e-06,
+      "loss": 123.8405,
+      "step": 90210
+    },
+    {
+      "epoch": 0.7463291558092402,
+      "grad_norm": 863.0435180664062,
+      "learning_rate": 2.07449673445428e-06,
+      "loss": 90.1876,
+      "step": 90220
+    },
+    {
+      "epoch": 0.7464118790586094,
+      "grad_norm": 1178.6463623046875,
+      "learning_rate": 2.0733530986127985e-06,
+      "loss": 101.5113,
+      "step": 90230
+    },
+    {
+      "epoch": 0.7464946023079787,
+      "grad_norm": 1176.33203125,
+      "learning_rate": 2.07220969563118e-06,
+      "loss": 90.9558,
+      "step": 90240
+    },
+    {
+      "epoch": 0.7465773255573479,
+      "grad_norm": 1168.3426513671875,
+      "learning_rate": 2.0710665256003994e-06,
+      "loss": 117.3465,
+      "step": 90250
+    },
+    {
+      "epoch": 0.7466600488067171,
+      "grad_norm": 598.4938354492188,
+      "learning_rate": 2.069923588611413e-06,
+      "loss": 88.4196,
+      "step": 90260
+    },
+    {
+      "epoch": 0.7467427720560864,
+      "grad_norm": 908.4249877929688,
+      "learning_rate": 2.068780884755161e-06,
+      "loss": 71.9552,
+      "step": 90270
+    },
+    {
+      "epoch": 0.7468254953054556,
+      "grad_norm": 610.5745849609375,
+      "learning_rate": 2.0676384141225586e-06,
+      "loss": 112.2541,
+      "step": 90280
+    },
+    {
+      "epoch": 0.7469082185548248,
+      "grad_norm": 722.8043212890625,
+      "learning_rate": 2.066496176804511e-06,
+      "loss": 77.0322,
+      "step": 90290
+    },
+    {
+      "epoch": 0.7469909418041941,
+      "grad_norm": 637.941650390625,
+      "learning_rate": 2.0653541728919002e-06,
+      "loss": 97.5822,
+      "step": 90300
+    },
+    {
+      "epoch": 0.7470736650535633,
+      "grad_norm": 1051.8499755859375,
+      "learning_rate": 2.0642124024755895e-06,
+      "loss": 78.8491,
+      "step": 90310
+    },
+    {
+      "epoch": 0.7471563883029325,
+      "grad_norm": 1052.2393798828125,
+      "learning_rate": 2.0630708656464245e-06,
+      "loss": 98.3979,
+      "step": 90320
+    },
+    {
+      "epoch": 0.7472391115523018,
+      "grad_norm": 490.3679504394531,
+      "learning_rate": 2.0619295624952318e-06,
+      "loss": 66.2834,
+      "step": 90330
+    },
+    {
+      "epoch": 0.747321834801671,
+      "grad_norm": 892.2313232421875,
+      "learning_rate": 2.0607884931128205e-06,
+      "loss": 76.3967,
+      "step": 90340
+    },
+    {
+      "epoch": 0.7474045580510402,
+      "grad_norm": 1203.5242919921875,
+      "learning_rate": 2.059647657589979e-06,
+      "loss": 84.8439,
+      "step": 90350
+    },
+    {
+      "epoch": 0.7474872813004095,
+      "grad_norm": 890.0416870117188,
+      "learning_rate": 2.0585070560174807e-06,
+      "loss": 74.8649,
+      "step": 90360
+    },
+    {
+      "epoch": 0.7475700045497787,
+      "grad_norm": 819.0018310546875,
+      "learning_rate": 2.057366688486073e-06,
+      "loss": 88.5144,
+      "step": 90370
+    },
+    {
+      "epoch": 0.7476527277991479,
+      "grad_norm": 1071.549072265625,
+      "learning_rate": 2.056226555086495e-06,
+      "loss": 79.6231,
+      "step": 90380
+    },
+    {
+      "epoch": 0.7477354510485172,
+      "grad_norm": 949.044677734375,
+      "learning_rate": 2.0550866559094597e-06,
+      "loss": 107.752,
+      "step": 90390
+    },
+    {
+      "epoch": 0.7478181742978864,
+      "grad_norm": 852.3134155273438,
+      "learning_rate": 2.053946991045664e-06,
+      "loss": 70.428,
+      "step": 90400
+    },
+    {
+      "epoch": 0.7479008975472556,
+      "grad_norm": 280.0954284667969,
+      "learning_rate": 2.0528075605857855e-06,
+      "loss": 89.7523,
+      "step": 90410
+    },
+    {
+      "epoch": 0.7479836207966248,
+      "grad_norm": 1142.2628173828125,
+      "learning_rate": 2.0516683646204836e-06,
+      "loss": 93.6377,
+      "step": 90420
+    },
+    {
+      "epoch": 0.7480663440459941,
+      "grad_norm": 561.314453125,
+      "learning_rate": 2.0505294032403987e-06,
+      "loss": 98.4227,
+      "step": 90430
+    },
+    {
+      "epoch": 0.7481490672953633,
+      "grad_norm": 538.2353515625,
+      "learning_rate": 2.0493906765361556e-06,
+      "loss": 81.2393,
+      "step": 90440
+    },
+    {
+      "epoch": 0.7482317905447325,
+      "grad_norm": 1096.204345703125,
+      "learning_rate": 2.0482521845983522e-06,
+      "loss": 92.642,
+      "step": 90450
+    },
+    {
+      "epoch": 0.7483145137941019,
+      "grad_norm": 855.7373046875,
+      "learning_rate": 2.047113927517576e-06,
+      "loss": 88.2958,
+      "step": 90460
+    },
+    {
+      "epoch": 0.748397237043471,
+      "grad_norm": 829.8861083984375,
+      "learning_rate": 2.0459759053843913e-06,
+      "loss": 84.8314,
+      "step": 90470
+    },
+    {
+      "epoch": 0.7484799602928403,
+      "grad_norm": 499.6122741699219,
+      "learning_rate": 2.0448381182893485e-06,
+      "loss": 87.0093,
+      "step": 90480
+    },
+    {
+      "epoch": 0.7485626835422096,
+      "grad_norm": 1019.7319946289062,
+      "learning_rate": 2.043700566322974e-06,
+      "loss": 91.0243,
+      "step": 90490
+    },
+    {
+      "epoch": 0.7486454067915788,
+      "grad_norm": 1175.784912109375,
+      "learning_rate": 2.0425632495757776e-06,
+      "loss": 85.5139,
+      "step": 90500
+    },
+    {
+      "epoch": 0.748728130040948,
+      "grad_norm": 1392.78759765625,
+      "learning_rate": 2.0414261681382507e-06,
+      "loss": 93.0289,
+      "step": 90510
+    },
+    {
+      "epoch": 0.7488108532903173,
+      "grad_norm": 986.0319213867188,
+      "learning_rate": 2.0402893221008657e-06,
+      "loss": 91.0319,
+      "step": 90520
+    },
+    {
+      "epoch": 0.7488935765396865,
+      "grad_norm": 1051.900146484375,
+      "learning_rate": 2.0391527115540777e-06,
+      "loss": 103.051,
+      "step": 90530
+    },
+    {
+      "epoch": 0.7489762997890557,
+      "grad_norm": 724.2633056640625,
+      "learning_rate": 2.0380163365883188e-06,
+      "loss": 93.0482,
+      "step": 90540
+    },
+    {
+      "epoch": 0.749059023038425,
+      "grad_norm": 1194.2574462890625,
+      "learning_rate": 2.0368801972940055e-06,
+      "loss": 113.0047,
+      "step": 90550
+    },
+    {
+      "epoch": 0.7491417462877942,
+      "grad_norm": 970.729736328125,
+      "learning_rate": 2.0357442937615367e-06,
+      "loss": 93.03,
+      "step": 90560
+    },
+    {
+      "epoch": 0.7492244695371634,
+      "grad_norm": 687.0914916992188,
+      "learning_rate": 2.034608626081288e-06,
+      "loss": 87.3318,
+      "step": 90570
+    },
+    {
+      "epoch": 0.7493071927865327,
+      "grad_norm": 732.1859130859375,
+      "learning_rate": 2.0334731943436235e-06,
+      "loss": 84.7605,
+      "step": 90580
+    },
+    {
+      "epoch": 0.7493899160359019,
+      "grad_norm": 1996.309814453125,
+      "learning_rate": 2.032337998638883e-06,
+      "loss": 117.3526,
+      "step": 90590
+    },
+    {
+      "epoch": 0.7494726392852711,
+      "grad_norm": 1726.9215087890625,
+      "learning_rate": 2.031203039057388e-06,
+      "loss": 91.3886,
+      "step": 90600
+    },
+    {
+      "epoch": 0.7495553625346404,
+      "grad_norm": 1001.2156372070312,
+      "learning_rate": 2.0300683156894435e-06,
+      "loss": 94.4224,
+      "step": 90610
+    },
+    {
+      "epoch": 0.7496380857840096,
+      "grad_norm": 1375.769775390625,
+      "learning_rate": 2.028933828625332e-06,
+      "loss": 69.2318,
+      "step": 90620
+    },
+    {
+      "epoch": 0.7497208090333788,
+      "grad_norm": 1265.17919921875,
+      "learning_rate": 2.0277995779553193e-06,
+      "loss": 77.7545,
+      "step": 90630
+    },
+    {
+      "epoch": 0.7498035322827481,
+      "grad_norm": 631.8235473632812,
+      "learning_rate": 2.026665563769655e-06,
+      "loss": 100.6061,
+      "step": 90640
+    },
+    {
+      "epoch": 0.7498862555321173,
+      "grad_norm": 845.4620971679688,
+      "learning_rate": 2.025531786158565e-06,
+      "loss": 123.6531,
+      "step": 90650
+    },
+    {
+      "epoch": 0.7499689787814865,
+      "grad_norm": 950.5982055664062,
+      "learning_rate": 2.02439824521226e-06,
+      "loss": 89.5225,
+      "step": 90660
+    },
+    {
+      "epoch": 0.7500517020308558,
+      "grad_norm": 1280.9688720703125,
+      "learning_rate": 2.023264941020929e-06,
+      "loss": 139.323,
+      "step": 90670
+    },
+    {
+      "epoch": 0.750134425280225,
+      "grad_norm": 989.0846557617188,
+      "learning_rate": 2.022131873674747e-06,
+      "loss": 88.5975,
+      "step": 90680
+    },
+    {
+      "epoch": 0.7502171485295942,
+      "grad_norm": 905.4840087890625,
+      "learning_rate": 2.020999043263865e-06,
+      "loss": 87.5264,
+      "step": 90690
+    },
+    {
+      "epoch": 0.7502998717789635,
+      "grad_norm": 810.5401611328125,
+      "learning_rate": 2.0198664498784194e-06,
+      "loss": 93.4054,
+      "step": 90700
+    },
+    {
+      "epoch": 0.7503825950283327,
+      "grad_norm": 836.0730590820312,
+      "learning_rate": 2.018734093608521e-06,
+      "loss": 88.2675,
+      "step": 90710
+    },
+    {
+      "epoch": 0.7504653182777019,
+      "grad_norm": 629.2024536132812,
+      "learning_rate": 2.017601974544269e-06,
+      "loss": 64.2829,
+      "step": 90720
+    },
+    {
+      "epoch": 0.7505480415270712,
+      "grad_norm": 597.7532348632812,
+      "learning_rate": 2.0164700927757407e-06,
+      "loss": 80.799,
+      "step": 90730
+    },
+    {
+      "epoch": 0.7506307647764404,
+      "grad_norm": 818.9828491210938,
+      "learning_rate": 2.0153384483929946e-06,
+      "loss": 89.5458,
+      "step": 90740
+    },
+    {
+      "epoch": 0.7507134880258096,
+      "grad_norm": 1135.253662109375,
+      "learning_rate": 2.0142070414860704e-06,
+      "loss": 113.3995,
+      "step": 90750
+    },
+    {
+      "epoch": 0.7507962112751789,
+      "grad_norm": 1244.027587890625,
+      "learning_rate": 2.0130758721449887e-06,
+      "loss": 104.453,
+      "step": 90760
+    },
+    {
+      "epoch": 0.7508789345245481,
+      "grad_norm": 731.18505859375,
+      "learning_rate": 2.01194494045975e-06,
+      "loss": 71.3096,
+      "step": 90770
+    },
+    {
+      "epoch": 0.7509616577739173,
+      "grad_norm": 613.3550415039062,
+      "learning_rate": 2.0108142465203413e-06,
+      "loss": 84.1263,
+      "step": 90780
+    },
+    {
+      "epoch": 0.7510443810232866,
+      "grad_norm": 1097.283935546875,
+      "learning_rate": 2.0096837904167252e-06,
+      "loss": 87.5751,
+      "step": 90790
+    },
+    {
+      "epoch": 0.7511271042726558,
+      "grad_norm": 1566.380615234375,
+      "learning_rate": 2.0085535722388454e-06,
+      "loss": 91.1146,
+      "step": 90800
+    },
+    {
+      "epoch": 0.751209827522025,
+      "grad_norm": 623.9690551757812,
+      "learning_rate": 2.007423592076629e-06,
+      "loss": 108.3401,
+      "step": 90810
+    },
+    {
+      "epoch": 0.7512925507713943,
+      "grad_norm": 744.1317749023438,
+      "learning_rate": 2.006293850019983e-06,
+      "loss": 98.0858,
+      "step": 90820
+    },
+    {
+      "epoch": 0.7513752740207635,
+      "grad_norm": 626.7324829101562,
+      "learning_rate": 2.005164346158796e-06,
+      "loss": 86.4027,
+      "step": 90830
+    },
+    {
+      "epoch": 0.7514579972701327,
+      "grad_norm": 633.4637451171875,
+      "learning_rate": 2.004035080582938e-06,
+      "loss": 84.082,
+      "step": 90840
+    },
+    {
+      "epoch": 0.751540720519502,
+      "grad_norm": 663.2689819335938,
+      "learning_rate": 2.002906053382258e-06,
+      "loss": 68.8682,
+      "step": 90850
+    },
+    {
+      "epoch": 0.7516234437688712,
+      "grad_norm": 998.2990112304688,
+      "learning_rate": 2.001777264646588e-06,
+      "loss": 89.2042,
+      "step": 90860
+    },
+    {
+      "epoch": 0.7517061670182404,
+      "grad_norm": 1075.44384765625,
+      "learning_rate": 2.000648714465744e-06,
+      "loss": 72.4316,
+      "step": 90870
+    },
+    {
+      "epoch": 0.7517888902676098,
+      "grad_norm": 646.1034545898438,
+      "learning_rate": 1.9995204029295147e-06,
+      "loss": 84.8476,
+      "step": 90880
+    },
+    {
+      "epoch": 0.751871613516979,
+      "grad_norm": 634.7570190429688,
+      "learning_rate": 1.9983923301276764e-06,
+      "loss": 76.6496,
+      "step": 90890
+    },
+    {
+      "epoch": 0.7519543367663482,
+      "grad_norm": 586.5011596679688,
+      "learning_rate": 1.9972644961499853e-06,
+      "loss": 68.339,
+      "step": 90900
+    },
+    {
+      "epoch": 0.7520370600157175,
+      "grad_norm": 1039.88037109375,
+      "learning_rate": 1.9961369010861777e-06,
+      "loss": 81.439,
+      "step": 90910
+    },
+    {
+      "epoch": 0.7521197832650867,
+      "grad_norm": 638.9922485351562,
+      "learning_rate": 1.995009545025971e-06,
+      "loss": 94.2341,
+      "step": 90920
+    },
+    {
+      "epoch": 0.7522025065144559,
+      "grad_norm": 710.9326171875,
+      "learning_rate": 1.9938824280590635e-06,
+      "loss": 81.596,
+      "step": 90930
+    },
+    {
+      "epoch": 0.7522852297638252,
+      "grad_norm": 1036.0745849609375,
+      "learning_rate": 1.992755550275135e-06,
+      "loss": 91.3788,
+      "step": 90940
+    },
+    {
+      "epoch": 0.7523679530131944,
+      "grad_norm": 930.1054077148438,
+      "learning_rate": 1.991628911763846e-06,
+      "loss": 102.8395,
+      "step": 90950
+    },
+    {
+      "epoch": 0.7524506762625636,
+      "grad_norm": 1212.5616455078125,
+      "learning_rate": 1.990502512614838e-06,
+      "loss": 120.0019,
+      "step": 90960
+    },
+    {
+      "epoch": 0.7525333995119329,
+      "grad_norm": 708.0130004882812,
+      "learning_rate": 1.989376352917733e-06,
+      "loss": 83.0712,
+      "step": 90970
+    },
+    {
+      "epoch": 0.7526161227613021,
+      "grad_norm": 1019.6633911132812,
+      "learning_rate": 1.988250432762135e-06,
+      "loss": 73.8178,
+      "step": 90980
+    },
+    {
+      "epoch": 0.7526988460106713,
+      "grad_norm": 602.7006225585938,
+      "learning_rate": 1.987124752237628e-06,
+      "loss": 84.8277,
+      "step": 90990
+    },
+    {
+      "epoch": 0.7527815692600406,
+      "grad_norm": 908.8017578125,
+      "learning_rate": 1.9859993114337773e-06,
+      "loss": 100.6673,
+      "step": 91000
+    },
+    {
+      "epoch": 0.7528642925094098,
+      "grad_norm": 3168.759765625,
+      "learning_rate": 1.984874110440129e-06,
+      "loss": 106.8103,
+      "step": 91010
+    },
+    {
+      "epoch": 0.752947015758779,
+      "grad_norm": 741.0905151367188,
+      "learning_rate": 1.9837491493462104e-06,
+      "loss": 112.0769,
+      "step": 91020
+    },
+    {
+      "epoch": 0.7530297390081483,
+      "grad_norm": 875.8446044921875,
+      "learning_rate": 1.9826244282415285e-06,
+      "loss": 77.1232,
+      "step": 91030
+    },
+    {
+      "epoch": 0.7531124622575175,
+      "grad_norm": 884.2897338867188,
+      "learning_rate": 1.9814999472155736e-06,
+      "loss": 76.4609,
+      "step": 91040
+    },
+    {
+      "epoch": 0.7531951855068867,
+      "grad_norm": 476.9480895996094,
+      "learning_rate": 1.9803757063578146e-06,
+      "loss": 84.1212,
+      "step": 91050
+    },
+    {
+      "epoch": 0.753277908756256,
+      "grad_norm": 548.7337646484375,
+      "learning_rate": 1.9792517057577026e-06,
+      "loss": 67.7463,
+      "step": 91060
+    },
+    {
+      "epoch": 0.7533606320056252,
+      "grad_norm": 1195.18505859375,
+      "learning_rate": 1.978127945504669e-06,
+      "loss": 96.1024,
+      "step": 91070
+    },
+    {
+      "epoch": 0.7534433552549944,
+      "grad_norm": 1155.169677734375,
+      "learning_rate": 1.977004425688126e-06,
+      "loss": 95.7761,
+      "step": 91080
+    },
+    {
+      "epoch": 0.7535260785043637,
+      "grad_norm": 591.861572265625,
+      "learning_rate": 1.9758811463974677e-06,
+      "loss": 64.2986,
+      "step": 91090
+    },
+    {
+      "epoch": 0.7536088017537329,
+      "grad_norm": 794.584228515625,
+      "learning_rate": 1.9747581077220675e-06,
+      "loss": 79.2889,
+      "step": 91100
+    },
+    {
+      "epoch": 0.7536915250031021,
+      "grad_norm": 891.2854614257812,
+      "learning_rate": 1.9736353097512802e-06,
+      "loss": 80.2451,
+      "step": 91110
+    },
+    {
+      "epoch": 0.7537742482524713,
+      "grad_norm": 474.7909851074219,
+      "learning_rate": 1.9725127525744423e-06,
+      "loss": 114.1455,
+      "step": 91120
+    },
+    {
+      "epoch": 0.7538569715018406,
+      "grad_norm": 843.3524169921875,
+      "learning_rate": 1.971390436280871e-06,
+      "loss": 97.2716,
+      "step": 91130
+    },
+    {
+      "epoch": 0.7539396947512098,
+      "grad_norm": 1710.1583251953125,
+      "learning_rate": 1.970268360959863e-06,
+      "loss": 77.38,
+      "step": 91140
+    },
+    {
+      "epoch": 0.754022418000579,
+      "grad_norm": 711.1730346679688,
+      "learning_rate": 1.9691465267006965e-06,
+      "loss": 80.5333,
+      "step": 91150
+    },
+    {
+      "epoch": 0.7541051412499483,
+      "grad_norm": 941.439697265625,
+      "learning_rate": 1.9680249335926314e-06,
+      "loss": 66.0788,
+      "step": 91160
+    },
+    {
+      "epoch": 0.7541878644993175,
+      "grad_norm": 922.4462890625,
+      "learning_rate": 1.9669035817249077e-06,
+      "loss": 88.3642,
+      "step": 91170
+    },
+    {
+      "epoch": 0.7542705877486867,
+      "grad_norm": 713.2926025390625,
+      "learning_rate": 1.9657824711867457e-06,
+      "loss": 68.5765,
+      "step": 91180
+    },
+    {
+      "epoch": 0.754353310998056,
+      "grad_norm": 617.7564697265625,
+      "learning_rate": 1.9646616020673474e-06,
+      "loss": 73.7422,
+      "step": 91190
+    },
+    {
+      "epoch": 0.7544360342474252,
+      "grad_norm": 1247.7928466796875,
+      "learning_rate": 1.9635409744558953e-06,
+      "loss": 80.8417,
+      "step": 91200
+    },
+    {
+      "epoch": 0.7545187574967944,
+      "grad_norm": 888.484375,
+      "learning_rate": 1.962420588441552e-06,
+      "loss": 80.5122,
+      "step": 91210
+    },
+    {
+      "epoch": 0.7546014807461637,
+      "grad_norm": 944.42626953125,
+      "learning_rate": 1.9613004441134635e-06,
+      "loss": 69.0574,
+      "step": 91220
+    },
+    {
+      "epoch": 0.7546842039955329,
+      "grad_norm": 731.5653076171875,
+      "learning_rate": 1.96018054156075e-06,
+      "loss": 108.5219,
+      "step": 91230
+    },
+    {
+      "epoch": 0.7547669272449021,
+      "grad_norm": 0.0,
+      "learning_rate": 1.9590608808725214e-06,
+      "loss": 79.5218,
+      "step": 91240
+    },
+    {
+      "epoch": 0.7548496504942714,
+      "grad_norm": 578.6743774414062,
+      "learning_rate": 1.9579414621378624e-06,
+      "loss": 90.3885,
+      "step": 91250
+    },
+    {
+      "epoch": 0.7549323737436406,
+      "grad_norm": 1231.1759033203125,
+      "learning_rate": 1.9568222854458403e-06,
+      "loss": 74.3492,
+      "step": 91260
+    },
+    {
+      "epoch": 0.7550150969930098,
+      "grad_norm": 1043.411376953125,
+      "learning_rate": 1.955703350885502e-06,
+      "loss": 93.9092,
+      "step": 91270
+    },
+    {
+      "epoch": 0.7550978202423791,
+      "grad_norm": 570.8914794921875,
+      "learning_rate": 1.954584658545877e-06,
+      "loss": 78.577,
+      "step": 91280
+    },
+    {
+      "epoch": 0.7551805434917483,
+      "grad_norm": 1726.863525390625,
+      "learning_rate": 1.9534662085159746e-06,
+      "loss": 119.8196,
+      "step": 91290
+    },
+    {
+      "epoch": 0.7552632667411175,
+      "grad_norm": 722.1109619140625,
+      "learning_rate": 1.9523480008847856e-06,
+      "loss": 83.1329,
+      "step": 91300
+    },
+    {
+      "epoch": 0.7553459899904869,
+      "grad_norm": 1024.208740234375,
+      "learning_rate": 1.9512300357412778e-06,
+      "loss": 82.1837,
+      "step": 91310
+    },
+    {
+      "epoch": 0.755428713239856,
+      "grad_norm": 743.8002319335938,
+      "learning_rate": 1.950112313174404e-06,
+      "loss": 95.0024,
+      "step": 91320
+    },
+    {
+      "epoch": 0.7555114364892253,
+      "grad_norm": 1056.7789306640625,
+      "learning_rate": 1.9489948332730945e-06,
+      "loss": 111.2676,
+      "step": 91330
+    },
+    {
+      "epoch": 0.7555941597385946,
+      "grad_norm": 948.68994140625,
+      "learning_rate": 1.947877596126266e-06,
+      "loss": 74.7832,
+      "step": 91340
+    },
+    {
+      "epoch": 0.7556768829879638,
+      "grad_norm": 746.3170166015625,
+      "learning_rate": 1.946760601822809e-06,
+      "loss": 77.3798,
+      "step": 91350
+    },
+    {
+      "epoch": 0.755759606237333,
+      "grad_norm": 783.4119873046875,
+      "learning_rate": 1.945643850451599e-06,
+      "loss": 91.6843,
+      "step": 91360
+    },
+    {
+      "epoch": 0.7558423294867023,
+      "grad_norm": 692.2002563476562,
+      "learning_rate": 1.9445273421014903e-06,
+      "loss": 82.1802,
+      "step": 91370
+    },
+    {
+      "epoch": 0.7559250527360715,
+      "grad_norm": 1240.4085693359375,
+      "learning_rate": 1.9434110768613184e-06,
+      "loss": 102.5016,
+      "step": 91380
+    },
+    {
+      "epoch": 0.7560077759854407,
+      "grad_norm": 1118.1903076171875,
+      "learning_rate": 1.9422950548199004e-06,
+      "loss": 82.6145,
+      "step": 91390
+    },
+    {
+      "epoch": 0.75609049923481,
+      "grad_norm": 552.0459594726562,
+      "learning_rate": 1.941179276066031e-06,
+      "loss": 90.4393,
+      "step": 91400
+    },
+    {
+      "epoch": 0.7561732224841792,
+      "grad_norm": 917.4848022460938,
+      "learning_rate": 1.9400637406884875e-06,
+      "loss": 90.3351,
+      "step": 91410
+    },
+    {
+      "epoch": 0.7562559457335484,
+      "grad_norm": 276.276611328125,
+      "learning_rate": 1.938948448776028e-06,
+      "loss": 111.6006,
+      "step": 91420
+    },
+    {
+      "epoch": 0.7563386689829177,
+      "grad_norm": 597.5177001953125,
+      "learning_rate": 1.9378334004173936e-06,
+      "loss": 78.6461,
+      "step": 91430
+    },
+    {
+      "epoch": 0.7564213922322869,
+      "grad_norm": 742.5587768554688,
+      "learning_rate": 1.9367185957013024e-06,
+      "loss": 89.7729,
+      "step": 91440
+    },
+    {
+      "epoch": 0.7565041154816561,
+      "grad_norm": 1106.80712890625,
+      "learning_rate": 1.9356040347164533e-06,
+      "loss": 76.2414,
+      "step": 91450
+    },
+    {
+      "epoch": 0.7565868387310254,
+      "grad_norm": 1360.2635498046875,
+      "learning_rate": 1.9344897175515283e-06,
+      "loss": 100.4726,
+      "step": 91460
+    },
+    {
+      "epoch": 0.7566695619803946,
+      "grad_norm": 615.8472290039062,
+      "learning_rate": 1.9333756442951886e-06,
+      "loss": 62.704,
+      "step": 91470
+    },
+    {
+      "epoch": 0.7567522852297638,
+      "grad_norm": 864.1071166992188,
+      "learning_rate": 1.9322618150360732e-06,
+      "loss": 108.7585,
+      "step": 91480
+    },
+    {
+      "epoch": 0.7568350084791331,
+      "grad_norm": 797.296142578125,
+      "learning_rate": 1.931148229862807e-06,
+      "loss": 66.7689,
+      "step": 91490
+    },
+    {
+      "epoch": 0.7569177317285023,
+      "grad_norm": 944.4542846679688,
+      "learning_rate": 1.9300348888639915e-06,
+      "loss": 90.2655,
+      "step": 91500
+    },
+    {
+      "epoch": 0.7570004549778715,
+      "grad_norm": 973.5896606445312,
+      "learning_rate": 1.9289217921282104e-06,
+      "loss": 81.9597,
+      "step": 91510
+    },
+    {
+      "epoch": 0.7570831782272408,
+      "grad_norm": 1060.939208984375,
+      "learning_rate": 1.927808939744027e-06,
+      "loss": 129.8845,
+      "step": 91520
+    },
+    {
+      "epoch": 0.75716590147661,
+      "grad_norm": 847.8220825195312,
+      "learning_rate": 1.9266963317999884e-06,
+      "loss": 77.1575,
+      "step": 91530
+    },
+    {
+      "epoch": 0.7572486247259792,
+      "grad_norm": 810.7078857421875,
+      "learning_rate": 1.9255839683846174e-06,
+      "loss": 90.8755,
+      "step": 91540
+    },
+    {
+      "epoch": 0.7573313479753485,
+      "grad_norm": 339.5457763671875,
+      "learning_rate": 1.9244718495864206e-06,
+      "loss": 58.7832,
+      "step": 91550
+    },
+    {
+      "epoch": 0.7574140712247177,
+      "grad_norm": 1120.6597900390625,
+      "learning_rate": 1.9233599754938857e-06,
+      "loss": 107.0604,
+      "step": 91560
+    },
+    {
+      "epoch": 0.7574967944740869,
+      "grad_norm": 1043.0989990234375,
+      "learning_rate": 1.922248346195477e-06,
+      "loss": 109.1071,
+      "step": 91570
+    },
+    {
+      "epoch": 0.7575795177234562,
+      "grad_norm": 790.0960693359375,
+      "learning_rate": 1.921136961779641e-06,
+      "loss": 90.9298,
+      "step": 91580
+    },
+    {
+      "epoch": 0.7576622409728254,
+      "grad_norm": 814.7366333007812,
+      "learning_rate": 1.9200258223348072e-06,
+      "loss": 87.8823,
+      "step": 91590
+    },
+    {
+      "epoch": 0.7577449642221946,
+      "grad_norm": 446.8509826660156,
+      "learning_rate": 1.918914927949384e-06,
+      "loss": 77.4138,
+      "step": 91600
+    },
+    {
+      "epoch": 0.7578276874715639,
+      "grad_norm": 1098.19091796875,
+      "learning_rate": 1.9178042787117594e-06,
+      "loss": 95.8741,
+      "step": 91610
+    },
+    {
+      "epoch": 0.7579104107209331,
+      "grad_norm": 479.6676940917969,
+      "learning_rate": 1.9166938747103013e-06,
+      "loss": 90.7424,
+      "step": 91620
+    },
+    {
+      "epoch": 0.7579931339703023,
+      "grad_norm": 925.6350708007812,
+      "learning_rate": 1.915583716033363e-06,
+      "loss": 101.3752,
+      "step": 91630
+    },
+    {
+      "epoch": 0.7580758572196716,
+      "grad_norm": 1013.79541015625,
+      "learning_rate": 1.9144738027692746e-06,
+      "loss": 101.7984,
+      "step": 91640
+    },
+    {
+      "epoch": 0.7581585804690408,
+      "grad_norm": 697.2052001953125,
+      "learning_rate": 1.913364135006343e-06,
+      "loss": 97.0362,
+      "step": 91650
+    },
+    {
+      "epoch": 0.75824130371841,
+      "grad_norm": 703.0621337890625,
+      "learning_rate": 1.9122547128328616e-06,
+      "loss": 85.1144,
+      "step": 91660
+    },
+    {
+      "epoch": 0.7583240269677793,
+      "grad_norm": 562.5040893554688,
+      "learning_rate": 1.9111455363371016e-06,
+      "loss": 90.2374,
+      "step": 91670
+    },
+    {
+      "epoch": 0.7584067502171485,
+      "grad_norm": 906.6265869140625,
+      "learning_rate": 1.910036605607316e-06,
+      "loss": 72.353,
+      "step": 91680
+    },
+    {
+      "epoch": 0.7584894734665177,
+      "grad_norm": 1488.5152587890625,
+      "learning_rate": 1.908927920731736e-06,
+      "loss": 77.0585,
+      "step": 91690
+    },
+    {
+      "epoch": 0.758572196715887,
+      "grad_norm": 573.968505859375,
+      "learning_rate": 1.9078194817985755e-06,
+      "loss": 68.6668,
+      "step": 91700
+    },
+    {
+      "epoch": 0.7586549199652562,
+      "grad_norm": 791.8029174804688,
+      "learning_rate": 1.9067112888960283e-06,
+      "loss": 73.1811,
+      "step": 91710
+    },
+    {
+      "epoch": 0.7587376432146254,
+      "grad_norm": 743.7737426757812,
+      "learning_rate": 1.905603342112265e-06,
+      "loss": 89.2683,
+      "step": 91720
+    },
+    {
+      "epoch": 0.7588203664639948,
+      "grad_norm": 863.6655883789062,
+      "learning_rate": 1.904495641535446e-06,
+      "loss": 93.7095,
+      "step": 91730
+    },
+    {
+      "epoch": 0.758903089713364,
+      "grad_norm": 533.572509765625,
+      "learning_rate": 1.9033881872537009e-06,
+      "loss": 94.9646,
+      "step": 91740
+    },
+    {
+      "epoch": 0.7589858129627332,
+      "grad_norm": 859.637939453125,
+      "learning_rate": 1.902280979355146e-06,
+      "loss": 75.9259,
+      "step": 91750
+    },
+    {
+      "epoch": 0.7590685362121025,
+      "grad_norm": 591.394287109375,
+      "learning_rate": 1.901174017927877e-06,
+      "loss": 85.4813,
+      "step": 91760
+    },
+    {
+      "epoch": 0.7591512594614717,
+      "grad_norm": 832.2922973632812,
+      "learning_rate": 1.9000673030599698e-06,
+      "loss": 91.3511,
+      "step": 91770
+    },
+    {
+      "epoch": 0.7592339827108409,
+      "grad_norm": 355.1169128417969,
+      "learning_rate": 1.89896083483948e-06,
+      "loss": 132.036,
+      "step": 91780
+    },
+    {
+      "epoch": 0.7593167059602102,
+      "grad_norm": 925.5943603515625,
+      "learning_rate": 1.897854613354445e-06,
+      "loss": 80.6945,
+      "step": 91790
+    },
+    {
+      "epoch": 0.7593994292095794,
+      "grad_norm": 1204.190185546875,
+      "learning_rate": 1.8967486386928819e-06,
+      "loss": 118.005,
+      "step": 91800
+    },
+    {
+      "epoch": 0.7594821524589486,
+      "grad_norm": 320.41064453125,
+      "learning_rate": 1.8956429109427855e-06,
+      "loss": 93.1453,
+      "step": 91810
+    },
+    {
+      "epoch": 0.7595648757083179,
+      "grad_norm": 1259.0718994140625,
+      "learning_rate": 1.8945374301921393e-06,
+      "loss": 104.1626,
+      "step": 91820
+    },
+    {
+      "epoch": 0.7596475989576871,
+      "grad_norm": 1038.4957275390625,
+      "learning_rate": 1.893432196528896e-06,
+      "loss": 71.0405,
+      "step": 91830
+    },
+    {
+      "epoch": 0.7597303222070563,
+      "grad_norm": 818.2511596679688,
+      "learning_rate": 1.892327210040995e-06,
+      "loss": 86.6824,
+      "step": 91840
+    },
+    {
+      "epoch": 0.7598130454564255,
+      "grad_norm": 662.8837280273438,
+      "learning_rate": 1.8912224708163561e-06,
+      "loss": 79.6246,
+      "step": 91850
+    },
+    {
+      "epoch": 0.7598957687057948,
+      "grad_norm": 484.0934143066406,
+      "learning_rate": 1.890117978942878e-06,
+      "loss": 66.859,
+      "step": 91860
+    },
+    {
+      "epoch": 0.759978491955164,
+      "grad_norm": 1255.5438232421875,
+      "learning_rate": 1.8890137345084392e-06,
+      "loss": 99.873,
+      "step": 91870
+    },
+    {
+      "epoch": 0.7600612152045332,
+      "grad_norm": 563.4220581054688,
+      "learning_rate": 1.8879097376009009e-06,
+      "loss": 68.9941,
+      "step": 91880
+    },
+    {
+      "epoch": 0.7601439384539025,
+      "grad_norm": 736.5455932617188,
+      "learning_rate": 1.8868059883081015e-06,
+      "loss": 69.8224,
+      "step": 91890
+    },
+    {
+      "epoch": 0.7602266617032717,
+      "grad_norm": 1109.2657470703125,
+      "learning_rate": 1.8857024867178625e-06,
+      "loss": 69.6023,
+      "step": 91900
+    },
+    {
+      "epoch": 0.7603093849526409,
+      "grad_norm": 710.873291015625,
+      "learning_rate": 1.8845992329179835e-06,
+      "loss": 75.4405,
+      "step": 91910
+    },
+    {
+      "epoch": 0.7603921082020102,
+      "grad_norm": 739.780517578125,
+      "learning_rate": 1.883496226996246e-06,
+      "loss": 71.1245,
+      "step": 91920
+    },
+    {
+      "epoch": 0.7604748314513794,
+      "grad_norm": 533.0693359375,
+      "learning_rate": 1.8823934690404106e-06,
+      "loss": 86.3577,
+      "step": 91930
+    },
+    {
+      "epoch": 0.7605575547007486,
+      "grad_norm": 1472.7764892578125,
+      "learning_rate": 1.8812909591382195e-06,
+      "loss": 94.3954,
+      "step": 91940
+    },
+    {
+      "epoch": 0.7606402779501179,
+      "grad_norm": 326.43096923828125,
+      "learning_rate": 1.8801886973773936e-06,
+      "loss": 101.5206,
+      "step": 91950
+    },
+    {
+      "epoch": 0.7607230011994871,
+      "grad_norm": 762.7881469726562,
+      "learning_rate": 1.8790866838456351e-06,
+      "loss": 72.6627,
+      "step": 91960
+    },
+    {
+      "epoch": 0.7608057244488563,
+      "grad_norm": 612.1148071289062,
+      "learning_rate": 1.877984918630626e-06,
+      "loss": 104.0263,
+      "step": 91970
+    },
+    {
+      "epoch": 0.7608884476982256,
+      "grad_norm": 1866.867919921875,
+      "learning_rate": 1.876883401820029e-06,
+      "loss": 111.1947,
+      "step": 91980
+    },
+    {
+      "epoch": 0.7609711709475948,
+      "grad_norm": 1281.453369140625,
+      "learning_rate": 1.8757821335014858e-06,
+      "loss": 104.3547,
+      "step": 91990
+    },
+    {
+      "epoch": 0.761053894196964,
+      "grad_norm": 890.40869140625,
+      "learning_rate": 1.8746811137626208e-06,
+      "loss": 82.3355,
+      "step": 92000
+    },
+    {
+      "epoch": 0.7611366174463333,
+      "grad_norm": 807.6094970703125,
+      "learning_rate": 1.8735803426910366e-06,
+      "loss": 93.4271,
+      "step": 92010
+    },
+    {
+      "epoch": 0.7612193406957025,
+      "grad_norm": 970.5708618164062,
+      "learning_rate": 1.8724798203743154e-06,
+      "loss": 91.6451,
+      "step": 92020
+    },
+    {
+      "epoch": 0.7613020639450717,
+      "grad_norm": 688.900146484375,
+      "learning_rate": 1.8713795469000218e-06,
+      "loss": 100.542,
+      "step": 92030
+    },
+    {
+      "epoch": 0.761384787194441,
+      "grad_norm": 870.4006958007812,
+      "learning_rate": 1.8702795223556992e-06,
+      "loss": 88.1513,
+      "step": 92040
+    },
+    {
+      "epoch": 0.7614675104438102,
+      "grad_norm": 558.2523803710938,
+      "learning_rate": 1.8691797468288713e-06,
+      "loss": 84.1693,
+      "step": 92050
+    },
+    {
+      "epoch": 0.7615502336931794,
+      "grad_norm": 627.4422607421875,
+      "learning_rate": 1.8680802204070432e-06,
+      "loss": 82.1566,
+      "step": 92060
+    },
+    {
+      "epoch": 0.7616329569425487,
+      "grad_norm": 792.007568359375,
+      "learning_rate": 1.8669809431776991e-06,
+      "loss": 103.5347,
+      "step": 92070
+    },
+    {
+      "epoch": 0.7617156801919179,
+      "grad_norm": 1743.6444091796875,
+      "learning_rate": 1.8658819152283003e-06,
+      "loss": 108.218,
+      "step": 92080
+    },
+    {
+      "epoch": 0.7617984034412871,
+      "grad_norm": 701.8472900390625,
+      "learning_rate": 1.8647831366462948e-06,
+      "loss": 75.0169,
+      "step": 92090
+    },
+    {
+      "epoch": 0.7618811266906564,
+      "grad_norm": 555.6596069335938,
+      "learning_rate": 1.8636846075191067e-06,
+      "loss": 91.5314,
+      "step": 92100
+    },
+    {
+      "epoch": 0.7619638499400256,
+      "grad_norm": 1169.95263671875,
+      "learning_rate": 1.8625863279341406e-06,
+      "loss": 90.8289,
+      "step": 92110
+    },
+    {
+      "epoch": 0.7620465731893948,
+      "grad_norm": 1040.67724609375,
+      "learning_rate": 1.8614882979787818e-06,
+      "loss": 75.7882,
+      "step": 92120
+    },
+    {
+      "epoch": 0.7621292964387641,
+      "grad_norm": 737.9378051757812,
+      "learning_rate": 1.8603905177403953e-06,
+      "loss": 109.6155,
+      "step": 92130
+    },
+    {
+      "epoch": 0.7622120196881333,
+      "grad_norm": 1175.499267578125,
+      "learning_rate": 1.8592929873063259e-06,
+      "loss": 77.9201,
+      "step": 92140
+    },
+    {
+      "epoch": 0.7622947429375025,
+      "grad_norm": 1110.6258544921875,
+      "learning_rate": 1.8581957067639e-06,
+      "loss": 110.9532,
+      "step": 92150
+    },
+    {
+      "epoch": 0.7623774661868719,
+      "grad_norm": 1796.15625,
+      "learning_rate": 1.8570986762004246e-06,
+      "loss": 90.3984,
+      "step": 92160
+    },
+    {
+      "epoch": 0.762460189436241,
+      "grad_norm": 778.6599731445312,
+      "learning_rate": 1.8560018957031816e-06,
+      "loss": 98.082,
+      "step": 92170
+    },
+    {
+      "epoch": 0.7625429126856103,
+      "grad_norm": 2528.74560546875,
+      "learning_rate": 1.8549053653594373e-06,
+      "loss": 116.1548,
+      "step": 92180
+    },
+    {
+      "epoch": 0.7626256359349796,
+      "grad_norm": 684.8113403320312,
+      "learning_rate": 1.8538090852564405e-06,
+      "loss": 72.5543,
+      "step": 92190
+    },
+    {
+      "epoch": 0.7627083591843488,
+      "grad_norm": 712.4026489257812,
+      "learning_rate": 1.852713055481416e-06,
+      "loss": 71.634,
+      "step": 92200
+    },
+    {
+      "epoch": 0.762791082433718,
+      "grad_norm": 856.0159301757812,
+      "learning_rate": 1.8516172761215695e-06,
+      "loss": 80.4302,
+      "step": 92210
+    },
+    {
+      "epoch": 0.7628738056830873,
+      "grad_norm": 690.3129272460938,
+      "learning_rate": 1.8505217472640868e-06,
+      "loss": 68.3549,
+      "step": 92220
+    },
+    {
+      "epoch": 0.7629565289324565,
+      "grad_norm": 787.6735229492188,
+      "learning_rate": 1.849426468996135e-06,
+      "loss": 87.8287,
+      "step": 92230
+    },
+    {
+      "epoch": 0.7630392521818257,
+      "grad_norm": 442.70745849609375,
+      "learning_rate": 1.8483314414048597e-06,
+      "loss": 102.3068,
+      "step": 92240
+    },
+    {
+      "epoch": 0.763121975431195,
+      "grad_norm": 812.5242919921875,
+      "learning_rate": 1.8472366645773892e-06,
+      "loss": 117.692,
+      "step": 92250
+    },
+    {
+      "epoch": 0.7632046986805642,
+      "grad_norm": 605.50927734375,
+      "learning_rate": 1.846142138600826e-06,
+      "loss": 69.203,
+      "step": 92260
+    },
+    {
+      "epoch": 0.7632874219299334,
+      "grad_norm": 907.07568359375,
+      "learning_rate": 1.8450478635622592e-06,
+      "loss": 139.9277,
+      "step": 92270
+    },
+    {
+      "epoch": 0.7633701451793027,
+      "grad_norm": 1189.9276123046875,
+      "learning_rate": 1.8439538395487528e-06,
+      "loss": 87.1735,
+      "step": 92280
+    },
+    {
+      "epoch": 0.7634528684286719,
+      "grad_norm": 575.260498046875,
+      "learning_rate": 1.842860066647356e-06,
+      "loss": 58.7681,
+      "step": 92290
+    },
+    {
+      "epoch": 0.7635355916780411,
+      "grad_norm": 1187.8050537109375,
+      "learning_rate": 1.841766544945095e-06,
+      "loss": 131.3646,
+      "step": 92300
+    },
+    {
+      "epoch": 0.7636183149274104,
+      "grad_norm": 811.5830078125,
+      "learning_rate": 1.8406732745289757e-06,
+      "loss": 83.9469,
+      "step": 92310
+    },
+    {
+      "epoch": 0.7637010381767796,
+      "grad_norm": 928.5191040039062,
+      "learning_rate": 1.839580255485985e-06,
+      "loss": 91.3141,
+      "step": 92320
+    },
+    {
+      "epoch": 0.7637837614261488,
+      "grad_norm": 1056.0318603515625,
+      "learning_rate": 1.83848748790309e-06,
+      "loss": 89.1773,
+      "step": 92330
+    },
+    {
+      "epoch": 0.7638664846755181,
+      "grad_norm": 1123.9605712890625,
+      "learning_rate": 1.8373949718672345e-06,
+      "loss": 61.8712,
+      "step": 92340
+    },
+    {
+      "epoch": 0.7639492079248873,
+      "grad_norm": 712.9464721679688,
+      "learning_rate": 1.8363027074653473e-06,
+      "loss": 84.6685,
+      "step": 92350
+    },
+    {
+      "epoch": 0.7640319311742565,
+      "grad_norm": 621.0189208984375,
+      "learning_rate": 1.835210694784334e-06,
+      "loss": 127.7506,
+      "step": 92360
+    },
+    {
+      "epoch": 0.7641146544236258,
+      "grad_norm": 916.6598510742188,
+      "learning_rate": 1.8341189339110793e-06,
+      "loss": 72.2045,
+      "step": 92370
+    },
+    {
+      "epoch": 0.764197377672995,
+      "grad_norm": 1478.23681640625,
+      "learning_rate": 1.8330274249324537e-06,
+      "loss": 96.8306,
+      "step": 92380
+    },
+    {
+      "epoch": 0.7642801009223642,
+      "grad_norm": 838.5194702148438,
+      "learning_rate": 1.831936167935301e-06,
+      "loss": 117.9247,
+      "step": 92390
+    },
+    {
+      "epoch": 0.7643628241717335,
+      "grad_norm": 834.4838256835938,
+      "learning_rate": 1.8308451630064484e-06,
+      "loss": 98.4779,
+      "step": 92400
+    },
+    {
+      "epoch": 0.7644455474211027,
+      "grad_norm": 1103.1943359375,
+      "learning_rate": 1.8297544102327014e-06,
+      "loss": 117.0596,
+      "step": 92410
+    },
+    {
+      "epoch": 0.7645282706704719,
+      "grad_norm": 605.728515625,
+      "learning_rate": 1.8286639097008484e-06,
+      "loss": 87.3383,
+      "step": 92420
+    },
+    {
+      "epoch": 0.7646109939198412,
+      "grad_norm": 826.8703002929688,
+      "learning_rate": 1.827573661497652e-06,
+      "loss": 79.3055,
+      "step": 92430
+    },
+    {
+      "epoch": 0.7646937171692104,
+      "grad_norm": 797.113525390625,
+      "learning_rate": 1.8264836657098595e-06,
+      "loss": 94.8799,
+      "step": 92440
+    },
+    {
+      "epoch": 0.7647764404185796,
+      "grad_norm": 1154.000244140625,
+      "learning_rate": 1.8253939224241974e-06,
+      "loss": 84.2387,
+      "step": 92450
+    },
+    {
+      "epoch": 0.7648591636679489,
+      "grad_norm": 1213.9759521484375,
+      "learning_rate": 1.8243044317273717e-06,
+      "loss": 61.1165,
+      "step": 92460
+    },
+    {
+      "epoch": 0.7649418869173181,
+      "grad_norm": 634.3380737304688,
+      "learning_rate": 1.823215193706066e-06,
+      "loss": 87.2058,
+      "step": 92470
+    },
+    {
+      "epoch": 0.7650246101666873,
+      "grad_norm": 807.1724853515625,
+      "learning_rate": 1.82212620844695e-06,
+      "loss": 88.5422,
+      "step": 92480
+    },
+    {
+      "epoch": 0.7651073334160566,
+      "grad_norm": 1034.462158203125,
+      "learning_rate": 1.8210374760366662e-06,
+      "loss": 87.1022,
+      "step": 92490
+    },
+    {
+      "epoch": 0.7651900566654258,
+      "grad_norm": 922.4021606445312,
+      "learning_rate": 1.8199489965618433e-06,
+      "loss": 101.4626,
+      "step": 92500
+    },
+    {
+      "epoch": 0.765272779914795,
+      "grad_norm": 820.8185424804688,
+      "learning_rate": 1.8188607701090827e-06,
+      "loss": 80.0996,
+      "step": 92510
+    },
+    {
+      "epoch": 0.7653555031641643,
+      "grad_norm": 701.537841796875,
+      "learning_rate": 1.8177727967649705e-06,
+      "loss": 83.0882,
+      "step": 92520
+    },
+    {
+      "epoch": 0.7654382264135335,
+      "grad_norm": 935.3859252929688,
+      "learning_rate": 1.816685076616073e-06,
+      "loss": 68.9125,
+      "step": 92530
+    },
+    {
+      "epoch": 0.7655209496629027,
+      "grad_norm": 758.0145874023438,
+      "learning_rate": 1.8155976097489342e-06,
+      "loss": 80.4841,
+      "step": 92540
+    },
+    {
+      "epoch": 0.765603672912272,
+      "grad_norm": 479.95880126953125,
+      "learning_rate": 1.8145103962500792e-06,
+      "loss": 108.154,
+      "step": 92550
+    },
+    {
+      "epoch": 0.7656863961616412,
+      "grad_norm": 1647.0076904296875,
+      "learning_rate": 1.8134234362060128e-06,
+      "loss": 97.0746,
+      "step": 92560
+    },
+    {
+      "epoch": 0.7657691194110104,
+      "grad_norm": 589.4857177734375,
+      "learning_rate": 1.8123367297032175e-06,
+      "loss": 90.745,
+      "step": 92570
+    },
+    {
+      "epoch": 0.7658518426603796,
+      "grad_norm": 579.7984008789062,
+      "learning_rate": 1.8112502768281608e-06,
+      "loss": 82.4168,
+      "step": 92580
+    },
+    {
+      "epoch": 0.765934565909749,
+      "grad_norm": 426.4167785644531,
+      "learning_rate": 1.810164077667287e-06,
+      "loss": 84.039,
+      "step": 92590
+    },
+    {
+      "epoch": 0.7660172891591182,
+      "grad_norm": 865.5641479492188,
+      "learning_rate": 1.809078132307016e-06,
+      "loss": 79.1781,
+      "step": 92600
+    },
+    {
+      "epoch": 0.7661000124084874,
+      "grad_norm": 1119.6689453125,
+      "learning_rate": 1.807992440833754e-06,
+      "loss": 72.2057,
+      "step": 92610
+    },
+    {
+      "epoch": 0.7661827356578567,
+      "grad_norm": 851.24365234375,
+      "learning_rate": 1.8069070033338842e-06,
+      "loss": 87.8442,
+      "step": 92620
+    },
+    {
+      "epoch": 0.7662654589072259,
+      "grad_norm": 1012.4996337890625,
+      "learning_rate": 1.8058218198937695e-06,
+      "loss": 104.0138,
+      "step": 92630
+    },
+    {
+      "epoch": 0.7663481821565951,
+      "grad_norm": 2512.142578125,
+      "learning_rate": 1.8047368905997536e-06,
+      "loss": 90.26,
+      "step": 92640
+    },
+    {
+      "epoch": 0.7664309054059644,
+      "grad_norm": 612.3758544921875,
+      "learning_rate": 1.8036522155381592e-06,
+      "loss": 140.536,
+      "step": 92650
+    },
+    {
+      "epoch": 0.7665136286553336,
+      "grad_norm": 825.9151611328125,
+      "learning_rate": 1.8025677947952879e-06,
+      "loss": 106.3884,
+      "step": 92660
+    },
+    {
+      "epoch": 0.7665963519047028,
+      "grad_norm": 806.4419555664062,
+      "learning_rate": 1.8014836284574223e-06,
+      "loss": 96.8457,
+      "step": 92670
+    },
+    {
+      "epoch": 0.7666790751540721,
+      "grad_norm": 366.78570556640625,
+      "learning_rate": 1.8003997166108278e-06,
+      "loss": 76.3395,
+      "step": 92680
+    },
+    {
+      "epoch": 0.7667617984034413,
+      "grad_norm": 1280.156982421875,
+      "learning_rate": 1.7993160593417424e-06,
+      "loss": 104.953,
+      "step": 92690
+    },
+    {
+      "epoch": 0.7668445216528105,
+      "grad_norm": 1448.958984375,
+      "learning_rate": 1.798232656736389e-06,
+      "loss": 86.1547,
+      "step": 92700
+    },
+    {
+      "epoch": 0.7669272449021798,
+      "grad_norm": 430.016845703125,
+      "learning_rate": 1.7971495088809688e-06,
+      "loss": 80.0066,
+      "step": 92710
+    },
+    {
+      "epoch": 0.767009968151549,
+      "grad_norm": 611.8375244140625,
+      "learning_rate": 1.796066615861663e-06,
+      "loss": 82.3525,
+      "step": 92720
+    },
+    {
+      "epoch": 0.7670926914009182,
+      "grad_norm": 163.59913635253906,
+      "learning_rate": 1.7949839777646327e-06,
+      "loss": 87.4548,
+      "step": 92730
+    },
+    {
+      "epoch": 0.7671754146502875,
+      "grad_norm": 830.4721069335938,
+      "learning_rate": 1.7939015946760186e-06,
+      "loss": 94.6959,
+      "step": 92740
+    },
+    {
+      "epoch": 0.7672581378996567,
+      "grad_norm": 602.9876098632812,
+      "learning_rate": 1.7928194666819398e-06,
+      "loss": 62.9848,
+      "step": 92750
+    },
+    {
+      "epoch": 0.7673408611490259,
+      "grad_norm": 942.3362426757812,
+      "learning_rate": 1.7917375938684979e-06,
+      "loss": 89.7564,
+      "step": 92760
+    },
+    {
+      "epoch": 0.7674235843983952,
+      "grad_norm": 1193.48486328125,
+      "learning_rate": 1.7906559763217713e-06,
+      "loss": 79.8312,
+      "step": 92770
+    },
+    {
+      "epoch": 0.7675063076477644,
+      "grad_norm": 819.6219482421875,
+      "learning_rate": 1.7895746141278198e-06,
+      "loss": 80.9784,
+      "step": 92780
+    },
+    {
+      "epoch": 0.7675890308971336,
+      "grad_norm": 716.1319580078125,
+      "learning_rate": 1.7884935073726822e-06,
+      "loss": 96.7796,
+      "step": 92790
+    },
+    {
+      "epoch": 0.7676717541465029,
+      "grad_norm": 862.4302368164062,
+      "learning_rate": 1.7874126561423771e-06,
+      "loss": 66.9127,
+      "step": 92800
+    },
+    {
+      "epoch": 0.7677544773958721,
+      "grad_norm": 323.0167236328125,
+      "learning_rate": 1.786332060522904e-06,
+      "loss": 83.0131,
+      "step": 92810
+    },
+    {
+      "epoch": 0.7678372006452413,
+      "grad_norm": 819.0519409179688,
+      "learning_rate": 1.7852517206002396e-06,
+      "loss": 93.2481,
+      "step": 92820
+    },
+    {
+      "epoch": 0.7679199238946106,
+      "grad_norm": 551.5955810546875,
+      "learning_rate": 1.7841716364603423e-06,
+      "loss": 84.1821,
+      "step": 92830
+    },
+    {
+      "epoch": 0.7680026471439798,
+      "grad_norm": 791.5858154296875,
+      "learning_rate": 1.783091808189149e-06,
+      "loss": 107.4551,
+      "step": 92840
+    },
+    {
+      "epoch": 0.768085370393349,
+      "grad_norm": 1241.2384033203125,
+      "learning_rate": 1.7820122358725772e-06,
+      "loss": 107.9027,
+      "step": 92850
+    },
+    {
+      "epoch": 0.7681680936427183,
+      "grad_norm": 390.0361633300781,
+      "learning_rate": 1.780932919596523e-06,
+      "loss": 86.1765,
+      "step": 92860
+    },
+    {
+      "epoch": 0.7682508168920875,
+      "grad_norm": 390.01922607421875,
+      "learning_rate": 1.779853859446863e-06,
+      "loss": 97.645,
+      "step": 92870
+    },
+    {
+      "epoch": 0.7683335401414567,
+      "grad_norm": 877.824462890625,
+      "learning_rate": 1.778775055509453e-06,
+      "loss": 121.0466,
+      "step": 92880
+    },
+    {
+      "epoch": 0.768416263390826,
+      "grad_norm": 679.44580078125,
+      "learning_rate": 1.777696507870128e-06,
+      "loss": 88.3246,
+      "step": 92890
+    },
+    {
+      "epoch": 0.7684989866401952,
+      "grad_norm": 1261.235595703125,
+      "learning_rate": 1.776618216614704e-06,
+      "loss": 91.5801,
+      "step": 92900
+    },
+    {
+      "epoch": 0.7685817098895644,
+      "grad_norm": 1093.8907470703125,
+      "learning_rate": 1.7755401818289748e-06,
+      "loss": 82.9775,
+      "step": 92910
+    },
+    {
+      "epoch": 0.7686644331389337,
+      "grad_norm": 567.2011108398438,
+      "learning_rate": 1.774462403598715e-06,
+      "loss": 104.8191,
+      "step": 92920
+    },
+    {
+      "epoch": 0.7687471563883029,
+      "grad_norm": 864.6884155273438,
+      "learning_rate": 1.7733848820096789e-06,
+      "loss": 72.2253,
+      "step": 92930
+    },
+    {
+      "epoch": 0.7688298796376721,
+      "grad_norm": 781.83740234375,
+      "learning_rate": 1.7723076171475995e-06,
+      "loss": 97.3354,
+      "step": 92940
+    },
+    {
+      "epoch": 0.7689126028870414,
+      "grad_norm": 622.6500854492188,
+      "learning_rate": 1.7712306090981896e-06,
+      "loss": 131.4019,
+      "step": 92950
+    },
+    {
+      "epoch": 0.7689953261364106,
+      "grad_norm": 603.605712890625,
+      "learning_rate": 1.7701538579471423e-06,
+      "loss": 64.7121,
+      "step": 92960
+    },
+    {
+      "epoch": 0.7690780493857798,
+      "grad_norm": 691.0977783203125,
+      "learning_rate": 1.7690773637801295e-06,
+      "loss": 89.4358,
+      "step": 92970
+    },
+    {
+      "epoch": 0.7691607726351491,
+      "grad_norm": 857.383056640625,
+      "learning_rate": 1.768001126682803e-06,
+      "loss": 98.0366,
+      "step": 92980
+    },
+    {
+      "epoch": 0.7692434958845183,
+      "grad_norm": 2012.945556640625,
+      "learning_rate": 1.7669251467407938e-06,
+      "loss": 76.4047,
+      "step": 92990
+    },
+    {
+      "epoch": 0.7693262191338875,
+      "grad_norm": 356.16473388671875,
+      "learning_rate": 1.7658494240397127e-06,
+      "loss": 73.393,
+      "step": 93000
+    },
+    {
+      "epoch": 0.7694089423832569,
+      "grad_norm": 853.9513549804688,
+      "learning_rate": 1.7647739586651508e-06,
+      "loss": 91.1541,
+      "step": 93010
+    },
+    {
+      "epoch": 0.769491665632626,
+      "grad_norm": 880.5347900390625,
+      "learning_rate": 1.7636987507026787e-06,
+      "loss": 117.1849,
+      "step": 93020
+    },
+    {
+      "epoch": 0.7695743888819953,
+      "grad_norm": 714.1282958984375,
+      "learning_rate": 1.762623800237841e-06,
+      "loss": 121.1683,
+      "step": 93030
+    },
+    {
+      "epoch": 0.7696571121313646,
+      "grad_norm": 2766.218994140625,
+      "learning_rate": 1.7615491073561714e-06,
+      "loss": 75.8225,
+      "step": 93040
+    },
+    {
+      "epoch": 0.7697398353807338,
+      "grad_norm": 723.0350952148438,
+      "learning_rate": 1.760474672143177e-06,
+      "loss": 83.1655,
+      "step": 93050
+    },
+    {
+      "epoch": 0.769822558630103,
+      "grad_norm": 864.4639892578125,
+      "learning_rate": 1.7594004946843458e-06,
+      "loss": 71.9456,
+      "step": 93060
+    },
+    {
+      "epoch": 0.7699052818794723,
+      "grad_norm": 1015.13330078125,
+      "learning_rate": 1.7583265750651446e-06,
+      "loss": 125.6941,
+      "step": 93070
+    },
+    {
+      "epoch": 0.7699880051288415,
+      "grad_norm": 551.4693603515625,
+      "learning_rate": 1.7572529133710204e-06,
+      "loss": 68.586,
+      "step": 93080
+    },
+    {
+      "epoch": 0.7700707283782107,
+      "grad_norm": 509.70941162109375,
+      "learning_rate": 1.7561795096874002e-06,
+      "loss": 66.0901,
+      "step": 93090
+    },
+    {
+      "epoch": 0.77015345162758,
+      "grad_norm": 1378.9339599609375,
+      "learning_rate": 1.755106364099689e-06,
+      "loss": 88.016,
+      "step": 93100
+    },
+    {
+      "epoch": 0.7702361748769492,
+      "grad_norm": 1190.9852294921875,
+      "learning_rate": 1.7540334766932738e-06,
+      "loss": 79.6336,
+      "step": 93110
+    },
+    {
+      "epoch": 0.7703188981263184,
+      "grad_norm": 805.7669677734375,
+      "learning_rate": 1.7529608475535165e-06,
+      "loss": 89.6155,
+      "step": 93120
+    },
+    {
+      "epoch": 0.7704016213756877,
+      "grad_norm": 520.67724609375,
+      "learning_rate": 1.7518884767657612e-06,
+      "loss": 62.7607,
+      "step": 93130
+    },
+    {
+      "epoch": 0.7704843446250569,
+      "grad_norm": 1119.815673828125,
+      "learning_rate": 1.7508163644153342e-06,
+      "loss": 97.3919,
+      "step": 93140
+    },
+    {
+      "epoch": 0.7705670678744261,
+      "grad_norm": 592.175048828125,
+      "learning_rate": 1.7497445105875377e-06,
+      "loss": 81.5456,
+      "step": 93150
+    },
+    {
+      "epoch": 0.7706497911237954,
+      "grad_norm": 603.5911254882812,
+      "learning_rate": 1.7486729153676536e-06,
+      "loss": 70.6593,
+      "step": 93160
+    },
+    {
+      "epoch": 0.7707325143731646,
+      "grad_norm": 762.583251953125,
+      "learning_rate": 1.7476015788409439e-06,
+      "loss": 93.7638,
+      "step": 93170
+    },
+    {
+      "epoch": 0.7708152376225338,
+      "grad_norm": 1937.641845703125,
+      "learning_rate": 1.7465305010926503e-06,
+      "loss": 100.976,
+      "step": 93180
+    },
+    {
+      "epoch": 0.7708979608719031,
+      "grad_norm": 1821.3037109375,
+      "learning_rate": 1.745459682207995e-06,
+      "loss": 98.9063,
+      "step": 93190
+    },
+    {
+      "epoch": 0.7709806841212723,
+      "grad_norm": 418.82037353515625,
+      "learning_rate": 1.7443891222721749e-06,
+      "loss": 96.1245,
+      "step": 93200
+    },
+    {
+      "epoch": 0.7710634073706415,
+      "grad_norm": 596.7721557617188,
+      "learning_rate": 1.7433188213703712e-06,
+      "loss": 83.4011,
+      "step": 93210
+    },
+    {
+      "epoch": 0.7711461306200108,
+      "grad_norm": 824.0478515625,
+      "learning_rate": 1.7422487795877424e-06,
+      "loss": 104.4605,
+      "step": 93220
+    },
+    {
+      "epoch": 0.77122885386938,
+      "grad_norm": 780.7681884765625,
+      "learning_rate": 1.7411789970094257e-06,
+      "loss": 109.6282,
+      "step": 93230
+    },
+    {
+      "epoch": 0.7713115771187492,
+      "grad_norm": 721.4907836914062,
+      "learning_rate": 1.7401094737205415e-06,
+      "loss": 127.5015,
+      "step": 93240
+    },
+    {
+      "epoch": 0.7713943003681185,
+      "grad_norm": 0.0,
+      "learning_rate": 1.739040209806186e-06,
+      "loss": 103.458,
+      "step": 93250
+    },
+    {
+      "epoch": 0.7714770236174877,
+      "grad_norm": 1141.449462890625,
+      "learning_rate": 1.7379712053514352e-06,
+      "loss": 91.0078,
+      "step": 93260
+    },
+    {
+      "epoch": 0.7715597468668569,
+      "grad_norm": 780.21337890625,
+      "learning_rate": 1.736902460441345e-06,
+      "loss": 83.8483,
+      "step": 93270
+    },
+    {
+      "epoch": 0.7716424701162262,
+      "grad_norm": 1003.1300048828125,
+      "learning_rate": 1.735833975160952e-06,
+      "loss": 91.4468,
+      "step": 93280
+    },
+    {
+      "epoch": 0.7717251933655954,
+      "grad_norm": 672.5834350585938,
+      "learning_rate": 1.7347657495952675e-06,
+      "loss": 95.7494,
+      "step": 93290
+    },
+    {
+      "epoch": 0.7718079166149646,
+      "grad_norm": 1039.33544921875,
+      "learning_rate": 1.7336977838292867e-06,
+      "loss": 83.7675,
+      "step": 93300
+    },
+    {
+      "epoch": 0.7718906398643338,
+      "grad_norm": 1989.511474609375,
+      "learning_rate": 1.7326300779479826e-06,
+      "loss": 78.2224,
+      "step": 93310
+    },
+    {
+      "epoch": 0.7719733631137031,
+      "grad_norm": 427.48193359375,
+      "learning_rate": 1.731562632036307e-06,
+      "loss": 89.7143,
+      "step": 93320
+    },
+    {
+      "epoch": 0.7720560863630723,
+      "grad_norm": 573.6722412109375,
+      "learning_rate": 1.730495446179194e-06,
+      "loss": 85.7981,
+      "step": 93330
+    },
+    {
+      "epoch": 0.7721388096124415,
+      "grad_norm": 937.4099731445312,
+      "learning_rate": 1.7294285204615536e-06,
+      "loss": 76.9562,
+      "step": 93340
+    },
+    {
+      "epoch": 0.7722215328618108,
+      "grad_norm": 931.5322265625,
+      "learning_rate": 1.7283618549682757e-06,
+      "loss": 73.166,
+      "step": 93350
+    },
+    {
+      "epoch": 0.77230425611118,
+      "grad_norm": 915.3590087890625,
+      "learning_rate": 1.727295449784232e-06,
+      "loss": 78.7844,
+      "step": 93360
+    },
+    {
+      "epoch": 0.7723869793605492,
+      "grad_norm": 1507.6265869140625,
+      "learning_rate": 1.726229304994268e-06,
+      "loss": 77.3544,
+      "step": 93370
+    },
+    {
+      "epoch": 0.7724697026099185,
+      "grad_norm": 809.1603393554688,
+      "learning_rate": 1.7251634206832135e-06,
+      "loss": 81.6114,
+      "step": 93380
+    },
+    {
+      "epoch": 0.7725524258592877,
+      "grad_norm": 410.7778015136719,
+      "learning_rate": 1.7240977969358757e-06,
+      "loss": 87.7018,
+      "step": 93390
+    },
+    {
+      "epoch": 0.7726351491086569,
+      "grad_norm": 805.246826171875,
+      "learning_rate": 1.7230324338370425e-06,
+      "loss": 80.5176,
+      "step": 93400
+    },
+    {
+      "epoch": 0.7727178723580262,
+      "grad_norm": 687.7548828125,
+      "learning_rate": 1.721967331471479e-06,
+      "loss": 79.3853,
+      "step": 93410
+    },
+    {
+      "epoch": 0.7728005956073954,
+      "grad_norm": 941.4087524414062,
+      "learning_rate": 1.7209024899239297e-06,
+      "loss": 81.4306,
+      "step": 93420
+    },
+    {
+      "epoch": 0.7728833188567646,
+      "grad_norm": 947.3975830078125,
+      "learning_rate": 1.7198379092791213e-06,
+      "loss": 79.1074,
+      "step": 93430
+    },
+    {
+      "epoch": 0.772966042106134,
+      "grad_norm": 1059.6492919921875,
+      "learning_rate": 1.7187735896217567e-06,
+      "loss": 75.3719,
+      "step": 93440
+    },
+    {
+      "epoch": 0.7730487653555032,
+      "grad_norm": 853.2528076171875,
+      "learning_rate": 1.7177095310365205e-06,
+      "loss": 69.0534,
+      "step": 93450
+    },
+    {
+      "epoch": 0.7731314886048724,
+      "grad_norm": 940.2296142578125,
+      "learning_rate": 1.7166457336080716e-06,
+      "loss": 70.2273,
+      "step": 93460
+    },
+    {
+      "epoch": 0.7732142118542417,
+      "grad_norm": 934.7996215820312,
+      "learning_rate": 1.715582197421053e-06,
+      "loss": 77.2213,
+      "step": 93470
+    },
+    {
+      "epoch": 0.7732969351036109,
+      "grad_norm": 550.6903076171875,
+      "learning_rate": 1.7145189225600856e-06,
+      "loss": 76.0427,
+      "step": 93480
+    },
+    {
+      "epoch": 0.7733796583529801,
+      "grad_norm": 1751.7384033203125,
+      "learning_rate": 1.7134559091097691e-06,
+      "loss": 95.2524,
+      "step": 93490
+    },
+    {
+      "epoch": 0.7734623816023494,
+      "grad_norm": 1242.5704345703125,
+      "learning_rate": 1.7123931571546826e-06,
+      "loss": 85.0259,
+      "step": 93500
+    },
+    {
+      "epoch": 0.7735451048517186,
+      "grad_norm": 1116.49365234375,
+      "learning_rate": 1.711330666779385e-06,
+      "loss": 93.0945,
+      "step": 93510
+    },
+    {
+      "epoch": 0.7736278281010878,
+      "grad_norm": 1309.2081298828125,
+      "learning_rate": 1.7102684380684109e-06,
+      "loss": 102.1733,
+      "step": 93520
+    },
+    {
+      "epoch": 0.7737105513504571,
+      "grad_norm": 610.8143920898438,
+      "learning_rate": 1.7092064711062816e-06,
+      "loss": 83.7248,
+      "step": 93530
+    },
+    {
+      "epoch": 0.7737932745998263,
+      "grad_norm": 656.4087524414062,
+      "learning_rate": 1.708144765977492e-06,
+      "loss": 83.5821,
+      "step": 93540
+    },
+    {
+      "epoch": 0.7738759978491955,
+      "grad_norm": 2868.773193359375,
+      "learning_rate": 1.7070833227665146e-06,
+      "loss": 86.4508,
+      "step": 93550
+    },
+    {
+      "epoch": 0.7739587210985648,
+      "grad_norm": 547.4402465820312,
+      "learning_rate": 1.7060221415578042e-06,
+      "loss": 96.0463,
+      "step": 93560
+    },
+    {
+      "epoch": 0.774041444347934,
+      "grad_norm": 983.4544067382812,
+      "learning_rate": 1.7049612224357954e-06,
+      "loss": 72.9841,
+      "step": 93570
+    },
+    {
+      "epoch": 0.7741241675973032,
+      "grad_norm": 447.38299560546875,
+      "learning_rate": 1.703900565484899e-06,
+      "loss": 75.4073,
+      "step": 93580
+    },
+    {
+      "epoch": 0.7742068908466725,
+      "grad_norm": 797.6534423828125,
+      "learning_rate": 1.7028401707895082e-06,
+      "loss": 72.9507,
+      "step": 93590
+    },
+    {
+      "epoch": 0.7742896140960417,
+      "grad_norm": 536.6268310546875,
+      "learning_rate": 1.7017800384339928e-06,
+      "loss": 96.6598,
+      "step": 93600
+    },
+    {
+      "epoch": 0.7743723373454109,
+      "grad_norm": 601.033935546875,
+      "learning_rate": 1.700720168502703e-06,
+      "loss": 94.863,
+      "step": 93610
+    },
+    {
+      "epoch": 0.7744550605947802,
+      "grad_norm": 1244.7003173828125,
+      "learning_rate": 1.6996605610799682e-06,
+      "loss": 89.0981,
+      "step": 93620
+    },
+    {
+      "epoch": 0.7745377838441494,
+      "grad_norm": 423.31646728515625,
+      "learning_rate": 1.6986012162500953e-06,
+      "loss": 90.755,
+      "step": 93630
+    },
+    {
+      "epoch": 0.7746205070935186,
+      "grad_norm": 583.6690673828125,
+      "learning_rate": 1.697542134097373e-06,
+      "loss": 102.8274,
+      "step": 93640
+    },
+    {
+      "epoch": 0.7747032303428879,
+      "grad_norm": 1013.60693359375,
+      "learning_rate": 1.6964833147060661e-06,
+      "loss": 76.9443,
+      "step": 93650
+    },
+    {
+      "epoch": 0.7747859535922571,
+      "grad_norm": 1234.383056640625,
+      "learning_rate": 1.6954247581604216e-06,
+      "loss": 84.067,
+      "step": 93660
+    },
+    {
+      "epoch": 0.7748686768416263,
+      "grad_norm": 1029.810791015625,
+      "learning_rate": 1.6943664645446622e-06,
+      "loss": 89.6185,
+      "step": 93670
+    },
+    {
+      "epoch": 0.7749514000909956,
+      "grad_norm": 1082.3262939453125,
+      "learning_rate": 1.6933084339429935e-06,
+      "loss": 88.7879,
+      "step": 93680
+    },
+    {
+      "epoch": 0.7750341233403648,
+      "grad_norm": 1071.9556884765625,
+      "learning_rate": 1.692250666439596e-06,
+      "loss": 78.9346,
+      "step": 93690
+    },
+    {
+      "epoch": 0.775116846589734,
+      "grad_norm": 677.65771484375,
+      "learning_rate": 1.6911931621186329e-06,
+      "loss": 82.8995,
+      "step": 93700
+    },
+    {
+      "epoch": 0.7751995698391033,
+      "grad_norm": 867.5599365234375,
+      "learning_rate": 1.6901359210642444e-06,
+      "loss": 78.1593,
+      "step": 93710
+    },
+    {
+      "epoch": 0.7752822930884725,
+      "grad_norm": 942.0477905273438,
+      "learning_rate": 1.6890789433605508e-06,
+      "loss": 94.003,
+      "step": 93720
+    },
+    {
+      "epoch": 0.7753650163378417,
+      "grad_norm": 519.2806396484375,
+      "learning_rate": 1.6880222290916503e-06,
+      "loss": 60.9571,
+      "step": 93730
+    },
+    {
+      "epoch": 0.775447739587211,
+      "grad_norm": 774.544921875,
+      "learning_rate": 1.686965778341621e-06,
+      "loss": 84.0111,
+      "step": 93740
+    },
+    {
+      "epoch": 0.7755304628365802,
+      "grad_norm": 667.4334716796875,
+      "learning_rate": 1.68590959119452e-06,
+      "loss": 76.5631,
+      "step": 93750
+    },
+    {
+      "epoch": 0.7756131860859494,
+      "grad_norm": 467.7774353027344,
+      "learning_rate": 1.6848536677343836e-06,
+      "loss": 60.4382,
+      "step": 93760
+    },
+    {
+      "epoch": 0.7756959093353187,
+      "grad_norm": 455.2456970214844,
+      "learning_rate": 1.683798008045226e-06,
+      "loss": 78.7938,
+      "step": 93770
+    },
+    {
+      "epoch": 0.7757786325846879,
+      "grad_norm": 1058.66943359375,
+      "learning_rate": 1.6827426122110412e-06,
+      "loss": 78.3543,
+      "step": 93780
+    },
+    {
+      "epoch": 0.7758613558340571,
+      "grad_norm": 635.2317504882812,
+      "learning_rate": 1.6816874803158034e-06,
+      "loss": 79.6428,
+      "step": 93790
+    },
+    {
+      "epoch": 0.7759440790834264,
+      "grad_norm": 562.6470947265625,
+      "learning_rate": 1.6806326124434634e-06,
+      "loss": 59.4104,
+      "step": 93800
+    },
+    {
+      "epoch": 0.7760268023327956,
+      "grad_norm": 832.704833984375,
+      "learning_rate": 1.679578008677953e-06,
+      "loss": 97.8018,
+      "step": 93810
+    },
+    {
+      "epoch": 0.7761095255821648,
+      "grad_norm": 826.8822631835938,
+      "learning_rate": 1.6785236691031808e-06,
+      "loss": 66.9371,
+      "step": 93820
+    },
+    {
+      "epoch": 0.7761922488315341,
+      "grad_norm": 717.45068359375,
+      "learning_rate": 1.6774695938030378e-06,
+      "loss": 60.8474,
+      "step": 93830
+    },
+    {
+      "epoch": 0.7762749720809033,
+      "grad_norm": 893.1639404296875,
+      "learning_rate": 1.6764157828613902e-06,
+      "loss": 94.3922,
+      "step": 93840
+    },
+    {
+      "epoch": 0.7763576953302725,
+      "grad_norm": 422.0210876464844,
+      "learning_rate": 1.675362236362086e-06,
+      "loss": 73.2281,
+      "step": 93850
+    },
+    {
+      "epoch": 0.7764404185796419,
+      "grad_norm": 989.2019653320312,
+      "learning_rate": 1.6743089543889502e-06,
+      "loss": 103.5385,
+      "step": 93860
+    },
+    {
+      "epoch": 0.776523141829011,
+      "grad_norm": 823.348388671875,
+      "learning_rate": 1.6732559370257884e-06,
+      "loss": 91.4836,
+      "step": 93870
+    },
+    {
+      "epoch": 0.7766058650783803,
+      "grad_norm": 1299.27392578125,
+      "learning_rate": 1.6722031843563836e-06,
+      "loss": 93.1182,
+      "step": 93880
+    },
+    {
+      "epoch": 0.7766885883277496,
+      "grad_norm": 577.8670654296875,
+      "learning_rate": 1.6711506964644992e-06,
+      "loss": 93.8182,
+      "step": 93890
+    },
+    {
+      "epoch": 0.7767713115771188,
+      "grad_norm": 502.0442810058594,
+      "learning_rate": 1.6700984734338765e-06,
+      "loss": 68.5818,
+      "step": 93900
+    },
+    {
+      "epoch": 0.776854034826488,
+      "grad_norm": 1039.2012939453125,
+      "learning_rate": 1.669046515348236e-06,
+      "loss": 85.6031,
+      "step": 93910
+    },
+    {
+      "epoch": 0.7769367580758573,
+      "grad_norm": 390.88385009765625,
+      "learning_rate": 1.6679948222912773e-06,
+      "loss": 79.0019,
+      "step": 93920
+    },
+    {
+      "epoch": 0.7770194813252265,
+      "grad_norm": 908.3690795898438,
+      "learning_rate": 1.6669433943466789e-06,
+      "loss": 125.9054,
+      "step": 93930
+    },
+    {
+      "epoch": 0.7771022045745957,
+      "grad_norm": 942.0368041992188,
+      "learning_rate": 1.6658922315980975e-06,
+      "loss": 101.4422,
+      "step": 93940
+    },
+    {
+      "epoch": 0.777184927823965,
+      "grad_norm": 817.743896484375,
+      "learning_rate": 1.6648413341291703e-06,
+      "loss": 79.908,
+      "step": 93950
+    },
+    {
+      "epoch": 0.7772676510733342,
+      "grad_norm": 882.8682250976562,
+      "learning_rate": 1.6637907020235117e-06,
+      "loss": 76.305,
+      "step": 93960
+    },
+    {
+      "epoch": 0.7773503743227034,
+      "grad_norm": 1031.4886474609375,
+      "learning_rate": 1.662740335364717e-06,
+      "loss": 126.5732,
+      "step": 93970
+    },
+    {
+      "epoch": 0.7774330975720727,
+      "grad_norm": 771.1046142578125,
+      "learning_rate": 1.661690234236355e-06,
+      "loss": 95.6885,
+      "step": 93980
+    },
+    {
+      "epoch": 0.7775158208214419,
+      "grad_norm": 716.6741943359375,
+      "learning_rate": 1.6606403987219815e-06,
+      "loss": 78.7987,
+      "step": 93990
+    },
+    {
+      "epoch": 0.7775985440708111,
+      "grad_norm": 630.847900390625,
+      "learning_rate": 1.6595908289051266e-06,
+      "loss": 90.0902,
+      "step": 94000
+    },
+    {
+      "epoch": 0.7776812673201804,
+      "grad_norm": 845.2430419921875,
+      "learning_rate": 1.6585415248692988e-06,
+      "loss": 88.3763,
+      "step": 94010
+    },
+    {
+      "epoch": 0.7777639905695496,
+      "grad_norm": 1075.101318359375,
+      "learning_rate": 1.6574924866979863e-06,
+      "loss": 54.1172,
+      "step": 94020
+    },
+    {
+      "epoch": 0.7778467138189188,
+      "grad_norm": 1161.794921875,
+      "learning_rate": 1.6564437144746564e-06,
+      "loss": 78.1358,
+      "step": 94030
+    },
+    {
+      "epoch": 0.777929437068288,
+      "grad_norm": 867.4984130859375,
+      "learning_rate": 1.6553952082827562e-06,
+      "loss": 96.2229,
+      "step": 94040
+    },
+    {
+      "epoch": 0.7780121603176573,
+      "grad_norm": 909.298583984375,
+      "learning_rate": 1.6543469682057105e-06,
+      "loss": 99.5179,
+      "step": 94050
+    },
+    {
+      "epoch": 0.7780948835670265,
+      "grad_norm": 684.060791015625,
+      "learning_rate": 1.6532989943269207e-06,
+      "loss": 81.0873,
+      "step": 94060
+    },
+    {
+      "epoch": 0.7781776068163957,
+      "grad_norm": 1213.9482421875,
+      "learning_rate": 1.6522512867297707e-06,
+      "loss": 106.0108,
+      "step": 94070
+    },
+    {
+      "epoch": 0.778260330065765,
+      "grad_norm": 980.2380981445312,
+      "learning_rate": 1.6512038454976198e-06,
+      "loss": 85.677,
+      "step": 94080
+    },
+    {
+      "epoch": 0.7783430533151342,
+      "grad_norm": 450.2137451171875,
+      "learning_rate": 1.6501566707138116e-06,
+      "loss": 65.2738,
+      "step": 94090
+    },
+    {
+      "epoch": 0.7784257765645034,
+      "grad_norm": 525.07177734375,
+      "learning_rate": 1.6491097624616637e-06,
+      "loss": 75.8729,
+      "step": 94100
+    },
+    {
+      "epoch": 0.7785084998138727,
+      "grad_norm": 908.7725219726562,
+      "learning_rate": 1.6480631208244735e-06,
+      "loss": 125.1627,
+      "step": 94110
+    },
+    {
+      "epoch": 0.7785912230632419,
+      "grad_norm": 468.6520690917969,
+      "learning_rate": 1.6470167458855174e-06,
+      "loss": 110.0471,
+      "step": 94120
+    },
+    {
+      "epoch": 0.7786739463126111,
+      "grad_norm": 1324.9775390625,
+      "learning_rate": 1.645970637728051e-06,
+      "loss": 73.7453,
+      "step": 94130
+    },
+    {
+      "epoch": 0.7787566695619804,
+      "grad_norm": 1378.573486328125,
+      "learning_rate": 1.6449247964353094e-06,
+      "loss": 115.2216,
+      "step": 94140
+    },
+    {
+      "epoch": 0.7788393928113496,
+      "grad_norm": 1270.413330078125,
+      "learning_rate": 1.643879222090502e-06,
+      "loss": 75.4719,
+      "step": 94150
+    },
+    {
+      "epoch": 0.7789221160607188,
+      "grad_norm": 823.2882690429688,
+      "learning_rate": 1.642833914776823e-06,
+      "loss": 99.8761,
+      "step": 94160
+    },
+    {
+      "epoch": 0.7790048393100881,
+      "grad_norm": 1631.2496337890625,
+      "learning_rate": 1.6417888745774418e-06,
+      "loss": 93.0183,
+      "step": 94170
+    },
+    {
+      "epoch": 0.7790875625594573,
+      "grad_norm": 618.7994384765625,
+      "learning_rate": 1.640744101575506e-06,
+      "loss": 87.0626,
+      "step": 94180
+    },
+    {
+      "epoch": 0.7791702858088265,
+      "grad_norm": 874.3128051757812,
+      "learning_rate": 1.6396995958541468e-06,
+      "loss": 78.908,
+      "step": 94190
+    },
+    {
+      "epoch": 0.7792530090581958,
+      "grad_norm": 1032.2872314453125,
+      "learning_rate": 1.6386553574964691e-06,
+      "loss": 87.541,
+      "step": 94200
+    },
+    {
+      "epoch": 0.779335732307565,
+      "grad_norm": 711.1810302734375,
+      "learning_rate": 1.6376113865855585e-06,
+      "loss": 65.8986,
+      "step": 94210
+    },
+    {
+      "epoch": 0.7794184555569342,
+      "grad_norm": 1084.5020751953125,
+      "learning_rate": 1.6365676832044796e-06,
+      "loss": 92.7198,
+      "step": 94220
+    },
+    {
+      "epoch": 0.7795011788063035,
+      "grad_norm": 877.0111694335938,
+      "learning_rate": 1.6355242474362732e-06,
+      "loss": 100.024,
+      "step": 94230
+    },
+    {
+      "epoch": 0.7795839020556727,
+      "grad_norm": 1269.5721435546875,
+      "learning_rate": 1.634481079363961e-06,
+      "loss": 84.5142,
+      "step": 94240
+    },
+    {
+      "epoch": 0.7796666253050419,
+      "grad_norm": 683.024169921875,
+      "learning_rate": 1.6334381790705439e-06,
+      "loss": 81.7544,
+      "step": 94250
+    },
+    {
+      "epoch": 0.7797493485544112,
+      "grad_norm": 611.2015991210938,
+      "learning_rate": 1.6323955466390001e-06,
+      "loss": 94.3467,
+      "step": 94260
+    },
+    {
+      "epoch": 0.7798320718037804,
+      "grad_norm": 724.1978149414062,
+      "learning_rate": 1.6313531821522876e-06,
+      "loss": 93.952,
+      "step": 94270
+    },
+    {
+      "epoch": 0.7799147950531496,
+      "grad_norm": 540.2195434570312,
+      "learning_rate": 1.6303110856933413e-06,
+      "loss": 78.9152,
+      "step": 94280
+    },
+    {
+      "epoch": 0.779997518302519,
+      "grad_norm": 1860.32666015625,
+      "learning_rate": 1.629269257345078e-06,
+      "loss": 131.9263,
+      "step": 94290
+    },
+    {
+      "epoch": 0.7800802415518882,
+      "grad_norm": 987.336181640625,
+      "learning_rate": 1.628227697190391e-06,
+      "loss": 98.8791,
+      "step": 94300
+    },
+    {
+      "epoch": 0.7801629648012574,
+      "grad_norm": 608.0608520507812,
+      "learning_rate": 1.6271864053121528e-06,
+      "loss": 141.6185,
+      "step": 94310
+    },
+    {
+      "epoch": 0.7802456880506267,
+      "grad_norm": 752.9619140625,
+      "learning_rate": 1.6261453817932122e-06,
+      "loss": 61.8438,
+      "step": 94320
+    },
+    {
+      "epoch": 0.7803284112999959,
+      "grad_norm": 849.4782104492188,
+      "learning_rate": 1.6251046267163988e-06,
+      "loss": 94.6276,
+      "step": 94330
+    },
+    {
+      "epoch": 0.7804111345493651,
+      "grad_norm": 788.5552368164062,
+      "learning_rate": 1.6240641401645224e-06,
+      "loss": 96.3306,
+      "step": 94340
+    },
+    {
+      "epoch": 0.7804938577987344,
+      "grad_norm": 507.6255187988281,
+      "learning_rate": 1.6230239222203687e-06,
+      "loss": 103.3219,
+      "step": 94350
+    },
+    {
+      "epoch": 0.7805765810481036,
+      "grad_norm": 920.311767578125,
+      "learning_rate": 1.621983972966703e-06,
+      "loss": 114.5354,
+      "step": 94360
+    },
+    {
+      "epoch": 0.7806593042974728,
+      "grad_norm": 1070.4920654296875,
+      "learning_rate": 1.6209442924862684e-06,
+      "loss": 107.5379,
+      "step": 94370
+    },
+    {
+      "epoch": 0.7807420275468421,
+      "grad_norm": 1888.6494140625,
+      "learning_rate": 1.6199048808617896e-06,
+      "loss": 89.8358,
+      "step": 94380
+    },
+    {
+      "epoch": 0.7808247507962113,
+      "grad_norm": 1030.0611572265625,
+      "learning_rate": 1.6188657381759676e-06,
+      "loss": 86.013,
+      "step": 94390
+    },
+    {
+      "epoch": 0.7809074740455805,
+      "grad_norm": 437.46112060546875,
+      "learning_rate": 1.6178268645114826e-06,
+      "loss": 73.3406,
+      "step": 94400
+    },
+    {
+      "epoch": 0.7809901972949498,
+      "grad_norm": 701.3662719726562,
+      "learning_rate": 1.6167882599509904e-06,
+      "loss": 88.5935,
+      "step": 94410
+    },
+    {
+      "epoch": 0.781072920544319,
+      "grad_norm": 639.0773315429688,
+      "learning_rate": 1.6157499245771296e-06,
+      "loss": 73.676,
+      "step": 94420
+    },
+    {
+      "epoch": 0.7811556437936882,
+      "grad_norm": 894.396484375,
+      "learning_rate": 1.6147118584725163e-06,
+      "loss": 90.9503,
+      "step": 94430
+    },
+    {
+      "epoch": 0.7812383670430575,
+      "grad_norm": 1062.1143798828125,
+      "learning_rate": 1.6136740617197433e-06,
+      "loss": 90.4783,
+      "step": 94440
+    },
+    {
+      "epoch": 0.7813210902924267,
+      "grad_norm": 959.5033569335938,
+      "learning_rate": 1.612636534401384e-06,
+      "loss": 61.4715,
+      "step": 94450
+    },
+    {
+      "epoch": 0.7814038135417959,
+      "grad_norm": 1169.7418212890625,
+      "learning_rate": 1.61159927659999e-06,
+      "loss": 77.9506,
+      "step": 94460
+    },
+    {
+      "epoch": 0.7814865367911652,
+      "grad_norm": 1380.6806640625,
+      "learning_rate": 1.6105622883980893e-06,
+      "loss": 80.2579,
+      "step": 94470
+    },
+    {
+      "epoch": 0.7815692600405344,
+      "grad_norm": 1116.7760009765625,
+      "learning_rate": 1.6095255698781954e-06,
+      "loss": 120.8729,
+      "step": 94480
+    },
+    {
+      "epoch": 0.7816519832899036,
+      "grad_norm": 709.1279907226562,
+      "learning_rate": 1.6084891211227899e-06,
+      "loss": 89.6381,
+      "step": 94490
+    },
+    {
+      "epoch": 0.7817347065392729,
+      "grad_norm": 626.1630859375,
+      "learning_rate": 1.6074529422143398e-06,
+      "loss": 82.1522,
+      "step": 94500
+    },
+    {
+      "epoch": 0.7818174297886421,
+      "grad_norm": 1719.0965576171875,
+      "learning_rate": 1.6064170332352897e-06,
+      "loss": 68.7441,
+      "step": 94510
+    },
+    {
+      "epoch": 0.7819001530380113,
+      "grad_norm": 716.7944946289062,
+      "learning_rate": 1.6053813942680618e-06,
+      "loss": 82.9248,
+      "step": 94520
+    },
+    {
+      "epoch": 0.7819828762873806,
+      "grad_norm": 828.8280029296875,
+      "learning_rate": 1.604346025395057e-06,
+      "loss": 79.9472,
+      "step": 94530
+    },
+    {
+      "epoch": 0.7820655995367498,
+      "grad_norm": 675.82568359375,
+      "learning_rate": 1.6033109266986552e-06,
+      "loss": 118.3841,
+      "step": 94540
+    },
+    {
+      "epoch": 0.782148322786119,
+      "grad_norm": 966.230224609375,
+      "learning_rate": 1.602276098261214e-06,
+      "loss": 103.3147,
+      "step": 94550
+    },
+    {
+      "epoch": 0.7822310460354883,
+      "grad_norm": 696.0429077148438,
+      "learning_rate": 1.6012415401650706e-06,
+      "loss": 98.5404,
+      "step": 94560
+    },
+    {
+      "epoch": 0.7823137692848575,
+      "grad_norm": 739.5772705078125,
+      "learning_rate": 1.6002072524925395e-06,
+      "loss": 78.0328,
+      "step": 94570
+    },
+    {
+      "epoch": 0.7823964925342267,
+      "grad_norm": 537.0844116210938,
+      "learning_rate": 1.5991732353259142e-06,
+      "loss": 76.4263,
+      "step": 94580
+    },
+    {
+      "epoch": 0.782479215783596,
+      "grad_norm": 462.9024658203125,
+      "learning_rate": 1.598139488747467e-06,
+      "loss": 95.2623,
+      "step": 94590
+    },
+    {
+      "epoch": 0.7825619390329652,
+      "grad_norm": 698.7216186523438,
+      "learning_rate": 1.5971060128394483e-06,
+      "loss": 85.2989,
+      "step": 94600
+    },
+    {
+      "epoch": 0.7826446622823344,
+      "grad_norm": 887.9057006835938,
+      "learning_rate": 1.596072807684087e-06,
+      "loss": 112.7182,
+      "step": 94610
+    },
+    {
+      "epoch": 0.7827273855317037,
+      "grad_norm": 981.1459350585938,
+      "learning_rate": 1.5950398733635903e-06,
+      "loss": 94.8544,
+      "step": 94620
+    },
+    {
+      "epoch": 0.7828101087810729,
+      "grad_norm": 1017.7367553710938,
+      "learning_rate": 1.5940072099601446e-06,
+      "loss": 69.3493,
+      "step": 94630
+    },
+    {
+      "epoch": 0.7828928320304421,
+      "grad_norm": 593.1333618164062,
+      "learning_rate": 1.5929748175559135e-06,
+      "loss": 60.2959,
+      "step": 94640
+    },
+    {
+      "epoch": 0.7829755552798114,
+      "grad_norm": 808.7359008789062,
+      "learning_rate": 1.5919426962330398e-06,
+      "loss": 85.7781,
+      "step": 94650
+    },
+    {
+      "epoch": 0.7830582785291806,
+      "grad_norm": 1296.2574462890625,
+      "learning_rate": 1.5909108460736455e-06,
+      "loss": 82.9248,
+      "step": 94660
+    },
+    {
+      "epoch": 0.7831410017785498,
+      "grad_norm": 622.1264038085938,
+      "learning_rate": 1.589879267159829e-06,
+      "loss": 74.3232,
+      "step": 94670
+    },
+    {
+      "epoch": 0.7832237250279191,
+      "grad_norm": 381.82867431640625,
+      "learning_rate": 1.5888479595736695e-06,
+      "loss": 91.2148,
+      "step": 94680
+    },
+    {
+      "epoch": 0.7833064482772883,
+      "grad_norm": 378.500244140625,
+      "learning_rate": 1.5878169233972218e-06,
+      "loss": 89.309,
+      "step": 94690
+    },
+    {
+      "epoch": 0.7833891715266575,
+      "grad_norm": 641.9385375976562,
+      "learning_rate": 1.5867861587125228e-06,
+      "loss": 81.5497,
+      "step": 94700
+    },
+    {
+      "epoch": 0.7834718947760269,
+      "grad_norm": 2622.712890625,
+      "learning_rate": 1.5857556656015837e-06,
+      "loss": 101.5679,
+      "step": 94710
+    },
+    {
+      "epoch": 0.783554618025396,
+      "grad_norm": 1032.7587890625,
+      "learning_rate": 1.5847254441463978e-06,
+      "loss": 78.4932,
+      "step": 94720
+    },
+    {
+      "epoch": 0.7836373412747653,
+      "grad_norm": 776.2950439453125,
+      "learning_rate": 1.583695494428934e-06,
+      "loss": 71.536,
+      "step": 94730
+    },
+    {
+      "epoch": 0.7837200645241346,
+      "grad_norm": 942.0223999023438,
+      "learning_rate": 1.5826658165311409e-06,
+      "loss": 88.8115,
+      "step": 94740
+    },
+    {
+      "epoch": 0.7838027877735038,
+      "grad_norm": 739.5013427734375,
+      "learning_rate": 1.5816364105349451e-06,
+      "loss": 117.7168,
+      "step": 94750
+    },
+    {
+      "epoch": 0.783885511022873,
+      "grad_norm": 477.8927001953125,
+      "learning_rate": 1.5806072765222524e-06,
+      "loss": 96.1067,
+      "step": 94760
+    },
+    {
+      "epoch": 0.7839682342722422,
+      "grad_norm": 751.9795532226562,
+      "learning_rate": 1.5795784145749453e-06,
+      "loss": 78.0177,
+      "step": 94770
+    },
+    {
+      "epoch": 0.7840509575216115,
+      "grad_norm": 756.887451171875,
+      "learning_rate": 1.5785498247748864e-06,
+      "loss": 82.0907,
+      "step": 94780
+    },
+    {
+      "epoch": 0.7841336807709807,
+      "grad_norm": 814.6382446289062,
+      "learning_rate": 1.5775215072039157e-06,
+      "loss": 118.0077,
+      "step": 94790
+    },
+    {
+      "epoch": 0.7842164040203499,
+      "grad_norm": 492.3858337402344,
+      "learning_rate": 1.5764934619438515e-06,
+      "loss": 83.478,
+      "step": 94800
+    },
+    {
+      "epoch": 0.7842991272697192,
+      "grad_norm": 1426.8245849609375,
+      "learning_rate": 1.5754656890764912e-06,
+      "loss": 82.622,
+      "step": 94810
+    },
+    {
+      "epoch": 0.7843818505190884,
+      "grad_norm": 838.7455444335938,
+      "learning_rate": 1.5744381886836091e-06,
+      "loss": 99.2792,
+      "step": 94820
+    },
+    {
+      "epoch": 0.7844645737684576,
+      "grad_norm": 848.1531982421875,
+      "learning_rate": 1.5734109608469612e-06,
+      "loss": 92.1673,
+      "step": 94830
+    },
+    {
+      "epoch": 0.7845472970178269,
+      "grad_norm": 401.6071472167969,
+      "learning_rate": 1.5723840056482731e-06,
+      "loss": 79.848,
+      "step": 94840
+    },
+    {
+      "epoch": 0.7846300202671961,
+      "grad_norm": 879.9978637695312,
+      "learning_rate": 1.5713573231692613e-06,
+      "loss": 59.712,
+      "step": 94850
+    },
+    {
+      "epoch": 0.7847127435165653,
+      "grad_norm": 913.5989990234375,
+      "learning_rate": 1.5703309134916116e-06,
+      "loss": 74.9016,
+      "step": 94860
+    },
+    {
+      "epoch": 0.7847954667659346,
+      "grad_norm": 1169.180908203125,
+      "learning_rate": 1.5693047766969916e-06,
+      "loss": 88.2039,
+      "step": 94870
+    },
+    {
+      "epoch": 0.7848781900153038,
+      "grad_norm": 809.9779663085938,
+      "learning_rate": 1.568278912867045e-06,
+      "loss": 110.5675,
+      "step": 94880
+    },
+    {
+      "epoch": 0.784960913264673,
+      "grad_norm": 501.2241516113281,
+      "learning_rate": 1.5672533220833962e-06,
+      "loss": 67.3911,
+      "step": 94890
+    },
+    {
+      "epoch": 0.7850436365140423,
+      "grad_norm": 1112.9195556640625,
+      "learning_rate": 1.5662280044276467e-06,
+      "loss": 94.5414,
+      "step": 94900
+    },
+    {
+      "epoch": 0.7851263597634115,
+      "grad_norm": 1013.2802734375,
+      "learning_rate": 1.5652029599813773e-06,
+      "loss": 106.9486,
+      "step": 94910
+    },
+    {
+      "epoch": 0.7852090830127807,
+      "grad_norm": 693.358154296875,
+      "learning_rate": 1.564178188826143e-06,
+      "loss": 78.1435,
+      "step": 94920
+    },
+    {
+      "epoch": 0.78529180626215,
+      "grad_norm": 1105.666259765625,
+      "learning_rate": 1.5631536910434807e-06,
+      "loss": 99.9668,
+      "step": 94930
+    },
+    {
+      "epoch": 0.7853745295115192,
+      "grad_norm": 1162.7235107421875,
+      "learning_rate": 1.5621294667149079e-06,
+      "loss": 87.6308,
+      "step": 94940
+    },
+    {
+      "epoch": 0.7854572527608884,
+      "grad_norm": 539.553955078125,
+      "learning_rate": 1.561105515921915e-06,
+      "loss": 91.6051,
+      "step": 94950
+    },
+    {
+      "epoch": 0.7855399760102577,
+      "grad_norm": 722.3108520507812,
+      "learning_rate": 1.5600818387459748e-06,
+      "loss": 64.7316,
+      "step": 94960
+    },
+    {
+      "epoch": 0.7856226992596269,
+      "grad_norm": 381.7198486328125,
+      "learning_rate": 1.559058435268535e-06,
+      "loss": 66.9805,
+      "step": 94970
+    },
+    {
+      "epoch": 0.7857054225089961,
+      "grad_norm": 635.7542724609375,
+      "learning_rate": 1.558035305571024e-06,
+      "loss": 114.3884,
+      "step": 94980
+    },
+    {
+      "epoch": 0.7857881457583654,
+      "grad_norm": 1400.4791259765625,
+      "learning_rate": 1.5570124497348466e-06,
+      "loss": 110.485,
+      "step": 94990
+    },
+    {
+      "epoch": 0.7858708690077346,
+      "grad_norm": 678.18896484375,
+      "learning_rate": 1.5559898678413898e-06,
+      "loss": 96.3298,
+      "step": 95000
+    },
+    {
+      "epoch": 0.7859535922571038,
+      "grad_norm": 1071.6373291015625,
+      "learning_rate": 1.554967559972011e-06,
+      "loss": 98.7113,
+      "step": 95010
+    },
+    {
+      "epoch": 0.7860363155064731,
+      "grad_norm": 1374.9266357421875,
+      "learning_rate": 1.5539455262080534e-06,
+      "loss": 118.118,
+      "step": 95020
+    },
+    {
+      "epoch": 0.7861190387558423,
+      "grad_norm": 759.0629272460938,
+      "learning_rate": 1.5529237666308333e-06,
+      "loss": 93.9589,
+      "step": 95030
+    },
+    {
+      "epoch": 0.7862017620052115,
+      "grad_norm": 854.2042236328125,
+      "learning_rate": 1.551902281321651e-06,
+      "loss": 77.0455,
+      "step": 95040
+    },
+    {
+      "epoch": 0.7862844852545808,
+      "grad_norm": 812.8806762695312,
+      "learning_rate": 1.5508810703617794e-06,
+      "loss": 79.7987,
+      "step": 95050
+    },
+    {
+      "epoch": 0.78636720850395,
+      "grad_norm": 709.544677734375,
+      "learning_rate": 1.5498601338324715e-06,
+      "loss": 79.4533,
+      "step": 95060
+    },
+    {
+      "epoch": 0.7864499317533192,
+      "grad_norm": 869.2487182617188,
+      "learning_rate": 1.5488394718149586e-06,
+      "loss": 70.6014,
+      "step": 95070
+    },
+    {
+      "epoch": 0.7865326550026885,
+      "grad_norm": 612.1897583007812,
+      "learning_rate": 1.5478190843904523e-06,
+      "loss": 80.7072,
+      "step": 95080
+    },
+    {
+      "epoch": 0.7866153782520577,
+      "grad_norm": 713.2626342773438,
+      "learning_rate": 1.546798971640136e-06,
+      "loss": 62.9009,
+      "step": 95090
+    },
+    {
+      "epoch": 0.7866981015014269,
+      "grad_norm": 761.9999389648438,
+      "learning_rate": 1.5457791336451777e-06,
+      "loss": 74.8367,
+      "step": 95100
+    },
+    {
+      "epoch": 0.7867808247507962,
+      "grad_norm": 1282.5313720703125,
+      "learning_rate": 1.5447595704867213e-06,
+      "loss": 114.0064,
+      "step": 95110
+    },
+    {
+      "epoch": 0.7868635480001654,
+      "grad_norm": 748.5642700195312,
+      "learning_rate": 1.543740282245888e-06,
+      "loss": 96.991,
+      "step": 95120
+    },
+    {
+      "epoch": 0.7869462712495346,
+      "grad_norm": 755.327392578125,
+      "learning_rate": 1.5427212690037774e-06,
+      "loss": 69.9441,
+      "step": 95130
+    },
+    {
+      "epoch": 0.787028994498904,
+      "grad_norm": 669.56201171875,
+      "learning_rate": 1.5417025308414695e-06,
+      "loss": 101.2025,
+      "step": 95140
+    },
+    {
+      "epoch": 0.7871117177482732,
+      "grad_norm": 710.7820434570312,
+      "learning_rate": 1.5406840678400204e-06,
+      "loss": 98.5515,
+      "step": 95150
+    },
+    {
+      "epoch": 0.7871944409976424,
+      "grad_norm": 1083.8662109375,
+      "learning_rate": 1.5396658800804632e-06,
+      "loss": 102.2657,
+      "step": 95160
+    },
+    {
+      "epoch": 0.7872771642470117,
+      "grad_norm": 1122.4599609375,
+      "learning_rate": 1.5386479676438132e-06,
+      "loss": 85.0763,
+      "step": 95170
+    },
+    {
+      "epoch": 0.7873598874963809,
+      "grad_norm": 1315.6075439453125,
+      "learning_rate": 1.5376303306110574e-06,
+      "loss": 93.9366,
+      "step": 95180
+    },
+    {
+      "epoch": 0.7874426107457501,
+      "grad_norm": 796.9031372070312,
+      "learning_rate": 1.536612969063166e-06,
+      "loss": 73.0806,
+      "step": 95190
+    },
+    {
+      "epoch": 0.7875253339951194,
+      "grad_norm": 571.8799438476562,
+      "learning_rate": 1.5355958830810858e-06,
+      "loss": 67.0068,
+      "step": 95200
+    },
+    {
+      "epoch": 0.7876080572444886,
+      "grad_norm": 558.7503051757812,
+      "learning_rate": 1.5345790727457416e-06,
+      "loss": 92.4161,
+      "step": 95210
+    },
+    {
+      "epoch": 0.7876907804938578,
+      "grad_norm": 538.6721801757812,
+      "learning_rate": 1.5335625381380364e-06,
+      "loss": 75.6145,
+      "step": 95220
+    },
+    {
+      "epoch": 0.7877735037432271,
+      "grad_norm": 1402.743408203125,
+      "learning_rate": 1.5325462793388502e-06,
+      "loss": 127.778,
+      "step": 95230
+    },
+    {
+      "epoch": 0.7878562269925963,
+      "grad_norm": 934.5457763671875,
+      "learning_rate": 1.531530296429044e-06,
+      "loss": 93.78,
+      "step": 95240
+    },
+    {
+      "epoch": 0.7879389502419655,
+      "grad_norm": 658.7655029296875,
+      "learning_rate": 1.5305145894894547e-06,
+      "loss": 99.2224,
+      "step": 95250
+    },
+    {
+      "epoch": 0.7880216734913348,
+      "grad_norm": 1240.9752197265625,
+      "learning_rate": 1.5294991586008977e-06,
+      "loss": 99.2145,
+      "step": 95260
+    },
+    {
+      "epoch": 0.788104396740704,
+      "grad_norm": 847.5444946289062,
+      "learning_rate": 1.528484003844164e-06,
+      "loss": 105.1305,
+      "step": 95270
+    },
+    {
+      "epoch": 0.7881871199900732,
+      "grad_norm": 1418.359130859375,
+      "learning_rate": 1.5274691253000257e-06,
+      "loss": 150.746,
+      "step": 95280
+    },
+    {
+      "epoch": 0.7882698432394425,
+      "grad_norm": 576.1387939453125,
+      "learning_rate": 1.5264545230492323e-06,
+      "loss": 72.2739,
+      "step": 95290
+    },
+    {
+      "epoch": 0.7883525664888117,
+      "grad_norm": 1523.2347412109375,
+      "learning_rate": 1.5254401971725114e-06,
+      "loss": 76.5713,
+      "step": 95300
+    },
+    {
+      "epoch": 0.7884352897381809,
+      "grad_norm": 850.4096069335938,
+      "learning_rate": 1.5244261477505678e-06,
+      "loss": 56.1657,
+      "step": 95310
+    },
+    {
+      "epoch": 0.7885180129875502,
+      "grad_norm": 620.0430908203125,
+      "learning_rate": 1.5234123748640834e-06,
+      "loss": 91.6503,
+      "step": 95320
+    },
+    {
+      "epoch": 0.7886007362369194,
+      "grad_norm": 732.49267578125,
+      "learning_rate": 1.5223988785937222e-06,
+      "loss": 91.2292,
+      "step": 95330
+    },
+    {
+      "epoch": 0.7886834594862886,
+      "grad_norm": 1129.179443359375,
+      "learning_rate": 1.5213856590201241e-06,
+      "loss": 67.9968,
+      "step": 95340
+    },
+    {
+      "epoch": 0.7887661827356579,
+      "grad_norm": 744.8907470703125,
+      "learning_rate": 1.520372716223903e-06,
+      "loss": 109.5926,
+      "step": 95350
+    },
+    {
+      "epoch": 0.7888489059850271,
+      "grad_norm": 1017.0743408203125,
+      "learning_rate": 1.5193600502856548e-06,
+      "loss": 108.5264,
+      "step": 95360
+    },
+    {
+      "epoch": 0.7889316292343963,
+      "grad_norm": 150.6322021484375,
+      "learning_rate": 1.5183476612859538e-06,
+      "loss": 100.5434,
+      "step": 95370
+    },
+    {
+      "epoch": 0.7890143524837656,
+      "grad_norm": 1154.6719970703125,
+      "learning_rate": 1.5173355493053509e-06,
+      "loss": 91.2678,
+      "step": 95380
+    },
+    {
+      "epoch": 0.7890970757331348,
+      "grad_norm": 906.0249633789062,
+      "learning_rate": 1.5163237144243754e-06,
+      "loss": 119.1058,
+      "step": 95390
+    },
+    {
+      "epoch": 0.789179798982504,
+      "grad_norm": 976.3828735351562,
+      "learning_rate": 1.5153121567235334e-06,
+      "loss": 90.547,
+      "step": 95400
+    },
+    {
+      "epoch": 0.7892625222318733,
+      "grad_norm": 1162.3525390625,
+      "learning_rate": 1.5143008762833112e-06,
+      "loss": 95.5285,
+      "step": 95410
+    },
+    {
+      "epoch": 0.7893452454812425,
+      "grad_norm": 1161.5557861328125,
+      "learning_rate": 1.5132898731841689e-06,
+      "loss": 82.0131,
+      "step": 95420
+    },
+    {
+      "epoch": 0.7894279687306117,
+      "grad_norm": 785.49755859375,
+      "learning_rate": 1.512279147506553e-06,
+      "loss": 79.7298,
+      "step": 95430
+    },
+    {
+      "epoch": 0.789510691979981,
+      "grad_norm": 1093.42333984375,
+      "learning_rate": 1.5112686993308768e-06,
+      "loss": 117.2086,
+      "step": 95440
+    },
+    {
+      "epoch": 0.7895934152293502,
+      "grad_norm": 1063.107177734375,
+      "learning_rate": 1.5102585287375394e-06,
+      "loss": 89.8321,
+      "step": 95450
+    },
+    {
+      "epoch": 0.7896761384787194,
+      "grad_norm": 987.2368774414062,
+      "learning_rate": 1.5092486358069154e-06,
+      "loss": 105.4375,
+      "step": 95460
+    },
+    {
+      "epoch": 0.7897588617280887,
+      "grad_norm": 788.5397338867188,
+      "learning_rate": 1.508239020619357e-06,
+      "loss": 104.0471,
+      "step": 95470
+    },
+    {
+      "epoch": 0.7898415849774579,
+      "grad_norm": 926.9537963867188,
+      "learning_rate": 1.5072296832551942e-06,
+      "loss": 72.8565,
+      "step": 95480
+    },
+    {
+      "epoch": 0.7899243082268271,
+      "grad_norm": 943.4337158203125,
+      "learning_rate": 1.5062206237947363e-06,
+      "loss": 100.017,
+      "step": 95490
+    },
+    {
+      "epoch": 0.7900070314761963,
+      "grad_norm": 588.101806640625,
+      "learning_rate": 1.5052118423182688e-06,
+      "loss": 69.1182,
+      "step": 95500
+    },
+    {
+      "epoch": 0.7900897547255656,
+      "grad_norm": 267.3822021484375,
+      "learning_rate": 1.5042033389060563e-06,
+      "loss": 62.8242,
+      "step": 95510
+    },
+    {
+      "epoch": 0.7901724779749348,
+      "grad_norm": 601.46875,
+      "learning_rate": 1.5031951136383406e-06,
+      "loss": 107.1168,
+      "step": 95520
+    },
+    {
+      "epoch": 0.790255201224304,
+      "grad_norm": 795.5377197265625,
+      "learning_rate": 1.5021871665953414e-06,
+      "loss": 70.5595,
+      "step": 95530
+    },
+    {
+      "epoch": 0.7903379244736733,
+      "grad_norm": 1379.524658203125,
+      "learning_rate": 1.5011794978572568e-06,
+      "loss": 84.3169,
+      "step": 95540
+    },
+    {
+      "epoch": 0.7904206477230425,
+      "grad_norm": 967.1463012695312,
+      "learning_rate": 1.500172107504262e-06,
+      "loss": 78.8489,
+      "step": 95550
+    },
+    {
+      "epoch": 0.7905033709724117,
+      "grad_norm": 608.6145629882812,
+      "learning_rate": 1.4991649956165105e-06,
+      "loss": 64.6662,
+      "step": 95560
+    },
+    {
+      "epoch": 0.790586094221781,
+      "grad_norm": 695.8089599609375,
+      "learning_rate": 1.4981581622741337e-06,
+      "loss": 158.1259,
+      "step": 95570
+    },
+    {
+      "epoch": 0.7906688174711503,
+      "grad_norm": 959.3106079101562,
+      "learning_rate": 1.4971516075572407e-06,
+      "loss": 74.8305,
+      "step": 95580
+    },
+    {
+      "epoch": 0.7907515407205195,
+      "grad_norm": 1071.234375,
+      "learning_rate": 1.4961453315459184e-06,
+      "loss": 65.4966,
+      "step": 95590
+    },
+    {
+      "epoch": 0.7908342639698888,
+      "grad_norm": 680.824951171875,
+      "learning_rate": 1.4951393343202314e-06,
+      "loss": 92.0542,
+      "step": 95600
+    },
+    {
+      "epoch": 0.790916987219258,
+      "grad_norm": 884.766357421875,
+      "learning_rate": 1.4941336159602227e-06,
+      "loss": 81.2589,
+      "step": 95610
+    },
+    {
+      "epoch": 0.7909997104686272,
+      "grad_norm": 1236.22900390625,
+      "learning_rate": 1.4931281765459122e-06,
+      "loss": 88.8313,
+      "step": 95620
+    },
+    {
+      "epoch": 0.7910824337179965,
+      "grad_norm": 1996.257568359375,
+      "learning_rate": 1.492123016157298e-06,
+      "loss": 86.4051,
+      "step": 95630
+    },
+    {
+      "epoch": 0.7911651569673657,
+      "grad_norm": 818.218994140625,
+      "learning_rate": 1.4911181348743569e-06,
+      "loss": 82.9814,
+      "step": 95640
+    },
+    {
+      "epoch": 0.7912478802167349,
+      "grad_norm": 925.0620727539062,
+      "learning_rate": 1.490113532777042e-06,
+      "loss": 140.5014,
+      "step": 95650
+    },
+    {
+      "epoch": 0.7913306034661042,
+      "grad_norm": 568.860595703125,
+      "learning_rate": 1.4891092099452853e-06,
+      "loss": 60.901,
+      "step": 95660
+    },
+    {
+      "epoch": 0.7914133267154734,
+      "grad_norm": 1018.9000244140625,
+      "learning_rate": 1.4881051664589958e-06,
+      "loss": 94.0601,
+      "step": 95670
+    },
+    {
+      "epoch": 0.7914960499648426,
+      "grad_norm": 646.0346069335938,
+      "learning_rate": 1.4871014023980607e-06,
+      "loss": 74.5163,
+      "step": 95680
+    },
+    {
+      "epoch": 0.7915787732142119,
+      "grad_norm": 749.203857421875,
+      "learning_rate": 1.486097917842345e-06,
+      "loss": 76.4487,
+      "step": 95690
+    },
+    {
+      "epoch": 0.7916614964635811,
+      "grad_norm": 519.8616333007812,
+      "learning_rate": 1.4850947128716914e-06,
+      "loss": 76.2586,
+      "step": 95700
+    },
+    {
+      "epoch": 0.7917442197129503,
+      "grad_norm": 1079.2755126953125,
+      "learning_rate": 1.4840917875659206e-06,
+      "loss": 85.6422,
+      "step": 95710
+    },
+    {
+      "epoch": 0.7918269429623196,
+      "grad_norm": 606.11474609375,
+      "learning_rate": 1.4830891420048298e-06,
+      "loss": 91.0621,
+      "step": 95720
+    },
+    {
+      "epoch": 0.7919096662116888,
+      "grad_norm": 1178.3045654296875,
+      "learning_rate": 1.482086776268196e-06,
+      "loss": 99.66,
+      "step": 95730
+    },
+    {
+      "epoch": 0.791992389461058,
+      "grad_norm": 980.57763671875,
+      "learning_rate": 1.4810846904357722e-06,
+      "loss": 96.6671,
+      "step": 95740
+    },
+    {
+      "epoch": 0.7920751127104273,
+      "grad_norm": 1058.7889404296875,
+      "learning_rate": 1.48008288458729e-06,
+      "loss": 109.5006,
+      "step": 95750
+    },
+    {
+      "epoch": 0.7921578359597965,
+      "grad_norm": 529.0836181640625,
+      "learning_rate": 1.4790813588024584e-06,
+      "loss": 88.0496,
+      "step": 95760
+    },
+    {
+      "epoch": 0.7922405592091657,
+      "grad_norm": 1165.0322265625,
+      "learning_rate": 1.4780801131609657e-06,
+      "loss": 111.1433,
+      "step": 95770
+    },
+    {
+      "epoch": 0.792323282458535,
+      "grad_norm": 773.7676391601562,
+      "learning_rate": 1.477079147742474e-06,
+      "loss": 106.7083,
+      "step": 95780
+    },
+    {
+      "epoch": 0.7924060057079042,
+      "grad_norm": 863.2052612304688,
+      "learning_rate": 1.476078462626624e-06,
+      "loss": 83.5902,
+      "step": 95790
+    },
+    {
+      "epoch": 0.7924887289572734,
+      "grad_norm": 709.7434692382812,
+      "learning_rate": 1.4750780578930402e-06,
+      "loss": 66.4846,
+      "step": 95800
+    },
+    {
+      "epoch": 0.7925714522066427,
+      "grad_norm": 703.7163696289062,
+      "learning_rate": 1.4740779336213178e-06,
+      "loss": 86.0505,
+      "step": 95810
+    },
+    {
+      "epoch": 0.7926541754560119,
+      "grad_norm": 1240.612060546875,
+      "learning_rate": 1.4730780898910329e-06,
+      "loss": 108.2829,
+      "step": 95820
+    },
+    {
+      "epoch": 0.7927368987053811,
+      "grad_norm": 702.4696655273438,
+      "learning_rate": 1.4720785267817378e-06,
+      "loss": 59.4141,
+      "step": 95830
+    },
+    {
+      "epoch": 0.7928196219547504,
+      "grad_norm": 966.9011840820312,
+      "learning_rate": 1.4710792443729633e-06,
+      "loss": 112.7589,
+      "step": 95840
+    },
+    {
+      "epoch": 0.7929023452041196,
+      "grad_norm": 1277.8905029296875,
+      "learning_rate": 1.470080242744218e-06,
+      "loss": 72.1038,
+      "step": 95850
+    },
+    {
+      "epoch": 0.7929850684534888,
+      "grad_norm": 1504.42529296875,
+      "learning_rate": 1.4690815219749887e-06,
+      "loss": 88.3772,
+      "step": 95860
+    },
+    {
+      "epoch": 0.7930677917028581,
+      "grad_norm": 663.6552734375,
+      "learning_rate": 1.4680830821447368e-06,
+      "loss": 80.6271,
+      "step": 95870
+    },
+    {
+      "epoch": 0.7931505149522273,
+      "grad_norm": 686.2540283203125,
+      "learning_rate": 1.4670849233329032e-06,
+      "loss": 85.9933,
+      "step": 95880
+    },
+    {
+      "epoch": 0.7932332382015965,
+      "grad_norm": 648.2404174804688,
+      "learning_rate": 1.4660870456189098e-06,
+      "loss": 91.8078,
+      "step": 95890
+    },
+    {
+      "epoch": 0.7933159614509658,
+      "grad_norm": 1184.7274169921875,
+      "learning_rate": 1.4650894490821515e-06,
+      "loss": 105.1896,
+      "step": 95900
+    },
+    {
+      "epoch": 0.793398684700335,
+      "grad_norm": 900.9071655273438,
+      "learning_rate": 1.464092133802003e-06,
+      "loss": 95.3142,
+      "step": 95910
+    },
+    {
+      "epoch": 0.7934814079497042,
+      "grad_norm": 786.411865234375,
+      "learning_rate": 1.4630950998578158e-06,
+      "loss": 71.5299,
+      "step": 95920
+    },
+    {
+      "epoch": 0.7935641311990735,
+      "grad_norm": 537.1038818359375,
+      "learning_rate": 1.4620983473289192e-06,
+      "loss": 106.0131,
+      "step": 95930
+    },
+    {
+      "epoch": 0.7936468544484427,
+      "grad_norm": 907.1212158203125,
+      "learning_rate": 1.4611018762946217e-06,
+      "loss": 67.942,
+      "step": 95940
+    },
+    {
+      "epoch": 0.7937295776978119,
+      "grad_norm": 532.5782470703125,
+      "learning_rate": 1.4601056868342051e-06,
+      "loss": 77.3442,
+      "step": 95950
+    },
+    {
+      "epoch": 0.7938123009471812,
+      "grad_norm": 1575.781005859375,
+      "learning_rate": 1.4591097790269333e-06,
+      "loss": 121.2645,
+      "step": 95960
+    },
+    {
+      "epoch": 0.7938950241965504,
+      "grad_norm": 569.5699462890625,
+      "learning_rate": 1.4581141529520455e-06,
+      "loss": 100.5707,
+      "step": 95970
+    },
+    {
+      "epoch": 0.7939777474459196,
+      "grad_norm": 680.08251953125,
+      "learning_rate": 1.4571188086887583e-06,
+      "loss": 87.8128,
+      "step": 95980
+    },
+    {
+      "epoch": 0.794060470695289,
+      "grad_norm": 462.587890625,
+      "learning_rate": 1.4561237463162693e-06,
+      "loss": 92.0846,
+      "step": 95990
+    },
+    {
+      "epoch": 0.7941431939446582,
+      "grad_norm": 552.5233154296875,
+      "learning_rate": 1.4551289659137497e-06,
+      "loss": 64.9093,
+      "step": 96000
+    },
+    {
+      "epoch": 0.7942259171940274,
+      "grad_norm": 878.8954467773438,
+      "learning_rate": 1.454134467560349e-06,
+      "loss": 90.1892,
+      "step": 96010
+    },
+    {
+      "epoch": 0.7943086404433967,
+      "grad_norm": 1012.231689453125,
+      "learning_rate": 1.453140251335196e-06,
+      "loss": 85.0141,
+      "step": 96020
+    },
+    {
+      "epoch": 0.7943913636927659,
+      "grad_norm": 762.498291015625,
+      "learning_rate": 1.4521463173173966e-06,
+      "loss": 85.905,
+      "step": 96030
+    },
+    {
+      "epoch": 0.7944740869421351,
+      "grad_norm": 857.3758544921875,
+      "learning_rate": 1.4511526655860309e-06,
+      "loss": 94.0118,
+      "step": 96040
+    },
+    {
+      "epoch": 0.7945568101915044,
+      "grad_norm": 752.3488159179688,
+      "learning_rate": 1.4501592962201604e-06,
+      "loss": 79.3199,
+      "step": 96050
+    },
+    {
+      "epoch": 0.7946395334408736,
+      "grad_norm": 1673.278564453125,
+      "learning_rate": 1.4491662092988234e-06,
+      "loss": 145.9322,
+      "step": 96060
+    },
+    {
+      "epoch": 0.7947222566902428,
+      "grad_norm": 1270.6055908203125,
+      "learning_rate": 1.448173404901035e-06,
+      "loss": 113.5627,
+      "step": 96070
+    },
+    {
+      "epoch": 0.7948049799396121,
+      "grad_norm": 464.31982421875,
+      "learning_rate": 1.447180883105786e-06,
+      "loss": 75.3027,
+      "step": 96080
+    },
+    {
+      "epoch": 0.7948877031889813,
+      "grad_norm": 786.0927124023438,
+      "learning_rate": 1.446188643992051e-06,
+      "loss": 95.312,
+      "step": 96090
+    },
+    {
+      "epoch": 0.7949704264383505,
+      "grad_norm": 571.1138916015625,
+      "learning_rate": 1.4451966876387752e-06,
+      "loss": 61.4562,
+      "step": 96100
+    },
+    {
+      "epoch": 0.7950531496877198,
+      "grad_norm": 684.5860595703125,
+      "learning_rate": 1.4442050141248853e-06,
+      "loss": 63.8922,
+      "step": 96110
+    },
+    {
+      "epoch": 0.795135872937089,
+      "grad_norm": 753.0285034179688,
+      "learning_rate": 1.4432136235292848e-06,
+      "loss": 78.2068,
+      "step": 96120
+    },
+    {
+      "epoch": 0.7952185961864582,
+      "grad_norm": 1021.8491821289062,
+      "learning_rate": 1.442222515930851e-06,
+      "loss": 99.4391,
+      "step": 96130
+    },
+    {
+      "epoch": 0.7953013194358275,
+      "grad_norm": 650.785888671875,
+      "learning_rate": 1.441231691408444e-06,
+      "loss": 96.3516,
+      "step": 96140
+    },
+    {
+      "epoch": 0.7953840426851967,
+      "grad_norm": 446.81951904296875,
+      "learning_rate": 1.4402411500408985e-06,
+      "loss": 60.6617,
+      "step": 96150
+    },
+    {
+      "epoch": 0.7954667659345659,
+      "grad_norm": 918.9302978515625,
+      "learning_rate": 1.439250891907028e-06,
+      "loss": 92.1007,
+      "step": 96160
+    },
+    {
+      "epoch": 0.7955494891839352,
+      "grad_norm": 1528.4639892578125,
+      "learning_rate": 1.4382609170856222e-06,
+      "loss": 94.0727,
+      "step": 96170
+    },
+    {
+      "epoch": 0.7956322124333044,
+      "grad_norm": 991.0545654296875,
+      "learning_rate": 1.4372712256554471e-06,
+      "loss": 74.275,
+      "step": 96180
+    },
+    {
+      "epoch": 0.7957149356826736,
+      "grad_norm": 408.3299560546875,
+      "learning_rate": 1.436281817695252e-06,
+      "loss": 93.5512,
+      "step": 96190
+    },
+    {
+      "epoch": 0.7957976589320428,
+      "grad_norm": 305.2838439941406,
+      "learning_rate": 1.4352926932837591e-06,
+      "loss": 84.8283,
+      "step": 96200
+    },
+    {
+      "epoch": 0.7958803821814121,
+      "grad_norm": 1160.2215576171875,
+      "learning_rate": 1.4343038524996645e-06,
+      "loss": 73.67,
+      "step": 96210
+    },
+    {
+      "epoch": 0.7959631054307813,
+      "grad_norm": 838.9818725585938,
+      "learning_rate": 1.4333152954216483e-06,
+      "loss": 85.1046,
+      "step": 96220
+    },
+    {
+      "epoch": 0.7960458286801505,
+      "grad_norm": 1157.9669189453125,
+      "learning_rate": 1.4323270221283653e-06,
+      "loss": 96.3205,
+      "step": 96230
+    },
+    {
+      "epoch": 0.7961285519295198,
+      "grad_norm": 1076.052978515625,
+      "learning_rate": 1.4313390326984478e-06,
+      "loss": 111.263,
+      "step": 96240
+    },
+    {
+      "epoch": 0.796211275178889,
+      "grad_norm": 715.1856689453125,
+      "learning_rate": 1.4303513272105057e-06,
+      "loss": 78.668,
+      "step": 96250
+    },
+    {
+      "epoch": 0.7962939984282582,
+      "grad_norm": 793.4249877929688,
+      "learning_rate": 1.4293639057431258e-06,
+      "loss": 78.2108,
+      "step": 96260
+    },
+    {
+      "epoch": 0.7963767216776275,
+      "grad_norm": 841.6565551757812,
+      "learning_rate": 1.4283767683748711e-06,
+      "loss": 103.3483,
+      "step": 96270
+    },
+    {
+      "epoch": 0.7964594449269967,
+      "grad_norm": 1000.8982543945312,
+      "learning_rate": 1.4273899151842873e-06,
+      "loss": 90.5414,
+      "step": 96280
+    },
+    {
+      "epoch": 0.7965421681763659,
+      "grad_norm": 751.5543212890625,
+      "learning_rate": 1.4264033462498932e-06,
+      "loss": 72.9708,
+      "step": 96290
+    },
+    {
+      "epoch": 0.7966248914257352,
+      "grad_norm": 768.4910278320312,
+      "learning_rate": 1.4254170616501828e-06,
+      "loss": 84.1671,
+      "step": 96300
+    },
+    {
+      "epoch": 0.7967076146751044,
+      "grad_norm": 1033.3453369140625,
+      "learning_rate": 1.424431061463632e-06,
+      "loss": 81.5555,
+      "step": 96310
+    },
+    {
+      "epoch": 0.7967903379244736,
+      "grad_norm": 957.588623046875,
+      "learning_rate": 1.4234453457686914e-06,
+      "loss": 105.335,
+      "step": 96320
+    },
+    {
+      "epoch": 0.7968730611738429,
+      "grad_norm": 1199.724365234375,
+      "learning_rate": 1.4224599146437906e-06,
+      "loss": 87.4864,
+      "step": 96330
+    },
+    {
+      "epoch": 0.7969557844232121,
+      "grad_norm": 529.3333740234375,
+      "learning_rate": 1.4214747681673362e-06,
+      "loss": 83.0788,
+      "step": 96340
+    },
+    {
+      "epoch": 0.7970385076725813,
+      "grad_norm": 629.4002075195312,
+      "learning_rate": 1.4204899064177107e-06,
+      "loss": 89.5744,
+      "step": 96350
+    },
+    {
+      "epoch": 0.7971212309219506,
+      "grad_norm": 1154.12255859375,
+      "learning_rate": 1.4195053294732757e-06,
+      "loss": 110.1019,
+      "step": 96360
+    },
+    {
+      "epoch": 0.7972039541713198,
+      "grad_norm": 886.4882202148438,
+      "learning_rate": 1.4185210374123698e-06,
+      "loss": 64.0015,
+      "step": 96370
+    },
+    {
+      "epoch": 0.797286677420689,
+      "grad_norm": 1252.8187255859375,
+      "learning_rate": 1.417537030313308e-06,
+      "loss": 88.5995,
+      "step": 96380
+    },
+    {
+      "epoch": 0.7973694006700583,
+      "grad_norm": 1106.0509033203125,
+      "learning_rate": 1.416553308254383e-06,
+      "loss": 107.5624,
+      "step": 96390
+    },
+    {
+      "epoch": 0.7974521239194275,
+      "grad_norm": 608.2451782226562,
+      "learning_rate": 1.415569871313866e-06,
+      "loss": 101.158,
+      "step": 96400
+    },
+    {
+      "epoch": 0.7975348471687967,
+      "grad_norm": 2912.268798828125,
+      "learning_rate": 1.4145867195700036e-06,
+      "loss": 84.9362,
+      "step": 96410
+    },
+    {
+      "epoch": 0.797617570418166,
+      "grad_norm": 1384.496337890625,
+      "learning_rate": 1.4136038531010216e-06,
+      "loss": 102.7532,
+      "step": 96420
+    },
+    {
+      "epoch": 0.7977002936675353,
+      "grad_norm": 765.7507934570312,
+      "learning_rate": 1.4126212719851211e-06,
+      "loss": 101.0161,
+      "step": 96430
+    },
+    {
+      "epoch": 0.7977830169169045,
+      "grad_norm": 559.5115356445312,
+      "learning_rate": 1.411638976300483e-06,
+      "loss": 59.2422,
+      "step": 96440
+    },
+    {
+      "epoch": 0.7978657401662738,
+      "grad_norm": 683.1123046875,
+      "learning_rate": 1.4106569661252623e-06,
+      "loss": 78.2508,
+      "step": 96450
+    },
+    {
+      "epoch": 0.797948463415643,
+      "grad_norm": 1658.7459716796875,
+      "learning_rate": 1.4096752415375941e-06,
+      "loss": 71.4444,
+      "step": 96460
+    },
+    {
+      "epoch": 0.7980311866650122,
+      "grad_norm": 790.0877075195312,
+      "learning_rate": 1.40869380261559e-06,
+      "loss": 91.7176,
+      "step": 96470
+    },
+    {
+      "epoch": 0.7981139099143815,
+      "grad_norm": 841.804931640625,
+      "learning_rate": 1.4077126494373379e-06,
+      "loss": 81.0083,
+      "step": 96480
+    },
+    {
+      "epoch": 0.7981966331637507,
+      "grad_norm": 936.5332641601562,
+      "learning_rate": 1.406731782080904e-06,
+      "loss": 73.6172,
+      "step": 96490
+    },
+    {
+      "epoch": 0.7982793564131199,
+      "grad_norm": 620.3352661132812,
+      "learning_rate": 1.4057512006243312e-06,
+      "loss": 113.0523,
+      "step": 96500
+    },
+    {
+      "epoch": 0.7983620796624892,
+      "grad_norm": 785.4129638671875,
+      "learning_rate": 1.4047709051456398e-06,
+      "loss": 82.846,
+      "step": 96510
+    },
+    {
+      "epoch": 0.7984448029118584,
+      "grad_norm": 1122.926025390625,
+      "learning_rate": 1.403790895722828e-06,
+      "loss": 73.4351,
+      "step": 96520
+    },
+    {
+      "epoch": 0.7985275261612276,
+      "grad_norm": 1235.810546875,
+      "learning_rate": 1.40281117243387e-06,
+      "loss": 113.9662,
+      "step": 96530
+    },
+    {
+      "epoch": 0.7986102494105969,
+      "grad_norm": 608.0697021484375,
+      "learning_rate": 1.4018317353567185e-06,
+      "loss": 91.6859,
+      "step": 96540
+    },
+    {
+      "epoch": 0.7986929726599661,
+      "grad_norm": 847.255615234375,
+      "learning_rate": 1.4008525845693022e-06,
+      "loss": 103.3336,
+      "step": 96550
+    },
+    {
+      "epoch": 0.7987756959093353,
+      "grad_norm": 1009.511474609375,
+      "learning_rate": 1.3998737201495277e-06,
+      "loss": 79.1723,
+      "step": 96560
+    },
+    {
+      "epoch": 0.7988584191587046,
+      "grad_norm": 701.9151611328125,
+      "learning_rate": 1.3988951421752789e-06,
+      "loss": 81.0391,
+      "step": 96570
+    },
+    {
+      "epoch": 0.7989411424080738,
+      "grad_norm": 777.0698852539062,
+      "learning_rate": 1.3979168507244172e-06,
+      "loss": 94.3455,
+      "step": 96580
+    },
+    {
+      "epoch": 0.799023865657443,
+      "grad_norm": 523.3616943359375,
+      "learning_rate": 1.3969388458747802e-06,
+      "loss": 78.6874,
+      "step": 96590
+    },
+    {
+      "epoch": 0.7991065889068123,
+      "grad_norm": 876.171142578125,
+      "learning_rate": 1.3959611277041834e-06,
+      "loss": 85.3403,
+      "step": 96600
+    },
+    {
+      "epoch": 0.7991893121561815,
+      "grad_norm": 538.2999267578125,
+      "learning_rate": 1.394983696290419e-06,
+      "loss": 94.9753,
+      "step": 96610
+    },
+    {
+      "epoch": 0.7992720354055507,
+      "grad_norm": 890.1272583007812,
+      "learning_rate": 1.3940065517112579e-06,
+      "loss": 90.0231,
+      "step": 96620
+    },
+    {
+      "epoch": 0.79935475865492,
+      "grad_norm": 1206.276123046875,
+      "learning_rate": 1.3930296940444472e-06,
+      "loss": 85.5307,
+      "step": 96630
+    },
+    {
+      "epoch": 0.7994374819042892,
+      "grad_norm": 1130.0594482421875,
+      "learning_rate": 1.392053123367707e-06,
+      "loss": 89.702,
+      "step": 96640
+    },
+    {
+      "epoch": 0.7995202051536584,
+      "grad_norm": 827.4190063476562,
+      "learning_rate": 1.3910768397587427e-06,
+      "loss": 137.461,
+      "step": 96650
+    },
+    {
+      "epoch": 0.7996029284030277,
+      "grad_norm": 794.8823852539062,
+      "learning_rate": 1.3901008432952323e-06,
+      "loss": 111.7997,
+      "step": 96660
+    },
+    {
+      "epoch": 0.7996856516523969,
+      "grad_norm": 721.3433837890625,
+      "learning_rate": 1.38912513405483e-06,
+      "loss": 81.388,
+      "step": 96670
+    },
+    {
+      "epoch": 0.7997683749017661,
+      "grad_norm": 998.39990234375,
+      "learning_rate": 1.3881497121151694e-06,
+      "loss": 63.9328,
+      "step": 96680
+    },
+    {
+      "epoch": 0.7998510981511354,
+      "grad_norm": 1102.243408203125,
+      "learning_rate": 1.3871745775538598e-06,
+      "loss": 95.1184,
+      "step": 96690
+    },
+    {
+      "epoch": 0.7999338214005046,
+      "grad_norm": 737.71240234375,
+      "learning_rate": 1.3861997304484887e-06,
+      "loss": 87.6006,
+      "step": 96700
+    },
+    {
+      "epoch": 0.8000165446498738,
+      "grad_norm": 1060.644775390625,
+      "learning_rate": 1.3852251708766195e-06,
+      "loss": 82.6589,
+      "step": 96710
+    },
+    {
+      "epoch": 0.8000992678992431,
+      "grad_norm": 946.8339233398438,
+      "learning_rate": 1.384250898915796e-06,
+      "loss": 100.851,
+      "step": 96720
+    },
+    {
+      "epoch": 0.8001819911486123,
+      "grad_norm": 1012.8794555664062,
+      "learning_rate": 1.3832769146435327e-06,
+      "loss": 131.0507,
+      "step": 96730
+    },
+    {
+      "epoch": 0.8002647143979815,
+      "grad_norm": 1018.9071655273438,
+      "learning_rate": 1.3823032181373253e-06,
+      "loss": 84.4321,
+      "step": 96740
+    },
+    {
+      "epoch": 0.8003474376473508,
+      "grad_norm": 2128.4169921875,
+      "learning_rate": 1.3813298094746491e-06,
+      "loss": 97.8866,
+      "step": 96750
+    },
+    {
+      "epoch": 0.80043016089672,
+      "grad_norm": 691.6203002929688,
+      "learning_rate": 1.3803566887329528e-06,
+      "loss": 69.4865,
+      "step": 96760
+    },
+    {
+      "epoch": 0.8005128841460892,
+      "grad_norm": 894.628662109375,
+      "learning_rate": 1.3793838559896628e-06,
+      "loss": 97.1931,
+      "step": 96770
+    },
+    {
+      "epoch": 0.8005956073954585,
+      "grad_norm": 844.1419067382812,
+      "learning_rate": 1.3784113113221826e-06,
+      "loss": 85.3941,
+      "step": 96780
+    },
+    {
+      "epoch": 0.8006783306448277,
+      "grad_norm": 966.1434326171875,
+      "learning_rate": 1.3774390548078942e-06,
+      "loss": 91.669,
+      "step": 96790
+    },
+    {
+      "epoch": 0.8007610538941969,
+      "grad_norm": 842.3434448242188,
+      "learning_rate": 1.3764670865241557e-06,
+      "loss": 80.8415,
+      "step": 96800
+    },
+    {
+      "epoch": 0.8008437771435662,
+      "grad_norm": 607.7752685546875,
+      "learning_rate": 1.3754954065483006e-06,
+      "loss": 110.9996,
+      "step": 96810
+    },
+    {
+      "epoch": 0.8009265003929354,
+      "grad_norm": 1030.109130859375,
+      "learning_rate": 1.374524014957641e-06,
+      "loss": 79.0626,
+      "step": 96820
+    },
+    {
+      "epoch": 0.8010092236423046,
+      "grad_norm": 892.157470703125,
+      "learning_rate": 1.373552911829466e-06,
+      "loss": 96.9693,
+      "step": 96830
+    },
+    {
+      "epoch": 0.801091946891674,
+      "grad_norm": 1113.1424560546875,
+      "learning_rate": 1.3725820972410437e-06,
+      "loss": 83.1945,
+      "step": 96840
+    },
+    {
+      "epoch": 0.8011746701410432,
+      "grad_norm": 726.9800415039062,
+      "learning_rate": 1.3716115712696166e-06,
+      "loss": 92.1727,
+      "step": 96850
+    },
+    {
+      "epoch": 0.8012573933904124,
+      "grad_norm": 1125.0447998046875,
+      "learning_rate": 1.3706413339924047e-06,
+      "loss": 102.9361,
+      "step": 96860
+    },
+    {
+      "epoch": 0.8013401166397817,
+      "grad_norm": 452.7906799316406,
+      "learning_rate": 1.369671385486605e-06,
+      "loss": 91.2214,
+      "step": 96870
+    },
+    {
+      "epoch": 0.8014228398891509,
+      "grad_norm": 509.42926025390625,
+      "learning_rate": 1.3687017258293928e-06,
+      "loss": 82.8149,
+      "step": 96880
+    },
+    {
+      "epoch": 0.8015055631385201,
+      "grad_norm": 1304.8404541015625,
+      "learning_rate": 1.36773235509792e-06,
+      "loss": 123.0008,
+      "step": 96890
+    },
+    {
+      "epoch": 0.8015882863878894,
+      "grad_norm": 618.9688720703125,
+      "learning_rate": 1.366763273369312e-06,
+      "loss": 85.5639,
+      "step": 96900
+    },
+    {
+      "epoch": 0.8016710096372586,
+      "grad_norm": 613.8262329101562,
+      "learning_rate": 1.3657944807206764e-06,
+      "loss": 64.9996,
+      "step": 96910
+    },
+    {
+      "epoch": 0.8017537328866278,
+      "grad_norm": 841.429931640625,
+      "learning_rate": 1.3648259772290957e-06,
+      "loss": 109.687,
+      "step": 96920
+    },
+    {
+      "epoch": 0.801836456135997,
+      "grad_norm": 544.7089233398438,
+      "learning_rate": 1.3638577629716265e-06,
+      "loss": 68.7118,
+      "step": 96930
+    },
+    {
+      "epoch": 0.8019191793853663,
+      "grad_norm": 917.5126342773438,
+      "learning_rate": 1.3628898380253092e-06,
+      "loss": 85.4107,
+      "step": 96940
+    },
+    {
+      "epoch": 0.8020019026347355,
+      "grad_norm": 790.5340576171875,
+      "learning_rate": 1.3619222024671557e-06,
+      "loss": 83.7136,
+      "step": 96950
+    },
+    {
+      "epoch": 0.8020846258841047,
+      "grad_norm": 1110.8216552734375,
+      "learning_rate": 1.360954856374156e-06,
+      "loss": 105.1742,
+      "step": 96960
+    },
+    {
+      "epoch": 0.802167349133474,
+      "grad_norm": 620.2880859375,
+      "learning_rate": 1.3599877998232768e-06,
+      "loss": 78.4277,
+      "step": 96970
+    },
+    {
+      "epoch": 0.8022500723828432,
+      "grad_norm": 1336.2269287109375,
+      "learning_rate": 1.359021032891465e-06,
+      "loss": 87.1557,
+      "step": 96980
+    },
+    {
+      "epoch": 0.8023327956322124,
+      "grad_norm": 723.6621704101562,
+      "learning_rate": 1.358054555655638e-06,
+      "loss": 84.3839,
+      "step": 96990
+    },
+    {
+      "epoch": 0.8024155188815817,
+      "grad_norm": 838.581298828125,
+      "learning_rate": 1.357088368192696e-06,
+      "loss": 111.4981,
+      "step": 97000
+    },
+    {
+      "epoch": 0.8024982421309509,
+      "grad_norm": 564.4741821289062,
+      "learning_rate": 1.356122470579514e-06,
+      "loss": 67.0325,
+      "step": 97010
+    },
+    {
+      "epoch": 0.8025809653803201,
+      "grad_norm": 722.4137573242188,
+      "learning_rate": 1.3551568628929434e-06,
+      "loss": 66.7703,
+      "step": 97020
+    },
+    {
+      "epoch": 0.8026636886296894,
+      "grad_norm": 958.3905029296875,
+      "learning_rate": 1.3541915452098126e-06,
+      "loss": 101.0116,
+      "step": 97030
+    },
+    {
+      "epoch": 0.8027464118790586,
+      "grad_norm": 1039.2890625,
+      "learning_rate": 1.3532265176069298e-06,
+      "loss": 88.899,
+      "step": 97040
+    },
+    {
+      "epoch": 0.8028291351284278,
+      "grad_norm": 715.7427368164062,
+      "learning_rate": 1.3522617801610767e-06,
+      "loss": 91.0962,
+      "step": 97050
+    },
+    {
+      "epoch": 0.8029118583777971,
+      "grad_norm": 973.7153930664062,
+      "learning_rate": 1.3512973329490137e-06,
+      "loss": 88.5474,
+      "step": 97060
+    },
+    {
+      "epoch": 0.8029945816271663,
+      "grad_norm": 1014.022705078125,
+      "learning_rate": 1.3503331760474759e-06,
+      "loss": 78.5856,
+      "step": 97070
+    },
+    {
+      "epoch": 0.8030773048765355,
+      "grad_norm": 1352.9793701171875,
+      "learning_rate": 1.3493693095331773e-06,
+      "loss": 103.864,
+      "step": 97080
+    },
+    {
+      "epoch": 0.8031600281259048,
+      "grad_norm": 384.0733642578125,
+      "learning_rate": 1.3484057334828088e-06,
+      "loss": 79.577,
+      "step": 97090
+    },
+    {
+      "epoch": 0.803242751375274,
+      "grad_norm": 1254.0048828125,
+      "learning_rate": 1.3474424479730375e-06,
+      "loss": 108.2203,
+      "step": 97100
+    },
+    {
+      "epoch": 0.8033254746246432,
+      "grad_norm": 641.9199829101562,
+      "learning_rate": 1.3464794530805076e-06,
+      "loss": 83.1312,
+      "step": 97110
+    },
+    {
+      "epoch": 0.8034081978740125,
+      "grad_norm": 770.1970825195312,
+      "learning_rate": 1.34551674888184e-06,
+      "loss": 89.1073,
+      "step": 97120
+    },
+    {
+      "epoch": 0.8034909211233817,
+      "grad_norm": 795.2720947265625,
+      "learning_rate": 1.3445543354536317e-06,
+      "loss": 134.2826,
+      "step": 97130
+    },
+    {
+      "epoch": 0.8035736443727509,
+      "grad_norm": 371.7003479003906,
+      "learning_rate": 1.3435922128724599e-06,
+      "loss": 85.9229,
+      "step": 97140
+    },
+    {
+      "epoch": 0.8036563676221202,
+      "grad_norm": 1355.1334228515625,
+      "learning_rate": 1.3426303812148766e-06,
+      "loss": 116.6411,
+      "step": 97150
+    },
+    {
+      "epoch": 0.8037390908714894,
+      "grad_norm": 1486.0423583984375,
+      "learning_rate": 1.3416688405574074e-06,
+      "loss": 83.7708,
+      "step": 97160
+    },
+    {
+      "epoch": 0.8038218141208586,
+      "grad_norm": 490.289794921875,
+      "learning_rate": 1.3407075909765593e-06,
+      "loss": 103.2564,
+      "step": 97170
+    },
+    {
+      "epoch": 0.8039045373702279,
+      "grad_norm": 699.4607543945312,
+      "learning_rate": 1.339746632548814e-06,
+      "loss": 106.1111,
+      "step": 97180
+    },
+    {
+      "epoch": 0.8039872606195971,
+      "grad_norm": 749.7698974609375,
+      "learning_rate": 1.3387859653506314e-06,
+      "loss": 96.1973,
+      "step": 97190
+    },
+    {
+      "epoch": 0.8040699838689663,
+      "grad_norm": 257.79119873046875,
+      "learning_rate": 1.3378255894584463e-06,
+      "loss": 93.4603,
+      "step": 97200
+    },
+    {
+      "epoch": 0.8041527071183356,
+      "grad_norm": 482.4723815917969,
+      "learning_rate": 1.3368655049486717e-06,
+      "loss": 114.9346,
+      "step": 97210
+    },
+    {
+      "epoch": 0.8042354303677048,
+      "grad_norm": 308.8788757324219,
+      "learning_rate": 1.3359057118976976e-06,
+      "loss": 88.0903,
+      "step": 97220
+    },
+    {
+      "epoch": 0.804318153617074,
+      "grad_norm": 932.05419921875,
+      "learning_rate": 1.3349462103818906e-06,
+      "loss": 76.6318,
+      "step": 97230
+    },
+    {
+      "epoch": 0.8044008768664433,
+      "grad_norm": 3546.006103515625,
+      "learning_rate": 1.3339870004775929e-06,
+      "loss": 82.4357,
+      "step": 97240
+    },
+    {
+      "epoch": 0.8044836001158125,
+      "grad_norm": 689.85302734375,
+      "learning_rate": 1.3330280822611246e-06,
+      "loss": 94.2522,
+      "step": 97250
+    },
+    {
+      "epoch": 0.8045663233651817,
+      "grad_norm": 322.86669921875,
+      "learning_rate": 1.3320694558087832e-06,
+      "loss": 76.2741,
+      "step": 97260
+    },
+    {
+      "epoch": 0.8046490466145511,
+      "grad_norm": 1147.7379150390625,
+      "learning_rate": 1.3311111211968414e-06,
+      "loss": 87.9512,
+      "step": 97270
+    },
+    {
+      "epoch": 0.8047317698639203,
+      "grad_norm": 778.215087890625,
+      "learning_rate": 1.3301530785015492e-06,
+      "loss": 68.0114,
+      "step": 97280
+    },
+    {
+      "epoch": 0.8048144931132895,
+      "grad_norm": 918.3253173828125,
+      "learning_rate": 1.329195327799135e-06,
+      "loss": 85.8742,
+      "step": 97290
+    },
+    {
+      "epoch": 0.8048972163626588,
+      "grad_norm": 1470.5067138671875,
+      "learning_rate": 1.3282378691658015e-06,
+      "loss": 80.2126,
+      "step": 97300
+    },
+    {
+      "epoch": 0.804979939612028,
+      "grad_norm": 682.7937622070312,
+      "learning_rate": 1.3272807026777302e-06,
+      "loss": 77.8562,
+      "step": 97310
+    },
+    {
+      "epoch": 0.8050626628613972,
+      "grad_norm": 711.0369262695312,
+      "learning_rate": 1.3263238284110769e-06,
+      "loss": 88.1487,
+      "step": 97320
+    },
+    {
+      "epoch": 0.8051453861107665,
+      "grad_norm": 711.9708251953125,
+      "learning_rate": 1.3253672464419776e-06,
+      "loss": 80.779,
+      "step": 97330
+    },
+    {
+      "epoch": 0.8052281093601357,
+      "grad_norm": 892.573486328125,
+      "learning_rate": 1.3244109568465414e-06,
+      "loss": 106.5878,
+      "step": 97340
+    },
+    {
+      "epoch": 0.8053108326095049,
+      "grad_norm": 895.7553100585938,
+      "learning_rate": 1.3234549597008572e-06,
+      "loss": 84.2314,
+      "step": 97350
+    },
+    {
+      "epoch": 0.8053935558588742,
+      "grad_norm": 486.6683349609375,
+      "learning_rate": 1.322499255080989e-06,
+      "loss": 66.8342,
+      "step": 97360
+    },
+    {
+      "epoch": 0.8054762791082434,
+      "grad_norm": 531.82763671875,
+      "learning_rate": 1.3215438430629774e-06,
+      "loss": 66.0651,
+      "step": 97370
+    },
+    {
+      "epoch": 0.8055590023576126,
+      "grad_norm": 902.6575317382812,
+      "learning_rate": 1.3205887237228399e-06,
+      "loss": 80.48,
+      "step": 97380
+    },
+    {
+      "epoch": 0.8056417256069819,
+      "grad_norm": 784.8746948242188,
+      "learning_rate": 1.3196338971365719e-06,
+      "loss": 80.4638,
+      "step": 97390
+    },
+    {
+      "epoch": 0.8057244488563511,
+      "grad_norm": 715.1857299804688,
+      "learning_rate": 1.3186793633801443e-06,
+      "loss": 91.9943,
+      "step": 97400
+    },
+    {
+      "epoch": 0.8058071721057203,
+      "grad_norm": 620.5697631835938,
+      "learning_rate": 1.317725122529504e-06,
+      "loss": 79.546,
+      "step": 97410
+    },
+    {
+      "epoch": 0.8058898953550896,
+      "grad_norm": 688.434326171875,
+      "learning_rate": 1.3167711746605771e-06,
+      "loss": 88.8246,
+      "step": 97420
+    },
+    {
+      "epoch": 0.8059726186044588,
+      "grad_norm": 629.0154418945312,
+      "learning_rate": 1.315817519849264e-06,
+      "loss": 78.7018,
+      "step": 97430
+    },
+    {
+      "epoch": 0.806055341853828,
+      "grad_norm": 568.8397216796875,
+      "learning_rate": 1.3148641581714421e-06,
+      "loss": 64.2857,
+      "step": 97440
+    },
+    {
+      "epoch": 0.8061380651031973,
+      "grad_norm": 1156.6614990234375,
+      "learning_rate": 1.3139110897029672e-06,
+      "loss": 80.8269,
+      "step": 97450
+    },
+    {
+      "epoch": 0.8062207883525665,
+      "grad_norm": 939.9249267578125,
+      "learning_rate": 1.3129583145196701e-06,
+      "loss": 98.3824,
+      "step": 97460
+    },
+    {
+      "epoch": 0.8063035116019357,
+      "grad_norm": 677.0330810546875,
+      "learning_rate": 1.3120058326973583e-06,
+      "loss": 80.2276,
+      "step": 97470
+    },
+    {
+      "epoch": 0.806386234851305,
+      "grad_norm": 733.0352172851562,
+      "learning_rate": 1.3110536443118172e-06,
+      "loss": 100.2774,
+      "step": 97480
+    },
+    {
+      "epoch": 0.8064689581006742,
+      "grad_norm": 895.4539794921875,
+      "learning_rate": 1.3101017494388074e-06,
+      "loss": 88.9246,
+      "step": 97490
+    },
+    {
+      "epoch": 0.8065516813500434,
+      "grad_norm": 1752.5474853515625,
+      "learning_rate": 1.3091501481540676e-06,
+      "loss": 83.0293,
+      "step": 97500
+    },
+    {
+      "epoch": 0.8066344045994127,
+      "grad_norm": 781.7203979492188,
+      "learning_rate": 1.3081988405333106e-06,
+      "loss": 95.9738,
+      "step": 97510
+    },
+    {
+      "epoch": 0.8067171278487819,
+      "grad_norm": 1117.3587646484375,
+      "learning_rate": 1.3072478266522298e-06,
+      "loss": 97.9509,
+      "step": 97520
+    },
+    {
+      "epoch": 0.8067998510981511,
+      "grad_norm": 1128.07177734375,
+      "learning_rate": 1.3062971065864915e-06,
+      "loss": 90.2064,
+      "step": 97530
+    },
+    {
+      "epoch": 0.8068825743475204,
+      "grad_norm": 563.0820922851562,
+      "learning_rate": 1.305346680411741e-06,
+      "loss": 97.338,
+      "step": 97540
+    },
+    {
+      "epoch": 0.8069652975968896,
+      "grad_norm": 1097.416748046875,
+      "learning_rate": 1.3043965482035987e-06,
+      "loss": 94.4323,
+      "step": 97550
+    },
+    {
+      "epoch": 0.8070480208462588,
+      "grad_norm": 1162.277587890625,
+      "learning_rate": 1.3034467100376624e-06,
+      "loss": 112.2583,
+      "step": 97560
+    },
+    {
+      "epoch": 0.8071307440956281,
+      "grad_norm": 708.5651245117188,
+      "learning_rate": 1.3024971659895069e-06,
+      "loss": 72.2733,
+      "step": 97570
+    },
+    {
+      "epoch": 0.8072134673449973,
+      "grad_norm": 877.0707397460938,
+      "learning_rate": 1.3015479161346839e-06,
+      "loss": 72.6177,
+      "step": 97580
+    },
+    {
+      "epoch": 0.8072961905943665,
+      "grad_norm": 622.7434692382812,
+      "learning_rate": 1.3005989605487168e-06,
+      "loss": 85.8667,
+      "step": 97590
+    },
+    {
+      "epoch": 0.8073789138437358,
+      "grad_norm": 1124.309326171875,
+      "learning_rate": 1.2996502993071137e-06,
+      "loss": 128.3939,
+      "step": 97600
+    },
+    {
+      "epoch": 0.807461637093105,
+      "grad_norm": 959.4961547851562,
+      "learning_rate": 1.2987019324853539e-06,
+      "loss": 86.4192,
+      "step": 97610
+    },
+    {
+      "epoch": 0.8075443603424742,
+      "grad_norm": 926.9354248046875,
+      "learning_rate": 1.2977538601588951e-06,
+      "loss": 83.3312,
+      "step": 97620
+    },
+    {
+      "epoch": 0.8076270835918435,
+      "grad_norm": 765.6702270507812,
+      "learning_rate": 1.2968060824031704e-06,
+      "loss": 100.2946,
+      "step": 97630
+    },
+    {
+      "epoch": 0.8077098068412127,
+      "grad_norm": 663.2040405273438,
+      "learning_rate": 1.29585859929359e-06,
+      "loss": 92.4886,
+      "step": 97640
+    },
+    {
+      "epoch": 0.8077925300905819,
+      "grad_norm": 1108.396484375,
+      "learning_rate": 1.2949114109055417e-06,
+      "loss": 64.0099,
+      "step": 97650
+    },
+    {
+      "epoch": 0.8078752533399511,
+      "grad_norm": 627.171142578125,
+      "learning_rate": 1.2939645173143894e-06,
+      "loss": 86.472,
+      "step": 97660
+    },
+    {
+      "epoch": 0.8079579765893204,
+      "grad_norm": 2170.154541015625,
+      "learning_rate": 1.293017918595471e-06,
+      "loss": 131.4853,
+      "step": 97670
+    },
+    {
+      "epoch": 0.8080406998386896,
+      "grad_norm": 752.8032836914062,
+      "learning_rate": 1.2920716148241036e-06,
+      "loss": 90.6649,
+      "step": 97680
+    },
+    {
+      "epoch": 0.8081234230880588,
+      "grad_norm": 707.9442138671875,
+      "learning_rate": 1.2911256060755794e-06,
+      "loss": 95.8097,
+      "step": 97690
+    },
+    {
+      "epoch": 0.8082061463374282,
+      "grad_norm": 716.3200073242188,
+      "learning_rate": 1.2901798924251712e-06,
+      "loss": 83.3109,
+      "step": 97700
+    },
+    {
+      "epoch": 0.8082888695867974,
+      "grad_norm": 769.82763671875,
+      "learning_rate": 1.2892344739481228e-06,
+      "loss": 91.1391,
+      "step": 97710
+    },
+    {
+      "epoch": 0.8083715928361666,
+      "grad_norm": 720.6710205078125,
+      "learning_rate": 1.288289350719657e-06,
+      "loss": 69.3394,
+      "step": 97720
+    },
+    {
+      "epoch": 0.8084543160855359,
+      "grad_norm": 1892.12451171875,
+      "learning_rate": 1.2873445228149733e-06,
+      "loss": 99.0175,
+      "step": 97730
+    },
+    {
+      "epoch": 0.8085370393349051,
+      "grad_norm": 1591.5841064453125,
+      "learning_rate": 1.2863999903092473e-06,
+      "loss": 87.3172,
+      "step": 97740
+    },
+    {
+      "epoch": 0.8086197625842743,
+      "grad_norm": 2348.568603515625,
+      "learning_rate": 1.2854557532776323e-06,
+      "loss": 88.4138,
+      "step": 97750
+    },
+    {
+      "epoch": 0.8087024858336436,
+      "grad_norm": 3484.744873046875,
+      "learning_rate": 1.2845118117952544e-06,
+      "loss": 120.2466,
+      "step": 97760
+    },
+    {
+      "epoch": 0.8087852090830128,
+      "grad_norm": 1318.188720703125,
+      "learning_rate": 1.2835681659372196e-06,
+      "loss": 66.7873,
+      "step": 97770
+    },
+    {
+      "epoch": 0.808867932332382,
+      "grad_norm": 612.43408203125,
+      "learning_rate": 1.282624815778608e-06,
+      "loss": 92.2864,
+      "step": 97780
+    },
+    {
+      "epoch": 0.8089506555817513,
+      "grad_norm": 984.0529174804688,
+      "learning_rate": 1.281681761394481e-06,
+      "loss": 67.4189,
+      "step": 97790
+    },
+    {
+      "epoch": 0.8090333788311205,
+      "grad_norm": 486.3183288574219,
+      "learning_rate": 1.2807390028598715e-06,
+      "loss": 81.0175,
+      "step": 97800
+    },
+    {
+      "epoch": 0.8091161020804897,
+      "grad_norm": 899.5704956054688,
+      "learning_rate": 1.2797965402497902e-06,
+      "loss": 67.1271,
+      "step": 97810
+    },
+    {
+      "epoch": 0.809198825329859,
+      "grad_norm": 440.5632019042969,
+      "learning_rate": 1.278854373639225e-06,
+      "loss": 75.3582,
+      "step": 97820
+    },
+    {
+      "epoch": 0.8092815485792282,
+      "grad_norm": 1131.36083984375,
+      "learning_rate": 1.2779125031031413e-06,
+      "loss": 95.9433,
+      "step": 97830
+    },
+    {
+      "epoch": 0.8093642718285974,
+      "grad_norm": 1148.0606689453125,
+      "learning_rate": 1.2769709287164755e-06,
+      "loss": 91.1547,
+      "step": 97840
+    },
+    {
+      "epoch": 0.8094469950779667,
+      "grad_norm": 1007.1502685546875,
+      "learning_rate": 1.2760296505541469e-06,
+      "loss": 128.5402,
+      "step": 97850
+    },
+    {
+      "epoch": 0.8095297183273359,
+      "grad_norm": 675.3720703125,
+      "learning_rate": 1.2750886686910485e-06,
+      "loss": 82.86,
+      "step": 97860
+    },
+    {
+      "epoch": 0.8096124415767051,
+      "grad_norm": 554.8408813476562,
+      "learning_rate": 1.2741479832020492e-06,
+      "loss": 103.5147,
+      "step": 97870
+    },
+    {
+      "epoch": 0.8096951648260744,
+      "grad_norm": 1001.0286865234375,
+      "learning_rate": 1.2732075941619948e-06,
+      "loss": 82.1101,
+      "step": 97880
+    },
+    {
+      "epoch": 0.8097778880754436,
+      "grad_norm": 778.3543090820312,
+      "learning_rate": 1.2722675016457091e-06,
+      "loss": 81.8874,
+      "step": 97890
+    },
+    {
+      "epoch": 0.8098606113248128,
+      "grad_norm": 948.7130737304688,
+      "learning_rate": 1.271327705727991e-06,
+      "loss": 99.7648,
+      "step": 97900
+    },
+    {
+      "epoch": 0.8099433345741821,
+      "grad_norm": 3270.607421875,
+      "learning_rate": 1.2703882064836142e-06,
+      "loss": 103.626,
+      "step": 97910
+    },
+    {
+      "epoch": 0.8100260578235513,
+      "grad_norm": 953.6922607421875,
+      "learning_rate": 1.2694490039873336e-06,
+      "loss": 93.171,
+      "step": 97920
+    },
+    {
+      "epoch": 0.8101087810729205,
+      "grad_norm": 762.2303466796875,
+      "learning_rate": 1.2685100983138731e-06,
+      "loss": 63.3839,
+      "step": 97930
+    },
+    {
+      "epoch": 0.8101915043222898,
+      "grad_norm": 976.2236328125,
+      "learning_rate": 1.2675714895379387e-06,
+      "loss": 94.6773,
+      "step": 97940
+    },
+    {
+      "epoch": 0.810274227571659,
+      "grad_norm": 606.363525390625,
+      "learning_rate": 1.2666331777342112e-06,
+      "loss": 81.9096,
+      "step": 97950
+    },
+    {
+      "epoch": 0.8103569508210282,
+      "grad_norm": 1317.268798828125,
+      "learning_rate": 1.265695162977348e-06,
+      "loss": 93.898,
+      "step": 97960
+    },
+    {
+      "epoch": 0.8104396740703975,
+      "grad_norm": 738.7415161132812,
+      "learning_rate": 1.2647574453419826e-06,
+      "loss": 63.2881,
+      "step": 97970
+    },
+    {
+      "epoch": 0.8105223973197667,
+      "grad_norm": 1187.66162109375,
+      "learning_rate": 1.2638200249027233e-06,
+      "loss": 84.0641,
+      "step": 97980
+    },
+    {
+      "epoch": 0.8106051205691359,
+      "grad_norm": 1476.5244140625,
+      "learning_rate": 1.2628829017341594e-06,
+      "loss": 133.0605,
+      "step": 97990
+    },
+    {
+      "epoch": 0.8106878438185052,
+      "grad_norm": 747.8192749023438,
+      "learning_rate": 1.2619460759108521e-06,
+      "loss": 98.8314,
+      "step": 98000
+    },
+    {
+      "epoch": 0.8107705670678744,
+      "grad_norm": 760.4857177734375,
+      "learning_rate": 1.2610095475073415e-06,
+      "loss": 64.8025,
+      "step": 98010
+    },
+    {
+      "epoch": 0.8108532903172436,
+      "grad_norm": 1347.1624755859375,
+      "learning_rate": 1.26007331659814e-06,
+      "loss": 89.3215,
+      "step": 98020
+    },
+    {
+      "epoch": 0.8109360135666129,
+      "grad_norm": 778.6011352539062,
+      "learning_rate": 1.2591373832577408e-06,
+      "loss": 83.764,
+      "step": 98030
+    },
+    {
+      "epoch": 0.8110187368159821,
+      "grad_norm": 825.260498046875,
+      "learning_rate": 1.2582017475606117e-06,
+      "loss": 106.7728,
+      "step": 98040
+    },
+    {
+      "epoch": 0.8111014600653513,
+      "grad_norm": 1110.2823486328125,
+      "learning_rate": 1.2572664095811976e-06,
+      "loss": 93.7398,
+      "step": 98050
+    },
+    {
+      "epoch": 0.8111841833147206,
+      "grad_norm": 764.0325317382812,
+      "learning_rate": 1.2563313693939177e-06,
+      "loss": 89.7206,
+      "step": 98060
+    },
+    {
+      "epoch": 0.8112669065640898,
+      "grad_norm": 744.2551879882812,
+      "learning_rate": 1.25539662707317e-06,
+      "loss": 72.9195,
+      "step": 98070
+    },
+    {
+      "epoch": 0.811349629813459,
+      "grad_norm": 857.3899536132812,
+      "learning_rate": 1.2544621826933257e-06,
+      "loss": 84.4764,
+      "step": 98080
+    },
+    {
+      "epoch": 0.8114323530628283,
+      "grad_norm": 1150.891845703125,
+      "learning_rate": 1.2535280363287388e-06,
+      "loss": 73.521,
+      "step": 98090
+    },
+    {
+      "epoch": 0.8115150763121975,
+      "grad_norm": 585.0152587890625,
+      "learning_rate": 1.2525941880537307e-06,
+      "loss": 74.5685,
+      "step": 98100
+    },
+    {
+      "epoch": 0.8115977995615667,
+      "grad_norm": 927.9107666015625,
+      "learning_rate": 1.2516606379426044e-06,
+      "loss": 58.3165,
+      "step": 98110
+    },
+    {
+      "epoch": 0.8116805228109361,
+      "grad_norm": 997.0220336914062,
+      "learning_rate": 1.250727386069639e-06,
+      "loss": 102.8923,
+      "step": 98120
+    },
+    {
+      "epoch": 0.8117632460603053,
+      "grad_norm": 918.0977783203125,
+      "learning_rate": 1.2497944325090882e-06,
+      "loss": 92.9588,
+      "step": 98130
+    },
+    {
+      "epoch": 0.8118459693096745,
+      "grad_norm": 437.514404296875,
+      "learning_rate": 1.248861777335184e-06,
+      "loss": 74.5204,
+      "step": 98140
+    },
+    {
+      "epoch": 0.8119286925590438,
+      "grad_norm": 873.174072265625,
+      "learning_rate": 1.247929420622132e-06,
+      "loss": 76.6309,
+      "step": 98150
+    },
+    {
+      "epoch": 0.812011415808413,
+      "grad_norm": 1936.2078857421875,
+      "learning_rate": 1.2469973624441168e-06,
+      "loss": 133.0517,
+      "step": 98160
+    },
+    {
+      "epoch": 0.8120941390577822,
+      "grad_norm": 568.5994262695312,
+      "learning_rate": 1.2460656028752976e-06,
+      "loss": 93.9224,
+      "step": 98170
+    },
+    {
+      "epoch": 0.8121768623071515,
+      "grad_norm": 882.7687377929688,
+      "learning_rate": 1.24513414198981e-06,
+      "loss": 120.2803,
+      "step": 98180
+    },
+    {
+      "epoch": 0.8122595855565207,
+      "grad_norm": 753.4219970703125,
+      "learning_rate": 1.244202979861766e-06,
+      "loss": 78.88,
+      "step": 98190
+    },
+    {
+      "epoch": 0.8123423088058899,
+      "grad_norm": 552.3338012695312,
+      "learning_rate": 1.2432721165652544e-06,
+      "loss": 56.6679,
+      "step": 98200
+    },
+    {
+      "epoch": 0.8124250320552592,
+      "grad_norm": 948.5594482421875,
+      "learning_rate": 1.2423415521743392e-06,
+      "loss": 77.7815,
+      "step": 98210
+    },
+    {
+      "epoch": 0.8125077553046284,
+      "grad_norm": 668.4248657226562,
+      "learning_rate": 1.2414112867630619e-06,
+      "loss": 82.7063,
+      "step": 98220
+    },
+    {
+      "epoch": 0.8125904785539976,
+      "grad_norm": 507.14691162109375,
+      "learning_rate": 1.2404813204054383e-06,
+      "loss": 77.7765,
+      "step": 98230
+    },
+    {
+      "epoch": 0.8126732018033669,
+      "grad_norm": 1215.5714111328125,
+      "learning_rate": 1.239551653175462e-06,
+      "loss": 132.3814,
+      "step": 98240
+    },
+    {
+      "epoch": 0.8127559250527361,
+      "grad_norm": 711.6063232421875,
+      "learning_rate": 1.238622285147103e-06,
+      "loss": 81.4849,
+      "step": 98250
+    },
+    {
+      "epoch": 0.8128386483021053,
+      "grad_norm": 1013.1691284179688,
+      "learning_rate": 1.237693216394306e-06,
+      "loss": 117.3906,
+      "step": 98260
+    },
+    {
+      "epoch": 0.8129213715514746,
+      "grad_norm": 735.0127563476562,
+      "learning_rate": 1.236764446990994e-06,
+      "loss": 75.7862,
+      "step": 98270
+    },
+    {
+      "epoch": 0.8130040948008438,
+      "grad_norm": 728.4838256835938,
+      "learning_rate": 1.2358359770110634e-06,
+      "loss": 64.463,
+      "step": 98280
+    },
+    {
+      "epoch": 0.813086818050213,
+      "grad_norm": 746.21044921875,
+      "learning_rate": 1.2349078065283886e-06,
+      "loss": 84.8162,
+      "step": 98290
+    },
+    {
+      "epoch": 0.8131695412995823,
+      "grad_norm": 745.414306640625,
+      "learning_rate": 1.2339799356168207e-06,
+      "loss": 91.913,
+      "step": 98300
+    },
+    {
+      "epoch": 0.8132522645489515,
+      "grad_norm": 788.8545532226562,
+      "learning_rate": 1.2330523643501858e-06,
+      "loss": 73.4357,
+      "step": 98310
+    },
+    {
+      "epoch": 0.8133349877983207,
+      "grad_norm": 1031.7391357421875,
+      "learning_rate": 1.2321250928022855e-06,
+      "loss": 94.2501,
+      "step": 98320
+    },
+    {
+      "epoch": 0.81341771104769,
+      "grad_norm": 723.92919921875,
+      "learning_rate": 1.2311981210468998e-06,
+      "loss": 80.1034,
+      "step": 98330
+    },
+    {
+      "epoch": 0.8135004342970592,
+      "grad_norm": 1015.01220703125,
+      "learning_rate": 1.2302714491577834e-06,
+      "loss": 126.1582,
+      "step": 98340
+    },
+    {
+      "epoch": 0.8135831575464284,
+      "grad_norm": 718.5742797851562,
+      "learning_rate": 1.2293450772086667e-06,
+      "loss": 60.2428,
+      "step": 98350
+    },
+    {
+      "epoch": 0.8136658807957977,
+      "grad_norm": 796.4895629882812,
+      "learning_rate": 1.2284190052732575e-06,
+      "loss": 75.6503,
+      "step": 98360
+    },
+    {
+      "epoch": 0.8137486040451669,
+      "grad_norm": 479.49462890625,
+      "learning_rate": 1.2274932334252387e-06,
+      "loss": 111.6113,
+      "step": 98370
+    },
+    {
+      "epoch": 0.8138313272945361,
+      "grad_norm": 723.4696655273438,
+      "learning_rate": 1.2265677617382698e-06,
+      "loss": 76.3011,
+      "step": 98380
+    },
+    {
+      "epoch": 0.8139140505439053,
+      "grad_norm": 791.28173828125,
+      "learning_rate": 1.2256425902859864e-06,
+      "loss": 78.6211,
+      "step": 98390
+    },
+    {
+      "epoch": 0.8139967737932746,
+      "grad_norm": 1258.62548828125,
+      "learning_rate": 1.224717719142e-06,
+      "loss": 89.4393,
+      "step": 98400
+    },
+    {
+      "epoch": 0.8140794970426438,
+      "grad_norm": 820.9175415039062,
+      "learning_rate": 1.2237931483798993e-06,
+      "loss": 107.2661,
+      "step": 98410
+    },
+    {
+      "epoch": 0.814162220292013,
+      "grad_norm": 1227.457275390625,
+      "learning_rate": 1.2228688780732463e-06,
+      "loss": 71.8221,
+      "step": 98420
+    },
+    {
+      "epoch": 0.8142449435413823,
+      "grad_norm": 548.6934204101562,
+      "learning_rate": 1.2219449082955825e-06,
+      "loss": 115.0666,
+      "step": 98430
+    },
+    {
+      "epoch": 0.8143276667907515,
+      "grad_norm": 997.916259765625,
+      "learning_rate": 1.2210212391204234e-06,
+      "loss": 82.2138,
+      "step": 98440
+    },
+    {
+      "epoch": 0.8144103900401207,
+      "grad_norm": 1067.23486328125,
+      "learning_rate": 1.2200978706212606e-06,
+      "loss": 70.915,
+      "step": 98450
+    },
+    {
+      "epoch": 0.81449311328949,
+      "grad_norm": 1409.0748291015625,
+      "learning_rate": 1.2191748028715632e-06,
+      "loss": 64.1739,
+      "step": 98460
+    },
+    {
+      "epoch": 0.8145758365388592,
+      "grad_norm": 1686.0634765625,
+      "learning_rate": 1.2182520359447753e-06,
+      "loss": 88.7429,
+      "step": 98470
+    },
+    {
+      "epoch": 0.8146585597882284,
+      "grad_norm": 990.0397338867188,
+      "learning_rate": 1.2173295699143172e-06,
+      "loss": 106.0889,
+      "step": 98480
+    },
+    {
+      "epoch": 0.8147412830375977,
+      "grad_norm": 991.7344360351562,
+      "learning_rate": 1.2164074048535846e-06,
+      "loss": 95.5769,
+      "step": 98490
+    },
+    {
+      "epoch": 0.8148240062869669,
+      "grad_norm": 1526.431396484375,
+      "learning_rate": 1.2154855408359507e-06,
+      "loss": 99.9895,
+      "step": 98500
+    },
+    {
+      "epoch": 0.8149067295363361,
+      "grad_norm": 802.585205078125,
+      "learning_rate": 1.2145639779347634e-06,
+      "loss": 83.8027,
+      "step": 98510
+    },
+    {
+      "epoch": 0.8149894527857054,
+      "grad_norm": 806.9586181640625,
+      "learning_rate": 1.2136427162233493e-06,
+      "loss": 82.421,
+      "step": 98520
+    },
+    {
+      "epoch": 0.8150721760350746,
+      "grad_norm": 810.6295166015625,
+      "learning_rate": 1.212721755775006e-06,
+      "loss": 98.6951,
+      "step": 98530
+    },
+    {
+      "epoch": 0.8151548992844438,
+      "grad_norm": 912.5408325195312,
+      "learning_rate": 1.2118010966630095e-06,
+      "loss": 83.3605,
+      "step": 98540
+    },
+    {
+      "epoch": 0.8152376225338132,
+      "grad_norm": 1080.7371826171875,
+      "learning_rate": 1.210880738960616e-06,
+      "loss": 84.3936,
+      "step": 98550
+    },
+    {
+      "epoch": 0.8153203457831824,
+      "grad_norm": 572.9903564453125,
+      "learning_rate": 1.2099606827410521e-06,
+      "loss": 64.3546,
+      "step": 98560
+    },
+    {
+      "epoch": 0.8154030690325516,
+      "grad_norm": 914.2869873046875,
+      "learning_rate": 1.2090409280775228e-06,
+      "loss": 94.1465,
+      "step": 98570
+    },
+    {
+      "epoch": 0.8154857922819209,
+      "grad_norm": 880.4005126953125,
+      "learning_rate": 1.208121475043209e-06,
+      "loss": 79.7238,
+      "step": 98580
+    },
+    {
+      "epoch": 0.8155685155312901,
+      "grad_norm": 1575.5872802734375,
+      "learning_rate": 1.2072023237112668e-06,
+      "loss": 102.0344,
+      "step": 98590
+    },
+    {
+      "epoch": 0.8156512387806593,
+      "grad_norm": 1086.993408203125,
+      "learning_rate": 1.2062834741548291e-06,
+      "loss": 90.5145,
+      "step": 98600
+    },
+    {
+      "epoch": 0.8157339620300286,
+      "grad_norm": 759.2196655273438,
+      "learning_rate": 1.2053649264470064e-06,
+      "loss": 93.4888,
+      "step": 98610
+    },
+    {
+      "epoch": 0.8158166852793978,
+      "grad_norm": 939.29736328125,
+      "learning_rate": 1.2044466806608794e-06,
+      "loss": 66.8826,
+      "step": 98620
+    },
+    {
+      "epoch": 0.815899408528767,
+      "grad_norm": 691.4633178710938,
+      "learning_rate": 1.203528736869511e-06,
+      "loss": 75.7453,
+      "step": 98630
+    },
+    {
+      "epoch": 0.8159821317781363,
+      "grad_norm": 843.3115234375,
+      "learning_rate": 1.2026110951459364e-06,
+      "loss": 104.9937,
+      "step": 98640
+    },
+    {
+      "epoch": 0.8160648550275055,
+      "grad_norm": 707.0924682617188,
+      "learning_rate": 1.2016937555631702e-06,
+      "loss": 66.0707,
+      "step": 98650
+    },
+    {
+      "epoch": 0.8161475782768747,
+      "grad_norm": 1021.8479614257812,
+      "learning_rate": 1.2007767181942003e-06,
+      "loss": 99.7831,
+      "step": 98660
+    },
+    {
+      "epoch": 0.816230301526244,
+      "grad_norm": 850.4284057617188,
+      "learning_rate": 1.1998599831119912e-06,
+      "loss": 75.5687,
+      "step": 98670
+    },
+    {
+      "epoch": 0.8163130247756132,
+      "grad_norm": 1095.51318359375,
+      "learning_rate": 1.1989435503894826e-06,
+      "loss": 101.9039,
+      "step": 98680
+    },
+    {
+      "epoch": 0.8163957480249824,
+      "grad_norm": 1368.3948974609375,
+      "learning_rate": 1.1980274200995928e-06,
+      "loss": 91.4454,
+      "step": 98690
+    },
+    {
+      "epoch": 0.8164784712743517,
+      "grad_norm": 969.6019287109375,
+      "learning_rate": 1.197111592315211e-06,
+      "loss": 84.7942,
+      "step": 98700
+    },
+    {
+      "epoch": 0.8165611945237209,
+      "grad_norm": 531.8544311523438,
+      "learning_rate": 1.1961960671092066e-06,
+      "loss": 84.7319,
+      "step": 98710
+    },
+    {
+      "epoch": 0.8166439177730901,
+      "grad_norm": 626.1632080078125,
+      "learning_rate": 1.1952808445544246e-06,
+      "loss": 70.9973,
+      "step": 98720
+    },
+    {
+      "epoch": 0.8167266410224594,
+      "grad_norm": 718.9005737304688,
+      "learning_rate": 1.1943659247236838e-06,
+      "loss": 81.7739,
+      "step": 98730
+    },
+    {
+      "epoch": 0.8168093642718286,
+      "grad_norm": 826.1453247070312,
+      "learning_rate": 1.1934513076897798e-06,
+      "loss": 77.8973,
+      "step": 98740
+    },
+    {
+      "epoch": 0.8168920875211978,
+      "grad_norm": 670.5861206054688,
+      "learning_rate": 1.1925369935254872e-06,
+      "loss": 70.1664,
+      "step": 98750
+    },
+    {
+      "epoch": 0.8169748107705671,
+      "grad_norm": 1031.8236083984375,
+      "learning_rate": 1.191622982303552e-06,
+      "loss": 96.2467,
+      "step": 98760
+    },
+    {
+      "epoch": 0.8170575340199363,
+      "grad_norm": 862.9549560546875,
+      "learning_rate": 1.1907092740966976e-06,
+      "loss": 95.5136,
+      "step": 98770
+    },
+    {
+      "epoch": 0.8171402572693055,
+      "grad_norm": 1296.6314697265625,
+      "learning_rate": 1.1897958689776256e-06,
+      "loss": 125.511,
+      "step": 98780
+    },
+    {
+      "epoch": 0.8172229805186748,
+      "grad_norm": 840.3079833984375,
+      "learning_rate": 1.1888827670190085e-06,
+      "loss": 103.1509,
+      "step": 98790
+    },
+    {
+      "epoch": 0.817305703768044,
+      "grad_norm": 822.6774291992188,
+      "learning_rate": 1.1879699682934993e-06,
+      "loss": 89.4122,
+      "step": 98800
+    },
+    {
+      "epoch": 0.8173884270174132,
+      "grad_norm": 5291.2255859375,
+      "learning_rate": 1.1870574728737244e-06,
+      "loss": 141.3245,
+      "step": 98810
+    },
+    {
+      "epoch": 0.8174711502667825,
+      "grad_norm": 973.1185302734375,
+      "learning_rate": 1.1861452808322877e-06,
+      "loss": 87.4708,
+      "step": 98820
+    },
+    {
+      "epoch": 0.8175538735161517,
+      "grad_norm": 890.1282348632812,
+      "learning_rate": 1.1852333922417658e-06,
+      "loss": 85.0632,
+      "step": 98830
+    },
+    {
+      "epoch": 0.8176365967655209,
+      "grad_norm": 1301.4791259765625,
+      "learning_rate": 1.1843218071747171e-06,
+      "loss": 96.202,
+      "step": 98840
+    },
+    {
+      "epoch": 0.8177193200148902,
+      "grad_norm": 659.366455078125,
+      "learning_rate": 1.183410525703671e-06,
+      "loss": 86.4163,
+      "step": 98850
+    },
+    {
+      "epoch": 0.8178020432642594,
+      "grad_norm": 824.1502685546875,
+      "learning_rate": 1.182499547901133e-06,
+      "loss": 80.8635,
+      "step": 98860
+    },
+    {
+      "epoch": 0.8178847665136286,
+      "grad_norm": 1438.4332275390625,
+      "learning_rate": 1.1815888738395882e-06,
+      "loss": 104.6398,
+      "step": 98870
+    },
+    {
+      "epoch": 0.8179674897629979,
+      "grad_norm": 850.4103393554688,
+      "learning_rate": 1.1806785035914903e-06,
+      "loss": 112.7476,
+      "step": 98880
+    },
+    {
+      "epoch": 0.8180502130123671,
+      "grad_norm": 1077.782470703125,
+      "learning_rate": 1.1797684372292762e-06,
+      "loss": 61.9891,
+      "step": 98890
+    },
+    {
+      "epoch": 0.8181329362617363,
+      "grad_norm": 528.4021606445312,
+      "learning_rate": 1.1788586748253545e-06,
+      "loss": 92.2062,
+      "step": 98900
+    },
+    {
+      "epoch": 0.8182156595111056,
+      "grad_norm": 739.4981079101562,
+      "learning_rate": 1.1779492164521117e-06,
+      "loss": 67.1746,
+      "step": 98910
+    },
+    {
+      "epoch": 0.8182983827604748,
+      "grad_norm": 935.0463256835938,
+      "learning_rate": 1.177040062181909e-06,
+      "loss": 65.3778,
+      "step": 98920
+    },
+    {
+      "epoch": 0.818381106009844,
+      "grad_norm": 1494.3590087890625,
+      "learning_rate": 1.1761312120870822e-06,
+      "loss": 85.324,
+      "step": 98930
+    },
+    {
+      "epoch": 0.8184638292592133,
+      "grad_norm": 514.237060546875,
+      "learning_rate": 1.175222666239947e-06,
+      "loss": 91.5604,
+      "step": 98940
+    },
+    {
+      "epoch": 0.8185465525085825,
+      "grad_norm": 856.9160766601562,
+      "learning_rate": 1.174314424712792e-06,
+      "loss": 86.7412,
+      "step": 98950
+    },
+    {
+      "epoch": 0.8186292757579517,
+      "grad_norm": 849.8342895507812,
+      "learning_rate": 1.1734064875778795e-06,
+      "loss": 81.7563,
+      "step": 98960
+    },
+    {
+      "epoch": 0.8187119990073211,
+      "grad_norm": 648.2426147460938,
+      "learning_rate": 1.1724988549074506e-06,
+      "loss": 84.3063,
+      "step": 98970
+    },
+    {
+      "epoch": 0.8187947222566903,
+      "grad_norm": 736.9127197265625,
+      "learning_rate": 1.1715915267737228e-06,
+      "loss": 123.6431,
+      "step": 98980
+    },
+    {
+      "epoch": 0.8188774455060595,
+      "grad_norm": 2442.851318359375,
+      "learning_rate": 1.1706845032488867e-06,
+      "loss": 78.3283,
+      "step": 98990
+    },
+    {
+      "epoch": 0.8189601687554288,
+      "grad_norm": 895.6693115234375,
+      "learning_rate": 1.1697777844051105e-06,
+      "loss": 67.0833,
+      "step": 99000
+    },
+    {
+      "epoch": 0.819042892004798,
+      "grad_norm": 616.1843872070312,
+      "learning_rate": 1.1688713703145377e-06,
+      "loss": 97.3783,
+      "step": 99010
+    },
+    {
+      "epoch": 0.8191256152541672,
+      "grad_norm": 753.047119140625,
+      "learning_rate": 1.1679652610492875e-06,
+      "loss": 78.6516,
+      "step": 99020
+    },
+    {
+      "epoch": 0.8192083385035365,
+      "grad_norm": 1777.462890625,
+      "learning_rate": 1.1670594566814536e-06,
+      "loss": 78.5895,
+      "step": 99030
+    },
+    {
+      "epoch": 0.8192910617529057,
+      "grad_norm": 898.646484375,
+      "learning_rate": 1.1661539572831105e-06,
+      "loss": 67.8556,
+      "step": 99040
+    },
+    {
+      "epoch": 0.8193737850022749,
+      "grad_norm": 1100.24267578125,
+      "learning_rate": 1.1652487629263003e-06,
+      "loss": 92.8213,
+      "step": 99050
+    },
+    {
+      "epoch": 0.8194565082516442,
+      "grad_norm": 727.767822265625,
+      "learning_rate": 1.1643438736830476e-06,
+      "loss": 87.5095,
+      "step": 99060
+    },
+    {
+      "epoch": 0.8195392315010134,
+      "grad_norm": 556.3981323242188,
+      "learning_rate": 1.1634392896253495e-06,
+      "loss": 98.1264,
+      "step": 99070
+    },
+    {
+      "epoch": 0.8196219547503826,
+      "grad_norm": 761.73828125,
+      "learning_rate": 1.1625350108251793e-06,
+      "loss": 85.1379,
+      "step": 99080
+    },
+    {
+      "epoch": 0.8197046779997519,
+      "grad_norm": 553.5272827148438,
+      "learning_rate": 1.1616310373544865e-06,
+      "loss": 72.9364,
+      "step": 99090
+    },
+    {
+      "epoch": 0.8197874012491211,
+      "grad_norm": 601.8910522460938,
+      "learning_rate": 1.1607273692851967e-06,
+      "loss": 87.1516,
+      "step": 99100
+    },
+    {
+      "epoch": 0.8198701244984903,
+      "grad_norm": 369.5881652832031,
+      "learning_rate": 1.1598240066892103e-06,
+      "loss": 66.5272,
+      "step": 99110
+    },
+    {
+      "epoch": 0.8199528477478595,
+      "grad_norm": 523.6664428710938,
+      "learning_rate": 1.1589209496384035e-06,
+      "loss": 84.0007,
+      "step": 99120
+    },
+    {
+      "epoch": 0.8200355709972288,
+      "grad_norm": 1775.257080078125,
+      "learning_rate": 1.1580181982046285e-06,
+      "loss": 100.826,
+      "step": 99130
+    },
+    {
+      "epoch": 0.820118294246598,
+      "grad_norm": 1454.958984375,
+      "learning_rate": 1.1571157524597137e-06,
+      "loss": 103.4577,
+      "step": 99140
+    },
+    {
+      "epoch": 0.8202010174959672,
+      "grad_norm": 673.6801147460938,
+      "learning_rate": 1.1562136124754613e-06,
+      "loss": 89.0967,
+      "step": 99150
+    },
+    {
+      "epoch": 0.8202837407453365,
+      "grad_norm": 514.0848388671875,
+      "learning_rate": 1.1553117783236516e-06,
+      "loss": 67.5358,
+      "step": 99160
+    },
+    {
+      "epoch": 0.8203664639947057,
+      "grad_norm": 888.429443359375,
+      "learning_rate": 1.154410250076039e-06,
+      "loss": 101.8155,
+      "step": 99170
+    },
+    {
+      "epoch": 0.8204491872440749,
+      "grad_norm": 515.0179443359375,
+      "learning_rate": 1.1535090278043538e-06,
+      "loss": 92.6118,
+      "step": 99180
+    },
+    {
+      "epoch": 0.8205319104934442,
+      "grad_norm": 659.7156372070312,
+      "learning_rate": 1.1526081115803017e-06,
+      "loss": 77.2248,
+      "step": 99190
+    },
+    {
+      "epoch": 0.8206146337428134,
+      "grad_norm": 680.0890502929688,
+      "learning_rate": 1.1517075014755657e-06,
+      "loss": 107.0121,
+      "step": 99200
+    },
+    {
+      "epoch": 0.8206973569921826,
+      "grad_norm": 1098.1448974609375,
+      "learning_rate": 1.1508071975618025e-06,
+      "loss": 96.3001,
+      "step": 99210
+    },
+    {
+      "epoch": 0.8207800802415519,
+      "grad_norm": 628.8194580078125,
+      "learning_rate": 1.1499071999106449e-06,
+      "loss": 94.3524,
+      "step": 99220
+    },
+    {
+      "epoch": 0.8208628034909211,
+      "grad_norm": 908.0435791015625,
+      "learning_rate": 1.149007508593702e-06,
+      "loss": 89.7701,
+      "step": 99230
+    },
+    {
+      "epoch": 0.8209455267402903,
+      "grad_norm": 635.8421630859375,
+      "learning_rate": 1.148108123682558e-06,
+      "loss": 69.7421,
+      "step": 99240
+    },
+    {
+      "epoch": 0.8210282499896596,
+      "grad_norm": 540.8094482421875,
+      "learning_rate": 1.1472090452487728e-06,
+      "loss": 83.0186,
+      "step": 99250
+    },
+    {
+      "epoch": 0.8211109732390288,
+      "grad_norm": 655.1044311523438,
+      "learning_rate": 1.146310273363882e-06,
+      "loss": 79.3086,
+      "step": 99260
+    },
+    {
+      "epoch": 0.821193696488398,
+      "grad_norm": 579.1731567382812,
+      "learning_rate": 1.1454118080993965e-06,
+      "loss": 86.8246,
+      "step": 99270
+    },
+    {
+      "epoch": 0.8212764197377673,
+      "grad_norm": 1405.385498046875,
+      "learning_rate": 1.1445136495268033e-06,
+      "loss": 124.9013,
+      "step": 99280
+    },
+    {
+      "epoch": 0.8213591429871365,
+      "grad_norm": 553.7492065429688,
+      "learning_rate": 1.143615797717565e-06,
+      "loss": 80.3261,
+      "step": 99290
+    },
+    {
+      "epoch": 0.8214418662365057,
+      "grad_norm": 2029.3385009765625,
+      "learning_rate": 1.1427182527431192e-06,
+      "loss": 95.4808,
+      "step": 99300
+    },
+    {
+      "epoch": 0.821524589485875,
+      "grad_norm": 756.5120849609375,
+      "learning_rate": 1.1418210146748792e-06,
+      "loss": 67.8008,
+      "step": 99310
+    },
+    {
+      "epoch": 0.8216073127352442,
+      "grad_norm": 396.6879577636719,
+      "learning_rate": 1.1409240835842344e-06,
+      "loss": 70.0907,
+      "step": 99320
+    },
+    {
+      "epoch": 0.8216900359846134,
+      "grad_norm": 836.3045654296875,
+      "learning_rate": 1.1400274595425499e-06,
+      "loss": 81.1893,
+      "step": 99330
+    },
+    {
+      "epoch": 0.8217727592339827,
+      "grad_norm": 831.0084838867188,
+      "learning_rate": 1.139131142621165e-06,
+      "loss": 108.7785,
+      "step": 99340
+    },
+    {
+      "epoch": 0.8218554824833519,
+      "grad_norm": 594.3251953125,
+      "learning_rate": 1.1382351328913964e-06,
+      "loss": 90.5988,
+      "step": 99350
+    },
+    {
+      "epoch": 0.8219382057327211,
+      "grad_norm": 997.0202026367188,
+      "learning_rate": 1.137339430424535e-06,
+      "loss": 98.3652,
+      "step": 99360
+    },
+    {
+      "epoch": 0.8220209289820904,
+      "grad_norm": 1159.4931640625,
+      "learning_rate": 1.1364440352918482e-06,
+      "loss": 93.9449,
+      "step": 99370
+    },
+    {
+      "epoch": 0.8221036522314596,
+      "grad_norm": 1038.8775634765625,
+      "learning_rate": 1.1355489475645798e-06,
+      "loss": 109.6431,
+      "step": 99380
+    },
+    {
+      "epoch": 0.8221863754808288,
+      "grad_norm": 1000.264404296875,
+      "learning_rate": 1.1346541673139428e-06,
+      "loss": 81.506,
+      "step": 99390
+    },
+    {
+      "epoch": 0.8222690987301982,
+      "grad_norm": 986.3641967773438,
+      "learning_rate": 1.1337596946111357e-06,
+      "loss": 91.9383,
+      "step": 99400
+    },
+    {
+      "epoch": 0.8223518219795674,
+      "grad_norm": 548.0112915039062,
+      "learning_rate": 1.132865529527326e-06,
+      "loss": 79.2564,
+      "step": 99410
+    },
+    {
+      "epoch": 0.8224345452289366,
+      "grad_norm": 912.9034423828125,
+      "learning_rate": 1.1319716721336587e-06,
+      "loss": 88.9765,
+      "step": 99420
+    },
+    {
+      "epoch": 0.8225172684783059,
+      "grad_norm": 463.36883544921875,
+      "learning_rate": 1.1310781225012535e-06,
+      "loss": 58.9931,
+      "step": 99430
+    },
+    {
+      "epoch": 0.8225999917276751,
+      "grad_norm": 1589.856201171875,
+      "learning_rate": 1.130184880701206e-06,
+      "loss": 102.2906,
+      "step": 99440
+    },
+    {
+      "epoch": 0.8226827149770443,
+      "grad_norm": 653.0153198242188,
+      "learning_rate": 1.1292919468045876e-06,
+      "loss": 83.1269,
+      "step": 99450
+    },
+    {
+      "epoch": 0.8227654382264136,
+      "grad_norm": 1476.3031005859375,
+      "learning_rate": 1.128399320882445e-06,
+      "loss": 81.7267,
+      "step": 99460
+    },
+    {
+      "epoch": 0.8228481614757828,
+      "grad_norm": 633.2077026367188,
+      "learning_rate": 1.1275070030058016e-06,
+      "loss": 74.4383,
+      "step": 99470
+    },
+    {
+      "epoch": 0.822930884725152,
+      "grad_norm": 1069.526123046875,
+      "learning_rate": 1.1266149932456516e-06,
+      "loss": 93.1175,
+      "step": 99480
+    },
+    {
+      "epoch": 0.8230136079745213,
+      "grad_norm": 513.8902587890625,
+      "learning_rate": 1.1257232916729693e-06,
+      "loss": 78.2727,
+      "step": 99490
+    },
+    {
+      "epoch": 0.8230963312238905,
+      "grad_norm": 552.6843872070312,
+      "learning_rate": 1.1248318983587052e-06,
+      "loss": 77.5065,
+      "step": 99500
+    },
+    {
+      "epoch": 0.8231790544732597,
+      "grad_norm": 878.9365234375,
+      "learning_rate": 1.1239408133737828e-06,
+      "loss": 77.3874,
+      "step": 99510
+    },
+    {
+      "epoch": 0.823261777722629,
+      "grad_norm": 1088.0330810546875,
+      "learning_rate": 1.123050036789101e-06,
+      "loss": 93.4705,
+      "step": 99520
+    },
+    {
+      "epoch": 0.8233445009719982,
+      "grad_norm": 916.3295288085938,
+      "learning_rate": 1.122159568675535e-06,
+      "loss": 86.091,
+      "step": 99530
+    },
+    {
+      "epoch": 0.8234272242213674,
+      "grad_norm": 595.7514038085938,
+      "learning_rate": 1.121269409103935e-06,
+      "loss": 110.6302,
+      "step": 99540
+    },
+    {
+      "epoch": 0.8235099474707367,
+      "grad_norm": 819.598388671875,
+      "learning_rate": 1.1203795581451288e-06,
+      "loss": 116.6621,
+      "step": 99550
+    },
+    {
+      "epoch": 0.8235926707201059,
+      "grad_norm": 669.3980712890625,
+      "learning_rate": 1.1194900158699146e-06,
+      "loss": 83.4266,
+      "step": 99560
+    },
+    {
+      "epoch": 0.8236753939694751,
+      "grad_norm": 630.958251953125,
+      "learning_rate": 1.1186007823490708e-06,
+      "loss": 86.0332,
+      "step": 99570
+    },
+    {
+      "epoch": 0.8237581172188444,
+      "grad_norm": 1063.6295166015625,
+      "learning_rate": 1.1177118576533492e-06,
+      "loss": 99.3058,
+      "step": 99580
+    },
+    {
+      "epoch": 0.8238408404682136,
+      "grad_norm": 1337.9732666015625,
+      "learning_rate": 1.116823241853477e-06,
+      "loss": 54.1832,
+      "step": 99590
+    },
+    {
+      "epoch": 0.8239235637175828,
+      "grad_norm": 680.459228515625,
+      "learning_rate": 1.1159349350201587e-06,
+      "loss": 86.0068,
+      "step": 99600
+    },
+    {
+      "epoch": 0.8240062869669521,
+      "grad_norm": 976.7091674804688,
+      "learning_rate": 1.1150469372240724e-06,
+      "loss": 85.8299,
+      "step": 99610
+    },
+    {
+      "epoch": 0.8240890102163213,
+      "grad_norm": 839.8695678710938,
+      "learning_rate": 1.114159248535872e-06,
+      "loss": 85.7322,
+      "step": 99620
+    },
+    {
+      "epoch": 0.8241717334656905,
+      "grad_norm": 524.162841796875,
+      "learning_rate": 1.1132718690261868e-06,
+      "loss": 87.2623,
+      "step": 99630
+    },
+    {
+      "epoch": 0.8242544567150598,
+      "grad_norm": 808.6865234375,
+      "learning_rate": 1.1123847987656221e-06,
+      "loss": 89.9084,
+      "step": 99640
+    },
+    {
+      "epoch": 0.824337179964429,
+      "grad_norm": 1346.7198486328125,
+      "learning_rate": 1.1114980378247565e-06,
+      "loss": 73.3104,
+      "step": 99650
+    },
+    {
+      "epoch": 0.8244199032137982,
+      "grad_norm": 860.0714111328125,
+      "learning_rate": 1.1106115862741457e-06,
+      "loss": 136.1401,
+      "step": 99660
+    },
+    {
+      "epoch": 0.8245026264631675,
+      "grad_norm": 874.57666015625,
+      "learning_rate": 1.1097254441843215e-06,
+      "loss": 112.3625,
+      "step": 99670
+    },
+    {
+      "epoch": 0.8245853497125367,
+      "grad_norm": 788.482666015625,
+      "learning_rate": 1.1088396116257893e-06,
+      "loss": 54.8236,
+      "step": 99680
+    },
+    {
+      "epoch": 0.8246680729619059,
+      "grad_norm": 807.5397338867188,
+      "learning_rate": 1.10795408866903e-06,
+      "loss": 79.1535,
+      "step": 99690
+    },
+    {
+      "epoch": 0.8247507962112752,
+      "grad_norm": 1106.181396484375,
+      "learning_rate": 1.1070688753845032e-06,
+      "loss": 117.5441,
+      "step": 99700
+    },
+    {
+      "epoch": 0.8248335194606444,
+      "grad_norm": 782.138427734375,
+      "learning_rate": 1.1061839718426399e-06,
+      "loss": 92.2893,
+      "step": 99710
+    },
+    {
+      "epoch": 0.8249162427100136,
+      "grad_norm": 1031.1993408203125,
+      "learning_rate": 1.1052993781138477e-06,
+      "loss": 80.9152,
+      "step": 99720
+    },
+    {
+      "epoch": 0.8249989659593829,
+      "grad_norm": 443.52801513671875,
+      "learning_rate": 1.1044150942685112e-06,
+      "loss": 104.3299,
+      "step": 99730
+    },
+    {
+      "epoch": 0.8250816892087521,
+      "grad_norm": 1952.7318115234375,
+      "learning_rate": 1.1035311203769855e-06,
+      "loss": 135.8231,
+      "step": 99740
+    },
+    {
+      "epoch": 0.8251644124581213,
+      "grad_norm": 863.5310668945312,
+      "learning_rate": 1.1026474565096068e-06,
+      "loss": 85.8996,
+      "step": 99750
+    },
+    {
+      "epoch": 0.8252471357074906,
+      "grad_norm": 756.4827880859375,
+      "learning_rate": 1.1017641027366832e-06,
+      "loss": 97.5026,
+      "step": 99760
+    },
+    {
+      "epoch": 0.8253298589568598,
+      "grad_norm": 881.189453125,
+      "learning_rate": 1.1008810591284997e-06,
+      "loss": 100.8524,
+      "step": 99770
+    },
+    {
+      "epoch": 0.825412582206229,
+      "grad_norm": 574.4223022460938,
+      "learning_rate": 1.0999983257553137e-06,
+      "loss": 82.259,
+      "step": 99780
+    },
+    {
+      "epoch": 0.8254953054555983,
+      "grad_norm": 1166.70166015625,
+      "learning_rate": 1.0991159026873643e-06,
+      "loss": 79.2827,
+      "step": 99790
+    },
+    {
+      "epoch": 0.8255780287049675,
+      "grad_norm": 712.7079467773438,
+      "learning_rate": 1.0982337899948603e-06,
+      "loss": 71.3755,
+      "step": 99800
+    },
+    {
+      "epoch": 0.8256607519543367,
+      "grad_norm": 921.7284545898438,
+      "learning_rate": 1.0973519877479878e-06,
+      "loss": 50.2585,
+      "step": 99810
+    },
+    {
+      "epoch": 0.8257434752037061,
+      "grad_norm": 480.2099914550781,
+      "learning_rate": 1.0964704960169054e-06,
+      "loss": 75.6909,
+      "step": 99820
+    },
+    {
+      "epoch": 0.8258261984530753,
+      "grad_norm": 659.5361328125,
+      "learning_rate": 1.0955893148717512e-06,
+      "loss": 92.5624,
+      "step": 99830
+    },
+    {
+      "epoch": 0.8259089217024445,
+      "grad_norm": 386.55810546875,
+      "learning_rate": 1.0947084443826361e-06,
+      "loss": 71.9917,
+      "step": 99840
+    },
+    {
+      "epoch": 0.8259916449518137,
+      "grad_norm": 968.19140625,
+      "learning_rate": 1.0938278846196471e-06,
+      "loss": 75.8837,
+      "step": 99850
+    },
+    {
+      "epoch": 0.826074368201183,
+      "grad_norm": 1425.4276123046875,
+      "learning_rate": 1.0929476356528469e-06,
+      "loss": 70.3875,
+      "step": 99860
+    },
+    {
+      "epoch": 0.8261570914505522,
+      "grad_norm": 895.6668090820312,
+      "learning_rate": 1.092067697552272e-06,
+      "loss": 91.4471,
+      "step": 99870
+    },
+    {
+      "epoch": 0.8262398146999214,
+      "grad_norm": 1033.364013671875,
+      "learning_rate": 1.091188070387934e-06,
+      "loss": 90.5695,
+      "step": 99880
+    },
+    {
+      "epoch": 0.8263225379492907,
+      "grad_norm": 660.748046875,
+      "learning_rate": 1.0903087542298241e-06,
+      "loss": 96.0515,
+      "step": 99890
+    },
+    {
+      "epoch": 0.8264052611986599,
+      "grad_norm": 480.1137390136719,
+      "learning_rate": 1.0894297491479044e-06,
+      "loss": 80.2592,
+      "step": 99900
+    },
+    {
+      "epoch": 0.8264879844480291,
+      "grad_norm": 643.3427124023438,
+      "learning_rate": 1.0885510552121115e-06,
+      "loss": 59.6039,
+      "step": 99910
+    },
+    {
+      "epoch": 0.8265707076973984,
+      "grad_norm": 1013.935546875,
+      "learning_rate": 1.0876726724923597e-06,
+      "loss": 77.1848,
+      "step": 99920
+    },
+    {
+      "epoch": 0.8266534309467676,
+      "grad_norm": 841.8418579101562,
+      "learning_rate": 1.086794601058538e-06,
+      "loss": 83.5782,
+      "step": 99930
+    },
+    {
+      "epoch": 0.8267361541961368,
+      "grad_norm": 337.9281311035156,
+      "learning_rate": 1.0859168409805109e-06,
+      "loss": 68.1758,
+      "step": 99940
+    },
+    {
+      "epoch": 0.8268188774455061,
+      "grad_norm": 665.4752807617188,
+      "learning_rate": 1.0850393923281176e-06,
+      "loss": 100.0561,
+      "step": 99950
+    },
+    {
+      "epoch": 0.8269016006948753,
+      "grad_norm": 569.7544555664062,
+      "learning_rate": 1.0841622551711728e-06,
+      "loss": 83.8357,
+      "step": 99960
+    },
+    {
+      "epoch": 0.8269843239442445,
+      "grad_norm": 661.4277954101562,
+      "learning_rate": 1.0832854295794659e-06,
+      "loss": 65.98,
+      "step": 99970
+    },
+    {
+      "epoch": 0.8270670471936138,
+      "grad_norm": 930.4299926757812,
+      "learning_rate": 1.0824089156227624e-06,
+      "loss": 77.1078,
+      "step": 99980
+    },
+    {
+      "epoch": 0.827149770442983,
+      "grad_norm": 1039.9451904296875,
+      "learning_rate": 1.0815327133708015e-06,
+      "loss": 49.8599,
+      "step": 99990
+    },
+    {
+      "epoch": 0.8272324936923522,
+      "grad_norm": 907.8119506835938,
+      "learning_rate": 1.0806568228932995e-06,
+      "loss": 87.3194,
+      "step": 100000
+    },
+    {
+      "epoch": 0.8273152169417215,
+      "grad_norm": 665.1284790039062,
+      "learning_rate": 1.079781244259947e-06,
+      "loss": 113.8395,
+      "step": 100010
+    },
+    {
+      "epoch": 0.8273979401910907,
+      "grad_norm": 1413.6357421875,
+      "learning_rate": 1.0789059775404093e-06,
+      "loss": 98.0291,
+      "step": 100020
+    },
+    {
+      "epoch": 0.8274806634404599,
+      "grad_norm": 738.0913696289062,
+      "learning_rate": 1.0780310228043278e-06,
+      "loss": 97.2634,
+      "step": 100030
+    },
+    {
+      "epoch": 0.8275633866898292,
+      "grad_norm": 610.05859375,
+      "learning_rate": 1.0771563801213186e-06,
+      "loss": 86.15,
+      "step": 100040
+    },
+    {
+      "epoch": 0.8276461099391984,
+      "grad_norm": 723.9077758789062,
+      "learning_rate": 1.076282049560972e-06,
+      "loss": 92.4268,
+      "step": 100050
+    },
+    {
+      "epoch": 0.8277288331885676,
+      "grad_norm": 939.378662109375,
+      "learning_rate": 1.075408031192856e-06,
+      "loss": 82.4279,
+      "step": 100060
+    },
+    {
+      "epoch": 0.8278115564379369,
+      "grad_norm": 1257.1422119140625,
+      "learning_rate": 1.0745343250865114e-06,
+      "loss": 73.5913,
+      "step": 100070
+    },
+    {
+      "epoch": 0.8278942796873061,
+      "grad_norm": 568.9317626953125,
+      "learning_rate": 1.073660931311455e-06,
+      "loss": 89.3267,
+      "step": 100080
+    },
+    {
+      "epoch": 0.8279770029366753,
+      "grad_norm": 454.4713134765625,
+      "learning_rate": 1.0727878499371786e-06,
+      "loss": 71.951,
+      "step": 100090
+    },
+    {
+      "epoch": 0.8280597261860446,
+      "grad_norm": 692.2919311523438,
+      "learning_rate": 1.0719150810331497e-06,
+      "loss": 70.2479,
+      "step": 100100
+    },
+    {
+      "epoch": 0.8281424494354138,
+      "grad_norm": 1330.4888916015625,
+      "learning_rate": 1.0710426246688105e-06,
+      "loss": 102.4429,
+      "step": 100110
+    },
+    {
+      "epoch": 0.828225172684783,
+      "grad_norm": 739.1103515625,
+      "learning_rate": 1.0701704809135782e-06,
+      "loss": 78.6834,
+      "step": 100120
+    },
+    {
+      "epoch": 0.8283078959341523,
+      "grad_norm": 964.680908203125,
+      "learning_rate": 1.0692986498368452e-06,
+      "loss": 103.2081,
+      "step": 100130
+    },
+    {
+      "epoch": 0.8283906191835215,
+      "grad_norm": 928.7930908203125,
+      "learning_rate": 1.0684271315079786e-06,
+      "loss": 75.7119,
+      "step": 100140
+    },
+    {
+      "epoch": 0.8284733424328907,
+      "grad_norm": 771.4544677734375,
+      "learning_rate": 1.0675559259963226e-06,
+      "loss": 91.3507,
+      "step": 100150
+    },
+    {
+      "epoch": 0.82855606568226,
+      "grad_norm": 1169.62109375,
+      "learning_rate": 1.066685033371193e-06,
+      "loss": 69.6342,
+      "step": 100160
+    },
+    {
+      "epoch": 0.8286387889316292,
+      "grad_norm": 1027.3876953125,
+      "learning_rate": 1.0658144537018844e-06,
+      "loss": 68.9951,
+      "step": 100170
+    },
+    {
+      "epoch": 0.8287215121809984,
+      "grad_norm": 561.0545043945312,
+      "learning_rate": 1.0649441870576644e-06,
+      "loss": 63.7628,
+      "step": 100180
+    },
+    {
+      "epoch": 0.8288042354303677,
+      "grad_norm": 2845.06689453125,
+      "learning_rate": 1.0640742335077758e-06,
+      "loss": 74.9551,
+      "step": 100190
+    },
+    {
+      "epoch": 0.8288869586797369,
+      "grad_norm": 514.4091186523438,
+      "learning_rate": 1.0632045931214369e-06,
+      "loss": 69.5426,
+      "step": 100200
+    },
+    {
+      "epoch": 0.8289696819291061,
+      "grad_norm": 830.9058837890625,
+      "learning_rate": 1.0623352659678415e-06,
+      "loss": 88.3192,
+      "step": 100210
+    },
+    {
+      "epoch": 0.8290524051784754,
+      "grad_norm": 681.1116333007812,
+      "learning_rate": 1.0614662521161573e-06,
+      "loss": 73.3991,
+      "step": 100220
+    },
+    {
+      "epoch": 0.8291351284278446,
+      "grad_norm": 1170.43212890625,
+      "learning_rate": 1.0605975516355277e-06,
+      "loss": 56.5079,
+      "step": 100230
+    },
+    {
+      "epoch": 0.8292178516772138,
+      "grad_norm": 640.3706665039062,
+      "learning_rate": 1.059729164595073e-06,
+      "loss": 76.578,
+      "step": 100240
+    },
+    {
+      "epoch": 0.8293005749265832,
+      "grad_norm": 1166.9835205078125,
+      "learning_rate": 1.0588610910638825e-06,
+      "loss": 109.2789,
+      "step": 100250
+    },
+    {
+      "epoch": 0.8293832981759524,
+      "grad_norm": 928.90185546875,
+      "learning_rate": 1.057993331111029e-06,
+      "loss": 89.8654,
+      "step": 100260
+    },
+    {
+      "epoch": 0.8294660214253216,
+      "grad_norm": 927.5479736328125,
+      "learning_rate": 1.057125884805555e-06,
+      "loss": 82.5449,
+      "step": 100270
+    },
+    {
+      "epoch": 0.8295487446746909,
+      "grad_norm": 980.9613037109375,
+      "learning_rate": 1.0562587522164792e-06,
+      "loss": 93.9806,
+      "step": 100280
+    },
+    {
+      "epoch": 0.8296314679240601,
+      "grad_norm": 702.5833740234375,
+      "learning_rate": 1.0553919334127943e-06,
+      "loss": 67.3568,
+      "step": 100290
+    },
+    {
+      "epoch": 0.8297141911734293,
+      "grad_norm": 1179.7244873046875,
+      "learning_rate": 1.0545254284634703e-06,
+      "loss": 96.2675,
+      "step": 100300
+    },
+    {
+      "epoch": 0.8297969144227986,
+      "grad_norm": 885.7655639648438,
+      "learning_rate": 1.053659237437451e-06,
+      "loss": 112.9435,
+      "step": 100310
+    },
+    {
+      "epoch": 0.8298796376721678,
+      "grad_norm": 940.1737060546875,
+      "learning_rate": 1.052793360403655e-06,
+      "loss": 80.4012,
+      "step": 100320
+    },
+    {
+      "epoch": 0.829962360921537,
+      "grad_norm": 1659.75341796875,
+      "learning_rate": 1.0519277974309771e-06,
+      "loss": 96.4695,
+      "step": 100330
+    },
+    {
+      "epoch": 0.8300450841709063,
+      "grad_norm": 809.1747436523438,
+      "learning_rate": 1.0510625485882825e-06,
+      "loss": 90.7687,
+      "step": 100340
+    },
+    {
+      "epoch": 0.8301278074202755,
+      "grad_norm": 565.280029296875,
+      "learning_rate": 1.0501976139444191e-06,
+      "loss": 96.0154,
+      "step": 100350
+    },
+    {
+      "epoch": 0.8302105306696447,
+      "grad_norm": 1022.1803588867188,
+      "learning_rate": 1.0493329935682045e-06,
+      "loss": 84.8389,
+      "step": 100360
+    },
+    {
+      "epoch": 0.830293253919014,
+      "grad_norm": 615.2613525390625,
+      "learning_rate": 1.0484686875284323e-06,
+      "loss": 68.2956,
+      "step": 100370
+    },
+    {
+      "epoch": 0.8303759771683832,
+      "grad_norm": 573.2850952148438,
+      "learning_rate": 1.0476046958938719e-06,
+      "loss": 95.2249,
+      "step": 100380
+    },
+    {
+      "epoch": 0.8304587004177524,
+      "grad_norm": 504.07470703125,
+      "learning_rate": 1.0467410187332667e-06,
+      "loss": 77.4282,
+      "step": 100390
+    },
+    {
+      "epoch": 0.8305414236671217,
+      "grad_norm": 1332.99169921875,
+      "learning_rate": 1.045877656115335e-06,
+      "loss": 117.6646,
+      "step": 100400
+    },
+    {
+      "epoch": 0.8306241469164909,
+      "grad_norm": 665.0089721679688,
+      "learning_rate": 1.045014608108773e-06,
+      "loss": 91.1985,
+      "step": 100410
+    },
+    {
+      "epoch": 0.8307068701658601,
+      "grad_norm": 1073.398681640625,
+      "learning_rate": 1.0441518747822466e-06,
+      "loss": 96.1906,
+      "step": 100420
+    },
+    {
+      "epoch": 0.8307895934152294,
+      "grad_norm": 774.0573120117188,
+      "learning_rate": 1.0432894562044004e-06,
+      "loss": 83.1879,
+      "step": 100430
+    },
+    {
+      "epoch": 0.8308723166645986,
+      "grad_norm": 1170.0047607421875,
+      "learning_rate": 1.0424273524438521e-06,
+      "loss": 98.2264,
+      "step": 100440
+    },
+    {
+      "epoch": 0.8309550399139678,
+      "grad_norm": 567.7711181640625,
+      "learning_rate": 1.041565563569198e-06,
+      "loss": 72.0893,
+      "step": 100450
+    },
+    {
+      "epoch": 0.8310377631633371,
+      "grad_norm": 946.6346435546875,
+      "learning_rate": 1.040704089649005e-06,
+      "loss": 101.0679,
+      "step": 100460
+    },
+    {
+      "epoch": 0.8311204864127063,
+      "grad_norm": 609.6173095703125,
+      "learning_rate": 1.039842930751817e-06,
+      "loss": 93.027,
+      "step": 100470
+    },
+    {
+      "epoch": 0.8312032096620755,
+      "grad_norm": 693.0861206054688,
+      "learning_rate": 1.0389820869461525e-06,
+      "loss": 75.2829,
+      "step": 100480
+    },
+    {
+      "epoch": 0.8312859329114448,
+      "grad_norm": 987.9539184570312,
+      "learning_rate": 1.0381215583005043e-06,
+      "loss": 80.7349,
+      "step": 100490
+    },
+    {
+      "epoch": 0.831368656160814,
+      "grad_norm": 645.7682495117188,
+      "learning_rate": 1.0372613448833429e-06,
+      "loss": 103.9739,
+      "step": 100500
+    },
+    {
+      "epoch": 0.8314513794101832,
+      "grad_norm": 983.28271484375,
+      "learning_rate": 1.0364014467631078e-06,
+      "loss": 79.8077,
+      "step": 100510
+    },
+    {
+      "epoch": 0.8315341026595525,
+      "grad_norm": 754.6640625,
+      "learning_rate": 1.03554186400822e-06,
+      "loss": 88.8747,
+      "step": 100520
+    },
+    {
+      "epoch": 0.8316168259089217,
+      "grad_norm": 1055.522705078125,
+      "learning_rate": 1.034682596687071e-06,
+      "loss": 90.0075,
+      "step": 100530
+    },
+    {
+      "epoch": 0.8316995491582909,
+      "grad_norm": 1082.3468017578125,
+      "learning_rate": 1.0338236448680283e-06,
+      "loss": 82.0571,
+      "step": 100540
+    },
+    {
+      "epoch": 0.8317822724076602,
+      "grad_norm": 1203.2752685546875,
+      "learning_rate": 1.0329650086194371e-06,
+      "loss": 92.2786,
+      "step": 100550
+    },
+    {
+      "epoch": 0.8318649956570294,
+      "grad_norm": 513.8129272460938,
+      "learning_rate": 1.0321066880096142e-06,
+      "loss": 100.5733,
+      "step": 100560
+    },
+    {
+      "epoch": 0.8319477189063986,
+      "grad_norm": 637.0855712890625,
+      "learning_rate": 1.0312486831068518e-06,
+      "loss": 104.5459,
+      "step": 100570
+    },
+    {
+      "epoch": 0.8320304421557678,
+      "grad_norm": 712.1045532226562,
+      "learning_rate": 1.0303909939794176e-06,
+      "loss": 69.6618,
+      "step": 100580
+    },
+    {
+      "epoch": 0.8321131654051371,
+      "grad_norm": 686.3661499023438,
+      "learning_rate": 1.0295336206955553e-06,
+      "loss": 88.7248,
+      "step": 100590
+    },
+    {
+      "epoch": 0.8321958886545063,
+      "grad_norm": 864.8832397460938,
+      "learning_rate": 1.0286765633234795e-06,
+      "loss": 83.1856,
+      "step": 100600
+    },
+    {
+      "epoch": 0.8322786119038755,
+      "grad_norm": 716.6763916015625,
+      "learning_rate": 1.027819821931384e-06,
+      "loss": 115.8508,
+      "step": 100610
+    },
+    {
+      "epoch": 0.8323613351532448,
+      "grad_norm": 1016.16748046875,
+      "learning_rate": 1.0269633965874348e-06,
+      "loss": 84.6865,
+      "step": 100620
+    },
+    {
+      "epoch": 0.832444058402614,
+      "grad_norm": 910.2161254882812,
+      "learning_rate": 1.0261072873597744e-06,
+      "loss": 85.6898,
+      "step": 100630
+    },
+    {
+      "epoch": 0.8325267816519832,
+      "grad_norm": 555.4244384765625,
+      "learning_rate": 1.0252514943165188e-06,
+      "loss": 95.8644,
+      "step": 100640
+    },
+    {
+      "epoch": 0.8326095049013525,
+      "grad_norm": 991.6724853515625,
+      "learning_rate": 1.0243960175257605e-06,
+      "loss": 90.1952,
+      "step": 100650
+    },
+    {
+      "epoch": 0.8326922281507217,
+      "grad_norm": 1144.724609375,
+      "learning_rate": 1.0235408570555661e-06,
+      "loss": 81.0206,
+      "step": 100660
+    },
+    {
+      "epoch": 0.832774951400091,
+      "grad_norm": 1066.9254150390625,
+      "learning_rate": 1.022686012973978e-06,
+      "loss": 65.6117,
+      "step": 100670
+    },
+    {
+      "epoch": 0.8328576746494603,
+      "grad_norm": 1321.929931640625,
+      "learning_rate": 1.0218314853490086e-06,
+      "loss": 115.5369,
+      "step": 100680
+    },
+    {
+      "epoch": 0.8329403978988295,
+      "grad_norm": 735.5868530273438,
+      "learning_rate": 1.0209772742486501e-06,
+      "loss": 69.8609,
+      "step": 100690
+    },
+    {
+      "epoch": 0.8330231211481987,
+      "grad_norm": 1540.8714599609375,
+      "learning_rate": 1.0201233797408694e-06,
+      "loss": 103.727,
+      "step": 100700
+    },
+    {
+      "epoch": 0.833105844397568,
+      "grad_norm": 1017.089111328125,
+      "learning_rate": 1.019269801893606e-06,
+      "loss": 84.9051,
+      "step": 100710
+    },
+    {
+      "epoch": 0.8331885676469372,
+      "grad_norm": 844.1512451171875,
+      "learning_rate": 1.0184165407747755e-06,
+      "loss": 62.8379,
+      "step": 100720
+    },
+    {
+      "epoch": 0.8332712908963064,
+      "grad_norm": 1083.6983642578125,
+      "learning_rate": 1.0175635964522661e-06,
+      "loss": 76.4931,
+      "step": 100730
+    },
+    {
+      "epoch": 0.8333540141456757,
+      "grad_norm": 1482.5361328125,
+      "learning_rate": 1.0167109689939459e-06,
+      "loss": 98.549,
+      "step": 100740
+    },
+    {
+      "epoch": 0.8334367373950449,
+      "grad_norm": 684.1696166992188,
+      "learning_rate": 1.0158586584676533e-06,
+      "loss": 69.7886,
+      "step": 100750
+    },
+    {
+      "epoch": 0.8335194606444141,
+      "grad_norm": 612.1447143554688,
+      "learning_rate": 1.0150066649412038e-06,
+      "loss": 65.4549,
+      "step": 100760
+    },
+    {
+      "epoch": 0.8336021838937834,
+      "grad_norm": 578.9083251953125,
+      "learning_rate": 1.0141549884823837e-06,
+      "loss": 81.0457,
+      "step": 100770
+    },
+    {
+      "epoch": 0.8336849071431526,
+      "grad_norm": 911.5106201171875,
+      "learning_rate": 1.0133036291589587e-06,
+      "loss": 87.0133,
+      "step": 100780
+    },
+    {
+      "epoch": 0.8337676303925218,
+      "grad_norm": 845.751708984375,
+      "learning_rate": 1.0124525870386676e-06,
+      "loss": 75.2599,
+      "step": 100790
+    },
+    {
+      "epoch": 0.8338503536418911,
+      "grad_norm": 547.631591796875,
+      "learning_rate": 1.0116018621892237e-06,
+      "loss": 111.8343,
+      "step": 100800
+    },
+    {
+      "epoch": 0.8339330768912603,
+      "grad_norm": 1028.4368896484375,
+      "learning_rate": 1.0107514546783154e-06,
+      "loss": 92.0812,
+      "step": 100810
+    },
+    {
+      "epoch": 0.8340158001406295,
+      "grad_norm": 653.3988037109375,
+      "learning_rate": 1.0099013645736056e-06,
+      "loss": 84.4328,
+      "step": 100820
+    },
+    {
+      "epoch": 0.8340985233899988,
+      "grad_norm": 479.5120544433594,
+      "learning_rate": 1.0090515919427308e-06,
+      "loss": 93.0216,
+      "step": 100830
+    },
+    {
+      "epoch": 0.834181246639368,
+      "grad_norm": 596.44677734375,
+      "learning_rate": 1.0082021368533078e-06,
+      "loss": 72.0628,
+      "step": 100840
+    },
+    {
+      "epoch": 0.8342639698887372,
+      "grad_norm": 1943.263916015625,
+      "learning_rate": 1.0073529993729191e-06,
+      "loss": 97.5844,
+      "step": 100850
+    },
+    {
+      "epoch": 0.8343466931381065,
+      "grad_norm": 601.1087646484375,
+      "learning_rate": 1.0065041795691289e-06,
+      "loss": 85.2571,
+      "step": 100860
+    },
+    {
+      "epoch": 0.8344294163874757,
+      "grad_norm": 903.16064453125,
+      "learning_rate": 1.0056556775094734e-06,
+      "loss": 106.5367,
+      "step": 100870
+    },
+    {
+      "epoch": 0.8345121396368449,
+      "grad_norm": 942.596923828125,
+      "learning_rate": 1.0048074932614637e-06,
+      "loss": 85.1813,
+      "step": 100880
+    },
+    {
+      "epoch": 0.8345948628862142,
+      "grad_norm": 885.8580322265625,
+      "learning_rate": 1.0039596268925867e-06,
+      "loss": 126.8213,
+      "step": 100890
+    },
+    {
+      "epoch": 0.8346775861355834,
+      "grad_norm": 819.2152709960938,
+      "learning_rate": 1.0031120784703025e-06,
+      "loss": 100.7738,
+      "step": 100900
+    },
+    {
+      "epoch": 0.8347603093849526,
+      "grad_norm": 1307.6453857421875,
+      "learning_rate": 1.0022648480620474e-06,
+      "loss": 91.3189,
+      "step": 100910
+    },
+    {
+      "epoch": 0.8348430326343219,
+      "grad_norm": 1590.2379150390625,
+      "learning_rate": 1.001417935735231e-06,
+      "loss": 102.9999,
+      "step": 100920
+    },
+    {
+      "epoch": 0.8349257558836911,
+      "grad_norm": 919.4576416015625,
+      "learning_rate": 1.0005713415572383e-06,
+      "loss": 103.8122,
+      "step": 100930
+    },
+    {
+      "epoch": 0.8350084791330603,
+      "grad_norm": 835.6314697265625,
+      "learning_rate": 9.99725065595429e-07,
+      "loss": 95.0695,
+      "step": 100940
+    },
+    {
+      "epoch": 0.8350912023824296,
+      "grad_norm": 1234.212646484375,
+      "learning_rate": 9.988791079171378e-07,
+      "loss": 119.7262,
+      "step": 100950
+    },
+    {
+      "epoch": 0.8351739256317988,
+      "grad_norm": 558.1318969726562,
+      "learning_rate": 9.98033468589673e-07,
+      "loss": 114.4958,
+      "step": 100960
+    },
+    {
+      "epoch": 0.835256648881168,
+      "grad_norm": 611.1209106445312,
+      "learning_rate": 9.971881476803185e-07,
+      "loss": 76.4033,
+      "step": 100970
+    },
+    {
+      "epoch": 0.8353393721305373,
+      "grad_norm": 1021.4954223632812,
+      "learning_rate": 9.963431452563331e-07,
+      "loss": 86.9654,
+      "step": 100980
+    },
+    {
+      "epoch": 0.8354220953799065,
+      "grad_norm": 538.6202392578125,
+      "learning_rate": 9.954984613849488e-07,
+      "loss": 93.7995,
+      "step": 100990
+    },
+    {
+      "epoch": 0.8355048186292757,
+      "grad_norm": 643.6083984375,
+      "learning_rate": 9.94654096133374e-07,
+      "loss": 80.0093,
+      "step": 101000
+    },
+    {
+      "epoch": 0.835587541878645,
+      "grad_norm": 822.85791015625,
+      "learning_rate": 9.938100495687907e-07,
+      "loss": 74.7564,
+      "step": 101010
+    },
+    {
+      "epoch": 0.8356702651280142,
+      "grad_norm": 732.5307006835938,
+      "learning_rate": 9.929663217583562e-07,
+      "loss": 98.5391,
+      "step": 101020
+    },
+    {
+      "epoch": 0.8357529883773834,
+      "grad_norm": 762.5067138671875,
+      "learning_rate": 9.92122912769201e-07,
+      "loss": 108.4045,
+      "step": 101030
+    },
+    {
+      "epoch": 0.8358357116267527,
+      "grad_norm": 957.1695556640625,
+      "learning_rate": 9.912798226684322e-07,
+      "loss": 86.2965,
+      "step": 101040
+    },
+    {
+      "epoch": 0.8359184348761219,
+      "grad_norm": 614.9725952148438,
+      "learning_rate": 9.90437051523131e-07,
+      "loss": 68.6436,
+      "step": 101050
+    },
+    {
+      "epoch": 0.8360011581254911,
+      "grad_norm": 594.4864501953125,
+      "learning_rate": 9.895945994003514e-07,
+      "loss": 76.7402,
+      "step": 101060
+    },
+    {
+      "epoch": 0.8360838813748604,
+      "grad_norm": 601.7037353515625,
+      "learning_rate": 9.887524663671243e-07,
+      "loss": 92.365,
+      "step": 101070
+    },
+    {
+      "epoch": 0.8361666046242296,
+      "grad_norm": 1368.8843994140625,
+      "learning_rate": 9.879106524904547e-07,
+      "loss": 100.576,
+      "step": 101080
+    },
+    {
+      "epoch": 0.8362493278735988,
+      "grad_norm": 971.6827392578125,
+      "learning_rate": 9.870691578373216e-07,
+      "loss": 72.7704,
+      "step": 101090
+    },
+    {
+      "epoch": 0.8363320511229682,
+      "grad_norm": 824.6112670898438,
+      "learning_rate": 9.862279824746784e-07,
+      "loss": 66.5189,
+      "step": 101100
+    },
+    {
+      "epoch": 0.8364147743723374,
+      "grad_norm": 811.2408447265625,
+      "learning_rate": 9.853871264694536e-07,
+      "loss": 101.336,
+      "step": 101110
+    },
+    {
+      "epoch": 0.8364974976217066,
+      "grad_norm": 577.3998413085938,
+      "learning_rate": 9.845465898885509e-07,
+      "loss": 91.787,
+      "step": 101120
+    },
+    {
+      "epoch": 0.8365802208710759,
+      "grad_norm": 623.9453735351562,
+      "learning_rate": 9.837063727988478e-07,
+      "loss": 64.0953,
+      "step": 101130
+    },
+    {
+      "epoch": 0.8366629441204451,
+      "grad_norm": 773.7106323242188,
+      "learning_rate": 9.828664752671963e-07,
+      "loss": 73.4519,
+      "step": 101140
+    },
+    {
+      "epoch": 0.8367456673698143,
+      "grad_norm": 1168.3428955078125,
+      "learning_rate": 9.82026897360423e-07,
+      "loss": 84.8574,
+      "step": 101150
+    },
+    {
+      "epoch": 0.8368283906191836,
+      "grad_norm": 1095.8922119140625,
+      "learning_rate": 9.811876391453296e-07,
+      "loss": 95.7534,
+      "step": 101160
+    },
+    {
+      "epoch": 0.8369111138685528,
+      "grad_norm": 932.7236938476562,
+      "learning_rate": 9.803487006886914e-07,
+      "loss": 102.7611,
+      "step": 101170
+    },
+    {
+      "epoch": 0.836993837117922,
+      "grad_norm": 983.7799072265625,
+      "learning_rate": 9.795100820572601e-07,
+      "loss": 72.2526,
+      "step": 101180
+    },
+    {
+      "epoch": 0.8370765603672913,
+      "grad_norm": 440.9442138671875,
+      "learning_rate": 9.78671783317761e-07,
+      "loss": 82.0223,
+      "step": 101190
+    },
+    {
+      "epoch": 0.8371592836166605,
+      "grad_norm": 806.7423706054688,
+      "learning_rate": 9.778338045368901e-07,
+      "loss": 94.5054,
+      "step": 101200
+    },
+    {
+      "epoch": 0.8372420068660297,
+      "grad_norm": 1106.8468017578125,
+      "learning_rate": 9.769961457813254e-07,
+      "loss": 93.8625,
+      "step": 101210
+    },
+    {
+      "epoch": 0.837324730115399,
+      "grad_norm": 983.9953002929688,
+      "learning_rate": 9.761588071177141e-07,
+      "loss": 109.6925,
+      "step": 101220
+    },
+    {
+      "epoch": 0.8374074533647682,
+      "grad_norm": 850.5468139648438,
+      "learning_rate": 9.753217886126797e-07,
+      "loss": 84.854,
+      "step": 101230
+    },
+    {
+      "epoch": 0.8374901766141374,
+      "grad_norm": 1075.742919921875,
+      "learning_rate": 9.7448509033282e-07,
+      "loss": 79.7995,
+      "step": 101240
+    },
+    {
+      "epoch": 0.8375728998635067,
+      "grad_norm": 621.546875,
+      "learning_rate": 9.73648712344707e-07,
+      "loss": 81.4679,
+      "step": 101250
+    },
+    {
+      "epoch": 0.8376556231128759,
+      "grad_norm": 409.54095458984375,
+      "learning_rate": 9.728126547148875e-07,
+      "loss": 86.3328,
+      "step": 101260
+    },
+    {
+      "epoch": 0.8377383463622451,
+      "grad_norm": 1011.2172241210938,
+      "learning_rate": 9.719769175098842e-07,
+      "loss": 70.703,
+      "step": 101270
+    },
+    {
+      "epoch": 0.8378210696116144,
+      "grad_norm": 919.0086669921875,
+      "learning_rate": 9.711415007961899e-07,
+      "loss": 94.6781,
+      "step": 101280
+    },
+    {
+      "epoch": 0.8379037928609836,
+      "grad_norm": 908.69189453125,
+      "learning_rate": 9.70306404640275e-07,
+      "loss": 133.2232,
+      "step": 101290
+    },
+    {
+      "epoch": 0.8379865161103528,
+      "grad_norm": 736.891845703125,
+      "learning_rate": 9.69471629108587e-07,
+      "loss": 87.2299,
+      "step": 101300
+    },
+    {
+      "epoch": 0.838069239359722,
+      "grad_norm": 376.5754699707031,
+      "learning_rate": 9.686371742675443e-07,
+      "loss": 78.6543,
+      "step": 101310
+    },
+    {
+      "epoch": 0.8381519626090913,
+      "grad_norm": 6189.4169921875,
+      "learning_rate": 9.678030401835399e-07,
+      "loss": 124.2069,
+      "step": 101320
+    },
+    {
+      "epoch": 0.8382346858584605,
+      "grad_norm": 1200.981689453125,
+      "learning_rate": 9.66969226922942e-07,
+      "loss": 96.2958,
+      "step": 101330
+    },
+    {
+      "epoch": 0.8383174091078297,
+      "grad_norm": 1382.705078125,
+      "learning_rate": 9.66135734552094e-07,
+      "loss": 82.9574,
+      "step": 101340
+    },
+    {
+      "epoch": 0.838400132357199,
+      "grad_norm": 671.8123168945312,
+      "learning_rate": 9.653025631373125e-07,
+      "loss": 62.2783,
+      "step": 101350
+    },
+    {
+      "epoch": 0.8384828556065682,
+      "grad_norm": 720.500732421875,
+      "learning_rate": 9.644697127448904e-07,
+      "loss": 67.8807,
+      "step": 101360
+    },
+    {
+      "epoch": 0.8385655788559374,
+      "grad_norm": 402.5303955078125,
+      "learning_rate": 9.636371834410918e-07,
+      "loss": 76.3787,
+      "step": 101370
+    },
+    {
+      "epoch": 0.8386483021053067,
+      "grad_norm": 698.8262329101562,
+      "learning_rate": 9.62804975292158e-07,
+      "loss": 90.8034,
+      "step": 101380
+    },
+    {
+      "epoch": 0.8387310253546759,
+      "grad_norm": 715.2416381835938,
+      "learning_rate": 9.619730883643026e-07,
+      "loss": 70.7118,
+      "step": 101390
+    },
+    {
+      "epoch": 0.8388137486040451,
+      "grad_norm": 1074.1124267578125,
+      "learning_rate": 9.611415227237181e-07,
+      "loss": 67.4193,
+      "step": 101400
+    },
+    {
+      "epoch": 0.8388964718534144,
+      "grad_norm": 1017.0978393554688,
+      "learning_rate": 9.603102784365664e-07,
+      "loss": 95.0126,
+      "step": 101410
+    },
+    {
+      "epoch": 0.8389791951027836,
+      "grad_norm": 695.4487915039062,
+      "learning_rate": 9.594793555689868e-07,
+      "loss": 113.1844,
+      "step": 101420
+    },
+    {
+      "epoch": 0.8390619183521528,
+      "grad_norm": 556.859130859375,
+      "learning_rate": 9.58648754187091e-07,
+      "loss": 102.4575,
+      "step": 101430
+    },
+    {
+      "epoch": 0.8391446416015221,
+      "grad_norm": 1115.348388671875,
+      "learning_rate": 9.578184743569662e-07,
+      "loss": 101.0772,
+      "step": 101440
+    },
+    {
+      "epoch": 0.8392273648508913,
+      "grad_norm": 611.0310668945312,
+      "learning_rate": 9.569885161446762e-07,
+      "loss": 75.4577,
+      "step": 101450
+    },
+    {
+      "epoch": 0.8393100881002605,
+      "grad_norm": 886.7157592773438,
+      "learning_rate": 9.56158879616253e-07,
+      "loss": 94.1332,
+      "step": 101460
+    },
+    {
+      "epoch": 0.8393928113496298,
+      "grad_norm": 683.062744140625,
+      "learning_rate": 9.553295648377097e-07,
+      "loss": 63.7114,
+      "step": 101470
+    },
+    {
+      "epoch": 0.839475534598999,
+      "grad_norm": 957.2645874023438,
+      "learning_rate": 9.5450057187503e-07,
+      "loss": 87.1377,
+      "step": 101480
+    },
+    {
+      "epoch": 0.8395582578483682,
+      "grad_norm": 845.6005249023438,
+      "learning_rate": 9.536719007941725e-07,
+      "loss": 78.0765,
+      "step": 101490
+    },
+    {
+      "epoch": 0.8396409810977375,
+      "grad_norm": 448.1517639160156,
+      "learning_rate": 9.528435516610729e-07,
+      "loss": 96.6448,
+      "step": 101500
+    },
+    {
+      "epoch": 0.8397237043471067,
+      "grad_norm": 999.4203491210938,
+      "learning_rate": 9.520155245416379e-07,
+      "loss": 84.303,
+      "step": 101510
+    },
+    {
+      "epoch": 0.839806427596476,
+      "grad_norm": 661.683349609375,
+      "learning_rate": 9.511878195017499e-07,
+      "loss": 71.9918,
+      "step": 101520
+    },
+    {
+      "epoch": 0.8398891508458453,
+      "grad_norm": 1106.2100830078125,
+      "learning_rate": 9.503604366072666e-07,
+      "loss": 80.9662,
+      "step": 101530
+    },
+    {
+      "epoch": 0.8399718740952145,
+      "grad_norm": 1053.9813232421875,
+      "learning_rate": 9.495333759240171e-07,
+      "loss": 99.4201,
+      "step": 101540
+    },
+    {
+      "epoch": 0.8400545973445837,
+      "grad_norm": 1770.6314697265625,
+      "learning_rate": 9.487066375178078e-07,
+      "loss": 97.2852,
+      "step": 101550
+    },
+    {
+      "epoch": 0.840137320593953,
+      "grad_norm": 831.5025634765625,
+      "learning_rate": 9.47880221454418e-07,
+      "loss": 87.7667,
+      "step": 101560
+    },
+    {
+      "epoch": 0.8402200438433222,
+      "grad_norm": 619.336181640625,
+      "learning_rate": 9.470541277996026e-07,
+      "loss": 105.6826,
+      "step": 101570
+    },
+    {
+      "epoch": 0.8403027670926914,
+      "grad_norm": 590.185302734375,
+      "learning_rate": 9.462283566190894e-07,
+      "loss": 116.2632,
+      "step": 101580
+    },
+    {
+      "epoch": 0.8403854903420607,
+      "grad_norm": 652.5518798828125,
+      "learning_rate": 9.454029079785809e-07,
+      "loss": 90.1779,
+      "step": 101590
+    },
+    {
+      "epoch": 0.8404682135914299,
+      "grad_norm": 699.6583251953125,
+      "learning_rate": 9.445777819437557e-07,
+      "loss": 91.0062,
+      "step": 101600
+    },
+    {
+      "epoch": 0.8405509368407991,
+      "grad_norm": 601.4156494140625,
+      "learning_rate": 9.437529785802647e-07,
+      "loss": 69.9341,
+      "step": 101610
+    },
+    {
+      "epoch": 0.8406336600901684,
+      "grad_norm": 309.7095642089844,
+      "learning_rate": 9.429284979537346e-07,
+      "loss": 72.0386,
+      "step": 101620
+    },
+    {
+      "epoch": 0.8407163833395376,
+      "grad_norm": 1362.3778076171875,
+      "learning_rate": 9.421043401297636e-07,
+      "loss": 86.1247,
+      "step": 101630
+    },
+    {
+      "epoch": 0.8407991065889068,
+      "grad_norm": 1056.7972412109375,
+      "learning_rate": 9.412805051739266e-07,
+      "loss": 94.5386,
+      "step": 101640
+    },
+    {
+      "epoch": 0.8408818298382761,
+      "grad_norm": 394.115478515625,
+      "learning_rate": 9.404569931517726e-07,
+      "loss": 57.8183,
+      "step": 101650
+    },
+    {
+      "epoch": 0.8409645530876453,
+      "grad_norm": 1883.9283447265625,
+      "learning_rate": 9.396338041288255e-07,
+      "loss": 83.5486,
+      "step": 101660
+    },
+    {
+      "epoch": 0.8410472763370145,
+      "grad_norm": 773.7887573242188,
+      "learning_rate": 9.388109381705817e-07,
+      "loss": 64.771,
+      "step": 101670
+    },
+    {
+      "epoch": 0.8411299995863838,
+      "grad_norm": 1021.2868041992188,
+      "learning_rate": 9.379883953425134e-07,
+      "loss": 84.9474,
+      "step": 101680
+    },
+    {
+      "epoch": 0.841212722835753,
+      "grad_norm": 808.8427734375,
+      "learning_rate": 9.371661757100648e-07,
+      "loss": 94.7058,
+      "step": 101690
+    },
+    {
+      "epoch": 0.8412954460851222,
+      "grad_norm": 811.6118774414062,
+      "learning_rate": 9.363442793386606e-07,
+      "loss": 95.701,
+      "step": 101700
+    },
+    {
+      "epoch": 0.8413781693344915,
+      "grad_norm": 923.9641723632812,
+      "learning_rate": 9.355227062936912e-07,
+      "loss": 101.2932,
+      "step": 101710
+    },
+    {
+      "epoch": 0.8414608925838607,
+      "grad_norm": 789.2193603515625,
+      "learning_rate": 9.34701456640526e-07,
+      "loss": 69.8859,
+      "step": 101720
+    },
+    {
+      "epoch": 0.8415436158332299,
+      "grad_norm": 676.7222900390625,
+      "learning_rate": 9.338805304445092e-07,
+      "loss": 77.5232,
+      "step": 101730
+    },
+    {
+      "epoch": 0.8416263390825992,
+      "grad_norm": 793.80615234375,
+      "learning_rate": 9.330599277709579e-07,
+      "loss": 83.3514,
+      "step": 101740
+    },
+    {
+      "epoch": 0.8417090623319684,
+      "grad_norm": 857.0948486328125,
+      "learning_rate": 9.322396486851626e-07,
+      "loss": 94.8482,
+      "step": 101750
+    },
+    {
+      "epoch": 0.8417917855813376,
+      "grad_norm": 867.2514038085938,
+      "learning_rate": 9.314196932523906e-07,
+      "loss": 96.5802,
+      "step": 101760
+    },
+    {
+      "epoch": 0.8418745088307069,
+      "grad_norm": 1006.9816284179688,
+      "learning_rate": 9.306000615378813e-07,
+      "loss": 89.8353,
+      "step": 101770
+    },
+    {
+      "epoch": 0.8419572320800761,
+      "grad_norm": 811.1846923828125,
+      "learning_rate": 9.297807536068476e-07,
+      "loss": 98.205,
+      "step": 101780
+    },
+    {
+      "epoch": 0.8420399553294453,
+      "grad_norm": 760.5066528320312,
+      "learning_rate": 9.289617695244818e-07,
+      "loss": 109.3908,
+      "step": 101790
+    },
+    {
+      "epoch": 0.8421226785788146,
+      "grad_norm": 855.790771484375,
+      "learning_rate": 9.281431093559439e-07,
+      "loss": 89.0774,
+      "step": 101800
+    },
+    {
+      "epoch": 0.8422054018281838,
+      "grad_norm": 901.5059814453125,
+      "learning_rate": 9.273247731663709e-07,
+      "loss": 77.9359,
+      "step": 101810
+    },
+    {
+      "epoch": 0.842288125077553,
+      "grad_norm": 856.0442504882812,
+      "learning_rate": 9.26506761020875e-07,
+      "loss": 103.1285,
+      "step": 101820
+    },
+    {
+      "epoch": 0.8423708483269223,
+      "grad_norm": 919.891357421875,
+      "learning_rate": 9.256890729845414e-07,
+      "loss": 96.1979,
+      "step": 101830
+    },
+    {
+      "epoch": 0.8424535715762915,
+      "grad_norm": 742.4312133789062,
+      "learning_rate": 9.248717091224291e-07,
+      "loss": 88.905,
+      "step": 101840
+    },
+    {
+      "epoch": 0.8425362948256607,
+      "grad_norm": 697.6671142578125,
+      "learning_rate": 9.240546694995733e-07,
+      "loss": 71.3341,
+      "step": 101850
+    },
+    {
+      "epoch": 0.84261901807503,
+      "grad_norm": 611.2695922851562,
+      "learning_rate": 9.23237954180981e-07,
+      "loss": 75.6786,
+      "step": 101860
+    },
+    {
+      "epoch": 0.8427017413243992,
+      "grad_norm": 730.123779296875,
+      "learning_rate": 9.224215632316346e-07,
+      "loss": 69.9281,
+      "step": 101870
+    },
+    {
+      "epoch": 0.8427844645737684,
+      "grad_norm": 824.8983764648438,
+      "learning_rate": 9.216054967164916e-07,
+      "loss": 95.3742,
+      "step": 101880
+    },
+    {
+      "epoch": 0.8428671878231377,
+      "grad_norm": 783.8893432617188,
+      "learning_rate": 9.207897547004812e-07,
+      "loss": 81.9008,
+      "step": 101890
+    },
+    {
+      "epoch": 0.8429499110725069,
+      "grad_norm": 996.7938842773438,
+      "learning_rate": 9.199743372485093e-07,
+      "loss": 117.712,
+      "step": 101900
+    },
+    {
+      "epoch": 0.8430326343218761,
+      "grad_norm": 677.0623779296875,
+      "learning_rate": 9.191592444254549e-07,
+      "loss": 76.7643,
+      "step": 101910
+    },
+    {
+      "epoch": 0.8431153575712454,
+      "grad_norm": 1035.1934814453125,
+      "learning_rate": 9.183444762961702e-07,
+      "loss": 91.2087,
+      "step": 101920
+    },
+    {
+      "epoch": 0.8431980808206146,
+      "grad_norm": 940.440185546875,
+      "learning_rate": 9.175300329254839e-07,
+      "loss": 106.9464,
+      "step": 101930
+    },
+    {
+      "epoch": 0.8432808040699838,
+      "grad_norm": 685.4142456054688,
+      "learning_rate": 9.167159143781967e-07,
+      "loss": 85.0712,
+      "step": 101940
+    },
+    {
+      "epoch": 0.8433635273193532,
+      "grad_norm": 931.5978393554688,
+      "learning_rate": 9.159021207190843e-07,
+      "loss": 80.0363,
+      "step": 101950
+    },
+    {
+      "epoch": 0.8434462505687224,
+      "grad_norm": 683.4605102539062,
+      "learning_rate": 9.150886520128966e-07,
+      "loss": 107.6678,
+      "step": 101960
+    },
+    {
+      "epoch": 0.8435289738180916,
+      "grad_norm": 1009.03369140625,
+      "learning_rate": 9.142755083243577e-07,
+      "loss": 84.9778,
+      "step": 101970
+    },
+    {
+      "epoch": 0.8436116970674609,
+      "grad_norm": 847.3112182617188,
+      "learning_rate": 9.134626897181659e-07,
+      "loss": 75.6202,
+      "step": 101980
+    },
+    {
+      "epoch": 0.8436944203168301,
+      "grad_norm": 609.8668823242188,
+      "learning_rate": 9.126501962589928e-07,
+      "loss": 90.4189,
+      "step": 101990
+    },
+    {
+      "epoch": 0.8437771435661993,
+      "grad_norm": 860.5015869140625,
+      "learning_rate": 9.118380280114858e-07,
+      "loss": 82.1258,
+      "step": 102000
+    },
+    {
+      "epoch": 0.8438598668155685,
+      "grad_norm": 771.5186157226562,
+      "learning_rate": 9.110261850402641e-07,
+      "loss": 77.5802,
+      "step": 102010
+    },
+    {
+      "epoch": 0.8439425900649378,
+      "grad_norm": 814.3161010742188,
+      "learning_rate": 9.102146674099232e-07,
+      "loss": 74.584,
+      "step": 102020
+    },
+    {
+      "epoch": 0.844025313314307,
+      "grad_norm": 399.5432434082031,
+      "learning_rate": 9.094034751850317e-07,
+      "loss": 119.564,
+      "step": 102030
+    },
+    {
+      "epoch": 0.8441080365636762,
+      "grad_norm": 472.89300537109375,
+      "learning_rate": 9.085926084301327e-07,
+      "loss": 73.9548,
+      "step": 102040
+    },
+    {
+      "epoch": 0.8441907598130455,
+      "grad_norm": 1127.3804931640625,
+      "learning_rate": 9.077820672097426e-07,
+      "loss": 69.2523,
+      "step": 102050
+    },
+    {
+      "epoch": 0.8442734830624147,
+      "grad_norm": 555.0116577148438,
+      "learning_rate": 9.069718515883524e-07,
+      "loss": 92.0813,
+      "step": 102060
+    },
+    {
+      "epoch": 0.8443562063117839,
+      "grad_norm": 739.4345703125,
+      "learning_rate": 9.06161961630428e-07,
+      "loss": 74.4284,
+      "step": 102070
+    },
+    {
+      "epoch": 0.8444389295611532,
+      "grad_norm": 1793.3900146484375,
+      "learning_rate": 9.053523974004075e-07,
+      "loss": 115.2249,
+      "step": 102080
+    },
+    {
+      "epoch": 0.8445216528105224,
+      "grad_norm": 1008.2157592773438,
+      "learning_rate": 9.045431589627052e-07,
+      "loss": 80.6738,
+      "step": 102090
+    },
+    {
+      "epoch": 0.8446043760598916,
+      "grad_norm": 867.9142456054688,
+      "learning_rate": 9.037342463817084e-07,
+      "loss": 84.0623,
+      "step": 102100
+    },
+    {
+      "epoch": 0.8446870993092609,
+      "grad_norm": 1016.9595336914062,
+      "learning_rate": 9.029256597217778e-07,
+      "loss": 97.6914,
+      "step": 102110
+    },
+    {
+      "epoch": 0.8447698225586301,
+      "grad_norm": 538.5012817382812,
+      "learning_rate": 9.021173990472498e-07,
+      "loss": 72.0225,
+      "step": 102120
+    },
+    {
+      "epoch": 0.8448525458079993,
+      "grad_norm": 910.0997314453125,
+      "learning_rate": 9.013094644224346e-07,
+      "loss": 87.6451,
+      "step": 102130
+    },
+    {
+      "epoch": 0.8449352690573686,
+      "grad_norm": 567.3370361328125,
+      "learning_rate": 9.005018559116135e-07,
+      "loss": 68.3961,
+      "step": 102140
+    },
+    {
+      "epoch": 0.8450179923067378,
+      "grad_norm": 592.2027587890625,
+      "learning_rate": 8.996945735790447e-07,
+      "loss": 95.2749,
+      "step": 102150
+    },
+    {
+      "epoch": 0.845100715556107,
+      "grad_norm": 781.4312133789062,
+      "learning_rate": 8.988876174889616e-07,
+      "loss": 62.1267,
+      "step": 102160
+    },
+    {
+      "epoch": 0.8451834388054763,
+      "grad_norm": 1090.7720947265625,
+      "learning_rate": 8.980809877055696e-07,
+      "loss": 76.1743,
+      "step": 102170
+    },
+    {
+      "epoch": 0.8452661620548455,
+      "grad_norm": 1055.83447265625,
+      "learning_rate": 8.97274684293048e-07,
+      "loss": 90.5386,
+      "step": 102180
+    },
+    {
+      "epoch": 0.8453488853042147,
+      "grad_norm": 443.5655822753906,
+      "learning_rate": 8.964687073155509e-07,
+      "loss": 100.6941,
+      "step": 102190
+    },
+    {
+      "epoch": 0.845431608553584,
+      "grad_norm": 919.127685546875,
+      "learning_rate": 8.95663056837206e-07,
+      "loss": 86.7603,
+      "step": 102200
+    },
+    {
+      "epoch": 0.8455143318029532,
+      "grad_norm": 703.267822265625,
+      "learning_rate": 8.948577329221153e-07,
+      "loss": 92.9238,
+      "step": 102210
+    },
+    {
+      "epoch": 0.8455970550523224,
+      "grad_norm": 597.8628540039062,
+      "learning_rate": 8.940527356343564e-07,
+      "loss": 80.7923,
+      "step": 102220
+    },
+    {
+      "epoch": 0.8456797783016917,
+      "grad_norm": 985.700927734375,
+      "learning_rate": 8.93248065037976e-07,
+      "loss": 72.7414,
+      "step": 102230
+    },
+    {
+      "epoch": 0.8457625015510609,
+      "grad_norm": 608.7626953125,
+      "learning_rate": 8.924437211969983e-07,
+      "loss": 95.9518,
+      "step": 102240
+    },
+    {
+      "epoch": 0.8458452248004301,
+      "grad_norm": 734.87841796875,
+      "learning_rate": 8.916397041754238e-07,
+      "loss": 93.5361,
+      "step": 102250
+    },
+    {
+      "epoch": 0.8459279480497994,
+      "grad_norm": 707.2564697265625,
+      "learning_rate": 8.90836014037223e-07,
+      "loss": 87.7497,
+      "step": 102260
+    },
+    {
+      "epoch": 0.8460106712991686,
+      "grad_norm": 466.18157958984375,
+      "learning_rate": 8.900326508463425e-07,
+      "loss": 99.7546,
+      "step": 102270
+    },
+    {
+      "epoch": 0.8460933945485378,
+      "grad_norm": 665.1851806640625,
+      "learning_rate": 8.892296146667018e-07,
+      "loss": 98.5748,
+      "step": 102280
+    },
+    {
+      "epoch": 0.8461761177979071,
+      "grad_norm": 1132.5751953125,
+      "learning_rate": 8.884269055621941e-07,
+      "loss": 91.5124,
+      "step": 102290
+    },
+    {
+      "epoch": 0.8462588410472763,
+      "grad_norm": 954.9869384765625,
+      "learning_rate": 8.876245235966884e-07,
+      "loss": 124.0621,
+      "step": 102300
+    },
+    {
+      "epoch": 0.8463415642966455,
+      "grad_norm": 1059.5760498046875,
+      "learning_rate": 8.868224688340277e-07,
+      "loss": 87.5652,
+      "step": 102310
+    },
+    {
+      "epoch": 0.8464242875460148,
+      "grad_norm": 886.4304809570312,
+      "learning_rate": 8.860207413380245e-07,
+      "loss": 93.7244,
+      "step": 102320
+    },
+    {
+      "epoch": 0.846507010795384,
+      "grad_norm": 954.0653076171875,
+      "learning_rate": 8.852193411724702e-07,
+      "loss": 94.8709,
+      "step": 102330
+    },
+    {
+      "epoch": 0.8465897340447532,
+      "grad_norm": 650.5189208984375,
+      "learning_rate": 8.844182684011276e-07,
+      "loss": 68.5645,
+      "step": 102340
+    },
+    {
+      "epoch": 0.8466724572941225,
+      "grad_norm": 680.576904296875,
+      "learning_rate": 8.83617523087737e-07,
+      "loss": 66.5604,
+      "step": 102350
+    },
+    {
+      "epoch": 0.8467551805434917,
+      "grad_norm": 767.3767700195312,
+      "learning_rate": 8.828171052960077e-07,
+      "loss": 114.503,
+      "step": 102360
+    },
+    {
+      "epoch": 0.846837903792861,
+      "grad_norm": 712.951171875,
+      "learning_rate": 8.820170150896268e-07,
+      "loss": 114.4278,
+      "step": 102370
+    },
+    {
+      "epoch": 0.8469206270422303,
+      "grad_norm": 753.8101806640625,
+      "learning_rate": 8.812172525322527e-07,
+      "loss": 66.3853,
+      "step": 102380
+    },
+    {
+      "epoch": 0.8470033502915995,
+      "grad_norm": 927.0662231445312,
+      "learning_rate": 8.8041781768752e-07,
+      "loss": 63.6904,
+      "step": 102390
+    },
+    {
+      "epoch": 0.8470860735409687,
+      "grad_norm": 1022.492919921875,
+      "learning_rate": 8.796187106190346e-07,
+      "loss": 80.9226,
+      "step": 102400
+    },
+    {
+      "epoch": 0.847168796790338,
+      "grad_norm": 647.39697265625,
+      "learning_rate": 8.788199313903778e-07,
+      "loss": 103.7165,
+      "step": 102410
+    },
+    {
+      "epoch": 0.8472515200397072,
+      "grad_norm": 835.044189453125,
+      "learning_rate": 8.78021480065106e-07,
+      "loss": 99.1959,
+      "step": 102420
+    },
+    {
+      "epoch": 0.8473342432890764,
+      "grad_norm": 1518.077880859375,
+      "learning_rate": 8.772233567067473e-07,
+      "loss": 88.389,
+      "step": 102430
+    },
+    {
+      "epoch": 0.8474169665384457,
+      "grad_norm": 717.4688110351562,
+      "learning_rate": 8.764255613788037e-07,
+      "loss": 77.5931,
+      "step": 102440
+    },
+    {
+      "epoch": 0.8474996897878149,
+      "grad_norm": 487.7947998046875,
+      "learning_rate": 8.756280941447554e-07,
+      "loss": 83.3511,
+      "step": 102450
+    },
+    {
+      "epoch": 0.8475824130371841,
+      "grad_norm": 1047.3568115234375,
+      "learning_rate": 8.748309550680506e-07,
+      "loss": 115.5404,
+      "step": 102460
+    },
+    {
+      "epoch": 0.8476651362865534,
+      "grad_norm": 601.6625366210938,
+      "learning_rate": 8.740341442121153e-07,
+      "loss": 80.3495,
+      "step": 102470
+    },
+    {
+      "epoch": 0.8477478595359226,
+      "grad_norm": 533.0794067382812,
+      "learning_rate": 8.732376616403487e-07,
+      "loss": 76.5713,
+      "step": 102480
+    },
+    {
+      "epoch": 0.8478305827852918,
+      "grad_norm": 918.6082153320312,
+      "learning_rate": 8.724415074161207e-07,
+      "loss": 88.8103,
+      "step": 102490
+    },
+    {
+      "epoch": 0.8479133060346611,
+      "grad_norm": 590.0209350585938,
+      "learning_rate": 8.716456816027791e-07,
+      "loss": 80.2414,
+      "step": 102500
+    },
+    {
+      "epoch": 0.8479960292840303,
+      "grad_norm": 869.5015869140625,
+      "learning_rate": 8.708501842636441e-07,
+      "loss": 91.9651,
+      "step": 102510
+    },
+    {
+      "epoch": 0.8480787525333995,
+      "grad_norm": 636.9760131835938,
+      "learning_rate": 8.700550154620091e-07,
+      "loss": 70.51,
+      "step": 102520
+    },
+    {
+      "epoch": 0.8481614757827688,
+      "grad_norm": 512.7169189453125,
+      "learning_rate": 8.692601752611435e-07,
+      "loss": 74.1003,
+      "step": 102530
+    },
+    {
+      "epoch": 0.848244199032138,
+      "grad_norm": 1306.0889892578125,
+      "learning_rate": 8.684656637242866e-07,
+      "loss": 87.0951,
+      "step": 102540
+    },
+    {
+      "epoch": 0.8483269222815072,
+      "grad_norm": 2022.1649169921875,
+      "learning_rate": 8.676714809146569e-07,
+      "loss": 118.8545,
+      "step": 102550
+    },
+    {
+      "epoch": 0.8484096455308765,
+      "grad_norm": 810.895751953125,
+      "learning_rate": 8.668776268954437e-07,
+      "loss": 87.6652,
+      "step": 102560
+    },
+    {
+      "epoch": 0.8484923687802457,
+      "grad_norm": 636.8590698242188,
+      "learning_rate": 8.660841017298082e-07,
+      "loss": 66.9613,
+      "step": 102570
+    },
+    {
+      "epoch": 0.8485750920296149,
+      "grad_norm": 1104.5308837890625,
+      "learning_rate": 8.652909054808884e-07,
+      "loss": 99.1564,
+      "step": 102580
+    },
+    {
+      "epoch": 0.8486578152789842,
+      "grad_norm": 802.7866821289062,
+      "learning_rate": 8.644980382117956e-07,
+      "loss": 89.1793,
+      "step": 102590
+    },
+    {
+      "epoch": 0.8487405385283534,
+      "grad_norm": 734.8566284179688,
+      "learning_rate": 8.637054999856148e-07,
+      "loss": 60.249,
+      "step": 102600
+    },
+    {
+      "epoch": 0.8488232617777226,
+      "grad_norm": 743.7099609375,
+      "learning_rate": 8.629132908654042e-07,
+      "loss": 92.0852,
+      "step": 102610
+    },
+    {
+      "epoch": 0.8489059850270919,
+      "grad_norm": 810.8662719726562,
+      "learning_rate": 8.621214109141962e-07,
+      "loss": 84.611,
+      "step": 102620
+    },
+    {
+      "epoch": 0.8489887082764611,
+      "grad_norm": 723.8987426757812,
+      "learning_rate": 8.613298601949971e-07,
+      "loss": 98.8016,
+      "step": 102630
+    },
+    {
+      "epoch": 0.8490714315258303,
+      "grad_norm": 1335.8057861328125,
+      "learning_rate": 8.605386387707865e-07,
+      "loss": 79.3897,
+      "step": 102640
+    },
+    {
+      "epoch": 0.8491541547751996,
+      "grad_norm": 916.5572509765625,
+      "learning_rate": 8.597477467045207e-07,
+      "loss": 84.2059,
+      "step": 102650
+    },
+    {
+      "epoch": 0.8492368780245688,
+      "grad_norm": 1256.606689453125,
+      "learning_rate": 8.589571840591243e-07,
+      "loss": 95.1835,
+      "step": 102660
+    },
+    {
+      "epoch": 0.849319601273938,
+      "grad_norm": 850.797119140625,
+      "learning_rate": 8.581669508975005e-07,
+      "loss": 84.3254,
+      "step": 102670
+    },
+    {
+      "epoch": 0.8494023245233073,
+      "grad_norm": 709.500244140625,
+      "learning_rate": 8.573770472825233e-07,
+      "loss": 73.6143,
+      "step": 102680
+    },
+    {
+      "epoch": 0.8494850477726765,
+      "grad_norm": 595.16650390625,
+      "learning_rate": 8.565874732770429e-07,
+      "loss": 72.705,
+      "step": 102690
+    },
+    {
+      "epoch": 0.8495677710220457,
+      "grad_norm": 871.81005859375,
+      "learning_rate": 8.55798228943881e-07,
+      "loss": 92.2849,
+      "step": 102700
+    },
+    {
+      "epoch": 0.849650494271415,
+      "grad_norm": 950.0880737304688,
+      "learning_rate": 8.550093143458355e-07,
+      "loss": 90.0403,
+      "step": 102710
+    },
+    {
+      "epoch": 0.8497332175207842,
+      "grad_norm": 855.10888671875,
+      "learning_rate": 8.542207295456751e-07,
+      "loss": 91.0368,
+      "step": 102720
+    },
+    {
+      "epoch": 0.8498159407701534,
+      "grad_norm": 846.3700561523438,
+      "learning_rate": 8.53432474606144e-07,
+      "loss": 101.0878,
+      "step": 102730
+    },
+    {
+      "epoch": 0.8498986640195226,
+      "grad_norm": 732.4088745117188,
+      "learning_rate": 8.526445495899627e-07,
+      "loss": 95.0499,
+      "step": 102740
+    },
+    {
+      "epoch": 0.8499813872688919,
+      "grad_norm": 723.0433349609375,
+      "learning_rate": 8.518569545598198e-07,
+      "loss": 65.2936,
+      "step": 102750
+    },
+    {
+      "epoch": 0.8500641105182611,
+      "grad_norm": 791.6629638671875,
+      "learning_rate": 8.510696895783821e-07,
+      "loss": 92.8047,
+      "step": 102760
+    },
+    {
+      "epoch": 0.8501468337676303,
+      "grad_norm": 669.09521484375,
+      "learning_rate": 8.502827547082876e-07,
+      "loss": 79.3404,
+      "step": 102770
+    },
+    {
+      "epoch": 0.8502295570169996,
+      "grad_norm": 292.19622802734375,
+      "learning_rate": 8.494961500121501e-07,
+      "loss": 74.0627,
+      "step": 102780
+    },
+    {
+      "epoch": 0.8503122802663688,
+      "grad_norm": 881.7793579101562,
+      "learning_rate": 8.487098755525552e-07,
+      "loss": 85.533,
+      "step": 102790
+    },
+    {
+      "epoch": 0.850395003515738,
+      "grad_norm": 861.0848999023438,
+      "learning_rate": 8.47923931392064e-07,
+      "loss": 66.5782,
+      "step": 102800
+    },
+    {
+      "epoch": 0.8504777267651074,
+      "grad_norm": 857.7160034179688,
+      "learning_rate": 8.471383175932102e-07,
+      "loss": 99.694,
+      "step": 102810
+    },
+    {
+      "epoch": 0.8505604500144766,
+      "grad_norm": 544.4356689453125,
+      "learning_rate": 8.463530342185011e-07,
+      "loss": 83.8572,
+      "step": 102820
+    },
+    {
+      "epoch": 0.8506431732638458,
+      "grad_norm": 837.5228881835938,
+      "learning_rate": 8.455680813304185e-07,
+      "loss": 65.1481,
+      "step": 102830
+    },
+    {
+      "epoch": 0.8507258965132151,
+      "grad_norm": 898.0479736328125,
+      "learning_rate": 8.447834589914172e-07,
+      "loss": 82.4079,
+      "step": 102840
+    },
+    {
+      "epoch": 0.8508086197625843,
+      "grad_norm": 715.7451782226562,
+      "learning_rate": 8.439991672639264e-07,
+      "loss": 89.7485,
+      "step": 102850
+    },
+    {
+      "epoch": 0.8508913430119535,
+      "grad_norm": 1106.7410888671875,
+      "learning_rate": 8.432152062103488e-07,
+      "loss": 75.6558,
+      "step": 102860
+    },
+    {
+      "epoch": 0.8509740662613228,
+      "grad_norm": 957.0466918945312,
+      "learning_rate": 8.424315758930596e-07,
+      "loss": 94.4857,
+      "step": 102870
+    },
+    {
+      "epoch": 0.851056789510692,
+      "grad_norm": 787.596435546875,
+      "learning_rate": 8.416482763744093e-07,
+      "loss": 130.3268,
+      "step": 102880
+    },
+    {
+      "epoch": 0.8511395127600612,
+      "grad_norm": 982.1759033203125,
+      "learning_rate": 8.408653077167217e-07,
+      "loss": 101.9882,
+      "step": 102890
+    },
+    {
+      "epoch": 0.8512222360094305,
+      "grad_norm": 746.855224609375,
+      "learning_rate": 8.400826699822933e-07,
+      "loss": 100.449,
+      "step": 102900
+    },
+    {
+      "epoch": 0.8513049592587997,
+      "grad_norm": 440.57867431640625,
+      "learning_rate": 8.393003632333957e-07,
+      "loss": 70.7138,
+      "step": 102910
+    },
+    {
+      "epoch": 0.8513876825081689,
+      "grad_norm": 402.388427734375,
+      "learning_rate": 8.385183875322733e-07,
+      "loss": 111.9875,
+      "step": 102920
+    },
+    {
+      "epoch": 0.8514704057575382,
+      "grad_norm": 730.9739990234375,
+      "learning_rate": 8.377367429411443e-07,
+      "loss": 68.7929,
+      "step": 102930
+    },
+    {
+      "epoch": 0.8515531290069074,
+      "grad_norm": 1171.7159423828125,
+      "learning_rate": 8.369554295221999e-07,
+      "loss": 123.8448,
+      "step": 102940
+    },
+    {
+      "epoch": 0.8516358522562766,
+      "grad_norm": 828.75048828125,
+      "learning_rate": 8.361744473376066e-07,
+      "loss": 78.1169,
+      "step": 102950
+    },
+    {
+      "epoch": 0.8517185755056459,
+      "grad_norm": 694.2513427734375,
+      "learning_rate": 8.353937964495029e-07,
+      "loss": 103.3453,
+      "step": 102960
+    },
+    {
+      "epoch": 0.8518012987550151,
+      "grad_norm": 1413.7930908203125,
+      "learning_rate": 8.346134769200021e-07,
+      "loss": 83.3088,
+      "step": 102970
+    },
+    {
+      "epoch": 0.8518840220043843,
+      "grad_norm": 860.365234375,
+      "learning_rate": 8.338334888111899e-07,
+      "loss": 77.3941,
+      "step": 102980
+    },
+    {
+      "epoch": 0.8519667452537536,
+      "grad_norm": 750.07177734375,
+      "learning_rate": 8.330538321851284e-07,
+      "loss": 126.0141,
+      "step": 102990
+    },
+    {
+      "epoch": 0.8520494685031228,
+      "grad_norm": 850.6483764648438,
+      "learning_rate": 8.322745071038474e-07,
+      "loss": 96.3906,
+      "step": 103000
+    },
+    {
+      "epoch": 0.852132191752492,
+      "grad_norm": 1176.2943115234375,
+      "learning_rate": 8.314955136293579e-07,
+      "loss": 80.5075,
+      "step": 103010
+    },
+    {
+      "epoch": 0.8522149150018613,
+      "grad_norm": 1178.8140869140625,
+      "learning_rate": 8.307168518236391e-07,
+      "loss": 113.7988,
+      "step": 103020
+    },
+    {
+      "epoch": 0.8522976382512305,
+      "grad_norm": 824.7714233398438,
+      "learning_rate": 8.299385217486466e-07,
+      "loss": 103.9782,
+      "step": 103030
+    },
+    {
+      "epoch": 0.8523803615005997,
+      "grad_norm": 816.1295776367188,
+      "learning_rate": 8.291605234663075e-07,
+      "loss": 102.6652,
+      "step": 103040
+    },
+    {
+      "epoch": 0.852463084749969,
+      "grad_norm": 942.5098876953125,
+      "learning_rate": 8.283828570385239e-07,
+      "loss": 83.2088,
+      "step": 103050
+    },
+    {
+      "epoch": 0.8525458079993382,
+      "grad_norm": 1104.326904296875,
+      "learning_rate": 8.276055225271718e-07,
+      "loss": 79.5488,
+      "step": 103060
+    },
+    {
+      "epoch": 0.8526285312487074,
+      "grad_norm": 845.7022705078125,
+      "learning_rate": 8.26828519994099e-07,
+      "loss": 79.7386,
+      "step": 103070
+    },
+    {
+      "epoch": 0.8527112544980767,
+      "grad_norm": 627.2059326171875,
+      "learning_rate": 8.260518495011299e-07,
+      "loss": 104.046,
+      "step": 103080
+    },
+    {
+      "epoch": 0.8527939777474459,
+      "grad_norm": 1893.8099365234375,
+      "learning_rate": 8.25275511110058e-07,
+      "loss": 90.7877,
+      "step": 103090
+    },
+    {
+      "epoch": 0.8528767009968151,
+      "grad_norm": 1187.659912109375,
+      "learning_rate": 8.244995048826532e-07,
+      "loss": 103.0294,
+      "step": 103100
+    },
+    {
+      "epoch": 0.8529594242461844,
+      "grad_norm": 673.2490844726562,
+      "learning_rate": 8.237238308806611e-07,
+      "loss": 96.1213,
+      "step": 103110
+    },
+    {
+      "epoch": 0.8530421474955536,
+      "grad_norm": 440.6217041015625,
+      "learning_rate": 8.229484891657974e-07,
+      "loss": 80.7047,
+      "step": 103120
+    },
+    {
+      "epoch": 0.8531248707449228,
+      "grad_norm": 816.9833984375,
+      "learning_rate": 8.221734797997522e-07,
+      "loss": 72.0308,
+      "step": 103130
+    },
+    {
+      "epoch": 0.8532075939942921,
+      "grad_norm": 1029.72900390625,
+      "learning_rate": 8.213988028441893e-07,
+      "loss": 90.9839,
+      "step": 103140
+    },
+    {
+      "epoch": 0.8532903172436613,
+      "grad_norm": 952.5328979492188,
+      "learning_rate": 8.20624458360747e-07,
+      "loss": 89.0053,
+      "step": 103150
+    },
+    {
+      "epoch": 0.8533730404930305,
+      "grad_norm": 1081.2255859375,
+      "learning_rate": 8.198504464110358e-07,
+      "loss": 122.348,
+      "step": 103160
+    },
+    {
+      "epoch": 0.8534557637423998,
+      "grad_norm": 1057.4853515625,
+      "learning_rate": 8.190767670566407e-07,
+      "loss": 92.9125,
+      "step": 103170
+    },
+    {
+      "epoch": 0.853538486991769,
+      "grad_norm": 975.9583129882812,
+      "learning_rate": 8.183034203591189e-07,
+      "loss": 117.357,
+      "step": 103180
+    },
+    {
+      "epoch": 0.8536212102411382,
+      "grad_norm": 674.9142456054688,
+      "learning_rate": 8.175304063800021e-07,
+      "loss": 82.4939,
+      "step": 103190
+    },
+    {
+      "epoch": 0.8537039334905075,
+      "grad_norm": 896.0360717773438,
+      "learning_rate": 8.167577251807951e-07,
+      "loss": 71.3201,
+      "step": 103200
+    },
+    {
+      "epoch": 0.8537866567398767,
+      "grad_norm": 779.8267211914062,
+      "learning_rate": 8.159853768229786e-07,
+      "loss": 78.3528,
+      "step": 103210
+    },
+    {
+      "epoch": 0.853869379989246,
+      "grad_norm": 607.8330078125,
+      "learning_rate": 8.152133613680035e-07,
+      "loss": 90.1494,
+      "step": 103220
+    },
+    {
+      "epoch": 0.8539521032386153,
+      "grad_norm": 634.585205078125,
+      "learning_rate": 8.144416788772957e-07,
+      "loss": 72.4637,
+      "step": 103230
+    },
+    {
+      "epoch": 0.8540348264879845,
+      "grad_norm": 672.9807739257812,
+      "learning_rate": 8.136703294122544e-07,
+      "loss": 74.1094,
+      "step": 103240
+    },
+    {
+      "epoch": 0.8541175497373537,
+      "grad_norm": 954.1567993164062,
+      "learning_rate": 8.128993130342538e-07,
+      "loss": 98.1181,
+      "step": 103250
+    },
+    {
+      "epoch": 0.854200272986723,
+      "grad_norm": 597.5408935546875,
+      "learning_rate": 8.121286298046372e-07,
+      "loss": 86.6912,
+      "step": 103260
+    },
+    {
+      "epoch": 0.8542829962360922,
+      "grad_norm": 522.8579711914062,
+      "learning_rate": 8.113582797847252e-07,
+      "loss": 96.0704,
+      "step": 103270
+    },
+    {
+      "epoch": 0.8543657194854614,
+      "grad_norm": 649.9797973632812,
+      "learning_rate": 8.105882630358125e-07,
+      "loss": 78.3974,
+      "step": 103280
+    },
+    {
+      "epoch": 0.8544484427348307,
+      "grad_norm": 538.1500244140625,
+      "learning_rate": 8.098185796191632e-07,
+      "loss": 62.7917,
+      "step": 103290
+    },
+    {
+      "epoch": 0.8545311659841999,
+      "grad_norm": 425.60430908203125,
+      "learning_rate": 8.090492295960206e-07,
+      "loss": 67.8472,
+      "step": 103300
+    },
+    {
+      "epoch": 0.8546138892335691,
+      "grad_norm": 498.4320068359375,
+      "learning_rate": 8.082802130275968e-07,
+      "loss": 80.5484,
+      "step": 103310
+    },
+    {
+      "epoch": 0.8546966124829384,
+      "grad_norm": 679.7952880859375,
+      "learning_rate": 8.075115299750797e-07,
+      "loss": 92.3465,
+      "step": 103320
+    },
+    {
+      "epoch": 0.8547793357323076,
+      "grad_norm": 643.3114013671875,
+      "learning_rate": 8.067431804996284e-07,
+      "loss": 77.1838,
+      "step": 103330
+    },
+    {
+      "epoch": 0.8548620589816768,
+      "grad_norm": 746.61865234375,
+      "learning_rate": 8.059751646623792e-07,
+      "loss": 91.1856,
+      "step": 103340
+    },
+    {
+      "epoch": 0.8549447822310461,
+      "grad_norm": 1265.4454345703125,
+      "learning_rate": 8.052074825244371e-07,
+      "loss": 99.7866,
+      "step": 103350
+    },
+    {
+      "epoch": 0.8550275054804153,
+      "grad_norm": 791.6976318359375,
+      "learning_rate": 8.044401341468839e-07,
+      "loss": 75.7532,
+      "step": 103360
+    },
+    {
+      "epoch": 0.8551102287297845,
+      "grad_norm": 963.1607055664062,
+      "learning_rate": 8.036731195907743e-07,
+      "loss": 87.9717,
+      "step": 103370
+    },
+    {
+      "epoch": 0.8551929519791538,
+      "grad_norm": 702.797119140625,
+      "learning_rate": 8.029064389171365e-07,
+      "loss": 77.9525,
+      "step": 103380
+    },
+    {
+      "epoch": 0.855275675228523,
+      "grad_norm": 794.2588500976562,
+      "learning_rate": 8.021400921869693e-07,
+      "loss": 62.4597,
+      "step": 103390
+    },
+    {
+      "epoch": 0.8553583984778922,
+      "grad_norm": 654.7899780273438,
+      "learning_rate": 8.013740794612512e-07,
+      "loss": 90.647,
+      "step": 103400
+    },
+    {
+      "epoch": 0.8554411217272615,
+      "grad_norm": 317.69580078125,
+      "learning_rate": 8.006084008009285e-07,
+      "loss": 68.5561,
+      "step": 103410
+    },
+    {
+      "epoch": 0.8555238449766307,
+      "grad_norm": 737.1478271484375,
+      "learning_rate": 7.998430562669234e-07,
+      "loss": 94.218,
+      "step": 103420
+    },
+    {
+      "epoch": 0.8556065682259999,
+      "grad_norm": 625.2897338867188,
+      "learning_rate": 7.990780459201291e-07,
+      "loss": 68.3049,
+      "step": 103430
+    },
+    {
+      "epoch": 0.8556892914753692,
+      "grad_norm": 955.1998291015625,
+      "learning_rate": 7.983133698214158e-07,
+      "loss": 98.8367,
+      "step": 103440
+    },
+    {
+      "epoch": 0.8557720147247384,
+      "grad_norm": 1487.3785400390625,
+      "learning_rate": 7.975490280316239e-07,
+      "loss": 124.9303,
+      "step": 103450
+    },
+    {
+      "epoch": 0.8558547379741076,
+      "grad_norm": 822.9546508789062,
+      "learning_rate": 7.96785020611569e-07,
+      "loss": 81.8008,
+      "step": 103460
+    },
+    {
+      "epoch": 0.8559374612234768,
+      "grad_norm": 357.85772705078125,
+      "learning_rate": 7.960213476220402e-07,
+      "loss": 73.6397,
+      "step": 103470
+    },
+    {
+      "epoch": 0.8560201844728461,
+      "grad_norm": 1243.547119140625,
+      "learning_rate": 7.952580091237993e-07,
+      "loss": 73.8588,
+      "step": 103480
+    },
+    {
+      "epoch": 0.8561029077222153,
+      "grad_norm": 1284.897705078125,
+      "learning_rate": 7.944950051775802e-07,
+      "loss": 114.3679,
+      "step": 103490
+    },
+    {
+      "epoch": 0.8561856309715845,
+      "grad_norm": 482.4344482421875,
+      "learning_rate": 7.937323358440935e-07,
+      "loss": 61.2101,
+      "step": 103500
+    },
+    {
+      "epoch": 0.8562683542209538,
+      "grad_norm": 644.4739990234375,
+      "learning_rate": 7.929700011840225e-07,
+      "loss": 59.5683,
+      "step": 103510
+    },
+    {
+      "epoch": 0.856351077470323,
+      "grad_norm": 894.75341796875,
+      "learning_rate": 7.922080012580191e-07,
+      "loss": 64.3507,
+      "step": 103520
+    },
+    {
+      "epoch": 0.8564338007196922,
+      "grad_norm": 336.1860046386719,
+      "learning_rate": 7.914463361267144e-07,
+      "loss": 77.5185,
+      "step": 103530
+    },
+    {
+      "epoch": 0.8565165239690615,
+      "grad_norm": 769.1115112304688,
+      "learning_rate": 7.906850058507098e-07,
+      "loss": 70.1408,
+      "step": 103540
+    },
+    {
+      "epoch": 0.8565992472184307,
+      "grad_norm": 759.4151000976562,
+      "learning_rate": 7.899240104905814e-07,
+      "loss": 62.1766,
+      "step": 103550
+    },
+    {
+      "epoch": 0.8566819704677999,
+      "grad_norm": 1259.5947265625,
+      "learning_rate": 7.891633501068774e-07,
+      "loss": 83.4048,
+      "step": 103560
+    },
+    {
+      "epoch": 0.8567646937171692,
+      "grad_norm": 948.8060302734375,
+      "learning_rate": 7.88403024760121e-07,
+      "loss": 82.6916,
+      "step": 103570
+    },
+    {
+      "epoch": 0.8568474169665384,
+      "grad_norm": 881.0368041992188,
+      "learning_rate": 7.876430345108072e-07,
+      "loss": 92.3865,
+      "step": 103580
+    },
+    {
+      "epoch": 0.8569301402159076,
+      "grad_norm": 609.7398681640625,
+      "learning_rate": 7.868833794194048e-07,
+      "loss": 55.702,
+      "step": 103590
+    },
+    {
+      "epoch": 0.8570128634652769,
+      "grad_norm": 967.415771484375,
+      "learning_rate": 7.861240595463565e-07,
+      "loss": 66.2445,
+      "step": 103600
+    },
+    {
+      "epoch": 0.8570955867146461,
+      "grad_norm": 916.1266479492188,
+      "learning_rate": 7.853650749520775e-07,
+      "loss": 95.8712,
+      "step": 103610
+    },
+    {
+      "epoch": 0.8571783099640153,
+      "grad_norm": 1089.77783203125,
+      "learning_rate": 7.846064256969571e-07,
+      "loss": 96.1446,
+      "step": 103620
+    },
+    {
+      "epoch": 0.8572610332133846,
+      "grad_norm": 1048.322265625,
+      "learning_rate": 7.838481118413571e-07,
+      "loss": 81.8681,
+      "step": 103630
+    },
+    {
+      "epoch": 0.8573437564627538,
+      "grad_norm": 375.7430419921875,
+      "learning_rate": 7.830901334456137e-07,
+      "loss": 88.0491,
+      "step": 103640
+    },
+    {
+      "epoch": 0.857426479712123,
+      "grad_norm": 1264.2864990234375,
+      "learning_rate": 7.823324905700352e-07,
+      "loss": 87.9994,
+      "step": 103650
+    },
+    {
+      "epoch": 0.8575092029614924,
+      "grad_norm": 961.72216796875,
+      "learning_rate": 7.815751832749035e-07,
+      "loss": 85.5537,
+      "step": 103660
+    },
+    {
+      "epoch": 0.8575919262108616,
+      "grad_norm": 1320.927978515625,
+      "learning_rate": 7.808182116204755e-07,
+      "loss": 114.779,
+      "step": 103670
+    },
+    {
+      "epoch": 0.8576746494602308,
+      "grad_norm": 738.6021728515625,
+      "learning_rate": 7.800615756669783e-07,
+      "loss": 100.2147,
+      "step": 103680
+    },
+    {
+      "epoch": 0.8577573727096001,
+      "grad_norm": 623.4457397460938,
+      "learning_rate": 7.793052754746144e-07,
+      "loss": 89.7,
+      "step": 103690
+    },
+    {
+      "epoch": 0.8578400959589693,
+      "grad_norm": 870.4207763671875,
+      "learning_rate": 7.785493111035597e-07,
+      "loss": 102.5554,
+      "step": 103700
+    },
+    {
+      "epoch": 0.8579228192083385,
+      "grad_norm": 1017.7730102539062,
+      "learning_rate": 7.777936826139626e-07,
+      "loss": 99.2322,
+      "step": 103710
+    },
+    {
+      "epoch": 0.8580055424577078,
+      "grad_norm": 630.4139404296875,
+      "learning_rate": 7.770383900659451e-07,
+      "loss": 73.928,
+      "step": 103720
+    },
+    {
+      "epoch": 0.858088265707077,
+      "grad_norm": 926.4790649414062,
+      "learning_rate": 7.762834335196013e-07,
+      "loss": 78.6408,
+      "step": 103730
+    },
+    {
+      "epoch": 0.8581709889564462,
+      "grad_norm": 1010.821044921875,
+      "learning_rate": 7.755288130350008e-07,
+      "loss": 70.0562,
+      "step": 103740
+    },
+    {
+      "epoch": 0.8582537122058155,
+      "grad_norm": 1223.12060546875,
+      "learning_rate": 7.747745286721852e-07,
+      "loss": 103.973,
+      "step": 103750
+    },
+    {
+      "epoch": 0.8583364354551847,
+      "grad_norm": 652.41796875,
+      "learning_rate": 7.740205804911693e-07,
+      "loss": 83.6676,
+      "step": 103760
+    },
+    {
+      "epoch": 0.8584191587045539,
+      "grad_norm": 588.4229736328125,
+      "learning_rate": 7.732669685519406e-07,
+      "loss": 99.9844,
+      "step": 103770
+    },
+    {
+      "epoch": 0.8585018819539232,
+      "grad_norm": 514.46728515625,
+      "learning_rate": 7.725136929144617e-07,
+      "loss": 77.1376,
+      "step": 103780
+    },
+    {
+      "epoch": 0.8585846052032924,
+      "grad_norm": 480.8714294433594,
+      "learning_rate": 7.717607536386662e-07,
+      "loss": 62.5537,
+      "step": 103790
+    },
+    {
+      "epoch": 0.8586673284526616,
+      "grad_norm": 789.140380859375,
+      "learning_rate": 7.71008150784463e-07,
+      "loss": 91.3066,
+      "step": 103800
+    },
+    {
+      "epoch": 0.8587500517020309,
+      "grad_norm": 514.1470947265625,
+      "learning_rate": 7.702558844117325e-07,
+      "loss": 103.8474,
+      "step": 103810
+    },
+    {
+      "epoch": 0.8588327749514001,
+      "grad_norm": 347.6160888671875,
+      "learning_rate": 7.695039545803295e-07,
+      "loss": 95.0633,
+      "step": 103820
+    },
+    {
+      "epoch": 0.8589154982007693,
+      "grad_norm": 526.0089721679688,
+      "learning_rate": 7.687523613500814e-07,
+      "loss": 83.8635,
+      "step": 103830
+    },
+    {
+      "epoch": 0.8589982214501386,
+      "grad_norm": 982.0831909179688,
+      "learning_rate": 7.680011047807894e-07,
+      "loss": 118.7206,
+      "step": 103840
+    },
+    {
+      "epoch": 0.8590809446995078,
+      "grad_norm": 723.9911499023438,
+      "learning_rate": 7.672501849322266e-07,
+      "loss": 83.4975,
+      "step": 103850
+    },
+    {
+      "epoch": 0.859163667948877,
+      "grad_norm": 953.7279663085938,
+      "learning_rate": 7.664996018641413e-07,
+      "loss": 72.9386,
+      "step": 103860
+    },
+    {
+      "epoch": 0.8592463911982463,
+      "grad_norm": 819.978271484375,
+      "learning_rate": 7.657493556362539e-07,
+      "loss": 84.9272,
+      "step": 103870
+    },
+    {
+      "epoch": 0.8593291144476155,
+      "grad_norm": 1022.4807739257812,
+      "learning_rate": 7.649994463082572e-07,
+      "loss": 93.3535,
+      "step": 103880
+    },
+    {
+      "epoch": 0.8594118376969847,
+      "grad_norm": 627.8077392578125,
+      "learning_rate": 7.642498739398185e-07,
+      "loss": 118.7882,
+      "step": 103890
+    },
+    {
+      "epoch": 0.859494560946354,
+      "grad_norm": 927.2117309570312,
+      "learning_rate": 7.63500638590578e-07,
+      "loss": 72.8843,
+      "step": 103900
+    },
+    {
+      "epoch": 0.8595772841957232,
+      "grad_norm": 905.3977661132812,
+      "learning_rate": 7.62751740320149e-07,
+      "loss": 96.9559,
+      "step": 103910
+    },
+    {
+      "epoch": 0.8596600074450924,
+      "grad_norm": 507.61529541015625,
+      "learning_rate": 7.620031791881172e-07,
+      "loss": 89.9055,
+      "step": 103920
+    },
+    {
+      "epoch": 0.8597427306944617,
+      "grad_norm": 831.4810180664062,
+      "learning_rate": 7.612549552540426e-07,
+      "loss": 68.1213,
+      "step": 103930
+    },
+    {
+      "epoch": 0.8598254539438309,
+      "grad_norm": 1053.16943359375,
+      "learning_rate": 7.605070685774596e-07,
+      "loss": 113.8502,
+      "step": 103940
+    },
+    {
+      "epoch": 0.8599081771932001,
+      "grad_norm": 908.6838989257812,
+      "learning_rate": 7.597595192178702e-07,
+      "loss": 100.7114,
+      "step": 103950
+    },
+    {
+      "epoch": 0.8599909004425694,
+      "grad_norm": 800.2693481445312,
+      "learning_rate": 7.590123072347566e-07,
+      "loss": 98.1212,
+      "step": 103960
+    },
+    {
+      "epoch": 0.8600736236919386,
+      "grad_norm": 559.2039794921875,
+      "learning_rate": 7.582654326875705e-07,
+      "loss": 84.4858,
+      "step": 103970
+    },
+    {
+      "epoch": 0.8601563469413078,
+      "grad_norm": 799.5015258789062,
+      "learning_rate": 7.575188956357371e-07,
+      "loss": 88.0666,
+      "step": 103980
+    },
+    {
+      "epoch": 0.8602390701906771,
+      "grad_norm": 413.585693359375,
+      "learning_rate": 7.567726961386546e-07,
+      "loss": 68.5193,
+      "step": 103990
+    },
+    {
+      "epoch": 0.8603217934400463,
+      "grad_norm": 689.4683837890625,
+      "learning_rate": 7.560268342556948e-07,
+      "loss": 68.5218,
+      "step": 104000
+    },
+    {
+      "epoch": 0.8604045166894155,
+      "grad_norm": 1106.870361328125,
+      "learning_rate": 7.552813100462025e-07,
+      "loss": 97.9523,
+      "step": 104010
+    },
+    {
+      "epoch": 0.8604872399387848,
+      "grad_norm": 949.8282470703125,
+      "learning_rate": 7.54536123569497e-07,
+      "loss": 75.4276,
+      "step": 104020
+    },
+    {
+      "epoch": 0.860569963188154,
+      "grad_norm": 1082.73828125,
+      "learning_rate": 7.537912748848669e-07,
+      "loss": 102.4075,
+      "step": 104030
+    },
+    {
+      "epoch": 0.8606526864375232,
+      "grad_norm": 1304.8778076171875,
+      "learning_rate": 7.530467640515782e-07,
+      "loss": 101.8359,
+      "step": 104040
+    },
+    {
+      "epoch": 0.8607354096868925,
+      "grad_norm": 732.41015625,
+      "learning_rate": 7.523025911288656e-07,
+      "loss": 90.7578,
+      "step": 104050
+    },
+    {
+      "epoch": 0.8608181329362617,
+      "grad_norm": 1173.3736572265625,
+      "learning_rate": 7.51558756175943e-07,
+      "loss": 85.3348,
+      "step": 104060
+    },
+    {
+      "epoch": 0.860900856185631,
+      "grad_norm": 673.0076904296875,
+      "learning_rate": 7.508152592519924e-07,
+      "loss": 98.3655,
+      "step": 104070
+    },
+    {
+      "epoch": 0.8609835794350003,
+      "grad_norm": 3371.377197265625,
+      "learning_rate": 7.500721004161709e-07,
+      "loss": 105.9698,
+      "step": 104080
+    },
+    {
+      "epoch": 0.8610663026843695,
+      "grad_norm": 945.2011108398438,
+      "learning_rate": 7.493292797276075e-07,
+      "loss": 106.223,
+      "step": 104090
+    },
+    {
+      "epoch": 0.8611490259337387,
+      "grad_norm": 777.4385375976562,
+      "learning_rate": 7.485867972454053e-07,
+      "loss": 71.3168,
+      "step": 104100
+    },
+    {
+      "epoch": 0.861231749183108,
+      "grad_norm": 1511.415283203125,
+      "learning_rate": 7.478446530286415e-07,
+      "loss": 109.106,
+      "step": 104110
+    },
+    {
+      "epoch": 0.8613144724324772,
+      "grad_norm": 477.33001708984375,
+      "learning_rate": 7.471028471363628e-07,
+      "loss": 119.4962,
+      "step": 104120
+    },
+    {
+      "epoch": 0.8613971956818464,
+      "grad_norm": 616.1210327148438,
+      "learning_rate": 7.463613796275921e-07,
+      "loss": 104.9887,
+      "step": 104130
+    },
+    {
+      "epoch": 0.8614799189312157,
+      "grad_norm": 676.0572509765625,
+      "learning_rate": 7.456202505613252e-07,
+      "loss": 79.0803,
+      "step": 104140
+    },
+    {
+      "epoch": 0.8615626421805849,
+      "grad_norm": 776.3839111328125,
+      "learning_rate": 7.448794599965286e-07,
+      "loss": 73.2689,
+      "step": 104150
+    },
+    {
+      "epoch": 0.8616453654299541,
+      "grad_norm": 814.0975341796875,
+      "learning_rate": 7.441390079921463e-07,
+      "loss": 71.7772,
+      "step": 104160
+    },
+    {
+      "epoch": 0.8617280886793234,
+      "grad_norm": 962.8041381835938,
+      "learning_rate": 7.433988946070913e-07,
+      "loss": 76.7952,
+      "step": 104170
+    },
+    {
+      "epoch": 0.8618108119286926,
+      "grad_norm": 901.0029296875,
+      "learning_rate": 7.426591199002514e-07,
+      "loss": 142.9842,
+      "step": 104180
+    },
+    {
+      "epoch": 0.8618935351780618,
+      "grad_norm": 527.9906005859375,
+      "learning_rate": 7.419196839304865e-07,
+      "loss": 70.0582,
+      "step": 104190
+    },
+    {
+      "epoch": 0.861976258427431,
+      "grad_norm": 820.9356079101562,
+      "learning_rate": 7.411805867566319e-07,
+      "loss": 69.29,
+      "step": 104200
+    },
+    {
+      "epoch": 0.8620589816768003,
+      "grad_norm": 788.0999145507812,
+      "learning_rate": 7.404418284374909e-07,
+      "loss": 103.0304,
+      "step": 104210
+    },
+    {
+      "epoch": 0.8621417049261695,
+      "grad_norm": 770.0556640625,
+      "learning_rate": 7.397034090318455e-07,
+      "loss": 82.4049,
+      "step": 104220
+    },
+    {
+      "epoch": 0.8622244281755387,
+      "grad_norm": 595.1284790039062,
+      "learning_rate": 7.389653285984471e-07,
+      "loss": 77.7128,
+      "step": 104230
+    },
+    {
+      "epoch": 0.862307151424908,
+      "grad_norm": 597.3800048828125,
+      "learning_rate": 7.382275871960215e-07,
+      "loss": 73.3765,
+      "step": 104240
+    },
+    {
+      "epoch": 0.8623898746742772,
+      "grad_norm": 1349.270263671875,
+      "learning_rate": 7.374901848832683e-07,
+      "loss": 98.555,
+      "step": 104250
+    },
+    {
+      "epoch": 0.8624725979236464,
+      "grad_norm": 962.1214599609375,
+      "learning_rate": 7.367531217188595e-07,
+      "loss": 66.6915,
+      "step": 104260
+    },
+    {
+      "epoch": 0.8625553211730157,
+      "grad_norm": 757.8329467773438,
+      "learning_rate": 7.360163977614388e-07,
+      "loss": 112.147,
+      "step": 104270
+    },
+    {
+      "epoch": 0.8626380444223849,
+      "grad_norm": 1464.066650390625,
+      "learning_rate": 7.352800130696253e-07,
+      "loss": 84.9319,
+      "step": 104280
+    },
+    {
+      "epoch": 0.8627207676717541,
+      "grad_norm": 862.40478515625,
+      "learning_rate": 7.345439677020077e-07,
+      "loss": 94.3516,
+      "step": 104290
+    },
+    {
+      "epoch": 0.8628034909211234,
+      "grad_norm": 282.1973876953125,
+      "learning_rate": 7.33808261717151e-07,
+      "loss": 82.4753,
+      "step": 104300
+    },
+    {
+      "epoch": 0.8628862141704926,
+      "grad_norm": 608.2290649414062,
+      "learning_rate": 7.330728951735916e-07,
+      "loss": 79.4648,
+      "step": 104310
+    },
+    {
+      "epoch": 0.8629689374198618,
+      "grad_norm": 794.7872924804688,
+      "learning_rate": 7.323378681298394e-07,
+      "loss": 76.8241,
+      "step": 104320
+    },
+    {
+      "epoch": 0.8630516606692311,
+      "grad_norm": 795.65380859375,
+      "learning_rate": 7.316031806443774e-07,
+      "loss": 68.2953,
+      "step": 104330
+    },
+    {
+      "epoch": 0.8631343839186003,
+      "grad_norm": 975.4625244140625,
+      "learning_rate": 7.308688327756591e-07,
+      "loss": 89.0946,
+      "step": 104340
+    },
+    {
+      "epoch": 0.8632171071679695,
+      "grad_norm": 435.83013916015625,
+      "learning_rate": 7.301348245821172e-07,
+      "loss": 83.3241,
+      "step": 104350
+    },
+    {
+      "epoch": 0.8632998304173388,
+      "grad_norm": 991.8388061523438,
+      "learning_rate": 7.294011561221503e-07,
+      "loss": 82.6674,
+      "step": 104360
+    },
+    {
+      "epoch": 0.863382553666708,
+      "grad_norm": 756.9240112304688,
+      "learning_rate": 7.286678274541358e-07,
+      "loss": 61.5385,
+      "step": 104370
+    },
+    {
+      "epoch": 0.8634652769160772,
+      "grad_norm": 836.5737915039062,
+      "learning_rate": 7.279348386364182e-07,
+      "loss": 88.6052,
+      "step": 104380
+    },
+    {
+      "epoch": 0.8635480001654465,
+      "grad_norm": 563.5185546875,
+      "learning_rate": 7.272021897273196e-07,
+      "loss": 77.5748,
+      "step": 104390
+    },
+    {
+      "epoch": 0.8636307234148157,
+      "grad_norm": 952.6004638671875,
+      "learning_rate": 7.264698807851328e-07,
+      "loss": 80.778,
+      "step": 104400
+    },
+    {
+      "epoch": 0.8637134466641849,
+      "grad_norm": 484.9126281738281,
+      "learning_rate": 7.257379118681251e-07,
+      "loss": 74.0515,
+      "step": 104410
+    },
+    {
+      "epoch": 0.8637961699135542,
+      "grad_norm": 680.9312133789062,
+      "learning_rate": 7.250062830345356e-07,
+      "loss": 96.0891,
+      "step": 104420
+    },
+    {
+      "epoch": 0.8638788931629234,
+      "grad_norm": 759.776123046875,
+      "learning_rate": 7.242749943425765e-07,
+      "loss": 110.3812,
+      "step": 104430
+    },
+    {
+      "epoch": 0.8639616164122926,
+      "grad_norm": 1156.90380859375,
+      "learning_rate": 7.235440458504317e-07,
+      "loss": 98.4978,
+      "step": 104440
+    },
+    {
+      "epoch": 0.8640443396616619,
+      "grad_norm": 910.2420654296875,
+      "learning_rate": 7.228134376162632e-07,
+      "loss": 121.0294,
+      "step": 104450
+    },
+    {
+      "epoch": 0.8641270629110311,
+      "grad_norm": 1492.252197265625,
+      "learning_rate": 7.22083169698199e-07,
+      "loss": 107.3472,
+      "step": 104460
+    },
+    {
+      "epoch": 0.8642097861604003,
+      "grad_norm": 1519.6361083984375,
+      "learning_rate": 7.21353242154344e-07,
+      "loss": 108.2701,
+      "step": 104470
+    },
+    {
+      "epoch": 0.8642925094097696,
+      "grad_norm": 1137.7952880859375,
+      "learning_rate": 7.206236550427747e-07,
+      "loss": 70.6361,
+      "step": 104480
+    },
+    {
+      "epoch": 0.8643752326591388,
+      "grad_norm": 538.6994018554688,
+      "learning_rate": 7.198944084215421e-07,
+      "loss": 59.6236,
+      "step": 104490
+    },
+    {
+      "epoch": 0.864457955908508,
+      "grad_norm": 511.9731140136719,
+      "learning_rate": 7.191655023486682e-07,
+      "loss": 84.1222,
+      "step": 104500
+    },
+    {
+      "epoch": 0.8645406791578774,
+      "grad_norm": 897.7769775390625,
+      "learning_rate": 7.184369368821486e-07,
+      "loss": 84.8233,
+      "step": 104510
+    },
+    {
+      "epoch": 0.8646234024072466,
+      "grad_norm": 597.1415405273438,
+      "learning_rate": 7.177087120799525e-07,
+      "loss": 68.9195,
+      "step": 104520
+    },
+    {
+      "epoch": 0.8647061256566158,
+      "grad_norm": 973.8169555664062,
+      "learning_rate": 7.169808280000213e-07,
+      "loss": 130.055,
+      "step": 104530
+    },
+    {
+      "epoch": 0.8647888489059851,
+      "grad_norm": 729.9190673828125,
+      "learning_rate": 7.16253284700269e-07,
+      "loss": 54.8949,
+      "step": 104540
+    },
+    {
+      "epoch": 0.8648715721553543,
+      "grad_norm": 1125.672119140625,
+      "learning_rate": 7.155260822385828e-07,
+      "loss": 103.3253,
+      "step": 104550
+    },
+    {
+      "epoch": 0.8649542954047235,
+      "grad_norm": 1073.4981689453125,
+      "learning_rate": 7.147992206728238e-07,
+      "loss": 69.4759,
+      "step": 104560
+    },
+    {
+      "epoch": 0.8650370186540928,
+      "grad_norm": 947.5101318359375,
+      "learning_rate": 7.140727000608239e-07,
+      "loss": 90.7275,
+      "step": 104570
+    },
+    {
+      "epoch": 0.865119741903462,
+      "grad_norm": 874.85302734375,
+      "learning_rate": 7.133465204603895e-07,
+      "loss": 91.6719,
+      "step": 104580
+    },
+    {
+      "epoch": 0.8652024651528312,
+      "grad_norm": 994.0018310546875,
+      "learning_rate": 7.126206819292997e-07,
+      "loss": 79.112,
+      "step": 104590
+    },
+    {
+      "epoch": 0.8652851884022005,
+      "grad_norm": 875.634521484375,
+      "learning_rate": 7.118951845253053e-07,
+      "loss": 75.0624,
+      "step": 104600
+    },
+    {
+      "epoch": 0.8653679116515697,
+      "grad_norm": 905.0895385742188,
+      "learning_rate": 7.111700283061318e-07,
+      "loss": 60.7013,
+      "step": 104610
+    },
+    {
+      "epoch": 0.8654506349009389,
+      "grad_norm": 526.9947509765625,
+      "learning_rate": 7.104452133294759e-07,
+      "loss": 88.9498,
+      "step": 104620
+    },
+    {
+      "epoch": 0.8655333581503082,
+      "grad_norm": 724.978515625,
+      "learning_rate": 7.097207396530081e-07,
+      "loss": 81.7083,
+      "step": 104630
+    },
+    {
+      "epoch": 0.8656160813996774,
+      "grad_norm": 626.593017578125,
+      "learning_rate": 7.089966073343712e-07,
+      "loss": 86.5442,
+      "step": 104640
+    },
+    {
+      "epoch": 0.8656988046490466,
+      "grad_norm": 718.5762329101562,
+      "learning_rate": 7.082728164311814e-07,
+      "loss": 77.1169,
+      "step": 104650
+    },
+    {
+      "epoch": 0.8657815278984159,
+      "grad_norm": 908.7042236328125,
+      "learning_rate": 7.07549367001027e-07,
+      "loss": 97.2458,
+      "step": 104660
+    },
+    {
+      "epoch": 0.8658642511477851,
+      "grad_norm": 863.9970092773438,
+      "learning_rate": 7.068262591014696e-07,
+      "loss": 97.336,
+      "step": 104670
+    },
+    {
+      "epoch": 0.8659469743971543,
+      "grad_norm": 558.906494140625,
+      "learning_rate": 7.06103492790044e-07,
+      "loss": 70.4578,
+      "step": 104680
+    },
+    {
+      "epoch": 0.8660296976465236,
+      "grad_norm": 552.6804809570312,
+      "learning_rate": 7.053810681242573e-07,
+      "loss": 70.8659,
+      "step": 104690
+    },
+    {
+      "epoch": 0.8661124208958928,
+      "grad_norm": 508.5573425292969,
+      "learning_rate": 7.046589851615893e-07,
+      "loss": 64.113,
+      "step": 104700
+    },
+    {
+      "epoch": 0.866195144145262,
+      "grad_norm": 806.9329833984375,
+      "learning_rate": 7.039372439594927e-07,
+      "loss": 76.6169,
+      "step": 104710
+    },
+    {
+      "epoch": 0.8662778673946313,
+      "grad_norm": 811.2073974609375,
+      "learning_rate": 7.032158445753934e-07,
+      "loss": 89.2829,
+      "step": 104720
+    },
+    {
+      "epoch": 0.8663605906440005,
+      "grad_norm": 810.32421875,
+      "learning_rate": 7.024947870666899e-07,
+      "loss": 82.497,
+      "step": 104730
+    },
+    {
+      "epoch": 0.8664433138933697,
+      "grad_norm": 758.9779052734375,
+      "learning_rate": 7.017740714907534e-07,
+      "loss": 82.4294,
+      "step": 104740
+    },
+    {
+      "epoch": 0.866526037142739,
+      "grad_norm": 794.656982421875,
+      "learning_rate": 7.010536979049277e-07,
+      "loss": 96.2241,
+      "step": 104750
+    },
+    {
+      "epoch": 0.8666087603921082,
+      "grad_norm": 494.71063232421875,
+      "learning_rate": 7.003336663665294e-07,
+      "loss": 83.7569,
+      "step": 104760
+    },
+    {
+      "epoch": 0.8666914836414774,
+      "grad_norm": 880.26318359375,
+      "learning_rate": 6.996139769328492e-07,
+      "loss": 84.9818,
+      "step": 104770
+    },
+    {
+      "epoch": 0.8667742068908467,
+      "grad_norm": 766.883544921875,
+      "learning_rate": 6.988946296611482e-07,
+      "loss": 80.0834,
+      "step": 104780
+    },
+    {
+      "epoch": 0.8668569301402159,
+      "grad_norm": 1070.8145751953125,
+      "learning_rate": 6.981756246086623e-07,
+      "loss": 81.7685,
+      "step": 104790
+    },
+    {
+      "epoch": 0.8669396533895851,
+      "grad_norm": 951.6563110351562,
+      "learning_rate": 6.974569618325993e-07,
+      "loss": 87.9841,
+      "step": 104800
+    },
+    {
+      "epoch": 0.8670223766389544,
+      "grad_norm": 1062.390869140625,
+      "learning_rate": 6.967386413901395e-07,
+      "loss": 88.5198,
+      "step": 104810
+    },
+    {
+      "epoch": 0.8671050998883236,
+      "grad_norm": 774.4631958007812,
+      "learning_rate": 6.96020663338437e-07,
+      "loss": 91.2737,
+      "step": 104820
+    },
+    {
+      "epoch": 0.8671878231376928,
+      "grad_norm": 432.4591979980469,
+      "learning_rate": 6.953030277346179e-07,
+      "loss": 77.8616,
+      "step": 104830
+    },
+    {
+      "epoch": 0.8672705463870621,
+      "grad_norm": 691.8195190429688,
+      "learning_rate": 6.945857346357804e-07,
+      "loss": 122.5289,
+      "step": 104840
+    },
+    {
+      "epoch": 0.8673532696364313,
+      "grad_norm": 952.2196044921875,
+      "learning_rate": 6.938687840989972e-07,
+      "loss": 63.169,
+      "step": 104850
+    },
+    {
+      "epoch": 0.8674359928858005,
+      "grad_norm": 0.0,
+      "learning_rate": 6.931521761813126e-07,
+      "loss": 88.435,
+      "step": 104860
+    },
+    {
+      "epoch": 0.8675187161351698,
+      "grad_norm": 733.5023193359375,
+      "learning_rate": 6.924359109397433e-07,
+      "loss": 74.1316,
+      "step": 104870
+    },
+    {
+      "epoch": 0.867601439384539,
+      "grad_norm": 641.714111328125,
+      "learning_rate": 6.917199884312809e-07,
+      "loss": 88.4582,
+      "step": 104880
+    },
+    {
+      "epoch": 0.8676841626339082,
+      "grad_norm": 790.0752563476562,
+      "learning_rate": 6.910044087128848e-07,
+      "loss": 84.1604,
+      "step": 104890
+    },
+    {
+      "epoch": 0.8677668858832776,
+      "grad_norm": 663.8312377929688,
+      "learning_rate": 6.902891718414916e-07,
+      "loss": 88.1632,
+      "step": 104900
+    },
+    {
+      "epoch": 0.8678496091326467,
+      "grad_norm": 591.2030639648438,
+      "learning_rate": 6.895742778740117e-07,
+      "loss": 81.408,
+      "step": 104910
+    },
+    {
+      "epoch": 0.867932332382016,
+      "grad_norm": 624.845458984375,
+      "learning_rate": 6.888597268673236e-07,
+      "loss": 62.2073,
+      "step": 104920
+    },
+    {
+      "epoch": 0.8680150556313851,
+      "grad_norm": 794.0993041992188,
+      "learning_rate": 6.881455188782821e-07,
+      "loss": 77.9982,
+      "step": 104930
+    },
+    {
+      "epoch": 0.8680977788807545,
+      "grad_norm": 666.7411499023438,
+      "learning_rate": 6.874316539637127e-07,
+      "loss": 97.2312,
+      "step": 104940
+    },
+    {
+      "epoch": 0.8681805021301237,
+      "grad_norm": 1049.414306640625,
+      "learning_rate": 6.867181321804145e-07,
+      "loss": 65.9971,
+      "step": 104950
+    },
+    {
+      "epoch": 0.8682632253794929,
+      "grad_norm": 1004.761474609375,
+      "learning_rate": 6.860049535851593e-07,
+      "loss": 101.2232,
+      "step": 104960
+    },
+    {
+      "epoch": 0.8683459486288622,
+      "grad_norm": 1056.7117919921875,
+      "learning_rate": 6.852921182346927e-07,
+      "loss": 89.3601,
+      "step": 104970
+    },
+    {
+      "epoch": 0.8684286718782314,
+      "grad_norm": 801.506103515625,
+      "learning_rate": 6.84579626185729e-07,
+      "loss": 115.8205,
+      "step": 104980
+    },
+    {
+      "epoch": 0.8685113951276006,
+      "grad_norm": 475.1383056640625,
+      "learning_rate": 6.838674774949594e-07,
+      "loss": 71.0523,
+      "step": 104990
+    },
+    {
+      "epoch": 0.8685941183769699,
+      "grad_norm": 2474.7802734375,
+      "learning_rate": 6.831556722190453e-07,
+      "loss": 73.0823,
+      "step": 105000
+    },
+    {
+      "epoch": 0.8686768416263391,
+      "grad_norm": 897.792724609375,
+      "learning_rate": 6.82444210414624e-07,
+      "loss": 100.9971,
+      "step": 105010
+    },
+    {
+      "epoch": 0.8687595648757083,
+      "grad_norm": 886.6505737304688,
+      "learning_rate": 6.817330921383014e-07,
+      "loss": 82.7632,
+      "step": 105020
+    },
+    {
+      "epoch": 0.8688422881250776,
+      "grad_norm": 768.843505859375,
+      "learning_rate": 6.81022317446659e-07,
+      "loss": 80.1767,
+      "step": 105030
+    },
+    {
+      "epoch": 0.8689250113744468,
+      "grad_norm": 497.87469482421875,
+      "learning_rate": 6.803118863962488e-07,
+      "loss": 81.78,
+      "step": 105040
+    },
+    {
+      "epoch": 0.869007734623816,
+      "grad_norm": 491.0223083496094,
+      "learning_rate": 6.796017990435977e-07,
+      "loss": 47.6318,
+      "step": 105050
+    },
+    {
+      "epoch": 0.8690904578731853,
+      "grad_norm": 710.4462890625,
+      "learning_rate": 6.788920554452044e-07,
+      "loss": 94.524,
+      "step": 105060
+    },
+    {
+      "epoch": 0.8691731811225545,
+      "grad_norm": 816.4830932617188,
+      "learning_rate": 6.781826556575377e-07,
+      "loss": 103.3359,
+      "step": 105070
+    },
+    {
+      "epoch": 0.8692559043719237,
+      "grad_norm": 654.7847290039062,
+      "learning_rate": 6.77473599737043e-07,
+      "loss": 90.9469,
+      "step": 105080
+    },
+    {
+      "epoch": 0.869338627621293,
+      "grad_norm": 850.1752319335938,
+      "learning_rate": 6.767648877401361e-07,
+      "loss": 75.9913,
+      "step": 105090
+    },
+    {
+      "epoch": 0.8694213508706622,
+      "grad_norm": 577.8858032226562,
+      "learning_rate": 6.76056519723205e-07,
+      "loss": 108.0124,
+      "step": 105100
+    },
+    {
+      "epoch": 0.8695040741200314,
+      "grad_norm": 1150.85205078125,
+      "learning_rate": 6.753484957426132e-07,
+      "loss": 108.4332,
+      "step": 105110
+    },
+    {
+      "epoch": 0.8695867973694007,
+      "grad_norm": 380.7832336425781,
+      "learning_rate": 6.746408158546947e-07,
+      "loss": 58.2194,
+      "step": 105120
+    },
+    {
+      "epoch": 0.8696695206187699,
+      "grad_norm": 683.9119873046875,
+      "learning_rate": 6.739334801157554e-07,
+      "loss": 94.9094,
+      "step": 105130
+    },
+    {
+      "epoch": 0.8697522438681391,
+      "grad_norm": 695.6708984375,
+      "learning_rate": 6.732264885820761e-07,
+      "loss": 90.9071,
+      "step": 105140
+    },
+    {
+      "epoch": 0.8698349671175084,
+      "grad_norm": 586.7554321289062,
+      "learning_rate": 6.725198413099071e-07,
+      "loss": 68.9317,
+      "step": 105150
+    },
+    {
+      "epoch": 0.8699176903668776,
+      "grad_norm": 791.099609375,
+      "learning_rate": 6.718135383554736e-07,
+      "loss": 86.9691,
+      "step": 105160
+    },
+    {
+      "epoch": 0.8700004136162468,
+      "grad_norm": 785.0557250976562,
+      "learning_rate": 6.711075797749733e-07,
+      "loss": 89.6558,
+      "step": 105170
+    },
+    {
+      "epoch": 0.8700831368656161,
+      "grad_norm": 899.1886596679688,
+      "learning_rate": 6.704019656245764e-07,
+      "loss": 93.6414,
+      "step": 105180
+    },
+    {
+      "epoch": 0.8701658601149853,
+      "grad_norm": 802.5755615234375,
+      "learning_rate": 6.696966959604234e-07,
+      "loss": 108.0222,
+      "step": 105190
+    },
+    {
+      "epoch": 0.8702485833643545,
+      "grad_norm": 1072.0621337890625,
+      "learning_rate": 6.689917708386317e-07,
+      "loss": 87.7261,
+      "step": 105200
+    },
+    {
+      "epoch": 0.8703313066137238,
+      "grad_norm": 519.7999267578125,
+      "learning_rate": 6.682871903152888e-07,
+      "loss": 99.5766,
+      "step": 105210
+    },
+    {
+      "epoch": 0.870414029863093,
+      "grad_norm": 964.565673828125,
+      "learning_rate": 6.675829544464535e-07,
+      "loss": 81.1996,
+      "step": 105220
+    },
+    {
+      "epoch": 0.8704967531124622,
+      "grad_norm": 692.7098388671875,
+      "learning_rate": 6.668790632881611e-07,
+      "loss": 78.7892,
+      "step": 105230
+    },
+    {
+      "epoch": 0.8705794763618315,
+      "grad_norm": 1007.2482299804688,
+      "learning_rate": 6.66175516896414e-07,
+      "loss": 140.3374,
+      "step": 105240
+    },
+    {
+      "epoch": 0.8706621996112007,
+      "grad_norm": 1265.0731201171875,
+      "learning_rate": 6.654723153271913e-07,
+      "loss": 86.9678,
+      "step": 105250
+    },
+    {
+      "epoch": 0.8707449228605699,
+      "grad_norm": 632.7286376953125,
+      "learning_rate": 6.64769458636444e-07,
+      "loss": 93.2092,
+      "step": 105260
+    },
+    {
+      "epoch": 0.8708276461099392,
+      "grad_norm": 836.0169067382812,
+      "learning_rate": 6.640669468800947e-07,
+      "loss": 92.5722,
+      "step": 105270
+    },
+    {
+      "epoch": 0.8709103693593084,
+      "grad_norm": 658.7850952148438,
+      "learning_rate": 6.633647801140391e-07,
+      "loss": 100.487,
+      "step": 105280
+    },
+    {
+      "epoch": 0.8709930926086776,
+      "grad_norm": 587.75146484375,
+      "learning_rate": 6.626629583941447e-07,
+      "loss": 68.9679,
+      "step": 105290
+    },
+    {
+      "epoch": 0.8710758158580469,
+      "grad_norm": 817.0744018554688,
+      "learning_rate": 6.619614817762537e-07,
+      "loss": 73.4202,
+      "step": 105300
+    },
+    {
+      "epoch": 0.8711585391074161,
+      "grad_norm": 1041.8173828125,
+      "learning_rate": 6.612603503161802e-07,
+      "loss": 64.3145,
+      "step": 105310
+    },
+    {
+      "epoch": 0.8712412623567853,
+      "grad_norm": 1404.5458984375,
+      "learning_rate": 6.605595640697071e-07,
+      "loss": 99.786,
+      "step": 105320
+    },
+    {
+      "epoch": 0.8713239856061546,
+      "grad_norm": 771.7532958984375,
+      "learning_rate": 6.598591230925943e-07,
+      "loss": 82.4178,
+      "step": 105330
+    },
+    {
+      "epoch": 0.8714067088555238,
+      "grad_norm": 751.2391357421875,
+      "learning_rate": 6.591590274405723e-07,
+      "loss": 122.4575,
+      "step": 105340
+    },
+    {
+      "epoch": 0.871489432104893,
+      "grad_norm": 911.8988037109375,
+      "learning_rate": 6.584592771693449e-07,
+      "loss": 81.5654,
+      "step": 105350
+    },
+    {
+      "epoch": 0.8715721553542624,
+      "grad_norm": 1637.146484375,
+      "learning_rate": 6.57759872334588e-07,
+      "loss": 97.9183,
+      "step": 105360
+    },
+    {
+      "epoch": 0.8716548786036316,
+      "grad_norm": 329.8784484863281,
+      "learning_rate": 6.570608129919492e-07,
+      "loss": 62.8334,
+      "step": 105370
+    },
+    {
+      "epoch": 0.8717376018530008,
+      "grad_norm": 1938.16455078125,
+      "learning_rate": 6.563620991970509e-07,
+      "loss": 76.7647,
+      "step": 105380
+    },
+    {
+      "epoch": 0.8718203251023701,
+      "grad_norm": 717.9798583984375,
+      "learning_rate": 6.556637310054842e-07,
+      "loss": 82.9026,
+      "step": 105390
+    },
+    {
+      "epoch": 0.8719030483517393,
+      "grad_norm": 1446.15966796875,
+      "learning_rate": 6.54965708472819e-07,
+      "loss": 78.1929,
+      "step": 105400
+    },
+    {
+      "epoch": 0.8719857716011085,
+      "grad_norm": 1189.690673828125,
+      "learning_rate": 6.542680316545902e-07,
+      "loss": 72.0123,
+      "step": 105410
+    },
+    {
+      "epoch": 0.8720684948504778,
+      "grad_norm": 613.6715698242188,
+      "learning_rate": 6.535707006063097e-07,
+      "loss": 61.0173,
+      "step": 105420
+    },
+    {
+      "epoch": 0.872151218099847,
+      "grad_norm": 1527.14990234375,
+      "learning_rate": 6.528737153834613e-07,
+      "loss": 130.8081,
+      "step": 105430
+    },
+    {
+      "epoch": 0.8722339413492162,
+      "grad_norm": 1041.12109375,
+      "learning_rate": 6.521770760415008e-07,
+      "loss": 93.4483,
+      "step": 105440
+    },
+    {
+      "epoch": 0.8723166645985855,
+      "grad_norm": 852.52197265625,
+      "learning_rate": 6.514807826358566e-07,
+      "loss": 90.9408,
+      "step": 105450
+    },
+    {
+      "epoch": 0.8723993878479547,
+      "grad_norm": 647.1378173828125,
+      "learning_rate": 6.507848352219299e-07,
+      "loss": 67.9603,
+      "step": 105460
+    },
+    {
+      "epoch": 0.8724821110973239,
+      "grad_norm": 1192.1534423828125,
+      "learning_rate": 6.500892338550929e-07,
+      "loss": 111.3489,
+      "step": 105470
+    },
+    {
+      "epoch": 0.8725648343466932,
+      "grad_norm": 814.5704345703125,
+      "learning_rate": 6.493939785906928e-07,
+      "loss": 87.0574,
+      "step": 105480
+    },
+    {
+      "epoch": 0.8726475575960624,
+      "grad_norm": 741.8197021484375,
+      "learning_rate": 6.486990694840467e-07,
+      "loss": 68.0563,
+      "step": 105490
+    },
+    {
+      "epoch": 0.8727302808454316,
+      "grad_norm": 948.8922729492188,
+      "learning_rate": 6.480045065904461e-07,
+      "loss": 89.7507,
+      "step": 105500
+    },
+    {
+      "epoch": 0.8728130040948009,
+      "grad_norm": 1155.1693115234375,
+      "learning_rate": 6.47310289965154e-07,
+      "loss": 81.7769,
+      "step": 105510
+    },
+    {
+      "epoch": 0.8728957273441701,
+      "grad_norm": 919.3316040039062,
+      "learning_rate": 6.466164196634056e-07,
+      "loss": 97.3234,
+      "step": 105520
+    },
+    {
+      "epoch": 0.8729784505935393,
+      "grad_norm": 987.2399291992188,
+      "learning_rate": 6.459228957404101e-07,
+      "loss": 97.3726,
+      "step": 105530
+    },
+    {
+      "epoch": 0.8730611738429086,
+      "grad_norm": 838.5607299804688,
+      "learning_rate": 6.452297182513468e-07,
+      "loss": 69.4488,
+      "step": 105540
+    },
+    {
+      "epoch": 0.8731438970922778,
+      "grad_norm": 1150.2685546875,
+      "learning_rate": 6.445368872513691e-07,
+      "loss": 72.3736,
+      "step": 105550
+    },
+    {
+      "epoch": 0.873226620341647,
+      "grad_norm": 687.7527465820312,
+      "learning_rate": 6.438444027956026e-07,
+      "loss": 68.8998,
+      "step": 105560
+    },
+    {
+      "epoch": 0.8733093435910163,
+      "grad_norm": 455.3255920410156,
+      "learning_rate": 6.431522649391447e-07,
+      "loss": 94.7914,
+      "step": 105570
+    },
+    {
+      "epoch": 0.8733920668403855,
+      "grad_norm": 599.0716552734375,
+      "learning_rate": 6.42460473737066e-07,
+      "loss": 77.5691,
+      "step": 105580
+    },
+    {
+      "epoch": 0.8734747900897547,
+      "grad_norm": 982.269775390625,
+      "learning_rate": 6.417690292444084e-07,
+      "loss": 88.5389,
+      "step": 105590
+    },
+    {
+      "epoch": 0.873557513339124,
+      "grad_norm": 304.9704895019531,
+      "learning_rate": 6.410779315161885e-07,
+      "loss": 74.6791,
+      "step": 105600
+    },
+    {
+      "epoch": 0.8736402365884932,
+      "grad_norm": 840.72021484375,
+      "learning_rate": 6.403871806073924e-07,
+      "loss": 66.5112,
+      "step": 105610
+    },
+    {
+      "epoch": 0.8737229598378624,
+      "grad_norm": 1154.091552734375,
+      "learning_rate": 6.396967765729806e-07,
+      "loss": 92.7399,
+      "step": 105620
+    },
+    {
+      "epoch": 0.8738056830872317,
+      "grad_norm": 875.4635009765625,
+      "learning_rate": 6.390067194678851e-07,
+      "loss": 72.3126,
+      "step": 105630
+    },
+    {
+      "epoch": 0.8738884063366009,
+      "grad_norm": 776.973876953125,
+      "learning_rate": 6.383170093470103e-07,
+      "loss": 86.9083,
+      "step": 105640
+    },
+    {
+      "epoch": 0.8739711295859701,
+      "grad_norm": 1360.440673828125,
+      "learning_rate": 6.376276462652342e-07,
+      "loss": 85.9056,
+      "step": 105650
+    },
+    {
+      "epoch": 0.8740538528353393,
+      "grad_norm": 1393.611572265625,
+      "learning_rate": 6.36938630277405e-07,
+      "loss": 97.7015,
+      "step": 105660
+    },
+    {
+      "epoch": 0.8741365760847086,
+      "grad_norm": 762.5498046875,
+      "learning_rate": 6.36249961438346e-07,
+      "loss": 111.3497,
+      "step": 105670
+    },
+    {
+      "epoch": 0.8742192993340778,
+      "grad_norm": 832.800048828125,
+      "learning_rate": 6.355616398028502e-07,
+      "loss": 81.6632,
+      "step": 105680
+    },
+    {
+      "epoch": 0.874302022583447,
+      "grad_norm": 661.8202514648438,
+      "learning_rate": 6.348736654256848e-07,
+      "loss": 86.8958,
+      "step": 105690
+    },
+    {
+      "epoch": 0.8743847458328163,
+      "grad_norm": 882.3600463867188,
+      "learning_rate": 6.341860383615889e-07,
+      "loss": 83.2556,
+      "step": 105700
+    },
+    {
+      "epoch": 0.8744674690821855,
+      "grad_norm": 867.7281494140625,
+      "learning_rate": 6.33498758665273e-07,
+      "loss": 94.6405,
+      "step": 105710
+    },
+    {
+      "epoch": 0.8745501923315547,
+      "grad_norm": 554.966796875,
+      "learning_rate": 6.328118263914218e-07,
+      "loss": 82.4325,
+      "step": 105720
+    },
+    {
+      "epoch": 0.874632915580924,
+      "grad_norm": 895.0468139648438,
+      "learning_rate": 6.321252415946904e-07,
+      "loss": 65.7037,
+      "step": 105730
+    },
+    {
+      "epoch": 0.8747156388302932,
+      "grad_norm": 812.9765014648438,
+      "learning_rate": 6.314390043297092e-07,
+      "loss": 88.4549,
+      "step": 105740
+    },
+    {
+      "epoch": 0.8747983620796624,
+      "grad_norm": 640.9434204101562,
+      "learning_rate": 6.307531146510754e-07,
+      "loss": 86.2904,
+      "step": 105750
+    },
+    {
+      "epoch": 0.8748810853290317,
+      "grad_norm": 553.9396362304688,
+      "learning_rate": 6.300675726133648e-07,
+      "loss": 75.9555,
+      "step": 105760
+    },
+    {
+      "epoch": 0.874963808578401,
+      "grad_norm": 419.89013671875,
+      "learning_rate": 6.293823782711222e-07,
+      "loss": 118.8286,
+      "step": 105770
+    },
+    {
+      "epoch": 0.8750465318277701,
+      "grad_norm": 733.7764892578125,
+      "learning_rate": 6.286975316788657e-07,
+      "loss": 60.0976,
+      "step": 105780
+    },
+    {
+      "epoch": 0.8751292550771395,
+      "grad_norm": 1776.192138671875,
+      "learning_rate": 6.280130328910849e-07,
+      "loss": 106.2316,
+      "step": 105790
+    },
+    {
+      "epoch": 0.8752119783265087,
+      "grad_norm": 659.7675170898438,
+      "learning_rate": 6.27328881962242e-07,
+      "loss": 86.0116,
+      "step": 105800
+    },
+    {
+      "epoch": 0.8752947015758779,
+      "grad_norm": 1190.6595458984375,
+      "learning_rate": 6.266450789467727e-07,
+      "loss": 82.2913,
+      "step": 105810
+    },
+    {
+      "epoch": 0.8753774248252472,
+      "grad_norm": 1189.8983154296875,
+      "learning_rate": 6.259616238990828e-07,
+      "loss": 50.2323,
+      "step": 105820
+    },
+    {
+      "epoch": 0.8754601480746164,
+      "grad_norm": 1355.994873046875,
+      "learning_rate": 6.252785168735537e-07,
+      "loss": 82.7705,
+      "step": 105830
+    },
+    {
+      "epoch": 0.8755428713239856,
+      "grad_norm": 752.8018798828125,
+      "learning_rate": 6.245957579245349e-07,
+      "loss": 67.2084,
+      "step": 105840
+    },
+    {
+      "epoch": 0.8756255945733549,
+      "grad_norm": 1385.89306640625,
+      "learning_rate": 6.239133471063502e-07,
+      "loss": 72.485,
+      "step": 105850
+    },
+    {
+      "epoch": 0.8757083178227241,
+      "grad_norm": 467.3131103515625,
+      "learning_rate": 6.23231284473298e-07,
+      "loss": 92.7437,
+      "step": 105860
+    },
+    {
+      "epoch": 0.8757910410720933,
+      "grad_norm": 937.4329833984375,
+      "learning_rate": 6.225495700796452e-07,
+      "loss": 98.6699,
+      "step": 105870
+    },
+    {
+      "epoch": 0.8758737643214626,
+      "grad_norm": 530.5836791992188,
+      "learning_rate": 6.218682039796343e-07,
+      "loss": 79.1359,
+      "step": 105880
+    },
+    {
+      "epoch": 0.8759564875708318,
+      "grad_norm": 652.8727416992188,
+      "learning_rate": 6.211871862274765e-07,
+      "loss": 91.6557,
+      "step": 105890
+    },
+    {
+      "epoch": 0.876039210820201,
+      "grad_norm": 818.7814331054688,
+      "learning_rate": 6.205065168773589e-07,
+      "loss": 101.5155,
+      "step": 105900
+    },
+    {
+      "epoch": 0.8761219340695703,
+      "grad_norm": 876.2640380859375,
+      "learning_rate": 6.198261959834384e-07,
+      "loss": 103.2004,
+      "step": 105910
+    },
+    {
+      "epoch": 0.8762046573189395,
+      "grad_norm": 516.7633056640625,
+      "learning_rate": 6.191462235998463e-07,
+      "loss": 77.5895,
+      "step": 105920
+    },
+    {
+      "epoch": 0.8762873805683087,
+      "grad_norm": 646.3327026367188,
+      "learning_rate": 6.184665997806832e-07,
+      "loss": 56.7936,
+      "step": 105930
+    },
+    {
+      "epoch": 0.876370103817678,
+      "grad_norm": 576.1641235351562,
+      "learning_rate": 6.177873245800237e-07,
+      "loss": 70.1931,
+      "step": 105940
+    },
+    {
+      "epoch": 0.8764528270670472,
+      "grad_norm": 1594.127685546875,
+      "learning_rate": 6.171083980519138e-07,
+      "loss": 94.1285,
+      "step": 105950
+    },
+    {
+      "epoch": 0.8765355503164164,
+      "grad_norm": 1294.6268310546875,
+      "learning_rate": 6.164298202503754e-07,
+      "loss": 101.6075,
+      "step": 105960
+    },
+    {
+      "epoch": 0.8766182735657857,
+      "grad_norm": 540.323974609375,
+      "learning_rate": 6.157515912293982e-07,
+      "loss": 87.1123,
+      "step": 105970
+    },
+    {
+      "epoch": 0.8767009968151549,
+      "grad_norm": 570.9243774414062,
+      "learning_rate": 6.150737110429461e-07,
+      "loss": 74.8245,
+      "step": 105980
+    },
+    {
+      "epoch": 0.8767837200645241,
+      "grad_norm": 906.1620483398438,
+      "learning_rate": 6.143961797449549e-07,
+      "loss": 71.5885,
+      "step": 105990
+    },
+    {
+      "epoch": 0.8768664433138934,
+      "grad_norm": 878.2374267578125,
+      "learning_rate": 6.137189973893331e-07,
+      "loss": 73.5436,
+      "step": 106000
+    },
+    {
+      "epoch": 0.8769491665632626,
+      "grad_norm": 873.0944213867188,
+      "learning_rate": 6.130421640299594e-07,
+      "loss": 61.5041,
+      "step": 106010
+    },
+    {
+      "epoch": 0.8770318898126318,
+      "grad_norm": 734.4179077148438,
+      "learning_rate": 6.123656797206873e-07,
+      "loss": 85.497,
+      "step": 106020
+    },
+    {
+      "epoch": 0.8771146130620011,
+      "grad_norm": 1206.48193359375,
+      "learning_rate": 6.116895445153415e-07,
+      "loss": 61.9635,
+      "step": 106030
+    },
+    {
+      "epoch": 0.8771973363113703,
+      "grad_norm": 976.5135498046875,
+      "learning_rate": 6.11013758467719e-07,
+      "loss": 101.1229,
+      "step": 106040
+    },
+    {
+      "epoch": 0.8772800595607395,
+      "grad_norm": 1087.355224609375,
+      "learning_rate": 6.103383216315883e-07,
+      "loss": 75.2539,
+      "step": 106050
+    },
+    {
+      "epoch": 0.8773627828101088,
+      "grad_norm": 674.3126220703125,
+      "learning_rate": 6.096632340606922e-07,
+      "loss": 77.9221,
+      "step": 106060
+    },
+    {
+      "epoch": 0.877445506059478,
+      "grad_norm": 753.2930297851562,
+      "learning_rate": 6.089884958087439e-07,
+      "loss": 87.5172,
+      "step": 106070
+    },
+    {
+      "epoch": 0.8775282293088472,
+      "grad_norm": 640.8326416015625,
+      "learning_rate": 6.083141069294285e-07,
+      "loss": 74.8427,
+      "step": 106080
+    },
+    {
+      "epoch": 0.8776109525582165,
+      "grad_norm": 772.4714965820312,
+      "learning_rate": 6.07640067476406e-07,
+      "loss": 132.3394,
+      "step": 106090
+    },
+    {
+      "epoch": 0.8776936758075857,
+      "grad_norm": 566.04248046875,
+      "learning_rate": 6.069663775033041e-07,
+      "loss": 65.6599,
+      "step": 106100
+    },
+    {
+      "epoch": 0.8777763990569549,
+      "grad_norm": 780.9393920898438,
+      "learning_rate": 6.06293037063726e-07,
+      "loss": 80.8205,
+      "step": 106110
+    },
+    {
+      "epoch": 0.8778591223063242,
+      "grad_norm": 654.2213745117188,
+      "learning_rate": 6.056200462112466e-07,
+      "loss": 98.3166,
+      "step": 106120
+    },
+    {
+      "epoch": 0.8779418455556934,
+      "grad_norm": 1010.9708862304688,
+      "learning_rate": 6.049474049994125e-07,
+      "loss": 77.5668,
+      "step": 106130
+    },
+    {
+      "epoch": 0.8780245688050626,
+      "grad_norm": 871.8776245117188,
+      "learning_rate": 6.042751134817431e-07,
+      "loss": 87.4441,
+      "step": 106140
+    },
+    {
+      "epoch": 0.8781072920544319,
+      "grad_norm": 801.5313110351562,
+      "learning_rate": 6.03603171711728e-07,
+      "loss": 70.5738,
+      "step": 106150
+    },
+    {
+      "epoch": 0.8781900153038011,
+      "grad_norm": 602.4784545898438,
+      "learning_rate": 6.029315797428331e-07,
+      "loss": 87.0834,
+      "step": 106160
+    },
+    {
+      "epoch": 0.8782727385531703,
+      "grad_norm": 798.4912109375,
+      "learning_rate": 6.02260337628493e-07,
+      "loss": 63.067,
+      "step": 106170
+    },
+    {
+      "epoch": 0.8783554618025396,
+      "grad_norm": 1035.9962158203125,
+      "learning_rate": 6.015894454221143e-07,
+      "loss": 102.9226,
+      "step": 106180
+    },
+    {
+      "epoch": 0.8784381850519088,
+      "grad_norm": 891.0628051757812,
+      "learning_rate": 6.00918903177078e-07,
+      "loss": 80.5977,
+      "step": 106190
+    },
+    {
+      "epoch": 0.878520908301278,
+      "grad_norm": 597.5098876953125,
+      "learning_rate": 6.002487109467347e-07,
+      "loss": 72.3878,
+      "step": 106200
+    },
+    {
+      "epoch": 0.8786036315506474,
+      "grad_norm": 920.417236328125,
+      "learning_rate": 5.995788687844095e-07,
+      "loss": 87.8151,
+      "step": 106210
+    },
+    {
+      "epoch": 0.8786863548000166,
+      "grad_norm": 683.3025512695312,
+      "learning_rate": 5.98909376743399e-07,
+      "loss": 95.9658,
+      "step": 106220
+    },
+    {
+      "epoch": 0.8787690780493858,
+      "grad_norm": 583.1339111328125,
+      "learning_rate": 5.982402348769706e-07,
+      "loss": 85.8737,
+      "step": 106230
+    },
+    {
+      "epoch": 0.8788518012987551,
+      "grad_norm": 894.7290649414062,
+      "learning_rate": 5.975714432383645e-07,
+      "loss": 148.6901,
+      "step": 106240
+    },
+    {
+      "epoch": 0.8789345245481243,
+      "grad_norm": 651.498291015625,
+      "learning_rate": 5.969030018807953e-07,
+      "loss": 69.2491,
+      "step": 106250
+    },
+    {
+      "epoch": 0.8790172477974935,
+      "grad_norm": 544.6094360351562,
+      "learning_rate": 5.962349108574478e-07,
+      "loss": 72.5222,
+      "step": 106260
+    },
+    {
+      "epoch": 0.8790999710468628,
+      "grad_norm": 1018.2208251953125,
+      "learning_rate": 5.955671702214765e-07,
+      "loss": 109.5223,
+      "step": 106270
+    },
+    {
+      "epoch": 0.879182694296232,
+      "grad_norm": 892.130615234375,
+      "learning_rate": 5.948997800260125e-07,
+      "loss": 109.8219,
+      "step": 106280
+    },
+    {
+      "epoch": 0.8792654175456012,
+      "grad_norm": 1923.7149658203125,
+      "learning_rate": 5.94232740324156e-07,
+      "loss": 79.6006,
+      "step": 106290
+    },
+    {
+      "epoch": 0.8793481407949705,
+      "grad_norm": 582.2818603515625,
+      "learning_rate": 5.935660511689805e-07,
+      "loss": 97.3011,
+      "step": 106300
+    },
+    {
+      "epoch": 0.8794308640443397,
+      "grad_norm": 1395.069580078125,
+      "learning_rate": 5.928997126135317e-07,
+      "loss": 101.4692,
+      "step": 106310
+    },
+    {
+      "epoch": 0.8795135872937089,
+      "grad_norm": 628.81298828125,
+      "learning_rate": 5.922337247108267e-07,
+      "loss": 88.9791,
+      "step": 106320
+    },
+    {
+      "epoch": 0.8795963105430782,
+      "grad_norm": 1247.3292236328125,
+      "learning_rate": 5.915680875138558e-07,
+      "loss": 81.8968,
+      "step": 106330
+    },
+    {
+      "epoch": 0.8796790337924474,
+      "grad_norm": 532.4841918945312,
+      "learning_rate": 5.909028010755786e-07,
+      "loss": 105.6303,
+      "step": 106340
+    },
+    {
+      "epoch": 0.8797617570418166,
+      "grad_norm": 748.1610717773438,
+      "learning_rate": 5.902378654489327e-07,
+      "loss": 93.8092,
+      "step": 106350
+    },
+    {
+      "epoch": 0.8798444802911859,
+      "grad_norm": 1042.5313720703125,
+      "learning_rate": 5.89573280686821e-07,
+      "loss": 82.0542,
+      "step": 106360
+    },
+    {
+      "epoch": 0.8799272035405551,
+      "grad_norm": 775.1285400390625,
+      "learning_rate": 5.889090468421216e-07,
+      "loss": 109.767,
+      "step": 106370
+    },
+    {
+      "epoch": 0.8800099267899243,
+      "grad_norm": 807.12744140625,
+      "learning_rate": 5.882451639676856e-07,
+      "loss": 85.814,
+      "step": 106380
+    },
+    {
+      "epoch": 0.8800926500392935,
+      "grad_norm": 712.1463012695312,
+      "learning_rate": 5.875816321163346e-07,
+      "loss": 80.7074,
+      "step": 106390
+    },
+    {
+      "epoch": 0.8801753732886628,
+      "grad_norm": 937.6873168945312,
+      "learning_rate": 5.869184513408633e-07,
+      "loss": 99.54,
+      "step": 106400
+    },
+    {
+      "epoch": 0.880258096538032,
+      "grad_norm": 647.0883178710938,
+      "learning_rate": 5.862556216940368e-07,
+      "loss": 66.8527,
+      "step": 106410
+    },
+    {
+      "epoch": 0.8803408197874012,
+      "grad_norm": 748.40673828125,
+      "learning_rate": 5.85593143228595e-07,
+      "loss": 90.8509,
+      "step": 106420
+    },
+    {
+      "epoch": 0.8804235430367705,
+      "grad_norm": 1141.7353515625,
+      "learning_rate": 5.849310159972466e-07,
+      "loss": 66.829,
+      "step": 106430
+    },
+    {
+      "epoch": 0.8805062662861397,
+      "grad_norm": 880.968994140625,
+      "learning_rate": 5.84269240052675e-07,
+      "loss": 92.991,
+      "step": 106440
+    },
+    {
+      "epoch": 0.8805889895355089,
+      "grad_norm": 803.4313354492188,
+      "learning_rate": 5.836078154475349e-07,
+      "loss": 126.6737,
+      "step": 106450
+    },
+    {
+      "epoch": 0.8806717127848782,
+      "grad_norm": 478.593994140625,
+      "learning_rate": 5.82946742234452e-07,
+      "loss": 79.7044,
+      "step": 106460
+    },
+    {
+      "epoch": 0.8807544360342474,
+      "grad_norm": 715.75146484375,
+      "learning_rate": 5.822860204660253e-07,
+      "loss": 70.8862,
+      "step": 106470
+    },
+    {
+      "epoch": 0.8808371592836166,
+      "grad_norm": 524.4168090820312,
+      "learning_rate": 5.816256501948259e-07,
+      "loss": 90.5394,
+      "step": 106480
+    },
+    {
+      "epoch": 0.8809198825329859,
+      "grad_norm": 752.4254150390625,
+      "learning_rate": 5.809656314733953e-07,
+      "loss": 86.0278,
+      "step": 106490
+    },
+    {
+      "epoch": 0.8810026057823551,
+      "grad_norm": 623.67529296875,
+      "learning_rate": 5.803059643542491e-07,
+      "loss": 67.2863,
+      "step": 106500
+    },
+    {
+      "epoch": 0.8810853290317243,
+      "grad_norm": 909.5820922851562,
+      "learning_rate": 5.796466488898733e-07,
+      "loss": 94.9997,
+      "step": 106510
+    },
+    {
+      "epoch": 0.8811680522810936,
+      "grad_norm": 1068.1527099609375,
+      "learning_rate": 5.789876851327275e-07,
+      "loss": 110.8501,
+      "step": 106520
+    },
+    {
+      "epoch": 0.8812507755304628,
+      "grad_norm": 2067.63671875,
+      "learning_rate": 5.783290731352415e-07,
+      "loss": 74.2425,
+      "step": 106530
+    },
+    {
+      "epoch": 0.881333498779832,
+      "grad_norm": 889.1475830078125,
+      "learning_rate": 5.776708129498188e-07,
+      "loss": 105.7229,
+      "step": 106540
+    },
+    {
+      "epoch": 0.8814162220292013,
+      "grad_norm": 828.7091674804688,
+      "learning_rate": 5.770129046288331e-07,
+      "loss": 111.1305,
+      "step": 106550
+    },
+    {
+      "epoch": 0.8814989452785705,
+      "grad_norm": 997.931640625,
+      "learning_rate": 5.76355348224632e-07,
+      "loss": 110.2826,
+      "step": 106560
+    },
+    {
+      "epoch": 0.8815816685279397,
+      "grad_norm": 1100.512451171875,
+      "learning_rate": 5.756981437895342e-07,
+      "loss": 103.0764,
+      "step": 106570
+    },
+    {
+      "epoch": 0.881664391777309,
+      "grad_norm": 890.1248779296875,
+      "learning_rate": 5.750412913758307e-07,
+      "loss": 87.1903,
+      "step": 106580
+    },
+    {
+      "epoch": 0.8817471150266782,
+      "grad_norm": 1222.4423828125,
+      "learning_rate": 5.743847910357836e-07,
+      "loss": 80.1742,
+      "step": 106590
+    },
+    {
+      "epoch": 0.8818298382760474,
+      "grad_norm": 637.3331298828125,
+      "learning_rate": 5.737286428216288e-07,
+      "loss": 63.0052,
+      "step": 106600
+    },
+    {
+      "epoch": 0.8819125615254167,
+      "grad_norm": 919.8414306640625,
+      "learning_rate": 5.730728467855695e-07,
+      "loss": 85.3765,
+      "step": 106610
+    },
+    {
+      "epoch": 0.881995284774786,
+      "grad_norm": 661.7064208984375,
+      "learning_rate": 5.724174029797886e-07,
+      "loss": 64.6699,
+      "step": 106620
+    },
+    {
+      "epoch": 0.8820780080241551,
+      "grad_norm": 944.8833618164062,
+      "learning_rate": 5.717623114564347e-07,
+      "loss": 80.0257,
+      "step": 106630
+    },
+    {
+      "epoch": 0.8821607312735245,
+      "grad_norm": 992.529296875,
+      "learning_rate": 5.711075722676312e-07,
+      "loss": 102.9055,
+      "step": 106640
+    },
+    {
+      "epoch": 0.8822434545228937,
+      "grad_norm": 691.0933227539062,
+      "learning_rate": 5.704531854654721e-07,
+      "loss": 75.375,
+      "step": 106650
+    },
+    {
+      "epoch": 0.8823261777722629,
+      "grad_norm": 519.4947509765625,
+      "learning_rate": 5.69799151102024e-07,
+      "loss": 76.9458,
+      "step": 106660
+    },
+    {
+      "epoch": 0.8824089010216322,
+      "grad_norm": 3443.64794921875,
+      "learning_rate": 5.691454692293258e-07,
+      "loss": 99.6053,
+      "step": 106670
+    },
+    {
+      "epoch": 0.8824916242710014,
+      "grad_norm": 666.6217041015625,
+      "learning_rate": 5.684921398993875e-07,
+      "loss": 58.0222,
+      "step": 106680
+    },
+    {
+      "epoch": 0.8825743475203706,
+      "grad_norm": 765.2488403320312,
+      "learning_rate": 5.678391631641933e-07,
+      "loss": 69.1236,
+      "step": 106690
+    },
+    {
+      "epoch": 0.8826570707697399,
+      "grad_norm": 797.377197265625,
+      "learning_rate": 5.671865390756948e-07,
+      "loss": 73.226,
+      "step": 106700
+    },
+    {
+      "epoch": 0.8827397940191091,
+      "grad_norm": 577.6158447265625,
+      "learning_rate": 5.665342676858182e-07,
+      "loss": 68.9044,
+      "step": 106710
+    },
+    {
+      "epoch": 0.8828225172684783,
+      "grad_norm": 503.404541015625,
+      "learning_rate": 5.658823490464638e-07,
+      "loss": 60.142,
+      "step": 106720
+    },
+    {
+      "epoch": 0.8829052405178476,
+      "grad_norm": 959.497314453125,
+      "learning_rate": 5.652307832095016e-07,
+      "loss": 98.6422,
+      "step": 106730
+    },
+    {
+      "epoch": 0.8829879637672168,
+      "grad_norm": 983.735595703125,
+      "learning_rate": 5.645795702267731e-07,
+      "loss": 76.1658,
+      "step": 106740
+    },
+    {
+      "epoch": 0.883070687016586,
+      "grad_norm": 817.6513671875,
+      "learning_rate": 5.639287101500923e-07,
+      "loss": 102.0763,
+      "step": 106750
+    },
+    {
+      "epoch": 0.8831534102659553,
+      "grad_norm": 574.2067260742188,
+      "learning_rate": 5.63278203031245e-07,
+      "loss": 80.8163,
+      "step": 106760
+    },
+    {
+      "epoch": 0.8832361335153245,
+      "grad_norm": 823.816650390625,
+      "learning_rate": 5.626280489219893e-07,
+      "loss": 62.7599,
+      "step": 106770
+    },
+    {
+      "epoch": 0.8833188567646937,
+      "grad_norm": 1204.035888671875,
+      "learning_rate": 5.619782478740565e-07,
+      "loss": 77.9808,
+      "step": 106780
+    },
+    {
+      "epoch": 0.883401580014063,
+      "grad_norm": 1444.1622314453125,
+      "learning_rate": 5.613287999391453e-07,
+      "loss": 104.0911,
+      "step": 106790
+    },
+    {
+      "epoch": 0.8834843032634322,
+      "grad_norm": 976.5982666015625,
+      "learning_rate": 5.606797051689294e-07,
+      "loss": 66.8928,
+      "step": 106800
+    },
+    {
+      "epoch": 0.8835670265128014,
+      "grad_norm": 785.950927734375,
+      "learning_rate": 5.600309636150575e-07,
+      "loss": 96.6358,
+      "step": 106810
+    },
+    {
+      "epoch": 0.8836497497621707,
+      "grad_norm": 736.114990234375,
+      "learning_rate": 5.593825753291443e-07,
+      "loss": 73.0294,
+      "step": 106820
+    },
+    {
+      "epoch": 0.8837324730115399,
+      "grad_norm": 567.788330078125,
+      "learning_rate": 5.587345403627803e-07,
+      "loss": 88.5388,
+      "step": 106830
+    },
+    {
+      "epoch": 0.8838151962609091,
+      "grad_norm": 1305.841064453125,
+      "learning_rate": 5.580868587675265e-07,
+      "loss": 75.9461,
+      "step": 106840
+    },
+    {
+      "epoch": 0.8838979195102784,
+      "grad_norm": 732.5675048828125,
+      "learning_rate": 5.574395305949148e-07,
+      "loss": 97.3696,
+      "step": 106850
+    },
+    {
+      "epoch": 0.8839806427596476,
+      "grad_norm": 1100.3834228515625,
+      "learning_rate": 5.567925558964532e-07,
+      "loss": 80.1282,
+      "step": 106860
+    },
+    {
+      "epoch": 0.8840633660090168,
+      "grad_norm": 1010.8685913085938,
+      "learning_rate": 5.561459347236148e-07,
+      "loss": 90.0034,
+      "step": 106870
+    },
+    {
+      "epoch": 0.8841460892583861,
+      "grad_norm": 371.01641845703125,
+      "learning_rate": 5.554996671278495e-07,
+      "loss": 85.3001,
+      "step": 106880
+    },
+    {
+      "epoch": 0.8842288125077553,
+      "grad_norm": 580.9638671875,
+      "learning_rate": 5.548537531605785e-07,
+      "loss": 85.6992,
+      "step": 106890
+    },
+    {
+      "epoch": 0.8843115357571245,
+      "grad_norm": 764.029052734375,
+      "learning_rate": 5.542081928731929e-07,
+      "loss": 57.3023,
+      "step": 106900
+    },
+    {
+      "epoch": 0.8843942590064938,
+      "grad_norm": 610.1265258789062,
+      "learning_rate": 5.535629863170594e-07,
+      "loss": 80.0151,
+      "step": 106910
+    },
+    {
+      "epoch": 0.884476982255863,
+      "grad_norm": 615.563720703125,
+      "learning_rate": 5.529181335435124e-07,
+      "loss": 81.8557,
+      "step": 106920
+    },
+    {
+      "epoch": 0.8845597055052322,
+      "grad_norm": 1012.834716796875,
+      "learning_rate": 5.522736346038598e-07,
+      "loss": 95.8173,
+      "step": 106930
+    },
+    {
+      "epoch": 0.8846424287546015,
+      "grad_norm": 1021.8489990234375,
+      "learning_rate": 5.516294895493824e-07,
+      "loss": 78.0766,
+      "step": 106940
+    },
+    {
+      "epoch": 0.8847251520039707,
+      "grad_norm": 827.3118286132812,
+      "learning_rate": 5.509856984313316e-07,
+      "loss": 80.5865,
+      "step": 106950
+    },
+    {
+      "epoch": 0.8848078752533399,
+      "grad_norm": 990.0963745117188,
+      "learning_rate": 5.503422613009296e-07,
+      "loss": 69.4347,
+      "step": 106960
+    },
+    {
+      "epoch": 0.8848905985027092,
+      "grad_norm": 593.1627197265625,
+      "learning_rate": 5.49699178209373e-07,
+      "loss": 66.5936,
+      "step": 106970
+    },
+    {
+      "epoch": 0.8849733217520784,
+      "grad_norm": 835.3819580078125,
+      "learning_rate": 5.490564492078287e-07,
+      "loss": 93.1555,
+      "step": 106980
+    },
+    {
+      "epoch": 0.8850560450014476,
+      "grad_norm": 934.9071044921875,
+      "learning_rate": 5.484140743474359e-07,
+      "loss": 92.1782,
+      "step": 106990
+    },
+    {
+      "epoch": 0.8851387682508169,
+      "grad_norm": 756.7869873046875,
+      "learning_rate": 5.477720536793035e-07,
+      "loss": 87.1874,
+      "step": 107000
+    },
+    {
+      "epoch": 0.8852214915001861,
+      "grad_norm": 781.40771484375,
+      "learning_rate": 5.471303872545175e-07,
+      "loss": 97.5913,
+      "step": 107010
+    },
+    {
+      "epoch": 0.8853042147495553,
+      "grad_norm": 991.211181640625,
+      "learning_rate": 5.464890751241303e-07,
+      "loss": 74.9666,
+      "step": 107020
+    },
+    {
+      "epoch": 0.8853869379989247,
+      "grad_norm": 1911.62109375,
+      "learning_rate": 5.458481173391694e-07,
+      "loss": 87.3059,
+      "step": 107030
+    },
+    {
+      "epoch": 0.8854696612482938,
+      "grad_norm": 1125.0557861328125,
+      "learning_rate": 5.452075139506314e-07,
+      "loss": 97.0107,
+      "step": 107040
+    },
+    {
+      "epoch": 0.885552384497663,
+      "grad_norm": 1039.9718017578125,
+      "learning_rate": 5.445672650094863e-07,
+      "loss": 114.9293,
+      "step": 107050
+    },
+    {
+      "epoch": 0.8856351077470324,
+      "grad_norm": 1257.056396484375,
+      "learning_rate": 5.43927370566677e-07,
+      "loss": 100.9882,
+      "step": 107060
+    },
+    {
+      "epoch": 0.8857178309964016,
+      "grad_norm": 506.9411926269531,
+      "learning_rate": 5.432878306731154e-07,
+      "loss": 72.05,
+      "step": 107070
+    },
+    {
+      "epoch": 0.8858005542457708,
+      "grad_norm": 1244.661376953125,
+      "learning_rate": 5.426486453796881e-07,
+      "loss": 103.0482,
+      "step": 107080
+    },
+    {
+      "epoch": 0.8858832774951401,
+      "grad_norm": 558.5795288085938,
+      "learning_rate": 5.420098147372515e-07,
+      "loss": 87.9579,
+      "step": 107090
+    },
+    {
+      "epoch": 0.8859660007445093,
+      "grad_norm": 582.4910278320312,
+      "learning_rate": 5.413713387966329e-07,
+      "loss": 90.7297,
+      "step": 107100
+    },
+    {
+      "epoch": 0.8860487239938785,
+      "grad_norm": 797.0523681640625,
+      "learning_rate": 5.407332176086367e-07,
+      "loss": 72.4927,
+      "step": 107110
+    },
+    {
+      "epoch": 0.8861314472432477,
+      "grad_norm": 889.9099731445312,
+      "learning_rate": 5.400954512240331e-07,
+      "loss": 59.4978,
+      "step": 107120
+    },
+    {
+      "epoch": 0.886214170492617,
+      "grad_norm": 705.6456909179688,
+      "learning_rate": 5.394580396935656e-07,
+      "loss": 71.2413,
+      "step": 107130
+    },
+    {
+      "epoch": 0.8862968937419862,
+      "grad_norm": 1448.95751953125,
+      "learning_rate": 5.388209830679508e-07,
+      "loss": 86.1324,
+      "step": 107140
+    },
+    {
+      "epoch": 0.8863796169913554,
+      "grad_norm": 619.7808227539062,
+      "learning_rate": 5.381842813978766e-07,
+      "loss": 88.696,
+      "step": 107150
+    },
+    {
+      "epoch": 0.8864623402407247,
+      "grad_norm": 811.8013916015625,
+      "learning_rate": 5.375479347340018e-07,
+      "loss": 69.7523,
+      "step": 107160
+    },
+    {
+      "epoch": 0.8865450634900939,
+      "grad_norm": 882.953369140625,
+      "learning_rate": 5.369119431269582e-07,
+      "loss": 73.939,
+      "step": 107170
+    },
+    {
+      "epoch": 0.8866277867394631,
+      "grad_norm": 563.3268432617188,
+      "learning_rate": 5.362763066273479e-07,
+      "loss": 89.5063,
+      "step": 107180
+    },
+    {
+      "epoch": 0.8867105099888324,
+      "grad_norm": 741.295166015625,
+      "learning_rate": 5.356410252857458e-07,
+      "loss": 136.3258,
+      "step": 107190
+    },
+    {
+      "epoch": 0.8867932332382016,
+      "grad_norm": 918.1881103515625,
+      "learning_rate": 5.350060991526996e-07,
+      "loss": 87.6655,
+      "step": 107200
+    },
+    {
+      "epoch": 0.8868759564875708,
+      "grad_norm": 571.861572265625,
+      "learning_rate": 5.343715282787271e-07,
+      "loss": 80.5383,
+      "step": 107210
+    },
+    {
+      "epoch": 0.8869586797369401,
+      "grad_norm": 1086.863037109375,
+      "learning_rate": 5.33737312714317e-07,
+      "loss": 95.8323,
+      "step": 107220
+    },
+    {
+      "epoch": 0.8870414029863093,
+      "grad_norm": 937.6774291992188,
+      "learning_rate": 5.33103452509931e-07,
+      "loss": 79.3969,
+      "step": 107230
+    },
+    {
+      "epoch": 0.8871241262356785,
+      "grad_norm": 990.2754516601562,
+      "learning_rate": 5.32469947716004e-07,
+      "loss": 83.5339,
+      "step": 107240
+    },
+    {
+      "epoch": 0.8872068494850478,
+      "grad_norm": 578.4818115234375,
+      "learning_rate": 5.318367983829393e-07,
+      "loss": 88.6761,
+      "step": 107250
+    },
+    {
+      "epoch": 0.887289572734417,
+      "grad_norm": 930.7289428710938,
+      "learning_rate": 5.312040045611144e-07,
+      "loss": 108.0735,
+      "step": 107260
+    },
+    {
+      "epoch": 0.8873722959837862,
+      "grad_norm": 776.857666015625,
+      "learning_rate": 5.305715663008781e-07,
+      "loss": 90.2431,
+      "step": 107270
+    },
+    {
+      "epoch": 0.8874550192331555,
+      "grad_norm": 1201.76171875,
+      "learning_rate": 5.299394836525507e-07,
+      "loss": 78.6439,
+      "step": 107280
+    },
+    {
+      "epoch": 0.8875377424825247,
+      "grad_norm": 609.2925415039062,
+      "learning_rate": 5.293077566664234e-07,
+      "loss": 98.0536,
+      "step": 107290
+    },
+    {
+      "epoch": 0.8876204657318939,
+      "grad_norm": 1422.3968505859375,
+      "learning_rate": 5.286763853927601e-07,
+      "loss": 97.6561,
+      "step": 107300
+    },
+    {
+      "epoch": 0.8877031889812632,
+      "grad_norm": 1007.32568359375,
+      "learning_rate": 5.280453698817961e-07,
+      "loss": 73.6183,
+      "step": 107310
+    },
+    {
+      "epoch": 0.8877859122306324,
+      "grad_norm": 1175.0106201171875,
+      "learning_rate": 5.27414710183739e-07,
+      "loss": 96.213,
+      "step": 107320
+    },
+    {
+      "epoch": 0.8878686354800016,
+      "grad_norm": 509.9411926269531,
+      "learning_rate": 5.26784406348767e-07,
+      "loss": 113.796,
+      "step": 107330
+    },
+    {
+      "epoch": 0.8879513587293709,
+      "grad_norm": 646.4354858398438,
+      "learning_rate": 5.261544584270301e-07,
+      "loss": 102.4905,
+      "step": 107340
+    },
+    {
+      "epoch": 0.8880340819787401,
+      "grad_norm": 795.3406372070312,
+      "learning_rate": 5.255248664686507e-07,
+      "loss": 91.493,
+      "step": 107350
+    },
+    {
+      "epoch": 0.8881168052281093,
+      "grad_norm": 746.2446899414062,
+      "learning_rate": 5.24895630523723e-07,
+      "loss": 75.7493,
+      "step": 107360
+    },
+    {
+      "epoch": 0.8881995284774786,
+      "grad_norm": 759.6536865234375,
+      "learning_rate": 5.242667506423122e-07,
+      "loss": 78.0097,
+      "step": 107370
+    },
+    {
+      "epoch": 0.8882822517268478,
+      "grad_norm": 706.584228515625,
+      "learning_rate": 5.236382268744544e-07,
+      "loss": 65.1131,
+      "step": 107380
+    },
+    {
+      "epoch": 0.888364974976217,
+      "grad_norm": 904.0101928710938,
+      "learning_rate": 5.230100592701598e-07,
+      "loss": 94.9032,
+      "step": 107390
+    },
+    {
+      "epoch": 0.8884476982255863,
+      "grad_norm": 1153.80908203125,
+      "learning_rate": 5.223822478794083e-07,
+      "loss": 92.7698,
+      "step": 107400
+    },
+    {
+      "epoch": 0.8885304214749555,
+      "grad_norm": 827.2999877929688,
+      "learning_rate": 5.217547927521515e-07,
+      "loss": 87.4861,
+      "step": 107410
+    },
+    {
+      "epoch": 0.8886131447243247,
+      "grad_norm": 802.9961547851562,
+      "learning_rate": 5.211276939383136e-07,
+      "loss": 74.549,
+      "step": 107420
+    },
+    {
+      "epoch": 0.888695867973694,
+      "grad_norm": 1088.9659423828125,
+      "learning_rate": 5.205009514877895e-07,
+      "loss": 94.3598,
+      "step": 107430
+    },
+    {
+      "epoch": 0.8887785912230632,
+      "grad_norm": 792.8547973632812,
+      "learning_rate": 5.198745654504472e-07,
+      "loss": 89.176,
+      "step": 107440
+    },
+    {
+      "epoch": 0.8888613144724324,
+      "grad_norm": 344.9277038574219,
+      "learning_rate": 5.19248535876124e-07,
+      "loss": 91.6492,
+      "step": 107450
+    },
+    {
+      "epoch": 0.8889440377218017,
+      "grad_norm": 0.0,
+      "learning_rate": 5.186228628146317e-07,
+      "loss": 69.1638,
+      "step": 107460
+    },
+    {
+      "epoch": 0.889026760971171,
+      "grad_norm": 1069.0447998046875,
+      "learning_rate": 5.179975463157511e-07,
+      "loss": 114.1318,
+      "step": 107470
+    },
+    {
+      "epoch": 0.8891094842205401,
+      "grad_norm": 491.0845642089844,
+      "learning_rate": 5.173725864292356e-07,
+      "loss": 81.6012,
+      "step": 107480
+    },
+    {
+      "epoch": 0.8891922074699095,
+      "grad_norm": 585.45654296875,
+      "learning_rate": 5.167479832048117e-07,
+      "loss": 101.9051,
+      "step": 107490
+    },
+    {
+      "epoch": 0.8892749307192787,
+      "grad_norm": 1258.58984375,
+      "learning_rate": 5.16123736692175e-07,
+      "loss": 72.9357,
+      "step": 107500
+    },
+    {
+      "epoch": 0.8893576539686479,
+      "grad_norm": 1067.1220703125,
+      "learning_rate": 5.154998469409945e-07,
+      "loss": 106.5081,
+      "step": 107510
+    },
+    {
+      "epoch": 0.8894403772180172,
+      "grad_norm": 908.0804443359375,
+      "learning_rate": 5.1487631400091e-07,
+      "loss": 80.5405,
+      "step": 107520
+    },
+    {
+      "epoch": 0.8895231004673864,
+      "grad_norm": 528.5298461914062,
+      "learning_rate": 5.142531379215338e-07,
+      "loss": 68.822,
+      "step": 107530
+    },
+    {
+      "epoch": 0.8896058237167556,
+      "grad_norm": 997.2656860351562,
+      "learning_rate": 5.136303187524478e-07,
+      "loss": 88.9139,
+      "step": 107540
+    },
+    {
+      "epoch": 0.8896885469661249,
+      "grad_norm": 1044.582275390625,
+      "learning_rate": 5.130078565432089e-07,
+      "loss": 82.6202,
+      "step": 107550
+    },
+    {
+      "epoch": 0.8897712702154941,
+      "grad_norm": 683.0887451171875,
+      "learning_rate": 5.123857513433406e-07,
+      "loss": 77.2961,
+      "step": 107560
+    },
+    {
+      "epoch": 0.8898539934648633,
+      "grad_norm": 307.61688232421875,
+      "learning_rate": 5.117640032023436e-07,
+      "loss": 82.64,
+      "step": 107570
+    },
+    {
+      "epoch": 0.8899367167142326,
+      "grad_norm": 549.88720703125,
+      "learning_rate": 5.111426121696866e-07,
+      "loss": 84.3351,
+      "step": 107580
+    },
+    {
+      "epoch": 0.8900194399636018,
+      "grad_norm": 685.7962646484375,
+      "learning_rate": 5.105215782948108e-07,
+      "loss": 86.4093,
+      "step": 107590
+    },
+    {
+      "epoch": 0.890102163212971,
+      "grad_norm": 4017.52880859375,
+      "learning_rate": 5.099009016271295e-07,
+      "loss": 93.5784,
+      "step": 107600
+    },
+    {
+      "epoch": 0.8901848864623403,
+      "grad_norm": 562.3822631835938,
+      "learning_rate": 5.092805822160262e-07,
+      "loss": 70.4941,
+      "step": 107610
+    },
+    {
+      "epoch": 0.8902676097117095,
+      "grad_norm": 987.6434936523438,
+      "learning_rate": 5.086606201108574e-07,
+      "loss": 84.2584,
+      "step": 107620
+    },
+    {
+      "epoch": 0.8903503329610787,
+      "grad_norm": 864.2491455078125,
+      "learning_rate": 5.080410153609511e-07,
+      "loss": 103.629,
+      "step": 107630
+    },
+    {
+      "epoch": 0.890433056210448,
+      "grad_norm": 553.91064453125,
+      "learning_rate": 5.074217680156062e-07,
+      "loss": 72.658,
+      "step": 107640
+    },
+    {
+      "epoch": 0.8905157794598172,
+      "grad_norm": 913.7005615234375,
+      "learning_rate": 5.068028781240925e-07,
+      "loss": 83.6153,
+      "step": 107650
+    },
+    {
+      "epoch": 0.8905985027091864,
+      "grad_norm": 643.3348388671875,
+      "learning_rate": 5.061843457356519e-07,
+      "loss": 92.5445,
+      "step": 107660
+    },
+    {
+      "epoch": 0.8906812259585557,
+      "grad_norm": 616.33984375,
+      "learning_rate": 5.055661708994996e-07,
+      "loss": 101.3816,
+      "step": 107670
+    },
+    {
+      "epoch": 0.8907639492079249,
+      "grad_norm": 926.9981079101562,
+      "learning_rate": 5.049483536648209e-07,
+      "loss": 75.4115,
+      "step": 107680
+    },
+    {
+      "epoch": 0.8908466724572941,
+      "grad_norm": 888.5517578125,
+      "learning_rate": 5.043308940807717e-07,
+      "loss": 90.7422,
+      "step": 107690
+    },
+    {
+      "epoch": 0.8909293957066634,
+      "grad_norm": 884.9143676757812,
+      "learning_rate": 5.037137921964814e-07,
+      "loss": 93.778,
+      "step": 107700
+    },
+    {
+      "epoch": 0.8910121189560326,
+      "grad_norm": 1310.1082763671875,
+      "learning_rate": 5.030970480610492e-07,
+      "loss": 92.942,
+      "step": 107710
+    },
+    {
+      "epoch": 0.8910948422054018,
+      "grad_norm": 1047.660888671875,
+      "learning_rate": 5.024806617235484e-07,
+      "loss": 86.2382,
+      "step": 107720
+    },
+    {
+      "epoch": 0.8911775654547711,
+      "grad_norm": 638.484619140625,
+      "learning_rate": 5.01864633233019e-07,
+      "loss": 103.3557,
+      "step": 107730
+    },
+    {
+      "epoch": 0.8912602887041403,
+      "grad_norm": 1222.545654296875,
+      "learning_rate": 5.01248962638477e-07,
+      "loss": 81.7074,
+      "step": 107740
+    },
+    {
+      "epoch": 0.8913430119535095,
+      "grad_norm": 648.8272705078125,
+      "learning_rate": 5.006336499889075e-07,
+      "loss": 87.4129,
+      "step": 107750
+    },
+    {
+      "epoch": 0.8914257352028788,
+      "grad_norm": 611.2053833007812,
+      "learning_rate": 5.000186953332709e-07,
+      "loss": 77.1965,
+      "step": 107760
+    },
+    {
+      "epoch": 0.891508458452248,
+      "grad_norm": 863.5994262695312,
+      "learning_rate": 4.99404098720494e-07,
+      "loss": 83.2366,
+      "step": 107770
+    },
+    {
+      "epoch": 0.8915911817016172,
+      "grad_norm": 1112.973388671875,
+      "learning_rate": 4.987898601994778e-07,
+      "loss": 94.6965,
+      "step": 107780
+    },
+    {
+      "epoch": 0.8916739049509865,
+      "grad_norm": 807.7054443359375,
+      "learning_rate": 4.981759798190949e-07,
+      "loss": 70.4171,
+      "step": 107790
+    },
+    {
+      "epoch": 0.8917566282003557,
+      "grad_norm": 845.5135498046875,
+      "learning_rate": 4.97562457628189e-07,
+      "loss": 95.9467,
+      "step": 107800
+    },
+    {
+      "epoch": 0.8918393514497249,
+      "grad_norm": 1642.2303466796875,
+      "learning_rate": 4.969492936755759e-07,
+      "loss": 71.8787,
+      "step": 107810
+    },
+    {
+      "epoch": 0.8919220746990941,
+      "grad_norm": 796.374755859375,
+      "learning_rate": 4.963364880100402e-07,
+      "loss": 104.153,
+      "step": 107820
+    },
+    {
+      "epoch": 0.8920047979484634,
+      "grad_norm": 420.7682800292969,
+      "learning_rate": 4.957240406803409e-07,
+      "loss": 78.5665,
+      "step": 107830
+    },
+    {
+      "epoch": 0.8920875211978326,
+      "grad_norm": 785.2117309570312,
+      "learning_rate": 4.951119517352082e-07,
+      "loss": 84.3331,
+      "step": 107840
+    },
+    {
+      "epoch": 0.8921702444472018,
+      "grad_norm": 341.3686828613281,
+      "learning_rate": 4.945002212233412e-07,
+      "loss": 83.6106,
+      "step": 107850
+    },
+    {
+      "epoch": 0.8922529676965711,
+      "grad_norm": 785.5858154296875,
+      "learning_rate": 4.938888491934158e-07,
+      "loss": 72.6205,
+      "step": 107860
+    },
+    {
+      "epoch": 0.8923356909459403,
+      "grad_norm": 1233.1529541015625,
+      "learning_rate": 4.932778356940743e-07,
+      "loss": 132.9701,
+      "step": 107870
+    },
+    {
+      "epoch": 0.8924184141953095,
+      "grad_norm": 772.5987548828125,
+      "learning_rate": 4.926671807739319e-07,
+      "loss": 69.4721,
+      "step": 107880
+    },
+    {
+      "epoch": 0.8925011374446788,
+      "grad_norm": 575.5115966796875,
+      "learning_rate": 4.920568844815776e-07,
+      "loss": 90.4714,
+      "step": 107890
+    },
+    {
+      "epoch": 0.892583860694048,
+      "grad_norm": 895.2702026367188,
+      "learning_rate": 4.914469468655675e-07,
+      "loss": 64.3026,
+      "step": 107900
+    },
+    {
+      "epoch": 0.8926665839434172,
+      "grad_norm": 973.0512084960938,
+      "learning_rate": 4.908373679744316e-07,
+      "loss": 69.2861,
+      "step": 107910
+    },
+    {
+      "epoch": 0.8927493071927866,
+      "grad_norm": 948.7655639648438,
+      "learning_rate": 4.902281478566728e-07,
+      "loss": 86.1326,
+      "step": 107920
+    },
+    {
+      "epoch": 0.8928320304421558,
+      "grad_norm": 656.8770751953125,
+      "learning_rate": 4.896192865607629e-07,
+      "loss": 78.8228,
+      "step": 107930
+    },
+    {
+      "epoch": 0.892914753691525,
+      "grad_norm": 1080.7999267578125,
+      "learning_rate": 4.890107841351466e-07,
+      "loss": 86.0108,
+      "step": 107940
+    },
+    {
+      "epoch": 0.8929974769408943,
+      "grad_norm": 1222.696533203125,
+      "learning_rate": 4.884026406282383e-07,
+      "loss": 82.2702,
+      "step": 107950
+    },
+    {
+      "epoch": 0.8930802001902635,
+      "grad_norm": 1044.7401123046875,
+      "learning_rate": 4.877948560884277e-07,
+      "loss": 83.305,
+      "step": 107960
+    },
+    {
+      "epoch": 0.8931629234396327,
+      "grad_norm": 748.3582153320312,
+      "learning_rate": 4.871874305640723e-07,
+      "loss": 77.1618,
+      "step": 107970
+    },
+    {
+      "epoch": 0.893245646689002,
+      "grad_norm": 833.1705932617188,
+      "learning_rate": 4.865803641035027e-07,
+      "loss": 83.6409,
+      "step": 107980
+    },
+    {
+      "epoch": 0.8933283699383712,
+      "grad_norm": 888.131591796875,
+      "learning_rate": 4.859736567550188e-07,
+      "loss": 99.5259,
+      "step": 107990
+    },
+    {
+      "epoch": 0.8934110931877404,
+      "grad_norm": 908.0950927734375,
+      "learning_rate": 4.853673085668947e-07,
+      "loss": 103.253,
+      "step": 108000
+    },
+    {
+      "epoch": 0.8934938164371097,
+      "grad_norm": 908.1394653320312,
+      "learning_rate": 4.84761319587374e-07,
+      "loss": 63.2051,
+      "step": 108010
+    },
+    {
+      "epoch": 0.8935765396864789,
+      "grad_norm": 761.0054931640625,
+      "learning_rate": 4.841556898646732e-07,
+      "loss": 73.0463,
+      "step": 108020
+    },
+    {
+      "epoch": 0.8936592629358481,
+      "grad_norm": 886.8889770507812,
+      "learning_rate": 4.835504194469792e-07,
+      "loss": 85.4193,
+      "step": 108030
+    },
+    {
+      "epoch": 0.8937419861852174,
+      "grad_norm": 672.312744140625,
+      "learning_rate": 4.829455083824508e-07,
+      "loss": 91.8696,
+      "step": 108040
+    },
+    {
+      "epoch": 0.8938247094345866,
+      "grad_norm": 700.14599609375,
+      "learning_rate": 4.823409567192172e-07,
+      "loss": 88.3726,
+      "step": 108050
+    },
+    {
+      "epoch": 0.8939074326839558,
+      "grad_norm": 1457.9686279296875,
+      "learning_rate": 4.817367645053806e-07,
+      "loss": 94.6632,
+      "step": 108060
+    },
+    {
+      "epoch": 0.8939901559333251,
+      "grad_norm": 1704.705810546875,
+      "learning_rate": 4.811329317890151e-07,
+      "loss": 94.3083,
+      "step": 108070
+    },
+    {
+      "epoch": 0.8940728791826943,
+      "grad_norm": 720.1951293945312,
+      "learning_rate": 4.805294586181624e-07,
+      "loss": 87.1297,
+      "step": 108080
+    },
+    {
+      "epoch": 0.8941556024320635,
+      "grad_norm": 581.7968139648438,
+      "learning_rate": 4.799263450408386e-07,
+      "loss": 66.5786,
+      "step": 108090
+    },
+    {
+      "epoch": 0.8942383256814328,
+      "grad_norm": 1700.38232421875,
+      "learning_rate": 4.79323591105032e-07,
+      "loss": 110.1113,
+      "step": 108100
+    },
+    {
+      "epoch": 0.894321048930802,
+      "grad_norm": 746.1917724609375,
+      "learning_rate": 4.787211968586996e-07,
+      "loss": 78.2092,
+      "step": 108110
+    },
+    {
+      "epoch": 0.8944037721801712,
+      "grad_norm": 1086.206298828125,
+      "learning_rate": 4.781191623497716e-07,
+      "loss": 102.1627,
+      "step": 108120
+    },
+    {
+      "epoch": 0.8944864954295405,
+      "grad_norm": 983.9944458007812,
+      "learning_rate": 4.775174876261496e-07,
+      "loss": 101.4622,
+      "step": 108130
+    },
+    {
+      "epoch": 0.8945692186789097,
+      "grad_norm": 724.5037231445312,
+      "learning_rate": 4.769161727357047e-07,
+      "loss": 98.8229,
+      "step": 108140
+    },
+    {
+      "epoch": 0.8946519419282789,
+      "grad_norm": 1058.3212890625,
+      "learning_rate": 4.763152177262836e-07,
+      "loss": 78.5956,
+      "step": 108150
+    },
+    {
+      "epoch": 0.8947346651776482,
+      "grad_norm": 1093.287841796875,
+      "learning_rate": 4.757146226456988e-07,
+      "loss": 128.6798,
+      "step": 108160
+    },
+    {
+      "epoch": 0.8948173884270174,
+      "grad_norm": 1089.8331298828125,
+      "learning_rate": 4.75114387541738e-07,
+      "loss": 102.0988,
+      "step": 108170
+    },
+    {
+      "epoch": 0.8949001116763866,
+      "grad_norm": 221.10438537597656,
+      "learning_rate": 4.7451451246215863e-07,
+      "loss": 87.4969,
+      "step": 108180
+    },
+    {
+      "epoch": 0.8949828349257559,
+      "grad_norm": 798.0653076171875,
+      "learning_rate": 4.7391499745469026e-07,
+      "loss": 77.1766,
+      "step": 108190
+    },
+    {
+      "epoch": 0.8950655581751251,
+      "grad_norm": 767.3575439453125,
+      "learning_rate": 4.733158425670342e-07,
+      "loss": 78.9256,
+      "step": 108200
+    },
+    {
+      "epoch": 0.8951482814244943,
+      "grad_norm": 1510.5670166015625,
+      "learning_rate": 4.727170478468612e-07,
+      "loss": 112.9961,
+      "step": 108210
+    },
+    {
+      "epoch": 0.8952310046738636,
+      "grad_norm": 2023.6324462890625,
+      "learning_rate": 4.721186133418154e-07,
+      "loss": 107.9993,
+      "step": 108220
+    },
+    {
+      "epoch": 0.8953137279232328,
+      "grad_norm": 434.2220458984375,
+      "learning_rate": 4.71520539099512e-07,
+      "loss": 86.038,
+      "step": 108230
+    },
+    {
+      "epoch": 0.895396451172602,
+      "grad_norm": 890.50048828125,
+      "learning_rate": 4.709228251675357e-07,
+      "loss": 90.7458,
+      "step": 108240
+    },
+    {
+      "epoch": 0.8954791744219713,
+      "grad_norm": 873.815673828125,
+      "learning_rate": 4.7032547159344466e-07,
+      "loss": 91.8802,
+      "step": 108250
+    },
+    {
+      "epoch": 0.8955618976713405,
+      "grad_norm": 899.265869140625,
+      "learning_rate": 4.6972847842476743e-07,
+      "loss": 121.3053,
+      "step": 108260
+    },
+    {
+      "epoch": 0.8956446209207097,
+      "grad_norm": 987.6566162109375,
+      "learning_rate": 4.6913184570900436e-07,
+      "loss": 125.955,
+      "step": 108270
+    },
+    {
+      "epoch": 0.895727344170079,
+      "grad_norm": 746.0098876953125,
+      "learning_rate": 4.685355734936264e-07,
+      "loss": 93.6544,
+      "step": 108280
+    },
+    {
+      "epoch": 0.8958100674194482,
+      "grad_norm": 577.643310546875,
+      "learning_rate": 4.6793966182607564e-07,
+      "loss": 85.8028,
+      "step": 108290
+    },
+    {
+      "epoch": 0.8958927906688174,
+      "grad_norm": 748.9794921875,
+      "learning_rate": 4.673441107537674e-07,
+      "loss": 79.7184,
+      "step": 108300
+    },
+    {
+      "epoch": 0.8959755139181868,
+      "grad_norm": 652.255126953125,
+      "learning_rate": 4.6674892032408605e-07,
+      "loss": 81.2928,
+      "step": 108310
+    },
+    {
+      "epoch": 0.896058237167556,
+      "grad_norm": 1240.67822265625,
+      "learning_rate": 4.661540905843881e-07,
+      "loss": 94.0372,
+      "step": 108320
+    },
+    {
+      "epoch": 0.8961409604169251,
+      "grad_norm": 807.7527465820312,
+      "learning_rate": 4.655596215820013e-07,
+      "loss": 78.2325,
+      "step": 108330
+    },
+    {
+      "epoch": 0.8962236836662945,
+      "grad_norm": 633.8656616210938,
+      "learning_rate": 4.649655133642256e-07,
+      "loss": 96.0483,
+      "step": 108340
+    },
+    {
+      "epoch": 0.8963064069156637,
+      "grad_norm": 622.1317749023438,
+      "learning_rate": 4.643717659783309e-07,
+      "loss": 88.281,
+      "step": 108350
+    },
+    {
+      "epoch": 0.8963891301650329,
+      "grad_norm": 937.6145629882812,
+      "learning_rate": 4.637783794715589e-07,
+      "loss": 81.1072,
+      "step": 108360
+    },
+    {
+      "epoch": 0.8964718534144022,
+      "grad_norm": 1903.2171630859375,
+      "learning_rate": 4.6318535389112296e-07,
+      "loss": 86.6768,
+      "step": 108370
+    },
+    {
+      "epoch": 0.8965545766637714,
+      "grad_norm": 1076.609619140625,
+      "learning_rate": 4.6259268928420753e-07,
+      "loss": 98.4922,
+      "step": 108380
+    },
+    {
+      "epoch": 0.8966372999131406,
+      "grad_norm": 897.179443359375,
+      "learning_rate": 4.620003856979671e-07,
+      "loss": 92.1115,
+      "step": 108390
+    },
+    {
+      "epoch": 0.8967200231625099,
+      "grad_norm": 866.747314453125,
+      "learning_rate": 4.6140844317953013e-07,
+      "loss": 92.1406,
+      "step": 108400
+    },
+    {
+      "epoch": 0.8968027464118791,
+      "grad_norm": 500.00457763671875,
+      "learning_rate": 4.6081686177599395e-07,
+      "loss": 66.3908,
+      "step": 108410
+    },
+    {
+      "epoch": 0.8968854696612483,
+      "grad_norm": 782.8357543945312,
+      "learning_rate": 4.602256415344275e-07,
+      "loss": 64.6097,
+      "step": 108420
+    },
+    {
+      "epoch": 0.8969681929106176,
+      "grad_norm": 1170.851318359375,
+      "learning_rate": 4.5963478250187266e-07,
+      "loss": 92.5552,
+      "step": 108430
+    },
+    {
+      "epoch": 0.8970509161599868,
+      "grad_norm": 1039.9666748046875,
+      "learning_rate": 4.5904428472534014e-07,
+      "loss": 93.8967,
+      "step": 108440
+    },
+    {
+      "epoch": 0.897133639409356,
+      "grad_norm": 638.94091796875,
+      "learning_rate": 4.58454148251814e-07,
+      "loss": 75.4576,
+      "step": 108450
+    },
+    {
+      "epoch": 0.8972163626587253,
+      "grad_norm": 805.0771484375,
+      "learning_rate": 4.578643731282484e-07,
+      "loss": 77.5966,
+      "step": 108460
+    },
+    {
+      "epoch": 0.8972990859080945,
+      "grad_norm": 896.1200561523438,
+      "learning_rate": 4.5727495940156906e-07,
+      "loss": 89.5274,
+      "step": 108470
+    },
+    {
+      "epoch": 0.8973818091574637,
+      "grad_norm": 995.3990478515625,
+      "learning_rate": 4.56685907118673e-07,
+      "loss": 81.9692,
+      "step": 108480
+    },
+    {
+      "epoch": 0.897464532406833,
+      "grad_norm": 967.8817749023438,
+      "learning_rate": 4.560972163264282e-07,
+      "loss": 85.0128,
+      "step": 108490
+    },
+    {
+      "epoch": 0.8975472556562022,
+      "grad_norm": 1025.315185546875,
+      "learning_rate": 4.5550888707167505e-07,
+      "loss": 98.1567,
+      "step": 108500
+    },
+    {
+      "epoch": 0.8976299789055714,
+      "grad_norm": 477.4637756347656,
+      "learning_rate": 4.549209194012216e-07,
+      "loss": 67.9016,
+      "step": 108510
+    },
+    {
+      "epoch": 0.8977127021549407,
+      "grad_norm": 873.9866943359375,
+      "learning_rate": 4.543333133618522e-07,
+      "loss": 98.0505,
+      "step": 108520
+    },
+    {
+      "epoch": 0.8977954254043099,
+      "grad_norm": 1049.532958984375,
+      "learning_rate": 4.537460690003198e-07,
+      "loss": 67.647,
+      "step": 108530
+    },
+    {
+      "epoch": 0.8978781486536791,
+      "grad_norm": 1115.316162109375,
+      "learning_rate": 4.531591863633478e-07,
+      "loss": 82.5628,
+      "step": 108540
+    },
+    {
+      "epoch": 0.8979608719030483,
+      "grad_norm": 532.7764892578125,
+      "learning_rate": 4.5257266549763203e-07,
+      "loss": 51.0,
+      "step": 108550
+    },
+    {
+      "epoch": 0.8980435951524176,
+      "grad_norm": 729.6134033203125,
+      "learning_rate": 4.5198650644983965e-07,
+      "loss": 114.1188,
+      "step": 108560
+    },
+    {
+      "epoch": 0.8981263184017868,
+      "grad_norm": 739.4089965820312,
+      "learning_rate": 4.514007092666084e-07,
+      "loss": 87.889,
+      "step": 108570
+    },
+    {
+      "epoch": 0.898209041651156,
+      "grad_norm": 662.3689575195312,
+      "learning_rate": 4.5081527399454814e-07,
+      "loss": 102.5013,
+      "step": 108580
+    },
+    {
+      "epoch": 0.8982917649005253,
+      "grad_norm": 909.9256591796875,
+      "learning_rate": 4.502302006802378e-07,
+      "loss": 82.6323,
+      "step": 108590
+    },
+    {
+      "epoch": 0.8983744881498945,
+      "grad_norm": 909.2214965820312,
+      "learning_rate": 4.496454893702301e-07,
+      "loss": 78.7108,
+      "step": 108600
+    },
+    {
+      "epoch": 0.8984572113992637,
+      "grad_norm": 580.1622314453125,
+      "learning_rate": 4.4906114011104616e-07,
+      "loss": 59.7331,
+      "step": 108610
+    },
+    {
+      "epoch": 0.898539934648633,
+      "grad_norm": 1207.3009033203125,
+      "learning_rate": 4.4847715294918215e-07,
+      "loss": 84.35,
+      "step": 108620
+    },
+    {
+      "epoch": 0.8986226578980022,
+      "grad_norm": 520.7291259765625,
+      "learning_rate": 4.478935279311031e-07,
+      "loss": 91.6451,
+      "step": 108630
+    },
+    {
+      "epoch": 0.8987053811473714,
+      "grad_norm": 703.8441162109375,
+      "learning_rate": 4.4731026510324406e-07,
+      "loss": 97.7126,
+      "step": 108640
+    },
+    {
+      "epoch": 0.8987881043967407,
+      "grad_norm": 707.9328002929688,
+      "learning_rate": 4.4672736451201347e-07,
+      "loss": 93.7411,
+      "step": 108650
+    },
+    {
+      "epoch": 0.8988708276461099,
+      "grad_norm": 912.7407836914062,
+      "learning_rate": 4.461448262037893e-07,
+      "loss": 107.5078,
+      "step": 108660
+    },
+    {
+      "epoch": 0.8989535508954791,
+      "grad_norm": 891.9956665039062,
+      "learning_rate": 4.455626502249233e-07,
+      "loss": 66.6602,
+      "step": 108670
+    },
+    {
+      "epoch": 0.8990362741448484,
+      "grad_norm": 736.6570434570312,
+      "learning_rate": 4.4498083662173396e-07,
+      "loss": 52.5633,
+      "step": 108680
+    },
+    {
+      "epoch": 0.8991189973942176,
+      "grad_norm": 798.45703125,
+      "learning_rate": 4.443993854405154e-07,
+      "loss": 95.3546,
+      "step": 108690
+    },
+    {
+      "epoch": 0.8992017206435868,
+      "grad_norm": 643.7161865234375,
+      "learning_rate": 4.4381829672752896e-07,
+      "loss": 111.5154,
+      "step": 108700
+    },
+    {
+      "epoch": 0.8992844438929561,
+      "grad_norm": 1088.9283447265625,
+      "learning_rate": 4.4323757052901153e-07,
+      "loss": 87.3449,
+      "step": 108710
+    },
+    {
+      "epoch": 0.8993671671423253,
+      "grad_norm": 626.2893676757812,
+      "learning_rate": 4.4265720689116776e-07,
+      "loss": 87.1247,
+      "step": 108720
+    },
+    {
+      "epoch": 0.8994498903916945,
+      "grad_norm": 608.726806640625,
+      "learning_rate": 4.420772058601747e-07,
+      "loss": 86.0277,
+      "step": 108730
+    },
+    {
+      "epoch": 0.8995326136410638,
+      "grad_norm": 433.5987243652344,
+      "learning_rate": 4.414975674821803e-07,
+      "loss": 73.3416,
+      "step": 108740
+    },
+    {
+      "epoch": 0.899615336890433,
+      "grad_norm": 1192.2998046875,
+      "learning_rate": 4.4091829180330503e-07,
+      "loss": 65.4931,
+      "step": 108750
+    },
+    {
+      "epoch": 0.8996980601398022,
+      "grad_norm": 438.9178466796875,
+      "learning_rate": 4.40339378869637e-07,
+      "loss": 76.4078,
+      "step": 108760
+    },
+    {
+      "epoch": 0.8997807833891716,
+      "grad_norm": 786.5850219726562,
+      "learning_rate": 4.3976082872723814e-07,
+      "loss": 68.2923,
+      "step": 108770
+    },
+    {
+      "epoch": 0.8998635066385408,
+      "grad_norm": 1577.2021484375,
+      "learning_rate": 4.3918264142214173e-07,
+      "loss": 98.498,
+      "step": 108780
+    },
+    {
+      "epoch": 0.89994622988791,
+      "grad_norm": 880.0924072265625,
+      "learning_rate": 4.3860481700035096e-07,
+      "loss": 84.358,
+      "step": 108790
+    },
+    {
+      "epoch": 0.9000289531372793,
+      "grad_norm": 2280.10693359375,
+      "learning_rate": 4.3802735550784014e-07,
+      "loss": 75.876,
+      "step": 108800
+    },
+    {
+      "epoch": 0.9001116763866485,
+      "grad_norm": 687.7626953125,
+      "learning_rate": 4.37450256990557e-07,
+      "loss": 121.7229,
+      "step": 108810
+    },
+    {
+      "epoch": 0.9001943996360177,
+      "grad_norm": 1037.4595947265625,
+      "learning_rate": 4.368735214944181e-07,
+      "loss": 87.6477,
+      "step": 108820
+    },
+    {
+      "epoch": 0.900277122885387,
+      "grad_norm": 5595.93359375,
+      "learning_rate": 4.362971490653106e-07,
+      "loss": 120.0495,
+      "step": 108830
+    },
+    {
+      "epoch": 0.9003598461347562,
+      "grad_norm": 1015.2440795898438,
+      "learning_rate": 4.357211397490951e-07,
+      "loss": 81.7864,
+      "step": 108840
+    },
+    {
+      "epoch": 0.9004425693841254,
+      "grad_norm": 632.1942749023438,
+      "learning_rate": 4.351454935916011e-07,
+      "loss": 83.9828,
+      "step": 108850
+    },
+    {
+      "epoch": 0.9005252926334947,
+      "grad_norm": 1923.382568359375,
+      "learning_rate": 4.3457021063862957e-07,
+      "loss": 84.9657,
+      "step": 108860
+    },
+    {
+      "epoch": 0.9006080158828639,
+      "grad_norm": 621.4093627929688,
+      "learning_rate": 4.339952909359546e-07,
+      "loss": 86.5398,
+      "step": 108870
+    },
+    {
+      "epoch": 0.9006907391322331,
+      "grad_norm": 1436.72021484375,
+      "learning_rate": 4.3342073452931845e-07,
+      "loss": 89.1011,
+      "step": 108880
+    },
+    {
+      "epoch": 0.9007734623816024,
+      "grad_norm": 388.8465881347656,
+      "learning_rate": 4.328465414644373e-07,
+      "loss": 75.2237,
+      "step": 108890
+    },
+    {
+      "epoch": 0.9008561856309716,
+      "grad_norm": 778.0177612304688,
+      "learning_rate": 4.322727117869951e-07,
+      "loss": 99.6206,
+      "step": 108900
+    },
+    {
+      "epoch": 0.9009389088803408,
+      "grad_norm": 582.366455078125,
+      "learning_rate": 4.3169924554265165e-07,
+      "loss": 82.2605,
+      "step": 108910
+    },
+    {
+      "epoch": 0.9010216321297101,
+      "grad_norm": 1931.376953125,
+      "learning_rate": 4.3112614277703304e-07,
+      "loss": 106.5273,
+      "step": 108920
+    },
+    {
+      "epoch": 0.9011043553790793,
+      "grad_norm": 899.9615478515625,
+      "learning_rate": 4.3055340353574004e-07,
+      "loss": 89.0014,
+      "step": 108930
+    },
+    {
+      "epoch": 0.9011870786284485,
+      "grad_norm": 740.9810791015625,
+      "learning_rate": 4.299810278643407e-07,
+      "loss": 104.6826,
+      "step": 108940
+    },
+    {
+      "epoch": 0.9012698018778178,
+      "grad_norm": 413.1753845214844,
+      "learning_rate": 4.2940901580837744e-07,
+      "loss": 94.5246,
+      "step": 108950
+    },
+    {
+      "epoch": 0.901352525127187,
+      "grad_norm": 717.9979858398438,
+      "learning_rate": 4.2883736741336277e-07,
+      "loss": 57.4638,
+      "step": 108960
+    },
+    {
+      "epoch": 0.9014352483765562,
+      "grad_norm": 962.7039184570312,
+      "learning_rate": 4.282660827247803e-07,
+      "loss": 105.8891,
+      "step": 108970
+    },
+    {
+      "epoch": 0.9015179716259255,
+      "grad_norm": 647.7130737304688,
+      "learning_rate": 4.2769516178808366e-07,
+      "loss": 62.1071,
+      "step": 108980
+    },
+    {
+      "epoch": 0.9016006948752947,
+      "grad_norm": 995.0176391601562,
+      "learning_rate": 4.2712460464869934e-07,
+      "loss": 63.1329,
+      "step": 108990
+    },
+    {
+      "epoch": 0.9016834181246639,
+      "grad_norm": 1008.2059326171875,
+      "learning_rate": 4.26554411352022e-07,
+      "loss": 88.6648,
+      "step": 109000
+    },
+    {
+      "epoch": 0.9017661413740332,
+      "grad_norm": 1263.0606689453125,
+      "learning_rate": 4.259845819434233e-07,
+      "loss": 86.4794,
+      "step": 109010
+    },
+    {
+      "epoch": 0.9018488646234024,
+      "grad_norm": 1270.1148681640625,
+      "learning_rate": 4.254151164682385e-07,
+      "loss": 106.5339,
+      "step": 109020
+    },
+    {
+      "epoch": 0.9019315878727716,
+      "grad_norm": 885.9868774414062,
+      "learning_rate": 4.248460149717781e-07,
+      "loss": 76.8595,
+      "step": 109030
+    },
+    {
+      "epoch": 0.9020143111221409,
+      "grad_norm": 845.4027099609375,
+      "learning_rate": 4.242772774993237e-07,
+      "loss": 83.7632,
+      "step": 109040
+    },
+    {
+      "epoch": 0.9020970343715101,
+      "grad_norm": 1098.0489501953125,
+      "learning_rate": 4.237089040961262e-07,
+      "loss": 64.5871,
+      "step": 109050
+    },
+    {
+      "epoch": 0.9021797576208793,
+      "grad_norm": 846.4615478515625,
+      "learning_rate": 4.2314089480740893e-07,
+      "loss": 79.8472,
+      "step": 109060
+    },
+    {
+      "epoch": 0.9022624808702486,
+      "grad_norm": 858.0891723632812,
+      "learning_rate": 4.2257324967836575e-07,
+      "loss": 93.2545,
+      "step": 109070
+    },
+    {
+      "epoch": 0.9023452041196178,
+      "grad_norm": 1338.1405029296875,
+      "learning_rate": 4.2200596875416165e-07,
+      "loss": 91.2463,
+      "step": 109080
+    },
+    {
+      "epoch": 0.902427927368987,
+      "grad_norm": 1132.8909912109375,
+      "learning_rate": 4.2143905207993217e-07,
+      "loss": 94.3591,
+      "step": 109090
+    },
+    {
+      "epoch": 0.9025106506183563,
+      "grad_norm": 1090.1063232421875,
+      "learning_rate": 4.2087249970078513e-07,
+      "loss": 80.4746,
+      "step": 109100
+    },
+    {
+      "epoch": 0.9025933738677255,
+      "grad_norm": 697.7597045898438,
+      "learning_rate": 4.2030631166179727e-07,
+      "loss": 69.0265,
+      "step": 109110
+    },
+    {
+      "epoch": 0.9026760971170947,
+      "grad_norm": 925.8981323242188,
+      "learning_rate": 4.197404880080186e-07,
+      "loss": 97.9635,
+      "step": 109120
+    },
+    {
+      "epoch": 0.902758820366464,
+      "grad_norm": 705.66552734375,
+      "learning_rate": 4.191750287844687e-07,
+      "loss": 96.6312,
+      "step": 109130
+    },
+    {
+      "epoch": 0.9028415436158332,
+      "grad_norm": 497.7852478027344,
+      "learning_rate": 4.186099340361383e-07,
+      "loss": 87.5341,
+      "step": 109140
+    },
+    {
+      "epoch": 0.9029242668652024,
+      "grad_norm": 537.1528930664062,
+      "learning_rate": 4.180452038079902e-07,
+      "loss": 82.3161,
+      "step": 109150
+    },
+    {
+      "epoch": 0.9030069901145718,
+      "grad_norm": 627.737548828125,
+      "learning_rate": 4.174808381449563e-07,
+      "loss": 84.3565,
+      "step": 109160
+    },
+    {
+      "epoch": 0.903089713363941,
+      "grad_norm": 743.4111328125,
+      "learning_rate": 4.1691683709194184e-07,
+      "loss": 62.4872,
+      "step": 109170
+    },
+    {
+      "epoch": 0.9031724366133101,
+      "grad_norm": 784.0700073242188,
+      "learning_rate": 4.163532006938209e-07,
+      "loss": 76.4367,
+      "step": 109180
+    },
+    {
+      "epoch": 0.9032551598626795,
+      "grad_norm": 770.558837890625,
+      "learning_rate": 4.1578992899543926e-07,
+      "loss": 90.6556,
+      "step": 109190
+    },
+    {
+      "epoch": 0.9033378831120487,
+      "grad_norm": 586.0490112304688,
+      "learning_rate": 4.1522702204161493e-07,
+      "loss": 84.0449,
+      "step": 109200
+    },
+    {
+      "epoch": 0.9034206063614179,
+      "grad_norm": 923.1376953125,
+      "learning_rate": 4.146644798771349e-07,
+      "loss": 118.2938,
+      "step": 109210
+    },
+    {
+      "epoch": 0.9035033296107872,
+      "grad_norm": 749.9999389648438,
+      "learning_rate": 4.141023025467583e-07,
+      "loss": 104.1386,
+      "step": 109220
+    },
+    {
+      "epoch": 0.9035860528601564,
+      "grad_norm": 1027.83349609375,
+      "learning_rate": 4.1354049009521504e-07,
+      "loss": 80.9892,
+      "step": 109230
+    },
+    {
+      "epoch": 0.9036687761095256,
+      "grad_norm": 886.0511474609375,
+      "learning_rate": 4.1297904256720646e-07,
+      "loss": 76.1879,
+      "step": 109240
+    },
+    {
+      "epoch": 0.9037514993588949,
+      "grad_norm": 880.5858764648438,
+      "learning_rate": 4.1241796000740296e-07,
+      "loss": 60.6476,
+      "step": 109250
+    },
+    {
+      "epoch": 0.9038342226082641,
+      "grad_norm": 1270.9674072265625,
+      "learning_rate": 4.118572424604489e-07,
+      "loss": 69.0217,
+      "step": 109260
+    },
+    {
+      "epoch": 0.9039169458576333,
+      "grad_norm": 513.490234375,
+      "learning_rate": 4.112968899709574e-07,
+      "loss": 70.7013,
+      "step": 109270
+    },
+    {
+      "epoch": 0.9039996691070025,
+      "grad_norm": 1045.45654296875,
+      "learning_rate": 4.1073690258351287e-07,
+      "loss": 71.1828,
+      "step": 109280
+    },
+    {
+      "epoch": 0.9040823923563718,
+      "grad_norm": 977.5629272460938,
+      "learning_rate": 4.101772803426707e-07,
+      "loss": 104.8837,
+      "step": 109290
+    },
+    {
+      "epoch": 0.904165115605741,
+      "grad_norm": 690.1253662109375,
+      "learning_rate": 4.0961802329295864e-07,
+      "loss": 70.6142,
+      "step": 109300
+    },
+    {
+      "epoch": 0.9042478388551102,
+      "grad_norm": 1407.6588134765625,
+      "learning_rate": 4.090591314788728e-07,
+      "loss": 83.4057,
+      "step": 109310
+    },
+    {
+      "epoch": 0.9043305621044795,
+      "grad_norm": 595.9920654296875,
+      "learning_rate": 4.085006049448825e-07,
+      "loss": 56.3994,
+      "step": 109320
+    },
+    {
+      "epoch": 0.9044132853538487,
+      "grad_norm": 965.2755126953125,
+      "learning_rate": 4.0794244373542736e-07,
+      "loss": 99.3913,
+      "step": 109330
+    },
+    {
+      "epoch": 0.9044960086032179,
+      "grad_norm": 831.5714111328125,
+      "learning_rate": 4.0738464789491673e-07,
+      "loss": 81.9469,
+      "step": 109340
+    },
+    {
+      "epoch": 0.9045787318525872,
+      "grad_norm": 581.9131469726562,
+      "learning_rate": 4.0682721746773346e-07,
+      "loss": 74.2238,
+      "step": 109350
+    },
+    {
+      "epoch": 0.9046614551019564,
+      "grad_norm": 733.6002197265625,
+      "learning_rate": 4.062701524982271e-07,
+      "loss": 83.6481,
+      "step": 109360
+    },
+    {
+      "epoch": 0.9047441783513256,
+      "grad_norm": 720.9714965820312,
+      "learning_rate": 4.057134530307233e-07,
+      "loss": 66.0066,
+      "step": 109370
+    },
+    {
+      "epoch": 0.9048269016006949,
+      "grad_norm": 1211.7706298828125,
+      "learning_rate": 4.0515711910951436e-07,
+      "loss": 103.6463,
+      "step": 109380
+    },
+    {
+      "epoch": 0.9049096248500641,
+      "grad_norm": 887.6861572265625,
+      "learning_rate": 4.0460115077886665e-07,
+      "loss": 74.1794,
+      "step": 109390
+    },
+    {
+      "epoch": 0.9049923480994333,
+      "grad_norm": 725.7745361328125,
+      "learning_rate": 4.0404554808301523e-07,
+      "loss": 86.6203,
+      "step": 109400
+    },
+    {
+      "epoch": 0.9050750713488026,
+      "grad_norm": 892.6699829101562,
+      "learning_rate": 4.0349031106616697e-07,
+      "loss": 83.9293,
+      "step": 109410
+    },
+    {
+      "epoch": 0.9051577945981718,
+      "grad_norm": 1044.3946533203125,
+      "learning_rate": 4.029354397724994e-07,
+      "loss": 81.0904,
+      "step": 109420
+    },
+    {
+      "epoch": 0.905240517847541,
+      "grad_norm": 426.7649230957031,
+      "learning_rate": 4.023809342461615e-07,
+      "loss": 81.0433,
+      "step": 109430
+    },
+    {
+      "epoch": 0.9053232410969103,
+      "grad_norm": 915.0441284179688,
+      "learning_rate": 4.0182679453127316e-07,
+      "loss": 117.5908,
+      "step": 109440
+    },
+    {
+      "epoch": 0.9054059643462795,
+      "grad_norm": 572.4617919921875,
+      "learning_rate": 4.0127302067192285e-07,
+      "loss": 74.3403,
+      "step": 109450
+    },
+    {
+      "epoch": 0.9054886875956487,
+      "grad_norm": 1104.9822998046875,
+      "learning_rate": 4.007196127121726e-07,
+      "loss": 92.4061,
+      "step": 109460
+    },
+    {
+      "epoch": 0.905571410845018,
+      "grad_norm": 1072.473876953125,
+      "learning_rate": 4.001665706960556e-07,
+      "loss": 92.3097,
+      "step": 109470
+    },
+    {
+      "epoch": 0.9056541340943872,
+      "grad_norm": 716.0008544921875,
+      "learning_rate": 3.996138946675737e-07,
+      "loss": 90.4532,
+      "step": 109480
+    },
+    {
+      "epoch": 0.9057368573437564,
+      "grad_norm": 509.80621337890625,
+      "learning_rate": 3.9906158467070187e-07,
+      "loss": 87.5559,
+      "step": 109490
+    },
+    {
+      "epoch": 0.9058195805931257,
+      "grad_norm": 1042.047607421875,
+      "learning_rate": 3.985096407493838e-07,
+      "loss": 98.2946,
+      "step": 109500
+    },
+    {
+      "epoch": 0.9059023038424949,
+      "grad_norm": 788.8461303710938,
+      "learning_rate": 3.97958062947536e-07,
+      "loss": 78.6691,
+      "step": 109510
+    },
+    {
+      "epoch": 0.9059850270918641,
+      "grad_norm": 560.147216796875,
+      "learning_rate": 3.9740685130904455e-07,
+      "loss": 60.2149,
+      "step": 109520
+    },
+    {
+      "epoch": 0.9060677503412334,
+      "grad_norm": 1752.591796875,
+      "learning_rate": 3.9685600587776815e-07,
+      "loss": 108.2573,
+      "step": 109530
+    },
+    {
+      "epoch": 0.9061504735906026,
+      "grad_norm": 870.2755737304688,
+      "learning_rate": 3.9630552669753243e-07,
+      "loss": 70.9846,
+      "step": 109540
+    },
+    {
+      "epoch": 0.9062331968399718,
+      "grad_norm": 1336.35400390625,
+      "learning_rate": 3.9575541381213776e-07,
+      "loss": 69.7969,
+      "step": 109550
+    },
+    {
+      "epoch": 0.9063159200893411,
+      "grad_norm": 897.2123413085938,
+      "learning_rate": 3.9520566726535367e-07,
+      "loss": 82.9407,
+      "step": 109560
+    },
+    {
+      "epoch": 0.9063986433387103,
+      "grad_norm": 1298.525390625,
+      "learning_rate": 3.9465628710092185e-07,
+      "loss": 114.8549,
+      "step": 109570
+    },
+    {
+      "epoch": 0.9064813665880795,
+      "grad_norm": 1084.1280517578125,
+      "learning_rate": 3.9410727336255396e-07,
+      "loss": 84.6156,
+      "step": 109580
+    },
+    {
+      "epoch": 0.9065640898374488,
+      "grad_norm": 580.8189697265625,
+      "learning_rate": 3.935586260939322e-07,
+      "loss": 58.7614,
+      "step": 109590
+    },
+    {
+      "epoch": 0.906646813086818,
+      "grad_norm": 892.2899169921875,
+      "learning_rate": 3.930103453387096e-07,
+      "loss": 102.3395,
+      "step": 109600
+    },
+    {
+      "epoch": 0.9067295363361872,
+      "grad_norm": 429.3028564453125,
+      "learning_rate": 3.9246243114051174e-07,
+      "loss": 77.4244,
+      "step": 109610
+    },
+    {
+      "epoch": 0.9068122595855566,
+      "grad_norm": 469.5216979980469,
+      "learning_rate": 3.919148835429315e-07,
+      "loss": 64.1674,
+      "step": 109620
+    },
+    {
+      "epoch": 0.9068949828349258,
+      "grad_norm": 779.257568359375,
+      "learning_rate": 3.913677025895357e-07,
+      "loss": 84.8305,
+      "step": 109630
+    },
+    {
+      "epoch": 0.906977706084295,
+      "grad_norm": 1248.734375,
+      "learning_rate": 3.9082088832386124e-07,
+      "loss": 85.1336,
+      "step": 109640
+    },
+    {
+      "epoch": 0.9070604293336643,
+      "grad_norm": 785.3364868164062,
+      "learning_rate": 3.9027444078941435e-07,
+      "loss": 110.6684,
+      "step": 109650
+    },
+    {
+      "epoch": 0.9071431525830335,
+      "grad_norm": 1085.676513671875,
+      "learning_rate": 3.897283600296753e-07,
+      "loss": 77.48,
+      "step": 109660
+    },
+    {
+      "epoch": 0.9072258758324027,
+      "grad_norm": 958.9440307617188,
+      "learning_rate": 3.8918264608809207e-07,
+      "loss": 78.6371,
+      "step": 109670
+    },
+    {
+      "epoch": 0.907308599081772,
+      "grad_norm": 1247.623046875,
+      "learning_rate": 3.886372990080856e-07,
+      "loss": 88.1574,
+      "step": 109680
+    },
+    {
+      "epoch": 0.9073913223311412,
+      "grad_norm": 860.1593017578125,
+      "learning_rate": 3.8809231883304544e-07,
+      "loss": 73.0786,
+      "step": 109690
+    },
+    {
+      "epoch": 0.9074740455805104,
+      "grad_norm": 1131.9002685546875,
+      "learning_rate": 3.875477056063343e-07,
+      "loss": 99.7562,
+      "step": 109700
+    },
+    {
+      "epoch": 0.9075567688298797,
+      "grad_norm": 578.1716918945312,
+      "learning_rate": 3.8700345937128346e-07,
+      "loss": 74.0123,
+      "step": 109710
+    },
+    {
+      "epoch": 0.9076394920792489,
+      "grad_norm": 990.4799194335938,
+      "learning_rate": 3.864595801711968e-07,
+      "loss": 96.1826,
+      "step": 109720
+    },
+    {
+      "epoch": 0.9077222153286181,
+      "grad_norm": 1159.52490234375,
+      "learning_rate": 3.8591606804934733e-07,
+      "loss": 113.0236,
+      "step": 109730
+    },
+    {
+      "epoch": 0.9078049385779874,
+      "grad_norm": 732.7988891601562,
+      "learning_rate": 3.853729230489811e-07,
+      "loss": 95.0779,
+      "step": 109740
+    },
+    {
+      "epoch": 0.9078876618273566,
+      "grad_norm": 947.6547241210938,
+      "learning_rate": 3.8483014521331184e-07,
+      "loss": 64.9452,
+      "step": 109750
+    },
+    {
+      "epoch": 0.9079703850767258,
+      "grad_norm": 954.159912109375,
+      "learning_rate": 3.8428773458552835e-07,
+      "loss": 96.2663,
+      "step": 109760
+    },
+    {
+      "epoch": 0.9080531083260951,
+      "grad_norm": 807.883056640625,
+      "learning_rate": 3.837456912087867e-07,
+      "loss": 105.7476,
+      "step": 109770
+    },
+    {
+      "epoch": 0.9081358315754643,
+      "grad_norm": 719.7168579101562,
+      "learning_rate": 3.8320401512621505e-07,
+      "loss": 90.5569,
+      "step": 109780
+    },
+    {
+      "epoch": 0.9082185548248335,
+      "grad_norm": 927.4055786132812,
+      "learning_rate": 3.826627063809113e-07,
+      "loss": 55.1157,
+      "step": 109790
+    },
+    {
+      "epoch": 0.9083012780742028,
+      "grad_norm": 900.8541870117188,
+      "learning_rate": 3.821217650159453e-07,
+      "loss": 66.976,
+      "step": 109800
+    },
+    {
+      "epoch": 0.908384001323572,
+      "grad_norm": 1139.78125,
+      "learning_rate": 3.8158119107435667e-07,
+      "loss": 83.4938,
+      "step": 109810
+    },
+    {
+      "epoch": 0.9084667245729412,
+      "grad_norm": 928.9935302734375,
+      "learning_rate": 3.810409845991575e-07,
+      "loss": 81.5093,
+      "step": 109820
+    },
+    {
+      "epoch": 0.9085494478223105,
+      "grad_norm": 866.55712890625,
+      "learning_rate": 3.805011456333285e-07,
+      "loss": 72.1152,
+      "step": 109830
+    },
+    {
+      "epoch": 0.9086321710716797,
+      "grad_norm": 1261.871826171875,
+      "learning_rate": 3.799616742198231e-07,
+      "loss": 81.8657,
+      "step": 109840
+    },
+    {
+      "epoch": 0.9087148943210489,
+      "grad_norm": 633.0125122070312,
+      "learning_rate": 3.794225704015636e-07,
+      "loss": 115.4469,
+      "step": 109850
+    },
+    {
+      "epoch": 0.9087976175704182,
+      "grad_norm": 1088.3365478515625,
+      "learning_rate": 3.7888383422144517e-07,
+      "loss": 96.6127,
+      "step": 109860
+    },
+    {
+      "epoch": 0.9088803408197874,
+      "grad_norm": 1705.669677734375,
+      "learning_rate": 3.7834546572233287e-07,
+      "loss": 90.1035,
+      "step": 109870
+    },
+    {
+      "epoch": 0.9089630640691566,
+      "grad_norm": 1340.4810791015625,
+      "learning_rate": 3.778074649470603e-07,
+      "loss": 97.6526,
+      "step": 109880
+    },
+    {
+      "epoch": 0.9090457873185259,
+      "grad_norm": 471.0425720214844,
+      "learning_rate": 3.772698319384349e-07,
+      "loss": 75.8749,
+      "step": 109890
+    },
+    {
+      "epoch": 0.9091285105678951,
+      "grad_norm": 664.8509521484375,
+      "learning_rate": 3.7673256673923356e-07,
+      "loss": 78.4327,
+      "step": 109900
+    },
+    {
+      "epoch": 0.9092112338172643,
+      "grad_norm": 1128.6776123046875,
+      "learning_rate": 3.7619566939220363e-07,
+      "loss": 109.4718,
+      "step": 109910
+    },
+    {
+      "epoch": 0.9092939570666336,
+      "grad_norm": 811.8029174804688,
+      "learning_rate": 3.7565913994006386e-07,
+      "loss": 80.7976,
+      "step": 109920
+    },
+    {
+      "epoch": 0.9093766803160028,
+      "grad_norm": 873.7833251953125,
+      "learning_rate": 3.751229784255039e-07,
+      "loss": 89.7391,
+      "step": 109930
+    },
+    {
+      "epoch": 0.909459403565372,
+      "grad_norm": 1071.7000732421875,
+      "learning_rate": 3.745871848911831e-07,
+      "loss": 93.3303,
+      "step": 109940
+    },
+    {
+      "epoch": 0.9095421268147413,
+      "grad_norm": 740.0640869140625,
+      "learning_rate": 3.7405175937973103e-07,
+      "loss": 81.3143,
+      "step": 109950
+    },
+    {
+      "epoch": 0.9096248500641105,
+      "grad_norm": 1029.9522705078125,
+      "learning_rate": 3.735167019337527e-07,
+      "loss": 94.7903,
+      "step": 109960
+    },
+    {
+      "epoch": 0.9097075733134797,
+      "grad_norm": 1091.040283203125,
+      "learning_rate": 3.7298201259581615e-07,
+      "loss": 78.4557,
+      "step": 109970
+    },
+    {
+      "epoch": 0.909790296562849,
+      "grad_norm": 554.0980834960938,
+      "learning_rate": 3.724476914084657e-07,
+      "loss": 78.7385,
+      "step": 109980
+    },
+    {
+      "epoch": 0.9098730198122182,
+      "grad_norm": 1428.4512939453125,
+      "learning_rate": 3.719137384142152e-07,
+      "loss": 104.3615,
+      "step": 109990
+    },
+    {
+      "epoch": 0.9099557430615874,
+      "grad_norm": 995.0647583007812,
+      "learning_rate": 3.7138015365554834e-07,
+      "loss": 110.6832,
+      "step": 110000
+    },
+    {
+      "epoch": 0.9100384663109566,
+      "grad_norm": 580.6385498046875,
+      "learning_rate": 3.7084693717492016e-07,
+      "loss": 56.6644,
+      "step": 110010
+    },
+    {
+      "epoch": 0.910121189560326,
+      "grad_norm": 1886.46240234375,
+      "learning_rate": 3.7031408901475605e-07,
+      "loss": 68.3266,
+      "step": 110020
+    },
+    {
+      "epoch": 0.9102039128096951,
+      "grad_norm": 573.3003540039062,
+      "learning_rate": 3.6978160921745277e-07,
+      "loss": 82.59,
+      "step": 110030
+    },
+    {
+      "epoch": 0.9102866360590643,
+      "grad_norm": 722.798828125,
+      "learning_rate": 3.692494978253769e-07,
+      "loss": 111.005,
+      "step": 110040
+    },
+    {
+      "epoch": 0.9103693593084337,
+      "grad_norm": 463.8318786621094,
+      "learning_rate": 3.6871775488086624e-07,
+      "loss": 60.8547,
+      "step": 110050
+    },
+    {
+      "epoch": 0.9104520825578029,
+      "grad_norm": 783.125,
+      "learning_rate": 3.681863804262292e-07,
+      "loss": 91.4471,
+      "step": 110060
+    },
+    {
+      "epoch": 0.910534805807172,
+      "grad_norm": 773.2374267578125,
+      "learning_rate": 3.676553745037448e-07,
+      "loss": 104.682,
+      "step": 110070
+    },
+    {
+      "epoch": 0.9106175290565414,
+      "grad_norm": 917.3580322265625,
+      "learning_rate": 3.671247371556624e-07,
+      "loss": 94.0389,
+      "step": 110080
+    },
+    {
+      "epoch": 0.9107002523059106,
+      "grad_norm": 892.2453002929688,
+      "learning_rate": 3.665944684242029e-07,
+      "loss": 98.2737,
+      "step": 110090
+    },
+    {
+      "epoch": 0.9107829755552798,
+      "grad_norm": 978.7361450195312,
+      "learning_rate": 3.660645683515568e-07,
+      "loss": 78.8953,
+      "step": 110100
+    },
+    {
+      "epoch": 0.9108656988046491,
+      "grad_norm": 538.931640625,
+      "learning_rate": 3.65535036979886e-07,
+      "loss": 89.924,
+      "step": 110110
+    },
+    {
+      "epoch": 0.9109484220540183,
+      "grad_norm": 516.4479370117188,
+      "learning_rate": 3.650058743513235e-07,
+      "loss": 87.3976,
+      "step": 110120
+    },
+    {
+      "epoch": 0.9110311453033875,
+      "grad_norm": 1158.6689453125,
+      "learning_rate": 3.644770805079717e-07,
+      "loss": 77.452,
+      "step": 110130
+    },
+    {
+      "epoch": 0.9111138685527568,
+      "grad_norm": 1315.6121826171875,
+      "learning_rate": 3.639486554919042e-07,
+      "loss": 106.9289,
+      "step": 110140
+    },
+    {
+      "epoch": 0.911196591802126,
+      "grad_norm": 816.3543701171875,
+      "learning_rate": 3.634205993451656e-07,
+      "loss": 82.4566,
+      "step": 110150
+    },
+    {
+      "epoch": 0.9112793150514952,
+      "grad_norm": 838.74267578125,
+      "learning_rate": 3.628929121097707e-07,
+      "loss": 87.7295,
+      "step": 110160
+    },
+    {
+      "epoch": 0.9113620383008645,
+      "grad_norm": 713.4924926757812,
+      "learning_rate": 3.6236559382770597e-07,
+      "loss": 71.8615,
+      "step": 110170
+    },
+    {
+      "epoch": 0.9114447615502337,
+      "grad_norm": 1672.6297607421875,
+      "learning_rate": 3.6183864454092664e-07,
+      "loss": 99.8394,
+      "step": 110180
+    },
+    {
+      "epoch": 0.9115274847996029,
+      "grad_norm": 982.43896484375,
+      "learning_rate": 3.6131206429135977e-07,
+      "loss": 80.5979,
+      "step": 110190
+    },
+    {
+      "epoch": 0.9116102080489722,
+      "grad_norm": 1182.8560791015625,
+      "learning_rate": 3.607858531209035e-07,
+      "loss": 89.0724,
+      "step": 110200
+    },
+    {
+      "epoch": 0.9116929312983414,
+      "grad_norm": 1181.8760986328125,
+      "learning_rate": 3.6026001107142606e-07,
+      "loss": 103.4322,
+      "step": 110210
+    },
+    {
+      "epoch": 0.9117756545477106,
+      "grad_norm": 418.0647277832031,
+      "learning_rate": 3.597345381847656e-07,
+      "loss": 89.5779,
+      "step": 110220
+    },
+    {
+      "epoch": 0.9118583777970799,
+      "grad_norm": 929.8257446289062,
+      "learning_rate": 3.592094345027325e-07,
+      "loss": 77.7989,
+      "step": 110230
+    },
+    {
+      "epoch": 0.9119411010464491,
+      "grad_norm": 1179.2841796875,
+      "learning_rate": 3.5868470006710564e-07,
+      "loss": 121.445,
+      "step": 110240
+    },
+    {
+      "epoch": 0.9120238242958183,
+      "grad_norm": 564.2163696289062,
+      "learning_rate": 3.581603349196372e-07,
+      "loss": 64.8469,
+      "step": 110250
+    },
+    {
+      "epoch": 0.9121065475451876,
+      "grad_norm": 596.6292114257812,
+      "learning_rate": 3.576363391020471e-07,
+      "loss": 93.1684,
+      "step": 110260
+    },
+    {
+      "epoch": 0.9121892707945568,
+      "grad_norm": 695.0462646484375,
+      "learning_rate": 3.5711271265602807e-07,
+      "loss": 77.7925,
+      "step": 110270
+    },
+    {
+      "epoch": 0.912271994043926,
+      "grad_norm": 707.590087890625,
+      "learning_rate": 3.56589455623243e-07,
+      "loss": 71.3893,
+      "step": 110280
+    },
+    {
+      "epoch": 0.9123547172932953,
+      "grad_norm": 0.0,
+      "learning_rate": 3.56066568045324e-07,
+      "loss": 64.9731,
+      "step": 110290
+    },
+    {
+      "epoch": 0.9124374405426645,
+      "grad_norm": 654.4577026367188,
+      "learning_rate": 3.555440499638768e-07,
+      "loss": 80.6185,
+      "step": 110300
+    },
+    {
+      "epoch": 0.9125201637920337,
+      "grad_norm": 691.3319702148438,
+      "learning_rate": 3.55021901420472e-07,
+      "loss": 61.5873,
+      "step": 110310
+    },
+    {
+      "epoch": 0.912602887041403,
+      "grad_norm": 966.6505737304688,
+      "learning_rate": 3.54500122456658e-07,
+      "loss": 73.3954,
+      "step": 110320
+    },
+    {
+      "epoch": 0.9126856102907722,
+      "grad_norm": 1062.859130859375,
+      "learning_rate": 3.5397871311394937e-07,
+      "loss": 88.4193,
+      "step": 110330
+    },
+    {
+      "epoch": 0.9127683335401414,
+      "grad_norm": 1225.390625,
+      "learning_rate": 3.534576734338324e-07,
+      "loss": 85.1301,
+      "step": 110340
+    },
+    {
+      "epoch": 0.9128510567895107,
+      "grad_norm": 604.2843017578125,
+      "learning_rate": 3.529370034577634e-07,
+      "loss": 74.7762,
+      "step": 110350
+    },
+    {
+      "epoch": 0.9129337800388799,
+      "grad_norm": 659.66357421875,
+      "learning_rate": 3.5241670322717025e-07,
+      "loss": 113.1879,
+      "step": 110360
+    },
+    {
+      "epoch": 0.9130165032882491,
+      "grad_norm": 746.4213256835938,
+      "learning_rate": 3.518967727834499e-07,
+      "loss": 81.8533,
+      "step": 110370
+    },
+    {
+      "epoch": 0.9130992265376184,
+      "grad_norm": 654.5377807617188,
+      "learning_rate": 3.513772121679715e-07,
+      "loss": 65.3405,
+      "step": 110380
+    },
+    {
+      "epoch": 0.9131819497869876,
+      "grad_norm": 734.1673583984375,
+      "learning_rate": 3.508580214220753e-07,
+      "loss": 102.2626,
+      "step": 110390
+    },
+    {
+      "epoch": 0.9132646730363568,
+      "grad_norm": 512.5213012695312,
+      "learning_rate": 3.503392005870687e-07,
+      "loss": 81.4534,
+      "step": 110400
+    },
+    {
+      "epoch": 0.9133473962857261,
+      "grad_norm": 852.9417114257812,
+      "learning_rate": 3.498207497042316e-07,
+      "loss": 85.8927,
+      "step": 110410
+    },
+    {
+      "epoch": 0.9134301195350953,
+      "grad_norm": 632.1807250976562,
+      "learning_rate": 3.4930266881481714e-07,
+      "loss": 124.6005,
+      "step": 110420
+    },
+    {
+      "epoch": 0.9135128427844645,
+      "grad_norm": 901.0861206054688,
+      "learning_rate": 3.487849579600455e-07,
+      "loss": 94.3462,
+      "step": 110430
+    },
+    {
+      "epoch": 0.9135955660338339,
+      "grad_norm": 660.1860961914062,
+      "learning_rate": 3.482676171811089e-07,
+      "loss": 86.6812,
+      "step": 110440
+    },
+    {
+      "epoch": 0.913678289283203,
+      "grad_norm": 1478.7293701171875,
+      "learning_rate": 3.4775064651916877e-07,
+      "loss": 68.7283,
+      "step": 110450
+    },
+    {
+      "epoch": 0.9137610125325722,
+      "grad_norm": 927.5498657226562,
+      "learning_rate": 3.472340460153595e-07,
+      "loss": 72.09,
+      "step": 110460
+    },
+    {
+      "epoch": 0.9138437357819416,
+      "grad_norm": 589.6502685546875,
+      "learning_rate": 3.4671781571078424e-07,
+      "loss": 97.7405,
+      "step": 110470
+    },
+    {
+      "epoch": 0.9139264590313108,
+      "grad_norm": 614.3450927734375,
+      "learning_rate": 3.462019556465157e-07,
+      "loss": 126.026,
+      "step": 110480
+    },
+    {
+      "epoch": 0.91400918228068,
+      "grad_norm": 394.06805419921875,
+      "learning_rate": 3.4568646586359944e-07,
+      "loss": 83.8727,
+      "step": 110490
+    },
+    {
+      "epoch": 0.9140919055300493,
+      "grad_norm": 824.491455078125,
+      "learning_rate": 3.4517134640305097e-07,
+      "loss": 100.4316,
+      "step": 110500
+    },
+    {
+      "epoch": 0.9141746287794185,
+      "grad_norm": 895.6577758789062,
+      "learning_rate": 3.446565973058552e-07,
+      "loss": 83.8836,
+      "step": 110510
+    },
+    {
+      "epoch": 0.9142573520287877,
+      "grad_norm": 1477.3956298828125,
+      "learning_rate": 3.441422186129689e-07,
+      "loss": 109.6496,
+      "step": 110520
+    },
+    {
+      "epoch": 0.914340075278157,
+      "grad_norm": 818.725830078125,
+      "learning_rate": 3.4362821036531936e-07,
+      "loss": 93.3673,
+      "step": 110530
+    },
+    {
+      "epoch": 0.9144227985275262,
+      "grad_norm": 812.5512084960938,
+      "learning_rate": 3.431145726038032e-07,
+      "loss": 72.2729,
+      "step": 110540
+    },
+    {
+      "epoch": 0.9145055217768954,
+      "grad_norm": 812.0310668945312,
+      "learning_rate": 3.426013053692878e-07,
+      "loss": 97.3781,
+      "step": 110550
+    },
+    {
+      "epoch": 0.9145882450262647,
+      "grad_norm": 372.3507385253906,
+      "learning_rate": 3.4208840870261326e-07,
+      "loss": 81.1851,
+      "step": 110560
+    },
+    {
+      "epoch": 0.9146709682756339,
+      "grad_norm": 1096.9320068359375,
+      "learning_rate": 3.415758826445864e-07,
+      "loss": 82.4156,
+      "step": 110570
+    },
+    {
+      "epoch": 0.9147536915250031,
+      "grad_norm": 1314.9417724609375,
+      "learning_rate": 3.410637272359868e-07,
+      "loss": 97.9378,
+      "step": 110580
+    },
+    {
+      "epoch": 0.9148364147743724,
+      "grad_norm": 835.7549438476562,
+      "learning_rate": 3.405519425175652e-07,
+      "loss": 110.182,
+      "step": 110590
+    },
+    {
+      "epoch": 0.9149191380237416,
+      "grad_norm": 819.3499145507812,
+      "learning_rate": 3.400405285300412e-07,
+      "loss": 71.8763,
+      "step": 110600
+    },
+    {
+      "epoch": 0.9150018612731108,
+      "grad_norm": 969.0990600585938,
+      "learning_rate": 3.3952948531410566e-07,
+      "loss": 73.1048,
+      "step": 110610
+    },
+    {
+      "epoch": 0.9150845845224801,
+      "grad_norm": 1323.8681640625,
+      "learning_rate": 3.390188129104205e-07,
+      "loss": 111.2788,
+      "step": 110620
+    },
+    {
+      "epoch": 0.9151673077718493,
+      "grad_norm": 1279.118408203125,
+      "learning_rate": 3.3850851135961814e-07,
+      "loss": 119.1711,
+      "step": 110630
+    },
+    {
+      "epoch": 0.9152500310212185,
+      "grad_norm": 767.4146118164062,
+      "learning_rate": 3.379985807023001e-07,
+      "loss": 123.3025,
+      "step": 110640
+    },
+    {
+      "epoch": 0.9153327542705878,
+      "grad_norm": 2297.595703125,
+      "learning_rate": 3.3748902097903936e-07,
+      "loss": 93.4889,
+      "step": 110650
+    },
+    {
+      "epoch": 0.915415477519957,
+      "grad_norm": 851.3316040039062,
+      "learning_rate": 3.369798322303786e-07,
+      "loss": 81.7054,
+      "step": 110660
+    },
+    {
+      "epoch": 0.9154982007693262,
+      "grad_norm": 671.036865234375,
+      "learning_rate": 3.364710144968325e-07,
+      "loss": 84.0587,
+      "step": 110670
+    },
+    {
+      "epoch": 0.9155809240186955,
+      "grad_norm": 1349.0550537109375,
+      "learning_rate": 3.359625678188849e-07,
+      "loss": 91.475,
+      "step": 110680
+    },
+    {
+      "epoch": 0.9156636472680647,
+      "grad_norm": 758.9757690429688,
+      "learning_rate": 3.3545449223699065e-07,
+      "loss": 79.7885,
+      "step": 110690
+    },
+    {
+      "epoch": 0.9157463705174339,
+      "grad_norm": 755.66064453125,
+      "learning_rate": 3.3494678779157464e-07,
+      "loss": 81.5341,
+      "step": 110700
+    },
+    {
+      "epoch": 0.9158290937668032,
+      "grad_norm": 864.9046630859375,
+      "learning_rate": 3.3443945452303337e-07,
+      "loss": 98.7411,
+      "step": 110710
+    },
+    {
+      "epoch": 0.9159118170161724,
+      "grad_norm": 536.4815063476562,
+      "learning_rate": 3.33932492471733e-07,
+      "loss": 98.7579,
+      "step": 110720
+    },
+    {
+      "epoch": 0.9159945402655416,
+      "grad_norm": 550.0960693359375,
+      "learning_rate": 3.334259016780106e-07,
+      "loss": 84.7401,
+      "step": 110730
+    },
+    {
+      "epoch": 0.9160772635149108,
+      "grad_norm": 1865.9971923828125,
+      "learning_rate": 3.3291968218217175e-07,
+      "loss": 84.6278,
+      "step": 110740
+    },
+    {
+      "epoch": 0.9161599867642801,
+      "grad_norm": 598.4445190429688,
+      "learning_rate": 3.324138340244948e-07,
+      "loss": 57.0423,
+      "step": 110750
+    },
+    {
+      "epoch": 0.9162427100136493,
+      "grad_norm": 871.3580932617188,
+      "learning_rate": 3.319083572452275e-07,
+      "loss": 72.3384,
+      "step": 110760
+    },
+    {
+      "epoch": 0.9163254332630185,
+      "grad_norm": 476.9991760253906,
+      "learning_rate": 3.314032518845889e-07,
+      "loss": 68.3785,
+      "step": 110770
+    },
+    {
+      "epoch": 0.9164081565123878,
+      "grad_norm": 693.8629760742188,
+      "learning_rate": 3.308985179827673e-07,
+      "loss": 114.7213,
+      "step": 110780
+    },
+    {
+      "epoch": 0.916490879761757,
+      "grad_norm": 797.1781616210938,
+      "learning_rate": 3.303941555799223e-07,
+      "loss": 86.4564,
+      "step": 110790
+    },
+    {
+      "epoch": 0.9165736030111262,
+      "grad_norm": 313.2153015136719,
+      "learning_rate": 3.298901647161834e-07,
+      "loss": 91.0484,
+      "step": 110800
+    },
+    {
+      "epoch": 0.9166563262604955,
+      "grad_norm": 865.4092407226562,
+      "learning_rate": 3.293865454316514e-07,
+      "loss": 105.9156,
+      "step": 110810
+    },
+    {
+      "epoch": 0.9167390495098647,
+      "grad_norm": 1219.156005859375,
+      "learning_rate": 3.2888329776639807e-07,
+      "loss": 95.9394,
+      "step": 110820
+    },
+    {
+      "epoch": 0.9168217727592339,
+      "grad_norm": 866.1614379882812,
+      "learning_rate": 3.283804217604619e-07,
+      "loss": 96.5862,
+      "step": 110830
+    },
+    {
+      "epoch": 0.9169044960086032,
+      "grad_norm": 650.52490234375,
+      "learning_rate": 3.27877917453856e-07,
+      "loss": 89.6044,
+      "step": 110840
+    },
+    {
+      "epoch": 0.9169872192579724,
+      "grad_norm": 1170.3040771484375,
+      "learning_rate": 3.273757848865622e-07,
+      "loss": 94.1275,
+      "step": 110850
+    },
+    {
+      "epoch": 0.9170699425073416,
+      "grad_norm": 1928.0880126953125,
+      "learning_rate": 3.2687402409853243e-07,
+      "loss": 109.4509,
+      "step": 110860
+    },
+    {
+      "epoch": 0.917152665756711,
+      "grad_norm": 754.0807495117188,
+      "learning_rate": 3.2637263512969033e-07,
+      "loss": 69.4093,
+      "step": 110870
+    },
+    {
+      "epoch": 0.9172353890060801,
+      "grad_norm": 695.7767333984375,
+      "learning_rate": 3.258716180199278e-07,
+      "loss": 82.0896,
+      "step": 110880
+    },
+    {
+      "epoch": 0.9173181122554493,
+      "grad_norm": 1156.6104736328125,
+      "learning_rate": 3.253709728091098e-07,
+      "loss": 72.8865,
+      "step": 110890
+    },
+    {
+      "epoch": 0.9174008355048187,
+      "grad_norm": 551.685546875,
+      "learning_rate": 3.2487069953706983e-07,
+      "loss": 74.3535,
+      "step": 110900
+    },
+    {
+      "epoch": 0.9174835587541879,
+      "grad_norm": 657.1426391601562,
+      "learning_rate": 3.243707982436123e-07,
+      "loss": 67.814,
+      "step": 110910
+    },
+    {
+      "epoch": 0.917566282003557,
+      "grad_norm": 796.87109375,
+      "learning_rate": 3.238712689685125e-07,
+      "loss": 61.9698,
+      "step": 110920
+    },
+    {
+      "epoch": 0.9176490052529264,
+      "grad_norm": 1190.6385498046875,
+      "learning_rate": 3.2337211175151484e-07,
+      "loss": 91.3083,
+      "step": 110930
+    },
+    {
+      "epoch": 0.9177317285022956,
+      "grad_norm": 977.4860229492188,
+      "learning_rate": 3.2287332663233527e-07,
+      "loss": 74.0361,
+      "step": 110940
+    },
+    {
+      "epoch": 0.9178144517516648,
+      "grad_norm": 757.3435668945312,
+      "learning_rate": 3.223749136506604e-07,
+      "loss": 80.6914,
+      "step": 110950
+    },
+    {
+      "epoch": 0.9178971750010341,
+      "grad_norm": 827.2467041015625,
+      "learning_rate": 3.218768728461458e-07,
+      "loss": 111.4018,
+      "step": 110960
+    },
+    {
+      "epoch": 0.9179798982504033,
+      "grad_norm": 609.2623291015625,
+      "learning_rate": 3.2137920425841907e-07,
+      "loss": 99.3807,
+      "step": 110970
+    },
+    {
+      "epoch": 0.9180626214997725,
+      "grad_norm": 1110.6324462890625,
+      "learning_rate": 3.2088190792707696e-07,
+      "loss": 80.4231,
+      "step": 110980
+    },
+    {
+      "epoch": 0.9181453447491418,
+      "grad_norm": 689.8076782226562,
+      "learning_rate": 3.2038498389168724e-07,
+      "loss": 65.6498,
+      "step": 110990
+    },
+    {
+      "epoch": 0.918228067998511,
+      "grad_norm": 941.228515625,
+      "learning_rate": 3.1988843219178776e-07,
+      "loss": 85.3875,
+      "step": 111000
+    },
+    {
+      "epoch": 0.9183107912478802,
+      "grad_norm": 887.41845703125,
+      "learning_rate": 3.193922528668869e-07,
+      "loss": 69.6305,
+      "step": 111010
+    },
+    {
+      "epoch": 0.9183935144972495,
+      "grad_norm": 713.7828369140625,
+      "learning_rate": 3.188964459564636e-07,
+      "loss": 81.9818,
+      "step": 111020
+    },
+    {
+      "epoch": 0.9184762377466187,
+      "grad_norm": 539.0556030273438,
+      "learning_rate": 3.184010114999664e-07,
+      "loss": 83.2385,
+      "step": 111030
+    },
+    {
+      "epoch": 0.9185589609959879,
+      "grad_norm": 917.137451171875,
+      "learning_rate": 3.179059495368153e-07,
+      "loss": 101.1152,
+      "step": 111040
+    },
+    {
+      "epoch": 0.9186416842453572,
+      "grad_norm": 666.9295043945312,
+      "learning_rate": 3.174112601064e-07,
+      "loss": 64.8105,
+      "step": 111050
+    },
+    {
+      "epoch": 0.9187244074947264,
+      "grad_norm": 525.5652465820312,
+      "learning_rate": 3.1691694324808063e-07,
+      "loss": 69.7607,
+      "step": 111060
+    },
+    {
+      "epoch": 0.9188071307440956,
+      "grad_norm": 1084.0616455078125,
+      "learning_rate": 3.1642299900118743e-07,
+      "loss": 113.7904,
+      "step": 111070
+    },
+    {
+      "epoch": 0.9188898539934649,
+      "grad_norm": 1473.717041015625,
+      "learning_rate": 3.159294274050212e-07,
+      "loss": 114.3284,
+      "step": 111080
+    },
+    {
+      "epoch": 0.9189725772428341,
+      "grad_norm": 789.0966186523438,
+      "learning_rate": 3.154362284988538e-07,
+      "loss": 85.0554,
+      "step": 111090
+    },
+    {
+      "epoch": 0.9190553004922033,
+      "grad_norm": 970.8892211914062,
+      "learning_rate": 3.1494340232192667e-07,
+      "loss": 77.95,
+      "step": 111100
+    },
+    {
+      "epoch": 0.9191380237415726,
+      "grad_norm": 856.5325927734375,
+      "learning_rate": 3.144509489134512e-07,
+      "loss": 85.2512,
+      "step": 111110
+    },
+    {
+      "epoch": 0.9192207469909418,
+      "grad_norm": 685.9720458984375,
+      "learning_rate": 3.139588683126099e-07,
+      "loss": 74.0177,
+      "step": 111120
+    },
+    {
+      "epoch": 0.919303470240311,
+      "grad_norm": 782.5280151367188,
+      "learning_rate": 3.134671605585554e-07,
+      "loss": 73.0808,
+      "step": 111130
+    },
+    {
+      "epoch": 0.9193861934896803,
+      "grad_norm": 421.8550720214844,
+      "learning_rate": 3.129758256904109e-07,
+      "loss": 67.5404,
+      "step": 111140
+    },
+    {
+      "epoch": 0.9194689167390495,
+      "grad_norm": 855.1493530273438,
+      "learning_rate": 3.1248486374726884e-07,
+      "loss": 86.1914,
+      "step": 111150
+    },
+    {
+      "epoch": 0.9195516399884187,
+      "grad_norm": 581.30029296875,
+      "learning_rate": 3.119942747681948e-07,
+      "loss": 95.9546,
+      "step": 111160
+    },
+    {
+      "epoch": 0.919634363237788,
+      "grad_norm": 637.724853515625,
+      "learning_rate": 3.1150405879221965e-07,
+      "loss": 97.2694,
+      "step": 111170
+    },
+    {
+      "epoch": 0.9197170864871572,
+      "grad_norm": 706.0328369140625,
+      "learning_rate": 3.110142158583496e-07,
+      "loss": 94.9882,
+      "step": 111180
+    },
+    {
+      "epoch": 0.9197998097365264,
+      "grad_norm": 560.701416015625,
+      "learning_rate": 3.1052474600555936e-07,
+      "loss": 66.4736,
+      "step": 111190
+    },
+    {
+      "epoch": 0.9198825329858957,
+      "grad_norm": 1017.1104125976562,
+      "learning_rate": 3.100356492727929e-07,
+      "loss": 84.1691,
+      "step": 111200
+    },
+    {
+      "epoch": 0.9199652562352649,
+      "grad_norm": 745.77587890625,
+      "learning_rate": 3.0954692569896585e-07,
+      "loss": 68.3992,
+      "step": 111210
+    },
+    {
+      "epoch": 0.9200479794846341,
+      "grad_norm": 499.56640625,
+      "learning_rate": 3.0905857532296414e-07,
+      "loss": 87.3954,
+      "step": 111220
+    },
+    {
+      "epoch": 0.9201307027340034,
+      "grad_norm": 1444.1029052734375,
+      "learning_rate": 3.085705981836423e-07,
+      "loss": 81.4259,
+      "step": 111230
+    },
+    {
+      "epoch": 0.9202134259833726,
+      "grad_norm": 508.9635314941406,
+      "learning_rate": 3.080829943198277e-07,
+      "loss": 89.5248,
+      "step": 111240
+    },
+    {
+      "epoch": 0.9202961492327418,
+      "grad_norm": 996.2418212890625,
+      "learning_rate": 3.0759576377031697e-07,
+      "loss": 69.8985,
+      "step": 111250
+    },
+    {
+      "epoch": 0.9203788724821111,
+      "grad_norm": 1550.9212646484375,
+      "learning_rate": 3.071089065738747e-07,
+      "loss": 69.4487,
+      "step": 111260
+    },
+    {
+      "epoch": 0.9204615957314803,
+      "grad_norm": 1241.609130859375,
+      "learning_rate": 3.0662242276923993e-07,
+      "loss": 112.2663,
+      "step": 111270
+    },
+    {
+      "epoch": 0.9205443189808495,
+      "grad_norm": 1453.5501708984375,
+      "learning_rate": 3.061363123951189e-07,
+      "loss": 87.4852,
+      "step": 111280
+    },
+    {
+      "epoch": 0.9206270422302189,
+      "grad_norm": 792.2757568359375,
+      "learning_rate": 3.0565057549019005e-07,
+      "loss": 89.3375,
+      "step": 111290
+    },
+    {
+      "epoch": 0.920709765479588,
+      "grad_norm": 365.10528564453125,
+      "learning_rate": 3.051652120931003e-07,
+      "loss": 77.1843,
+      "step": 111300
+    },
+    {
+      "epoch": 0.9207924887289572,
+      "grad_norm": 1038.219482421875,
+      "learning_rate": 3.0468022224246886e-07,
+      "loss": 100.4181,
+      "step": 111310
+    },
+    {
+      "epoch": 0.9208752119783266,
+      "grad_norm": 1549.9176025390625,
+      "learning_rate": 3.04195605976883e-07,
+      "loss": 101.2689,
+      "step": 111320
+    },
+    {
+      "epoch": 0.9209579352276958,
+      "grad_norm": 1094.44873046875,
+      "learning_rate": 3.0371136333490315e-07,
+      "loss": 62.7427,
+      "step": 111330
+    },
+    {
+      "epoch": 0.921040658477065,
+      "grad_norm": 764.5133056640625,
+      "learning_rate": 3.0322749435505563e-07,
+      "loss": 117.9781,
+      "step": 111340
+    },
+    {
+      "epoch": 0.9211233817264343,
+      "grad_norm": 922.7494506835938,
+      "learning_rate": 3.027439990758418e-07,
+      "loss": 87.756,
+      "step": 111350
+    },
+    {
+      "epoch": 0.9212061049758035,
+      "grad_norm": 2155.5322265625,
+      "learning_rate": 3.022608775357294e-07,
+      "loss": 109.2952,
+      "step": 111360
+    },
+    {
+      "epoch": 0.9212888282251727,
+      "grad_norm": 639.3640747070312,
+      "learning_rate": 3.017781297731598e-07,
+      "loss": 65.4102,
+      "step": 111370
+    },
+    {
+      "epoch": 0.921371551474542,
+      "grad_norm": 401.3136901855469,
+      "learning_rate": 3.012957558265428e-07,
+      "loss": 55.7189,
+      "step": 111380
+    },
+    {
+      "epoch": 0.9214542747239112,
+      "grad_norm": 815.2987060546875,
+      "learning_rate": 3.008137557342583e-07,
+      "loss": 89.766,
+      "step": 111390
+    },
+    {
+      "epoch": 0.9215369979732804,
+      "grad_norm": 848.1900024414062,
+      "learning_rate": 3.003321295346573e-07,
+      "loss": 70.3423,
+      "step": 111400
+    },
+    {
+      "epoch": 0.9216197212226497,
+      "grad_norm": 1112.6951904296875,
+      "learning_rate": 2.9985087726605965e-07,
+      "loss": 96.3883,
+      "step": 111410
+    },
+    {
+      "epoch": 0.9217024444720189,
+      "grad_norm": 863.7384643554688,
+      "learning_rate": 2.9936999896675757e-07,
+      "loss": 77.8842,
+      "step": 111420
+    },
+    {
+      "epoch": 0.9217851677213881,
+      "grad_norm": 549.550537109375,
+      "learning_rate": 2.988894946750115e-07,
+      "loss": 97.5944,
+      "step": 111430
+    },
+    {
+      "epoch": 0.9218678909707574,
+      "grad_norm": 348.6497497558594,
+      "learning_rate": 2.9840936442905253e-07,
+      "loss": 81.6903,
+      "step": 111440
+    },
+    {
+      "epoch": 0.9219506142201266,
+      "grad_norm": 1831.6419677734375,
+      "learning_rate": 2.979296082670835e-07,
+      "loss": 112.7743,
+      "step": 111450
+    },
+    {
+      "epoch": 0.9220333374694958,
+      "grad_norm": 490.64111328125,
+      "learning_rate": 2.974502262272749e-07,
+      "loss": 80.2675,
+      "step": 111460
+    },
+    {
+      "epoch": 0.922116060718865,
+      "grad_norm": 768.78076171875,
+      "learning_rate": 2.969712183477713e-07,
+      "loss": 74.6032,
+      "step": 111470
+    },
+    {
+      "epoch": 0.9221987839682343,
+      "grad_norm": 886.2664184570312,
+      "learning_rate": 2.964925846666833e-07,
+      "loss": 52.6146,
+      "step": 111480
+    },
+    {
+      "epoch": 0.9222815072176035,
+      "grad_norm": 648.3212280273438,
+      "learning_rate": 2.960143252220943e-07,
+      "loss": 88.3323,
+      "step": 111490
+    },
+    {
+      "epoch": 0.9223642304669727,
+      "grad_norm": 646.9230346679688,
+      "learning_rate": 2.955364400520583e-07,
+      "loss": 85.1518,
+      "step": 111500
+    },
+    {
+      "epoch": 0.922446953716342,
+      "grad_norm": 783.1369018554688,
+      "learning_rate": 2.950589291945954e-07,
+      "loss": 83.9216,
+      "step": 111510
+    },
+    {
+      "epoch": 0.9225296769657112,
+      "grad_norm": 624.5678100585938,
+      "learning_rate": 2.9458179268770147e-07,
+      "loss": 67.9207,
+      "step": 111520
+    },
+    {
+      "epoch": 0.9226124002150804,
+      "grad_norm": 933.9120483398438,
+      "learning_rate": 2.941050305693394e-07,
+      "loss": 96.648,
+      "step": 111530
+    },
+    {
+      "epoch": 0.9226951234644497,
+      "grad_norm": 759.1103515625,
+      "learning_rate": 2.9362864287744266e-07,
+      "loss": 79.1249,
+      "step": 111540
+    },
+    {
+      "epoch": 0.9227778467138189,
+      "grad_norm": 709.6221923828125,
+      "learning_rate": 2.931526296499154e-07,
+      "loss": 75.1609,
+      "step": 111550
+    },
+    {
+      "epoch": 0.9228605699631881,
+      "grad_norm": 1090.783203125,
+      "learning_rate": 2.926769909246313e-07,
+      "loss": 90.4893,
+      "step": 111560
+    },
+    {
+      "epoch": 0.9229432932125574,
+      "grad_norm": 1291.6949462890625,
+      "learning_rate": 2.9220172673943615e-07,
+      "loss": 121.0514,
+      "step": 111570
+    },
+    {
+      "epoch": 0.9230260164619266,
+      "grad_norm": 1319.6103515625,
+      "learning_rate": 2.9172683713214354e-07,
+      "loss": 81.1168,
+      "step": 111580
+    },
+    {
+      "epoch": 0.9231087397112958,
+      "grad_norm": 887.0220947265625,
+      "learning_rate": 2.9125232214053946e-07,
+      "loss": 77.7259,
+      "step": 111590
+    },
+    {
+      "epoch": 0.9231914629606651,
+      "grad_norm": 823.8740234375,
+      "learning_rate": 2.9077818180237693e-07,
+      "loss": 83.7634,
+      "step": 111600
+    },
+    {
+      "epoch": 0.9232741862100343,
+      "grad_norm": 1067.777099609375,
+      "learning_rate": 2.90304416155382e-07,
+      "loss": 77.3156,
+      "step": 111610
+    },
+    {
+      "epoch": 0.9233569094594035,
+      "grad_norm": 533.4111328125,
+      "learning_rate": 2.898310252372505e-07,
+      "loss": 81.9363,
+      "step": 111620
+    },
+    {
+      "epoch": 0.9234396327087728,
+      "grad_norm": 800.60888671875,
+      "learning_rate": 2.893580090856474e-07,
+      "loss": 112.345,
+      "step": 111630
+    },
+    {
+      "epoch": 0.923522355958142,
+      "grad_norm": 962.5048217773438,
+      "learning_rate": 2.888853677382081e-07,
+      "loss": 88.3361,
+      "step": 111640
+    },
+    {
+      "epoch": 0.9236050792075112,
+      "grad_norm": 889.5928344726562,
+      "learning_rate": 2.8841310123253865e-07,
+      "loss": 81.9884,
+      "step": 111650
+    },
+    {
+      "epoch": 0.9236878024568805,
+      "grad_norm": 697.1818237304688,
+      "learning_rate": 2.879412096062162e-07,
+      "loss": 121.3172,
+      "step": 111660
+    },
+    {
+      "epoch": 0.9237705257062497,
+      "grad_norm": 667.9248657226562,
+      "learning_rate": 2.874696928967863e-07,
+      "loss": 97.8055,
+      "step": 111670
+    },
+    {
+      "epoch": 0.9238532489556189,
+      "grad_norm": 442.2999572753906,
+      "learning_rate": 2.869985511417661e-07,
+      "loss": 54.4177,
+      "step": 111680
+    },
+    {
+      "epoch": 0.9239359722049882,
+      "grad_norm": 686.0079956054688,
+      "learning_rate": 2.8652778437864016e-07,
+      "loss": 87.2172,
+      "step": 111690
+    },
+    {
+      "epoch": 0.9240186954543574,
+      "grad_norm": 684.615966796875,
+      "learning_rate": 2.8605739264486733e-07,
+      "loss": 89.1697,
+      "step": 111700
+    },
+    {
+      "epoch": 0.9241014187037266,
+      "grad_norm": 1551.7012939453125,
+      "learning_rate": 2.8558737597787324e-07,
+      "loss": 115.9352,
+      "step": 111710
+    },
+    {
+      "epoch": 0.924184141953096,
+      "grad_norm": 1213.67236328125,
+      "learning_rate": 2.851177344150552e-07,
+      "loss": 97.0064,
+      "step": 111720
+    },
+    {
+      "epoch": 0.9242668652024651,
+      "grad_norm": 768.0172729492188,
+      "learning_rate": 2.846484679937811e-07,
+      "loss": 81.57,
+      "step": 111730
+    },
+    {
+      "epoch": 0.9243495884518343,
+      "grad_norm": 486.9347229003906,
+      "learning_rate": 2.841795767513877e-07,
+      "loss": 72.622,
+      "step": 111740
+    },
+    {
+      "epoch": 0.9244323117012037,
+      "grad_norm": 799.0153198242188,
+      "learning_rate": 2.8371106072518194e-07,
+      "loss": 95.9192,
+      "step": 111750
+    },
+    {
+      "epoch": 0.9245150349505729,
+      "grad_norm": 778.710205078125,
+      "learning_rate": 2.8324291995244333e-07,
+      "loss": 74.0421,
+      "step": 111760
+    },
+    {
+      "epoch": 0.924597758199942,
+      "grad_norm": 1426.642822265625,
+      "learning_rate": 2.8277515447041827e-07,
+      "loss": 101.7019,
+      "step": 111770
+    },
+    {
+      "epoch": 0.9246804814493114,
+      "grad_norm": 934.9050903320312,
+      "learning_rate": 2.8230776431632523e-07,
+      "loss": 93.3299,
+      "step": 111780
+    },
+    {
+      "epoch": 0.9247632046986806,
+      "grad_norm": 839.4944458007812,
+      "learning_rate": 2.8184074952735176e-07,
+      "loss": 76.1551,
+      "step": 111790
+    },
+    {
+      "epoch": 0.9248459279480498,
+      "grad_norm": 733.8991088867188,
+      "learning_rate": 2.813741101406564e-07,
+      "loss": 83.5848,
+      "step": 111800
+    },
+    {
+      "epoch": 0.9249286511974191,
+      "grad_norm": 1104.2376708984375,
+      "learning_rate": 2.8090784619336784e-07,
+      "loss": 109.4974,
+      "step": 111810
+    },
+    {
+      "epoch": 0.9250113744467883,
+      "grad_norm": 541.585205078125,
+      "learning_rate": 2.804419577225842e-07,
+      "loss": 99.5181,
+      "step": 111820
+    },
+    {
+      "epoch": 0.9250940976961575,
+      "grad_norm": 1171.6307373046875,
+      "learning_rate": 2.7997644476537353e-07,
+      "loss": 105.5711,
+      "step": 111830
+    },
+    {
+      "epoch": 0.9251768209455268,
+      "grad_norm": 1111.163818359375,
+      "learning_rate": 2.7951130735877576e-07,
+      "loss": 100.552,
+      "step": 111840
+    },
+    {
+      "epoch": 0.925259544194896,
+      "grad_norm": 465.4991149902344,
+      "learning_rate": 2.790465455397989e-07,
+      "loss": 91.685,
+      "step": 111850
+    },
+    {
+      "epoch": 0.9253422674442652,
+      "grad_norm": 768.08740234375,
+      "learning_rate": 2.785821593454224e-07,
+      "loss": 105.3449,
+      "step": 111860
+    },
+    {
+      "epoch": 0.9254249906936345,
+      "grad_norm": 497.09466552734375,
+      "learning_rate": 2.7811814881259503e-07,
+      "loss": 67.444,
+      "step": 111870
+    },
+    {
+      "epoch": 0.9255077139430037,
+      "grad_norm": 693.8814697265625,
+      "learning_rate": 2.776545139782361e-07,
+      "loss": 61.2311,
+      "step": 111880
+    },
+    {
+      "epoch": 0.9255904371923729,
+      "grad_norm": 892.4473266601562,
+      "learning_rate": 2.771912548792344e-07,
+      "loss": 87.7277,
+      "step": 111890
+    },
+    {
+      "epoch": 0.9256731604417422,
+      "grad_norm": 1138.6724853515625,
+      "learning_rate": 2.767283715524505e-07,
+      "loss": 122.5125,
+      "step": 111900
+    },
+    {
+      "epoch": 0.9257558836911114,
+      "grad_norm": 755.6520385742188,
+      "learning_rate": 2.762658640347127e-07,
+      "loss": 69.8308,
+      "step": 111910
+    },
+    {
+      "epoch": 0.9258386069404806,
+      "grad_norm": 983.39990234375,
+      "learning_rate": 2.75803732362821e-07,
+      "loss": 58.8145,
+      "step": 111920
+    },
+    {
+      "epoch": 0.9259213301898499,
+      "grad_norm": 895.5469360351562,
+      "learning_rate": 2.7534197657354543e-07,
+      "loss": 91.3707,
+      "step": 111930
+    },
+    {
+      "epoch": 0.9260040534392191,
+      "grad_norm": 1143.3978271484375,
+      "learning_rate": 2.7488059670362553e-07,
+      "loss": 80.0223,
+      "step": 111940
+    },
+    {
+      "epoch": 0.9260867766885883,
+      "grad_norm": 666.8565673828125,
+      "learning_rate": 2.744195927897708e-07,
+      "loss": 119.4018,
+      "step": 111950
+    },
+    {
+      "epoch": 0.9261694999379576,
+      "grad_norm": 1519.14404296875,
+      "learning_rate": 2.7395896486866193e-07,
+      "loss": 69.0608,
+      "step": 111960
+    },
+    {
+      "epoch": 0.9262522231873268,
+      "grad_norm": 1167.140625,
+      "learning_rate": 2.7349871297694895e-07,
+      "loss": 95.0523,
+      "step": 111970
+    },
+    {
+      "epoch": 0.926334946436696,
+      "grad_norm": 715.0806274414062,
+      "learning_rate": 2.73038837151251e-07,
+      "loss": 73.4228,
+      "step": 111980
+    },
+    {
+      "epoch": 0.9264176696860653,
+      "grad_norm": 602.8383178710938,
+      "learning_rate": 2.725793374281593e-07,
+      "loss": 103.7122,
+      "step": 111990
+    },
+    {
+      "epoch": 0.9265003929354345,
+      "grad_norm": 575.5548706054688,
+      "learning_rate": 2.7212021384423415e-07,
+      "loss": 59.8035,
+      "step": 112000
+    },
+    {
+      "epoch": 0.9265831161848037,
+      "grad_norm": 687.127197265625,
+      "learning_rate": 2.716614664360051e-07,
+      "loss": 53.2006,
+      "step": 112010
+    },
+    {
+      "epoch": 0.926665839434173,
+      "grad_norm": 912.4276123046875,
+      "learning_rate": 2.71203095239973e-07,
+      "loss": 71.3291,
+      "step": 112020
+    },
+    {
+      "epoch": 0.9267485626835422,
+      "grad_norm": 720.695068359375,
+      "learning_rate": 2.7074510029260814e-07,
+      "loss": 92.2435,
+      "step": 112030
+    },
+    {
+      "epoch": 0.9268312859329114,
+      "grad_norm": 881.0599975585938,
+      "learning_rate": 2.702874816303519e-07,
+      "loss": 83.2978,
+      "step": 112040
+    },
+    {
+      "epoch": 0.9269140091822807,
+      "grad_norm": 1031.5977783203125,
+      "learning_rate": 2.6983023928961406e-07,
+      "loss": 91.2726,
+      "step": 112050
+    },
+    {
+      "epoch": 0.9269967324316499,
+      "grad_norm": 976.7826538085938,
+      "learning_rate": 2.6937337330677547e-07,
+      "loss": 74.995,
+      "step": 112060
+    },
+    {
+      "epoch": 0.9270794556810191,
+      "grad_norm": 916.8348388671875,
+      "learning_rate": 2.689168837181866e-07,
+      "loss": 101.2523,
+      "step": 112070
+    },
+    {
+      "epoch": 0.9271621789303884,
+      "grad_norm": 889.0336303710938,
+      "learning_rate": 2.684607705601688e-07,
+      "loss": 84.4417,
+      "step": 112080
+    },
+    {
+      "epoch": 0.9272449021797576,
+      "grad_norm": 1171.827392578125,
+      "learning_rate": 2.680050338690132e-07,
+      "loss": 133.0743,
+      "step": 112090
+    },
+    {
+      "epoch": 0.9273276254291268,
+      "grad_norm": 868.6085205078125,
+      "learning_rate": 2.675496736809796e-07,
+      "loss": 84.3478,
+      "step": 112100
+    },
+    {
+      "epoch": 0.9274103486784961,
+      "grad_norm": 571.0438842773438,
+      "learning_rate": 2.6709469003230013e-07,
+      "loss": 91.2716,
+      "step": 112110
+    },
+    {
+      "epoch": 0.9274930719278653,
+      "grad_norm": 457.09393310546875,
+      "learning_rate": 2.6664008295917364e-07,
+      "loss": 72.2116,
+      "step": 112120
+    },
+    {
+      "epoch": 0.9275757951772345,
+      "grad_norm": 1358.0347900390625,
+      "learning_rate": 2.661858524977734e-07,
+      "loss": 93.3441,
+      "step": 112130
+    },
+    {
+      "epoch": 0.9276585184266039,
+      "grad_norm": 667.2669067382812,
+      "learning_rate": 2.657319986842394e-07,
+      "loss": 79.6386,
+      "step": 112140
+    },
+    {
+      "epoch": 0.927741241675973,
+      "grad_norm": 1060.2071533203125,
+      "learning_rate": 2.6527852155468327e-07,
+      "loss": 94.3927,
+      "step": 112150
+    },
+    {
+      "epoch": 0.9278239649253422,
+      "grad_norm": 1009.9804077148438,
+      "learning_rate": 2.648254211451856e-07,
+      "loss": 63.4351,
+      "step": 112160
+    },
+    {
+      "epoch": 0.9279066881747116,
+      "grad_norm": 910.8653564453125,
+      "learning_rate": 2.643726974917976e-07,
+      "loss": 94.569,
+      "step": 112170
+    },
+    {
+      "epoch": 0.9279894114240808,
+      "grad_norm": 451.7293701171875,
+      "learning_rate": 2.6392035063054043e-07,
+      "loss": 98.4698,
+      "step": 112180
+    },
+    {
+      "epoch": 0.92807213467345,
+      "grad_norm": 976.0929565429688,
+      "learning_rate": 2.634683805974059e-07,
+      "loss": 69.7292,
+      "step": 112190
+    },
+    {
+      "epoch": 0.9281548579228192,
+      "grad_norm": 999.5797119140625,
+      "learning_rate": 2.63016787428354e-07,
+      "loss": 100.5,
+      "step": 112200
+    },
+    {
+      "epoch": 0.9282375811721885,
+      "grad_norm": 620.88623046875,
+      "learning_rate": 2.6256557115931613e-07,
+      "loss": 76.7702,
+      "step": 112210
+    },
+    {
+      "epoch": 0.9283203044215577,
+      "grad_norm": 725.8551635742188,
+      "learning_rate": 2.6211473182619405e-07,
+      "loss": 74.561,
+      "step": 112220
+    },
+    {
+      "epoch": 0.9284030276709269,
+      "grad_norm": 1008.1248168945312,
+      "learning_rate": 2.616642694648591e-07,
+      "loss": 58.9289,
+      "step": 112230
+    },
+    {
+      "epoch": 0.9284857509202962,
+      "grad_norm": 1029.631591796875,
+      "learning_rate": 2.612141841111521e-07,
+      "loss": 83.1139,
+      "step": 112240
+    },
+    {
+      "epoch": 0.9285684741696654,
+      "grad_norm": 921.6142578125,
+      "learning_rate": 2.6076447580088426e-07,
+      "loss": 60.3248,
+      "step": 112250
+    },
+    {
+      "epoch": 0.9286511974190346,
+      "grad_norm": 684.411376953125,
+      "learning_rate": 2.603151445698371e-07,
+      "loss": 82.2432,
+      "step": 112260
+    },
+    {
+      "epoch": 0.9287339206684039,
+      "grad_norm": 1127.3673095703125,
+      "learning_rate": 2.598661904537619e-07,
+      "loss": 102.584,
+      "step": 112270
+    },
+    {
+      "epoch": 0.9288166439177731,
+      "grad_norm": 949.6649780273438,
+      "learning_rate": 2.5941761348837966e-07,
+      "loss": 110.4898,
+      "step": 112280
+    },
+    {
+      "epoch": 0.9288993671671423,
+      "grad_norm": 1033.648193359375,
+      "learning_rate": 2.5896941370938177e-07,
+      "loss": 92.9099,
+      "step": 112290
+    },
+    {
+      "epoch": 0.9289820904165116,
+      "grad_norm": 798.4220581054688,
+      "learning_rate": 2.5852159115242857e-07,
+      "loss": 71.5822,
+      "step": 112300
+    },
+    {
+      "epoch": 0.9290648136658808,
+      "grad_norm": 766.9046630859375,
+      "learning_rate": 2.580741458531516e-07,
+      "loss": 79.5017,
+      "step": 112310
+    },
+    {
+      "epoch": 0.92914753691525,
+      "grad_norm": 547.7767944335938,
+      "learning_rate": 2.576270778471529e-07,
+      "loss": 78.4615,
+      "step": 112320
+    },
+    {
+      "epoch": 0.9292302601646193,
+      "grad_norm": 636.5359497070312,
+      "learning_rate": 2.571803871700029e-07,
+      "loss": 64.6448,
+      "step": 112330
+    },
+    {
+      "epoch": 0.9293129834139885,
+      "grad_norm": 1199.5556640625,
+      "learning_rate": 2.567340738572427e-07,
+      "loss": 102.5773,
+      "step": 112340
+    },
+    {
+      "epoch": 0.9293957066633577,
+      "grad_norm": 1170.7523193359375,
+      "learning_rate": 2.5628813794438434e-07,
+      "loss": 80.3686,
+      "step": 112350
+    },
+    {
+      "epoch": 0.929478429912727,
+      "grad_norm": 1347.653564453125,
+      "learning_rate": 2.5584257946690836e-07,
+      "loss": 81.7604,
+      "step": 112360
+    },
+    {
+      "epoch": 0.9295611531620962,
+      "grad_norm": 786.822509765625,
+      "learning_rate": 2.553973984602648e-07,
+      "loss": 84.814,
+      "step": 112370
+    },
+    {
+      "epoch": 0.9296438764114654,
+      "grad_norm": 1360.4071044921875,
+      "learning_rate": 2.5495259495987576e-07,
+      "loss": 107.1304,
+      "step": 112380
+    },
+    {
+      "epoch": 0.9297265996608347,
+      "grad_norm": 1530.94873046875,
+      "learning_rate": 2.545081690011314e-07,
+      "loss": 98.4278,
+      "step": 112390
+    },
+    {
+      "epoch": 0.9298093229102039,
+      "grad_norm": 977.4533081054688,
+      "learning_rate": 2.540641206193939e-07,
+      "loss": 78.7725,
+      "step": 112400
+    },
+    {
+      "epoch": 0.9298920461595731,
+      "grad_norm": 600.151611328125,
+      "learning_rate": 2.536204498499922e-07,
+      "loss": 96.0168,
+      "step": 112410
+    },
+    {
+      "epoch": 0.9299747694089424,
+      "grad_norm": 640.1029052734375,
+      "learning_rate": 2.5317715672822927e-07,
+      "loss": 85.9019,
+      "step": 112420
+    },
+    {
+      "epoch": 0.9300574926583116,
+      "grad_norm": 747.0780029296875,
+      "learning_rate": 2.527342412893746e-07,
+      "loss": 62.0823,
+      "step": 112430
+    },
+    {
+      "epoch": 0.9301402159076808,
+      "grad_norm": 1375.25146484375,
+      "learning_rate": 2.5229170356867005e-07,
+      "loss": 96.1561,
+      "step": 112440
+    },
+    {
+      "epoch": 0.9302229391570501,
+      "grad_norm": 782.2132568359375,
+      "learning_rate": 2.5184954360132574e-07,
+      "loss": 89.7107,
+      "step": 112450
+    },
+    {
+      "epoch": 0.9303056624064193,
+      "grad_norm": 2209.59521484375,
+      "learning_rate": 2.514077614225219e-07,
+      "loss": 105.6186,
+      "step": 112460
+    },
+    {
+      "epoch": 0.9303883856557885,
+      "grad_norm": 583.348388671875,
+      "learning_rate": 2.509663570674087e-07,
+      "loss": 80.0686,
+      "step": 112470
+    },
+    {
+      "epoch": 0.9304711089051578,
+      "grad_norm": 459.78302001953125,
+      "learning_rate": 2.5052533057110805e-07,
+      "loss": 93.2552,
+      "step": 112480
+    },
+    {
+      "epoch": 0.930553832154527,
+      "grad_norm": 617.0430908203125,
+      "learning_rate": 2.500846819687097e-07,
+      "loss": 93.8821,
+      "step": 112490
+    },
+    {
+      "epoch": 0.9306365554038962,
+      "grad_norm": 0.0,
+      "learning_rate": 2.4964441129527337e-07,
+      "loss": 83.9168,
+      "step": 112500
+    },
+    {
+      "epoch": 0.9307192786532655,
+      "grad_norm": 957.689453125,
+      "learning_rate": 2.4920451858582997e-07,
+      "loss": 81.6168,
+      "step": 112510
+    },
+    {
+      "epoch": 0.9308020019026347,
+      "grad_norm": 1347.580810546875,
+      "learning_rate": 2.487650038753803e-07,
+      "loss": 84.5013,
+      "step": 112520
+    },
+    {
+      "epoch": 0.9308847251520039,
+      "grad_norm": 840.2437744140625,
+      "learning_rate": 2.483258671988942e-07,
+      "loss": 90.0643,
+      "step": 112530
+    },
+    {
+      "epoch": 0.9309674484013732,
+      "grad_norm": 803.8649291992188,
+      "learning_rate": 2.4788710859131203e-07,
+      "loss": 95.1621,
+      "step": 112540
+    },
+    {
+      "epoch": 0.9310501716507424,
+      "grad_norm": 775.64599609375,
+      "learning_rate": 2.474487280875426e-07,
+      "loss": 82.0533,
+      "step": 112550
+    },
+    {
+      "epoch": 0.9311328949001116,
+      "grad_norm": 648.1033325195312,
+      "learning_rate": 2.470107257224669e-07,
+      "loss": 69.88,
+      "step": 112560
+    },
+    {
+      "epoch": 0.931215618149481,
+      "grad_norm": 769.5602416992188,
+      "learning_rate": 2.4657310153093475e-07,
+      "loss": 101.4659,
+      "step": 112570
+    },
+    {
+      "epoch": 0.9312983413988501,
+      "grad_norm": 654.4227905273438,
+      "learning_rate": 2.46135855547765e-07,
+      "loss": 79.7583,
+      "step": 112580
+    },
+    {
+      "epoch": 0.9313810646482193,
+      "grad_norm": 909.7418823242188,
+      "learning_rate": 2.4569898780774816e-07,
+      "loss": 77.9152,
+      "step": 112590
+    },
+    {
+      "epoch": 0.9314637878975887,
+      "grad_norm": 475.4342041015625,
+      "learning_rate": 2.452624983456431e-07,
+      "loss": 87.7683,
+      "step": 112600
+    },
+    {
+      "epoch": 0.9315465111469579,
+      "grad_norm": 750.6339111328125,
+      "learning_rate": 2.4482638719618037e-07,
+      "loss": 95.7583,
+      "step": 112610
+    },
+    {
+      "epoch": 0.931629234396327,
+      "grad_norm": 719.8361206054688,
+      "learning_rate": 2.443906543940594e-07,
+      "loss": 89.19,
+      "step": 112620
+    },
+    {
+      "epoch": 0.9317119576456964,
+      "grad_norm": 788.4232788085938,
+      "learning_rate": 2.439552999739475e-07,
+      "loss": 64.0623,
+      "step": 112630
+    },
+    {
+      "epoch": 0.9317946808950656,
+      "grad_norm": 1180.428955078125,
+      "learning_rate": 2.4352032397048584e-07,
+      "loss": 77.4437,
+      "step": 112640
+    },
+    {
+      "epoch": 0.9318774041444348,
+      "grad_norm": 702.971923828125,
+      "learning_rate": 2.4308572641828234e-07,
+      "loss": 86.0467,
+      "step": 112650
+    },
+    {
+      "epoch": 0.9319601273938041,
+      "grad_norm": 467.1812438964844,
+      "learning_rate": 2.4265150735191644e-07,
+      "loss": 76.0462,
+      "step": 112660
+    },
+    {
+      "epoch": 0.9320428506431733,
+      "grad_norm": 1010.5810546875,
+      "learning_rate": 2.422176668059367e-07,
+      "loss": 71.2109,
+      "step": 112670
+    },
+    {
+      "epoch": 0.9321255738925425,
+      "grad_norm": 1030.048583984375,
+      "learning_rate": 2.417842048148622e-07,
+      "loss": 98.5793,
+      "step": 112680
+    },
+    {
+      "epoch": 0.9322082971419118,
+      "grad_norm": 608.4972534179688,
+      "learning_rate": 2.4135112141318084e-07,
+      "loss": 86.8852,
+      "step": 112690
+    },
+    {
+      "epoch": 0.932291020391281,
+      "grad_norm": 1069.7056884765625,
+      "learning_rate": 2.409184166353512e-07,
+      "loss": 73.6224,
+      "step": 112700
+    },
+    {
+      "epoch": 0.9323737436406502,
+      "grad_norm": 620.8689575195312,
+      "learning_rate": 2.4048609051580295e-07,
+      "loss": 73.3166,
+      "step": 112710
+    },
+    {
+      "epoch": 0.9324564668900195,
+      "grad_norm": 1390.774169921875,
+      "learning_rate": 2.4005414308893304e-07,
+      "loss": 93.2582,
+      "step": 112720
+    },
+    {
+      "epoch": 0.9325391901393887,
+      "grad_norm": 1095.233642578125,
+      "learning_rate": 2.396225743891095e-07,
+      "loss": 79.1661,
+      "step": 112730
+    },
+    {
+      "epoch": 0.9326219133887579,
+      "grad_norm": 651.0794677734375,
+      "learning_rate": 2.3919138445067045e-07,
+      "loss": 74.3248,
+      "step": 112740
+    },
+    {
+      "epoch": 0.9327046366381272,
+      "grad_norm": 694.2014770507812,
+      "learning_rate": 2.3876057330792344e-07,
+      "loss": 72.9786,
+      "step": 112750
+    },
+    {
+      "epoch": 0.9327873598874964,
+      "grad_norm": 845.7319946289062,
+      "learning_rate": 2.3833014099514716e-07,
+      "loss": 67.2363,
+      "step": 112760
+    },
+    {
+      "epoch": 0.9328700831368656,
+      "grad_norm": 579.4317016601562,
+      "learning_rate": 2.3790008754658811e-07,
+      "loss": 75.9026,
+      "step": 112770
+    },
+    {
+      "epoch": 0.9329528063862349,
+      "grad_norm": 1336.7401123046875,
+      "learning_rate": 2.3747041299646389e-07,
+      "loss": 100.3919,
+      "step": 112780
+    },
+    {
+      "epoch": 0.9330355296356041,
+      "grad_norm": 782.5462646484375,
+      "learning_rate": 2.3704111737896218e-07,
+      "loss": 66.6842,
+      "step": 112790
+    },
+    {
+      "epoch": 0.9331182528849733,
+      "grad_norm": 1084.5948486328125,
+      "learning_rate": 2.3661220072823953e-07,
+      "loss": 86.8441,
+      "step": 112800
+    },
+    {
+      "epoch": 0.9332009761343426,
+      "grad_norm": 873.84228515625,
+      "learning_rate": 2.3618366307842312e-07,
+      "loss": 80.4352,
+      "step": 112810
+    },
+    {
+      "epoch": 0.9332836993837118,
+      "grad_norm": 935.0689697265625,
+      "learning_rate": 2.3575550446360952e-07,
+      "loss": 74.7977,
+      "step": 112820
+    },
+    {
+      "epoch": 0.933366422633081,
+      "grad_norm": 1264.7552490234375,
+      "learning_rate": 2.3532772491786537e-07,
+      "loss": 62.4919,
+      "step": 112830
+    },
+    {
+      "epoch": 0.9334491458824503,
+      "grad_norm": 760.6544799804688,
+      "learning_rate": 2.3490032447522792e-07,
+      "loss": 78.2148,
+      "step": 112840
+    },
+    {
+      "epoch": 0.9335318691318195,
+      "grad_norm": 3588.122314453125,
+      "learning_rate": 2.3447330316970218e-07,
+      "loss": 120.6314,
+      "step": 112850
+    },
+    {
+      "epoch": 0.9336145923811887,
+      "grad_norm": 653.4447021484375,
+      "learning_rate": 2.3404666103526542e-07,
+      "loss": 97.4798,
+      "step": 112860
+    },
+    {
+      "epoch": 0.933697315630558,
+      "grad_norm": 705.934814453125,
+      "learning_rate": 2.3362039810586267e-07,
+      "loss": 94.1974,
+      "step": 112870
+    },
+    {
+      "epoch": 0.9337800388799272,
+      "grad_norm": 616.7807006835938,
+      "learning_rate": 2.3319451441541018e-07,
+      "loss": 67.858,
+      "step": 112880
+    },
+    {
+      "epoch": 0.9338627621292964,
+      "grad_norm": 934.6239624023438,
+      "learning_rate": 2.3276900999779305e-07,
+      "loss": 85.3062,
+      "step": 112890
+    },
+    {
+      "epoch": 0.9339454853786656,
+      "grad_norm": 563.1575317382812,
+      "learning_rate": 2.323438848868681e-07,
+      "loss": 75.0296,
+      "step": 112900
+    },
+    {
+      "epoch": 0.9340282086280349,
+      "grad_norm": 1033.3599853515625,
+      "learning_rate": 2.319191391164588e-07,
+      "loss": 86.1594,
+      "step": 112910
+    },
+    {
+      "epoch": 0.9341109318774041,
+      "grad_norm": 551.6663818359375,
+      "learning_rate": 2.3149477272036146e-07,
+      "loss": 72.7008,
+      "step": 112920
+    },
+    {
+      "epoch": 0.9341936551267733,
+      "grad_norm": 884.0999755859375,
+      "learning_rate": 2.3107078573234077e-07,
+      "loss": 88.6581,
+      "step": 112930
+    },
+    {
+      "epoch": 0.9342763783761426,
+      "grad_norm": 766.043701171875,
+      "learning_rate": 2.306471781861308e-07,
+      "loss": 75.3607,
+      "step": 112940
+    },
+    {
+      "epoch": 0.9343591016255118,
+      "grad_norm": 6711.1845703125,
+      "learning_rate": 2.3022395011543687e-07,
+      "loss": 123.3523,
+      "step": 112950
+    },
+    {
+      "epoch": 0.934441824874881,
+      "grad_norm": 771.6279296875,
+      "learning_rate": 2.2980110155393253e-07,
+      "loss": 97.1575,
+      "step": 112960
+    },
+    {
+      "epoch": 0.9345245481242503,
+      "grad_norm": 1057.9940185546875,
+      "learning_rate": 2.293786325352626e-07,
+      "loss": 113.2081,
+      "step": 112970
+    },
+    {
+      "epoch": 0.9346072713736195,
+      "grad_norm": 1071.353271484375,
+      "learning_rate": 2.289565430930407e-07,
+      "loss": 87.6715,
+      "step": 112980
+    },
+    {
+      "epoch": 0.9346899946229887,
+      "grad_norm": 862.3544921875,
+      "learning_rate": 2.2853483326085002e-07,
+      "loss": 79.9122,
+      "step": 112990
+    },
+    {
+      "epoch": 0.934772717872358,
+      "grad_norm": 895.1743774414062,
+      "learning_rate": 2.2811350307224534e-07,
+      "loss": 113.0911,
+      "step": 113000
+    },
+    {
+      "epoch": 0.9348554411217272,
+      "grad_norm": 753.2420654296875,
+      "learning_rate": 2.2769255256074874e-07,
+      "loss": 84.552,
+      "step": 113010
+    },
+    {
+      "epoch": 0.9349381643710964,
+      "grad_norm": 1342.393798828125,
+      "learning_rate": 2.27271981759854e-07,
+      "loss": 97.5154,
+      "step": 113020
+    },
+    {
+      "epoch": 0.9350208876204658,
+      "grad_norm": 1412.8348388671875,
+      "learning_rate": 2.2685179070302377e-07,
+      "loss": 82.4936,
+      "step": 113030
+    },
+    {
+      "epoch": 0.935103610869835,
+      "grad_norm": 668.9525756835938,
+      "learning_rate": 2.2643197942369022e-07,
+      "loss": 94.1694,
+      "step": 113040
+    },
+    {
+      "epoch": 0.9351863341192042,
+      "grad_norm": 772.8831176757812,
+      "learning_rate": 2.2601254795525774e-07,
+      "loss": 81.6293,
+      "step": 113050
+    },
+    {
+      "epoch": 0.9352690573685735,
+      "grad_norm": 1103.180908203125,
+      "learning_rate": 2.2559349633109629e-07,
+      "loss": 83.7612,
+      "step": 113060
+    },
+    {
+      "epoch": 0.9353517806179427,
+      "grad_norm": 819.5826416015625,
+      "learning_rate": 2.2517482458454808e-07,
+      "loss": 88.8579,
+      "step": 113070
+    },
+    {
+      "epoch": 0.9354345038673119,
+      "grad_norm": 2683.924072265625,
+      "learning_rate": 2.2475653274892594e-07,
+      "loss": 102.4598,
+      "step": 113080
+    },
+    {
+      "epoch": 0.9355172271166812,
+      "grad_norm": 808.376220703125,
+      "learning_rate": 2.2433862085751157e-07,
+      "loss": 86.0561,
+      "step": 113090
+    },
+    {
+      "epoch": 0.9355999503660504,
+      "grad_norm": 618.5714111328125,
+      "learning_rate": 2.2392108894355557e-07,
+      "loss": 45.9369,
+      "step": 113100
+    },
+    {
+      "epoch": 0.9356826736154196,
+      "grad_norm": 823.1011962890625,
+      "learning_rate": 2.2350393704027917e-07,
+      "loss": 98.3395,
+      "step": 113110
+    },
+    {
+      "epoch": 0.9357653968647889,
+      "grad_norm": 572.7033081054688,
+      "learning_rate": 2.230871651808736e-07,
+      "loss": 94.4214,
+      "step": 113120
+    },
+    {
+      "epoch": 0.9358481201141581,
+      "grad_norm": 2472.20947265625,
+      "learning_rate": 2.226707733984995e-07,
+      "loss": 115.9153,
+      "step": 113130
+    },
+    {
+      "epoch": 0.9359308433635273,
+      "grad_norm": 665.7183837890625,
+      "learning_rate": 2.2225476172628714e-07,
+      "loss": 60.923,
+      "step": 113140
+    },
+    {
+      "epoch": 0.9360135666128966,
+      "grad_norm": 926.3880615234375,
+      "learning_rate": 2.2183913019733605e-07,
+      "loss": 104.7108,
+      "step": 113150
+    },
+    {
+      "epoch": 0.9360962898622658,
+      "grad_norm": 866.1433715820312,
+      "learning_rate": 2.2142387884471593e-07,
+      "loss": 74.1606,
+      "step": 113160
+    },
+    {
+      "epoch": 0.936179013111635,
+      "grad_norm": 2463.356201171875,
+      "learning_rate": 2.210090077014676e-07,
+      "loss": 73.1271,
+      "step": 113170
+    },
+    {
+      "epoch": 0.9362617363610043,
+      "grad_norm": 2465.159912109375,
+      "learning_rate": 2.2059451680059962e-07,
+      "loss": 136.0056,
+      "step": 113180
+    },
+    {
+      "epoch": 0.9363444596103735,
+      "grad_norm": 1126.7469482421875,
+      "learning_rate": 2.2018040617509174e-07,
+      "loss": 108.5757,
+      "step": 113190
+    },
+    {
+      "epoch": 0.9364271828597427,
+      "grad_norm": 704.0745849609375,
+      "learning_rate": 2.1976667585789257e-07,
+      "loss": 93.8709,
+      "step": 113200
+    },
+    {
+      "epoch": 0.936509906109112,
+      "grad_norm": 623.0414428710938,
+      "learning_rate": 2.193533258819208e-07,
+      "loss": 100.2674,
+      "step": 113210
+    },
+    {
+      "epoch": 0.9365926293584812,
+      "grad_norm": 956.225830078125,
+      "learning_rate": 2.1894035628006517e-07,
+      "loss": 93.8058,
+      "step": 113220
+    },
+    {
+      "epoch": 0.9366753526078504,
+      "grad_norm": 1069.036376953125,
+      "learning_rate": 2.1852776708518265e-07,
+      "loss": 84.0381,
+      "step": 113230
+    },
+    {
+      "epoch": 0.9367580758572197,
+      "grad_norm": 1028.188720703125,
+      "learning_rate": 2.18115558330102e-07,
+      "loss": 62.8961,
+      "step": 113240
+    },
+    {
+      "epoch": 0.9368407991065889,
+      "grad_norm": 656.3768310546875,
+      "learning_rate": 2.1770373004762035e-07,
+      "loss": 98.3898,
+      "step": 113250
+    },
+    {
+      "epoch": 0.9369235223559581,
+      "grad_norm": 1091.6727294921875,
+      "learning_rate": 2.1729228227050426e-07,
+      "loss": 91.2749,
+      "step": 113260
+    },
+    {
+      "epoch": 0.9370062456053274,
+      "grad_norm": 1016.352783203125,
+      "learning_rate": 2.1688121503149195e-07,
+      "loss": 81.7884,
+      "step": 113270
+    },
+    {
+      "epoch": 0.9370889688546966,
+      "grad_norm": 699.8380737304688,
+      "learning_rate": 2.1647052836329065e-07,
+      "loss": 81.0783,
+      "step": 113280
+    },
+    {
+      "epoch": 0.9371716921040658,
+      "grad_norm": 887.7781372070312,
+      "learning_rate": 2.1606022229857525e-07,
+      "loss": 126.7601,
+      "step": 113290
+    },
+    {
+      "epoch": 0.9372544153534351,
+      "grad_norm": 768.95166015625,
+      "learning_rate": 2.1565029686999306e-07,
+      "loss": 73.0002,
+      "step": 113300
+    },
+    {
+      "epoch": 0.9373371386028043,
+      "grad_norm": 584.2427368164062,
+      "learning_rate": 2.1524075211016014e-07,
+      "loss": 74.3468,
+      "step": 113310
+    },
+    {
+      "epoch": 0.9374198618521735,
+      "grad_norm": 1104.49755859375,
+      "learning_rate": 2.148315880516605e-07,
+      "loss": 71.5521,
+      "step": 113320
+    },
+    {
+      "epoch": 0.9375025851015428,
+      "grad_norm": 663.4890747070312,
+      "learning_rate": 2.144228047270508e-07,
+      "loss": 82.779,
+      "step": 113330
+    },
+    {
+      "epoch": 0.937585308350912,
+      "grad_norm": 746.1346435546875,
+      "learning_rate": 2.140144021688556e-07,
+      "loss": 62.6374,
+      "step": 113340
+    },
+    {
+      "epoch": 0.9376680316002812,
+      "grad_norm": 1149.3975830078125,
+      "learning_rate": 2.1360638040957004e-07,
+      "loss": 114.1942,
+      "step": 113350
+    },
+    {
+      "epoch": 0.9377507548496505,
+      "grad_norm": 617.9911499023438,
+      "learning_rate": 2.1319873948165704e-07,
+      "loss": 67.7726,
+      "step": 113360
+    },
+    {
+      "epoch": 0.9378334780990197,
+      "grad_norm": 4376.74658203125,
+      "learning_rate": 2.1279147941755284e-07,
+      "loss": 123.5628,
+      "step": 113370
+    },
+    {
+      "epoch": 0.9379162013483889,
+      "grad_norm": 1298.1807861328125,
+      "learning_rate": 2.123846002496599e-07,
+      "loss": 91.0785,
+      "step": 113380
+    },
+    {
+      "epoch": 0.9379989245977582,
+      "grad_norm": 737.0429077148438,
+      "learning_rate": 2.119781020103523e-07,
+      "loss": 78.1234,
+      "step": 113390
+    },
+    {
+      "epoch": 0.9380816478471274,
+      "grad_norm": 1079.584716796875,
+      "learning_rate": 2.1157198473197417e-07,
+      "loss": 83.1591,
+      "step": 113400
+    },
+    {
+      "epoch": 0.9381643710964966,
+      "grad_norm": 733.552001953125,
+      "learning_rate": 2.111662484468363e-07,
+      "loss": 92.7453,
+      "step": 113410
+    },
+    {
+      "epoch": 0.938247094345866,
+      "grad_norm": 328.62188720703125,
+      "learning_rate": 2.1076089318722237e-07,
+      "loss": 84.1476,
+      "step": 113420
+    },
+    {
+      "epoch": 0.9383298175952351,
+      "grad_norm": 764.5359497070312,
+      "learning_rate": 2.1035591898538432e-07,
+      "loss": 115.1742,
+      "step": 113430
+    },
+    {
+      "epoch": 0.9384125408446043,
+      "grad_norm": 606.0637817382812,
+      "learning_rate": 2.0995132587354416e-07,
+      "loss": 91.7363,
+      "step": 113440
+    },
+    {
+      "epoch": 0.9384952640939737,
+      "grad_norm": 631.8139038085938,
+      "learning_rate": 2.0954711388389392e-07,
+      "loss": 70.5418,
+      "step": 113450
+    },
+    {
+      "epoch": 0.9385779873433429,
+      "grad_norm": 742.7326049804688,
+      "learning_rate": 2.09143283048594e-07,
+      "loss": 83.6848,
+      "step": 113460
+    },
+    {
+      "epoch": 0.938660710592712,
+      "grad_norm": 670.2506103515625,
+      "learning_rate": 2.08739833399777e-07,
+      "loss": 64.4276,
+      "step": 113470
+    },
+    {
+      "epoch": 0.9387434338420814,
+      "grad_norm": 569.9981079101562,
+      "learning_rate": 2.0833676496954225e-07,
+      "loss": 73.9018,
+      "step": 113480
+    },
+    {
+      "epoch": 0.9388261570914506,
+      "grad_norm": 843.3419189453125,
+      "learning_rate": 2.0793407778996021e-07,
+      "loss": 66.5153,
+      "step": 113490
+    },
+    {
+      "epoch": 0.9389088803408198,
+      "grad_norm": 1188.6783447265625,
+      "learning_rate": 2.0753177189307138e-07,
+      "loss": 82.2647,
+      "step": 113500
+    },
+    {
+      "epoch": 0.9389916035901891,
+      "grad_norm": 814.5333862304688,
+      "learning_rate": 2.071298473108846e-07,
+      "loss": 97.839,
+      "step": 113510
+    },
+    {
+      "epoch": 0.9390743268395583,
+      "grad_norm": 490.64227294921875,
+      "learning_rate": 2.0672830407537925e-07,
+      "loss": 70.4371,
+      "step": 113520
+    },
+    {
+      "epoch": 0.9391570500889275,
+      "grad_norm": 819.4855346679688,
+      "learning_rate": 2.0632714221850536e-07,
+      "loss": 79.6486,
+      "step": 113530
+    },
+    {
+      "epoch": 0.9392397733382968,
+      "grad_norm": 1092.7366943359375,
+      "learning_rate": 2.0592636177218017e-07,
+      "loss": 80.8918,
+      "step": 113540
+    },
+    {
+      "epoch": 0.939322496587666,
+      "grad_norm": 1289.69775390625,
+      "learning_rate": 2.055259627682926e-07,
+      "loss": 89.2574,
+      "step": 113550
+    },
+    {
+      "epoch": 0.9394052198370352,
+      "grad_norm": 651.2507934570312,
+      "learning_rate": 2.051259452387e-07,
+      "loss": 77.3755,
+      "step": 113560
+    },
+    {
+      "epoch": 0.9394879430864045,
+      "grad_norm": 588.5090942382812,
+      "learning_rate": 2.0472630921523185e-07,
+      "loss": 90.6757,
+      "step": 113570
+    },
+    {
+      "epoch": 0.9395706663357737,
+      "grad_norm": 806.9398193359375,
+      "learning_rate": 2.043270547296833e-07,
+      "loss": 84.2983,
+      "step": 113580
+    },
+    {
+      "epoch": 0.9396533895851429,
+      "grad_norm": 961.4072265625,
+      "learning_rate": 2.0392818181382168e-07,
+      "loss": 69.7272,
+      "step": 113590
+    },
+    {
+      "epoch": 0.9397361128345122,
+      "grad_norm": 765.0098876953125,
+      "learning_rate": 2.0352969049938332e-07,
+      "loss": 92.7051,
+      "step": 113600
+    },
+    {
+      "epoch": 0.9398188360838814,
+      "grad_norm": 1120.79833984375,
+      "learning_rate": 2.0313158081807504e-07,
+      "loss": 84.8201,
+      "step": 113610
+    },
+    {
+      "epoch": 0.9399015593332506,
+      "grad_norm": 834.6403198242188,
+      "learning_rate": 2.027338528015721e-07,
+      "loss": 77.9415,
+      "step": 113620
+    },
+    {
+      "epoch": 0.9399842825826198,
+      "grad_norm": 1245.4927978515625,
+      "learning_rate": 2.0233650648152026e-07,
+      "loss": 106.3016,
+      "step": 113630
+    },
+    {
+      "epoch": 0.9400670058319891,
+      "grad_norm": 606.9644165039062,
+      "learning_rate": 2.0193954188953425e-07,
+      "loss": 116.0985,
+      "step": 113640
+    },
+    {
+      "epoch": 0.9401497290813583,
+      "grad_norm": 788.4069213867188,
+      "learning_rate": 2.015429590571988e-07,
+      "loss": 84.2174,
+      "step": 113650
+    },
+    {
+      "epoch": 0.9402324523307275,
+      "grad_norm": 560.9490356445312,
+      "learning_rate": 2.0114675801606754e-07,
+      "loss": 73.4865,
+      "step": 113660
+    },
+    {
+      "epoch": 0.9403151755800968,
+      "grad_norm": 1249.0743408203125,
+      "learning_rate": 2.0075093879766584e-07,
+      "loss": 97.4712,
+      "step": 113670
+    },
+    {
+      "epoch": 0.940397898829466,
+      "grad_norm": 814.6300048828125,
+      "learning_rate": 2.003555014334857e-07,
+      "loss": 74.9081,
+      "step": 113680
+    },
+    {
+      "epoch": 0.9404806220788352,
+      "grad_norm": 929.8789672851562,
+      "learning_rate": 1.9996044595499142e-07,
+      "loss": 101.6634,
+      "step": 113690
+    },
+    {
+      "epoch": 0.9405633453282045,
+      "grad_norm": 495.4698486328125,
+      "learning_rate": 1.9956577239361507e-07,
+      "loss": 79.1617,
+      "step": 113700
+    },
+    {
+      "epoch": 0.9406460685775737,
+      "grad_norm": 496.3013916015625,
+      "learning_rate": 1.9917148078075876e-07,
+      "loss": 68.7562,
+      "step": 113710
+    },
+    {
+      "epoch": 0.9407287918269429,
+      "grad_norm": 884.4699096679688,
+      "learning_rate": 1.9877757114779517e-07,
+      "loss": 103.0708,
+      "step": 113720
+    },
+    {
+      "epoch": 0.9408115150763122,
+      "grad_norm": 874.1282348632812,
+      "learning_rate": 1.983840435260659e-07,
+      "loss": 84.1153,
+      "step": 113730
+    },
+    {
+      "epoch": 0.9408942383256814,
+      "grad_norm": 956.4913330078125,
+      "learning_rate": 1.9799089794688197e-07,
+      "loss": 78.011,
+      "step": 113740
+    },
+    {
+      "epoch": 0.9409769615750506,
+      "grad_norm": 651.1774291992188,
+      "learning_rate": 1.9759813444152342e-07,
+      "loss": 55.0541,
+      "step": 113750
+    },
+    {
+      "epoch": 0.9410596848244199,
+      "grad_norm": 874.5100708007812,
+      "learning_rate": 1.9720575304124135e-07,
+      "loss": 61.1703,
+      "step": 113760
+    },
+    {
+      "epoch": 0.9411424080737891,
+      "grad_norm": 646.1497802734375,
+      "learning_rate": 1.9681375377725631e-07,
+      "loss": 53.9226,
+      "step": 113770
+    },
+    {
+      "epoch": 0.9412251313231583,
+      "grad_norm": 452.9364318847656,
+      "learning_rate": 1.9642213668075673e-07,
+      "loss": 136.8662,
+      "step": 113780
+    },
+    {
+      "epoch": 0.9413078545725276,
+      "grad_norm": 505.2906494140625,
+      "learning_rate": 1.9603090178290207e-07,
+      "loss": 78.0005,
+      "step": 113790
+    },
+    {
+      "epoch": 0.9413905778218968,
+      "grad_norm": 842.5142211914062,
+      "learning_rate": 1.9564004911482192e-07,
+      "loss": 59.6014,
+      "step": 113800
+    },
+    {
+      "epoch": 0.941473301071266,
+      "grad_norm": 860.8002319335938,
+      "learning_rate": 1.9524957870761364e-07,
+      "loss": 103.618,
+      "step": 113810
+    },
+    {
+      "epoch": 0.9415560243206353,
+      "grad_norm": 587.0081176757812,
+      "learning_rate": 1.9485949059234567e-07,
+      "loss": 97.7175,
+      "step": 113820
+    },
+    {
+      "epoch": 0.9416387475700045,
+      "grad_norm": 842.8802490234375,
+      "learning_rate": 1.944697848000554e-07,
+      "loss": 126.7808,
+      "step": 113830
+    },
+    {
+      "epoch": 0.9417214708193737,
+      "grad_norm": 915.0126342773438,
+      "learning_rate": 1.9408046136174975e-07,
+      "loss": 79.9683,
+      "step": 113840
+    },
+    {
+      "epoch": 0.941804194068743,
+      "grad_norm": 609.2335815429688,
+      "learning_rate": 1.9369152030840553e-07,
+      "loss": 92.4454,
+      "step": 113850
+    },
+    {
+      "epoch": 0.9418869173181122,
+      "grad_norm": 582.4105224609375,
+      "learning_rate": 1.9330296167096972e-07,
+      "loss": 94.1964,
+      "step": 113860
+    },
+    {
+      "epoch": 0.9419696405674814,
+      "grad_norm": 1090.697265625,
+      "learning_rate": 1.9291478548035703e-07,
+      "loss": 78.0568,
+      "step": 113870
+    },
+    {
+      "epoch": 0.9420523638168508,
+      "grad_norm": 755.0408325195312,
+      "learning_rate": 1.9252699176745326e-07,
+      "loss": 85.2529,
+      "step": 113880
+    },
+    {
+      "epoch": 0.94213508706622,
+      "grad_norm": 657.3673095703125,
+      "learning_rate": 1.9213958056311376e-07,
+      "loss": 65.6693,
+      "step": 113890
+    },
+    {
+      "epoch": 0.9422178103155892,
+      "grad_norm": 1077.51904296875,
+      "learning_rate": 1.917525518981622e-07,
+      "loss": 100.6788,
+      "step": 113900
+    },
+    {
+      "epoch": 0.9423005335649585,
+      "grad_norm": 780.7044067382812,
+      "learning_rate": 1.91365905803394e-07,
+      "loss": 88.6271,
+      "step": 113910
+    },
+    {
+      "epoch": 0.9423832568143277,
+      "grad_norm": 674.3640747070312,
+      "learning_rate": 1.9097964230957112e-07,
+      "loss": 72.1338,
+      "step": 113920
+    },
+    {
+      "epoch": 0.9424659800636969,
+      "grad_norm": 830.9854125976562,
+      "learning_rate": 1.9059376144742792e-07,
+      "loss": 82.4124,
+      "step": 113930
+    },
+    {
+      "epoch": 0.9425487033130662,
+      "grad_norm": 555.4501342773438,
+      "learning_rate": 1.9020826324766707e-07,
+      "loss": 94.9241,
+      "step": 113940
+    },
+    {
+      "epoch": 0.9426314265624354,
+      "grad_norm": 762.3797607421875,
+      "learning_rate": 1.8982314774096067e-07,
+      "loss": 81.5409,
+      "step": 113950
+    },
+    {
+      "epoch": 0.9427141498118046,
+      "grad_norm": 840.9845581054688,
+      "learning_rate": 1.894384149579509e-07,
+      "loss": 81.5919,
+      "step": 113960
+    },
+    {
+      "epoch": 0.9427968730611739,
+      "grad_norm": 520.0264892578125,
+      "learning_rate": 1.8905406492924884e-07,
+      "loss": 82.138,
+      "step": 113970
+    },
+    {
+      "epoch": 0.9428795963105431,
+      "grad_norm": 1191.1163330078125,
+      "learning_rate": 1.8867009768543554e-07,
+      "loss": 92.1979,
+      "step": 113980
+    },
+    {
+      "epoch": 0.9429623195599123,
+      "grad_norm": 850.5126342773438,
+      "learning_rate": 1.8828651325706159e-07,
+      "loss": 124.8058,
+      "step": 113990
+    },
+    {
+      "epoch": 0.9430450428092816,
+      "grad_norm": 1242.6998291015625,
+      "learning_rate": 1.8790331167464758e-07,
+      "loss": 74.3944,
+      "step": 114000
+    },
+    {
+      "epoch": 0.9431277660586508,
+      "grad_norm": 937.7474975585938,
+      "learning_rate": 1.875204929686819e-07,
+      "loss": 78.7968,
+      "step": 114010
+    },
+    {
+      "epoch": 0.94321048930802,
+      "grad_norm": 811.911376953125,
+      "learning_rate": 1.8713805716962408e-07,
+      "loss": 81.0114,
+      "step": 114020
+    },
+    {
+      "epoch": 0.9432932125573893,
+      "grad_norm": 887.4229736328125,
+      "learning_rate": 1.867560043079031e-07,
+      "loss": 91.9161,
+      "step": 114030
+    },
+    {
+      "epoch": 0.9433759358067585,
+      "grad_norm": 556.4613037109375,
+      "learning_rate": 1.8637433441391739e-07,
+      "loss": 81.6592,
+      "step": 114040
+    },
+    {
+      "epoch": 0.9434586590561277,
+      "grad_norm": 785.0360107421875,
+      "learning_rate": 1.859930475180338e-07,
+      "loss": 88.5488,
+      "step": 114050
+    },
+    {
+      "epoch": 0.943541382305497,
+      "grad_norm": 1435.7166748046875,
+      "learning_rate": 1.8561214365059033e-07,
+      "loss": 75.1003,
+      "step": 114060
+    },
+    {
+      "epoch": 0.9436241055548662,
+      "grad_norm": 584.7539672851562,
+      "learning_rate": 1.8523162284189377e-07,
+      "loss": 85.8287,
+      "step": 114070
+    },
+    {
+      "epoch": 0.9437068288042354,
+      "grad_norm": 553.2955932617188,
+      "learning_rate": 1.848514851222205e-07,
+      "loss": 86.0015,
+      "step": 114080
+    },
+    {
+      "epoch": 0.9437895520536047,
+      "grad_norm": 628.0018310546875,
+      "learning_rate": 1.8447173052181577e-07,
+      "loss": 81.3418,
+      "step": 114090
+    },
+    {
+      "epoch": 0.9438722753029739,
+      "grad_norm": 708.7710571289062,
+      "learning_rate": 1.8409235907089484e-07,
+      "loss": 91.275,
+      "step": 114100
+    },
+    {
+      "epoch": 0.9439549985523431,
+      "grad_norm": 712.0585327148438,
+      "learning_rate": 1.8371337079964303e-07,
+      "loss": 99.5175,
+      "step": 114110
+    },
+    {
+      "epoch": 0.9440377218017124,
+      "grad_norm": 1265.697021484375,
+      "learning_rate": 1.8333476573821395e-07,
+      "loss": 103.0046,
+      "step": 114120
+    },
+    {
+      "epoch": 0.9441204450510816,
+      "grad_norm": 937.801025390625,
+      "learning_rate": 1.8295654391673245e-07,
+      "loss": 89.4391,
+      "step": 114130
+    },
+    {
+      "epoch": 0.9442031683004508,
+      "grad_norm": 484.9341125488281,
+      "learning_rate": 1.8257870536529167e-07,
+      "loss": 118.8798,
+      "step": 114140
+    },
+    {
+      "epoch": 0.9442858915498201,
+      "grad_norm": 604.8045654296875,
+      "learning_rate": 1.8220125011395419e-07,
+      "loss": 78.082,
+      "step": 114150
+    },
+    {
+      "epoch": 0.9443686147991893,
+      "grad_norm": 624.2005615234375,
+      "learning_rate": 1.8182417819275266e-07,
+      "loss": 70.6287,
+      "step": 114160
+    },
+    {
+      "epoch": 0.9444513380485585,
+      "grad_norm": 622.5794067382812,
+      "learning_rate": 1.8144748963168924e-07,
+      "loss": 80.9847,
+      "step": 114170
+    },
+    {
+      "epoch": 0.9445340612979278,
+      "grad_norm": 371.8035888671875,
+      "learning_rate": 1.8107118446073492e-07,
+      "loss": 83.7294,
+      "step": 114180
+    },
+    {
+      "epoch": 0.944616784547297,
+      "grad_norm": 911.8489379882812,
+      "learning_rate": 1.806952627098296e-07,
+      "loss": 69.7044,
+      "step": 114190
+    },
+    {
+      "epoch": 0.9446995077966662,
+      "grad_norm": 649.6644287109375,
+      "learning_rate": 1.8031972440888556e-07,
+      "loss": 99.0398,
+      "step": 114200
+    },
+    {
+      "epoch": 0.9447822310460355,
+      "grad_norm": 1058.6116943359375,
+      "learning_rate": 1.799445695877805e-07,
+      "loss": 105.9399,
+      "step": 114210
+    },
+    {
+      "epoch": 0.9448649542954047,
+      "grad_norm": 1170.114013671875,
+      "learning_rate": 1.7956979827636556e-07,
+      "loss": 95.6862,
+      "step": 114220
+    },
+    {
+      "epoch": 0.9449476775447739,
+      "grad_norm": 1229.6246337890625,
+      "learning_rate": 1.791954105044591e-07,
+      "loss": 83.8179,
+      "step": 114230
+    },
+    {
+      "epoch": 0.9450304007941432,
+      "grad_norm": 673.4191284179688,
+      "learning_rate": 1.788214063018495e-07,
+      "loss": 78.5252,
+      "step": 114240
+    },
+    {
+      "epoch": 0.9451131240435124,
+      "grad_norm": 1112.168212890625,
+      "learning_rate": 1.7844778569829412e-07,
+      "loss": 80.5072,
+      "step": 114250
+    },
+    {
+      "epoch": 0.9451958472928816,
+      "grad_norm": 1105.3406982421875,
+      "learning_rate": 1.7807454872352137e-07,
+      "loss": 76.1349,
+      "step": 114260
+    },
+    {
+      "epoch": 0.945278570542251,
+      "grad_norm": 727.6226806640625,
+      "learning_rate": 1.7770169540722638e-07,
+      "loss": 81.8822,
+      "step": 114270
+    },
+    {
+      "epoch": 0.9453612937916201,
+      "grad_norm": 932.0345458984375,
+      "learning_rate": 1.7732922577907595e-07,
+      "loss": 89.935,
+      "step": 114280
+    },
+    {
+      "epoch": 0.9454440170409893,
+      "grad_norm": 780.9658813476562,
+      "learning_rate": 1.769571398687059e-07,
+      "loss": 95.7571,
+      "step": 114290
+    },
+    {
+      "epoch": 0.9455267402903587,
+      "grad_norm": 883.4459838867188,
+      "learning_rate": 1.765854377057219e-07,
+      "loss": 72.5025,
+      "step": 114300
+    },
+    {
+      "epoch": 0.9456094635397279,
+      "grad_norm": 925.8482055664062,
+      "learning_rate": 1.76214119319697e-07,
+      "loss": 86.9446,
+      "step": 114310
+    },
+    {
+      "epoch": 0.9456921867890971,
+      "grad_norm": 770.7780151367188,
+      "learning_rate": 1.758431847401776e-07,
+      "loss": 78.9522,
+      "step": 114320
+    },
+    {
+      "epoch": 0.9457749100384664,
+      "grad_norm": 670.9049682617188,
+      "learning_rate": 1.7547263399667558e-07,
+      "loss": 84.8401,
+      "step": 114330
+    },
+    {
+      "epoch": 0.9458576332878356,
+      "grad_norm": 637.699951171875,
+      "learning_rate": 1.7510246711867572e-07,
+      "loss": 91.2389,
+      "step": 114340
+    },
+    {
+      "epoch": 0.9459403565372048,
+      "grad_norm": 908.4950561523438,
+      "learning_rate": 1.7473268413562837e-07,
+      "loss": 92.7157,
+      "step": 114350
+    },
+    {
+      "epoch": 0.946023079786574,
+      "grad_norm": 778.84765625,
+      "learning_rate": 1.743632850769561e-07,
+      "loss": 84.7941,
+      "step": 114360
+    },
+    {
+      "epoch": 0.9461058030359433,
+      "grad_norm": 945.8580322265625,
+      "learning_rate": 1.739942699720504e-07,
+      "loss": 124.0557,
+      "step": 114370
+    },
+    {
+      "epoch": 0.9461885262853125,
+      "grad_norm": 716.9354248046875,
+      "learning_rate": 1.7362563885027272e-07,
+      "loss": 101.1986,
+      "step": 114380
+    },
+    {
+      "epoch": 0.9462712495346817,
+      "grad_norm": 920.9649047851562,
+      "learning_rate": 1.7325739174095302e-07,
+      "loss": 72.5662,
+      "step": 114390
+    },
+    {
+      "epoch": 0.946353972784051,
+      "grad_norm": 918.943115234375,
+      "learning_rate": 1.728895286733906e-07,
+      "loss": 63.3512,
+      "step": 114400
+    },
+    {
+      "epoch": 0.9464366960334202,
+      "grad_norm": 959.2852172851562,
+      "learning_rate": 1.7252204967685427e-07,
+      "loss": 78.725,
+      "step": 114410
+    },
+    {
+      "epoch": 0.9465194192827894,
+      "grad_norm": 519.7755126953125,
+      "learning_rate": 1.7215495478058397e-07,
+      "loss": 80.2883,
+      "step": 114420
+    },
+    {
+      "epoch": 0.9466021425321587,
+      "grad_norm": 1290.6314697265625,
+      "learning_rate": 1.7178824401378802e-07,
+      "loss": 74.9698,
+      "step": 114430
+    },
+    {
+      "epoch": 0.9466848657815279,
+      "grad_norm": 840.70068359375,
+      "learning_rate": 1.7142191740564196e-07,
+      "loss": 72.1562,
+      "step": 114440
+    },
+    {
+      "epoch": 0.9467675890308971,
+      "grad_norm": 853.8070068359375,
+      "learning_rate": 1.7105597498529358e-07,
+      "loss": 80.7684,
+      "step": 114450
+    },
+    {
+      "epoch": 0.9468503122802664,
+      "grad_norm": 1219.2557373046875,
+      "learning_rate": 1.7069041678186017e-07,
+      "loss": 85.5803,
+      "step": 114460
+    },
+    {
+      "epoch": 0.9469330355296356,
+      "grad_norm": 652.1727294921875,
+      "learning_rate": 1.7032524282442618e-07,
+      "loss": 74.4581,
+      "step": 114470
+    },
+    {
+      "epoch": 0.9470157587790048,
+      "grad_norm": 469.0629577636719,
+      "learning_rate": 1.6996045314204734e-07,
+      "loss": 71.7418,
+      "step": 114480
+    },
+    {
+      "epoch": 0.9470984820283741,
+      "grad_norm": 666.7960815429688,
+      "learning_rate": 1.6959604776374871e-07,
+      "loss": 87.4047,
+      "step": 114490
+    },
+    {
+      "epoch": 0.9471812052777433,
+      "grad_norm": 587.7761840820312,
+      "learning_rate": 1.6923202671852379e-07,
+      "loss": 122.6015,
+      "step": 114500
+    },
+    {
+      "epoch": 0.9472639285271125,
+      "grad_norm": 1696.044677734375,
+      "learning_rate": 1.688683900353366e-07,
+      "loss": 96.9965,
+      "step": 114510
+    },
+    {
+      "epoch": 0.9473466517764818,
+      "grad_norm": 692.3433837890625,
+      "learning_rate": 1.6850513774311906e-07,
+      "loss": 80.0998,
+      "step": 114520
+    },
+    {
+      "epoch": 0.947429375025851,
+      "grad_norm": 755.7346801757812,
+      "learning_rate": 1.6814226987077464e-07,
+      "loss": 77.0792,
+      "step": 114530
+    },
+    {
+      "epoch": 0.9475120982752202,
+      "grad_norm": 481.3829345703125,
+      "learning_rate": 1.6777978644717474e-07,
+      "loss": 63.8102,
+      "step": 114540
+    },
+    {
+      "epoch": 0.9475948215245895,
+      "grad_norm": 654.6975708007812,
+      "learning_rate": 1.6741768750116017e-07,
+      "loss": 56.0552,
+      "step": 114550
+    },
+    {
+      "epoch": 0.9476775447739587,
+      "grad_norm": 745.0943603515625,
+      "learning_rate": 1.670559730615412e-07,
+      "loss": 76.7988,
+      "step": 114560
+    },
+    {
+      "epoch": 0.9477602680233279,
+      "grad_norm": 604.7420654296875,
+      "learning_rate": 1.6669464315709872e-07,
+      "loss": 75.703,
+      "step": 114570
+    },
+    {
+      "epoch": 0.9478429912726972,
+      "grad_norm": 1340.7713623046875,
+      "learning_rate": 1.6633369781658137e-07,
+      "loss": 78.7956,
+      "step": 114580
+    },
+    {
+      "epoch": 0.9479257145220664,
+      "grad_norm": 1017.1603393554688,
+      "learning_rate": 1.6597313706870842e-07,
+      "loss": 88.8706,
+      "step": 114590
+    },
+    {
+      "epoch": 0.9480084377714356,
+      "grad_norm": 465.18914794921875,
+      "learning_rate": 1.656129609421675e-07,
+      "loss": 90.0202,
+      "step": 114600
+    },
+    {
+      "epoch": 0.9480911610208049,
+      "grad_norm": 746.8944091796875,
+      "learning_rate": 1.6525316946561675e-07,
+      "loss": 69.2716,
+      "step": 114610
+    },
+    {
+      "epoch": 0.9481738842701741,
+      "grad_norm": 557.8377075195312,
+      "learning_rate": 1.648937626676822e-07,
+      "loss": 73.0987,
+      "step": 114620
+    },
+    {
+      "epoch": 0.9482566075195433,
+      "grad_norm": 1114.826171875,
+      "learning_rate": 1.6453474057696152e-07,
+      "loss": 73.0719,
+      "step": 114630
+    },
+    {
+      "epoch": 0.9483393307689126,
+      "grad_norm": 1133.66796875,
+      "learning_rate": 1.6417610322201904e-07,
+      "loss": 84.8226,
+      "step": 114640
+    },
+    {
+      "epoch": 0.9484220540182818,
+      "grad_norm": 638.8233032226562,
+      "learning_rate": 1.6381785063139144e-07,
+      "loss": 58.4258,
+      "step": 114650
+    },
+    {
+      "epoch": 0.948504777267651,
+      "grad_norm": 581.8013916015625,
+      "learning_rate": 1.6345998283358145e-07,
+      "loss": 90.1874,
+      "step": 114660
+    },
+    {
+      "epoch": 0.9485875005170203,
+      "grad_norm": 584.7073974609375,
+      "learning_rate": 1.631024998570646e-07,
+      "loss": 108.2388,
+      "step": 114670
+    },
+    {
+      "epoch": 0.9486702237663895,
+      "grad_norm": 1126.181884765625,
+      "learning_rate": 1.6274540173028318e-07,
+      "loss": 92.8387,
+      "step": 114680
+    },
+    {
+      "epoch": 0.9487529470157587,
+      "grad_norm": 823.510986328125,
+      "learning_rate": 1.6238868848165056e-07,
+      "loss": 86.9546,
+      "step": 114690
+    },
+    {
+      "epoch": 0.948835670265128,
+      "grad_norm": 1618.146728515625,
+      "learning_rate": 1.6203236013954792e-07,
+      "loss": 113.7917,
+      "step": 114700
+    },
+    {
+      "epoch": 0.9489183935144972,
+      "grad_norm": 6492.607421875,
+      "learning_rate": 1.6167641673232703e-07,
+      "loss": 122.2078,
+      "step": 114710
+    },
+    {
+      "epoch": 0.9490011167638664,
+      "grad_norm": 1291.8187255859375,
+      "learning_rate": 1.613208582883091e-07,
+      "loss": 75.8952,
+      "step": 114720
+    },
+    {
+      "epoch": 0.9490838400132358,
+      "grad_norm": 712.4785766601562,
+      "learning_rate": 1.609656848357838e-07,
+      "loss": 69.3616,
+      "step": 114730
+    },
+    {
+      "epoch": 0.949166563262605,
+      "grad_norm": 1505.3465576171875,
+      "learning_rate": 1.6061089640301063e-07,
+      "loss": 88.8558,
+      "step": 114740
+    },
+    {
+      "epoch": 0.9492492865119742,
+      "grad_norm": 827.9591064453125,
+      "learning_rate": 1.6025649301821877e-07,
+      "loss": 93.01,
+      "step": 114750
+    },
+    {
+      "epoch": 0.9493320097613435,
+      "grad_norm": 877.7481689453125,
+      "learning_rate": 1.599024747096062e-07,
+      "loss": 93.6232,
+      "step": 114760
+    },
+    {
+      "epoch": 0.9494147330107127,
+      "grad_norm": 1036.861083984375,
+      "learning_rate": 1.595488415053409e-07,
+      "loss": 90.6146,
+      "step": 114770
+    },
+    {
+      "epoch": 0.9494974562600819,
+      "grad_norm": 577.421875,
+      "learning_rate": 1.591955934335593e-07,
+      "loss": 68.6489,
+      "step": 114780
+    },
+    {
+      "epoch": 0.9495801795094512,
+      "grad_norm": 877.7052612304688,
+      "learning_rate": 1.588427305223683e-07,
+      "loss": 94.7773,
+      "step": 114790
+    },
+    {
+      "epoch": 0.9496629027588204,
+      "grad_norm": 977.225341796875,
+      "learning_rate": 1.584902527998433e-07,
+      "loss": 80.5525,
+      "step": 114800
+    },
+    {
+      "epoch": 0.9497456260081896,
+      "grad_norm": 1442.789794921875,
+      "learning_rate": 1.5813816029402963e-07,
+      "loss": 83.8474,
+      "step": 114810
+    },
+    {
+      "epoch": 0.9498283492575589,
+      "grad_norm": 750.039306640625,
+      "learning_rate": 1.5778645303294094e-07,
+      "loss": 82.7841,
+      "step": 114820
+    },
+    {
+      "epoch": 0.9499110725069281,
+      "grad_norm": 716.677490234375,
+      "learning_rate": 1.5743513104456154e-07,
+      "loss": 83.8299,
+      "step": 114830
+    },
+    {
+      "epoch": 0.9499937957562973,
+      "grad_norm": 688.27685546875,
+      "learning_rate": 1.5708419435684463e-07,
+      "loss": 93.2964,
+      "step": 114840
+    },
+    {
+      "epoch": 0.9500765190056666,
+      "grad_norm": 907.2855834960938,
+      "learning_rate": 1.5673364299771177e-07,
+      "loss": 100.0956,
+      "step": 114850
+    },
+    {
+      "epoch": 0.9501592422550358,
+      "grad_norm": 1261.3101806640625,
+      "learning_rate": 1.5638347699505673e-07,
+      "loss": 93.222,
+      "step": 114860
+    },
+    {
+      "epoch": 0.950241965504405,
+      "grad_norm": 672.2671508789062,
+      "learning_rate": 1.5603369637673727e-07,
+      "loss": 119.3111,
+      "step": 114870
+    },
+    {
+      "epoch": 0.9503246887537743,
+      "grad_norm": 998.6853637695312,
+      "learning_rate": 1.5568430117058718e-07,
+      "loss": 69.6595,
+      "step": 114880
+    },
+    {
+      "epoch": 0.9504074120031435,
+      "grad_norm": 602.7067260742188,
+      "learning_rate": 1.553352914044043e-07,
+      "loss": 88.5864,
+      "step": 114890
+    },
+    {
+      "epoch": 0.9504901352525127,
+      "grad_norm": 914.7874145507812,
+      "learning_rate": 1.5498666710595855e-07,
+      "loss": 66.891,
+      "step": 114900
+    },
+    {
+      "epoch": 0.950572858501882,
+      "grad_norm": 843.3950805664062,
+      "learning_rate": 1.5463842830298782e-07,
+      "loss": 80.3389,
+      "step": 114910
+    },
+    {
+      "epoch": 0.9506555817512512,
+      "grad_norm": 720.8048706054688,
+      "learning_rate": 1.5429057502320045e-07,
+      "loss": 82.6339,
+      "step": 114920
+    },
+    {
+      "epoch": 0.9507383050006204,
+      "grad_norm": 1675.1365966796875,
+      "learning_rate": 1.5394310729427265e-07,
+      "loss": 100.8808,
+      "step": 114930
+    },
+    {
+      "epoch": 0.9508210282499897,
+      "grad_norm": 751.2517700195312,
+      "learning_rate": 1.535960251438523e-07,
+      "loss": 80.3977,
+      "step": 114940
+    },
+    {
+      "epoch": 0.9509037514993589,
+      "grad_norm": 557.0956420898438,
+      "learning_rate": 1.53249328599554e-07,
+      "loss": 75.274,
+      "step": 114950
+    },
+    {
+      "epoch": 0.9509864747487281,
+      "grad_norm": 1328.832275390625,
+      "learning_rate": 1.5290301768896287e-07,
+      "loss": 63.9589,
+      "step": 114960
+    },
+    {
+      "epoch": 0.9510691979980974,
+      "grad_norm": 811.1527709960938,
+      "learning_rate": 1.5255709243963246e-07,
+      "loss": 86.3641,
+      "step": 114970
+    },
+    {
+      "epoch": 0.9511519212474666,
+      "grad_norm": 1016.0829467773438,
+      "learning_rate": 1.5221155287908851e-07,
+      "loss": 80.3929,
+      "step": 114980
+    },
+    {
+      "epoch": 0.9512346444968358,
+      "grad_norm": 1659.81787109375,
+      "learning_rate": 1.518663990348229e-07,
+      "loss": 106.4811,
+      "step": 114990
+    },
+    {
+      "epoch": 0.9513173677462051,
+      "grad_norm": 682.6334838867188,
+      "learning_rate": 1.5152163093429762e-07,
+      "loss": 82.3017,
+      "step": 115000
+    },
+    {
+      "epoch": 0.9514000909955743,
+      "grad_norm": 1205.90576171875,
+      "learning_rate": 1.5117724860494509e-07,
+      "loss": 84.5935,
+      "step": 115010
+    },
+    {
+      "epoch": 0.9514828142449435,
+      "grad_norm": 691.462890625,
+      "learning_rate": 1.5083325207416565e-07,
+      "loss": 72.7825,
+      "step": 115020
+    },
+    {
+      "epoch": 0.9515655374943128,
+      "grad_norm": 810.3004150390625,
+      "learning_rate": 1.504896413693302e-07,
+      "loss": 85.171,
+      "step": 115030
+    },
+    {
+      "epoch": 0.951648260743682,
+      "grad_norm": 750.6644897460938,
+      "learning_rate": 1.501464165177774e-07,
+      "loss": 94.0332,
+      "step": 115040
+    },
+    {
+      "epoch": 0.9517309839930512,
+      "grad_norm": 463.6679992675781,
+      "learning_rate": 1.4980357754681595e-07,
+      "loss": 81.951,
+      "step": 115050
+    },
+    {
+      "epoch": 0.9518137072424205,
+      "grad_norm": 824.5008544921875,
+      "learning_rate": 1.4946112448372462e-07,
+      "loss": 96.1971,
+      "step": 115060
+    },
+    {
+      "epoch": 0.9518964304917897,
+      "grad_norm": 629.86572265625,
+      "learning_rate": 1.491190573557505e-07,
+      "loss": 87.0432,
+      "step": 115070
+    },
+    {
+      "epoch": 0.9519791537411589,
+      "grad_norm": 1388.0849609375,
+      "learning_rate": 1.4877737619011067e-07,
+      "loss": 94.4698,
+      "step": 115080
+    },
+    {
+      "epoch": 0.9520618769905281,
+      "grad_norm": 826.2018432617188,
+      "learning_rate": 1.4843608101399065e-07,
+      "loss": 83.7667,
+      "step": 115090
+    },
+    {
+      "epoch": 0.9521446002398974,
+      "grad_norm": 675.1743774414062,
+      "learning_rate": 1.4809517185454646e-07,
+      "loss": 67.9461,
+      "step": 115100
+    },
+    {
+      "epoch": 0.9522273234892666,
+      "grad_norm": 530.15380859375,
+      "learning_rate": 1.4775464873890256e-07,
+      "loss": 54.279,
+      "step": 115110
+    },
+    {
+      "epoch": 0.9523100467386358,
+      "grad_norm": 862.5327758789062,
+      "learning_rate": 1.4741451169415165e-07,
+      "loss": 95.6716,
+      "step": 115120
+    },
+    {
+      "epoch": 0.9523927699880052,
+      "grad_norm": 1480.7520751953125,
+      "learning_rate": 1.4707476074735772e-07,
+      "loss": 112.3083,
+      "step": 115130
+    },
+    {
+      "epoch": 0.9524754932373743,
+      "grad_norm": 440.8155517578125,
+      "learning_rate": 1.4673539592555354e-07,
+      "loss": 67.7535,
+      "step": 115140
+    },
+    {
+      "epoch": 0.9525582164867435,
+      "grad_norm": 760.7919311523438,
+      "learning_rate": 1.4639641725574028e-07,
+      "loss": 81.5349,
+      "step": 115150
+    },
+    {
+      "epoch": 0.9526409397361129,
+      "grad_norm": 711.9234008789062,
+      "learning_rate": 1.460578247648886e-07,
+      "loss": 81.3426,
+      "step": 115160
+    },
+    {
+      "epoch": 0.9527236629854821,
+      "grad_norm": 728.8473510742188,
+      "learning_rate": 1.4571961847993977e-07,
+      "loss": 85.5738,
+      "step": 115170
+    },
+    {
+      "epoch": 0.9528063862348513,
+      "grad_norm": 426.4129943847656,
+      "learning_rate": 1.453817984278022e-07,
+      "loss": 90.5035,
+      "step": 115180
+    },
+    {
+      "epoch": 0.9528891094842206,
+      "grad_norm": 623.3121337890625,
+      "learning_rate": 1.450443646353561e-07,
+      "loss": 79.0887,
+      "step": 115190
+    },
+    {
+      "epoch": 0.9529718327335898,
+      "grad_norm": 748.9600219726562,
+      "learning_rate": 1.4470731712944885e-07,
+      "loss": 110.1946,
+      "step": 115200
+    },
+    {
+      "epoch": 0.953054555982959,
+      "grad_norm": 567.2506713867188,
+      "learning_rate": 1.443706559368968e-07,
+      "loss": 76.4982,
+      "step": 115210
+    },
+    {
+      "epoch": 0.9531372792323283,
+      "grad_norm": 406.186279296875,
+      "learning_rate": 1.4403438108448742e-07,
+      "loss": 120.5117,
+      "step": 115220
+    },
+    {
+      "epoch": 0.9532200024816975,
+      "grad_norm": 800.6393432617188,
+      "learning_rate": 1.436984925989765e-07,
+      "loss": 77.8392,
+      "step": 115230
+    },
+    {
+      "epoch": 0.9533027257310667,
+      "grad_norm": 632.5950927734375,
+      "learning_rate": 1.4336299050708935e-07,
+      "loss": 69.4117,
+      "step": 115240
+    },
+    {
+      "epoch": 0.953385448980436,
+      "grad_norm": 855.23974609375,
+      "learning_rate": 1.4302787483551962e-07,
+      "loss": 80.8834,
+      "step": 115250
+    },
+    {
+      "epoch": 0.9534681722298052,
+      "grad_norm": 749.250732421875,
+      "learning_rate": 1.426931456109315e-07,
+      "loss": 87.1514,
+      "step": 115260
+    },
+    {
+      "epoch": 0.9535508954791744,
+      "grad_norm": 503.9537658691406,
+      "learning_rate": 1.4235880285995762e-07,
+      "loss": 58.2232,
+      "step": 115270
+    },
+    {
+      "epoch": 0.9536336187285437,
+      "grad_norm": 932.1503295898438,
+      "learning_rate": 1.4202484660920057e-07,
+      "loss": 87.4663,
+      "step": 115280
+    },
+    {
+      "epoch": 0.9537163419779129,
+      "grad_norm": 973.0960083007812,
+      "learning_rate": 1.4169127688523187e-07,
+      "loss": 86.1386,
+      "step": 115290
+    },
+    {
+      "epoch": 0.9537990652272821,
+      "grad_norm": 702.0614013671875,
+      "learning_rate": 1.413580937145914e-07,
+      "loss": 77.9156,
+      "step": 115300
+    },
+    {
+      "epoch": 0.9538817884766514,
+      "grad_norm": 870.7654418945312,
+      "learning_rate": 1.410252971237891e-07,
+      "loss": 92.6622,
+      "step": 115310
+    },
+    {
+      "epoch": 0.9539645117260206,
+      "grad_norm": 1024.184326171875,
+      "learning_rate": 1.406928871393043e-07,
+      "loss": 81.5768,
+      "step": 115320
+    },
+    {
+      "epoch": 0.9540472349753898,
+      "grad_norm": 1577.88525390625,
+      "learning_rate": 1.4036086378758474e-07,
+      "loss": 82.3757,
+      "step": 115330
+    },
+    {
+      "epoch": 0.9541299582247591,
+      "grad_norm": 652.8252563476562,
+      "learning_rate": 1.4002922709504874e-07,
+      "loss": 72.8365,
+      "step": 115340
+    },
+    {
+      "epoch": 0.9542126814741283,
+      "grad_norm": 1175.390625,
+      "learning_rate": 1.3969797708808296e-07,
+      "loss": 84.131,
+      "step": 115350
+    },
+    {
+      "epoch": 0.9542954047234975,
+      "grad_norm": 1300.7655029296875,
+      "learning_rate": 1.39367113793043e-07,
+      "loss": 82.6223,
+      "step": 115360
+    },
+    {
+      "epoch": 0.9543781279728668,
+      "grad_norm": 706.9264526367188,
+      "learning_rate": 1.390366372362556e-07,
+      "loss": 95.0086,
+      "step": 115370
+    },
+    {
+      "epoch": 0.954460851222236,
+      "grad_norm": 700.4502563476562,
+      "learning_rate": 1.3870654744401358e-07,
+      "loss": 83.8916,
+      "step": 115380
+    },
+    {
+      "epoch": 0.9545435744716052,
+      "grad_norm": 775.755859375,
+      "learning_rate": 1.3837684444258092e-07,
+      "loss": 76.1233,
+      "step": 115390
+    },
+    {
+      "epoch": 0.9546262977209745,
+      "grad_norm": 778.3601684570312,
+      "learning_rate": 1.3804752825819113e-07,
+      "loss": 90.3711,
+      "step": 115400
+    },
+    {
+      "epoch": 0.9547090209703437,
+      "grad_norm": 1078.059326171875,
+      "learning_rate": 1.3771859891704653e-07,
+      "loss": 66.5463,
+      "step": 115410
+    },
+    {
+      "epoch": 0.9547917442197129,
+      "grad_norm": 954.205322265625,
+      "learning_rate": 1.373900564453179e-07,
+      "loss": 80.5711,
+      "step": 115420
+    },
+    {
+      "epoch": 0.9548744674690822,
+      "grad_norm": 1166.7353515625,
+      "learning_rate": 1.3706190086914595e-07,
+      "loss": 74.4393,
+      "step": 115430
+    },
+    {
+      "epoch": 0.9549571907184514,
+      "grad_norm": 567.34423828125,
+      "learning_rate": 1.3673413221464039e-07,
+      "loss": 75.5963,
+      "step": 115440
+    },
+    {
+      "epoch": 0.9550399139678206,
+      "grad_norm": 1016.5427856445312,
+      "learning_rate": 1.3640675050788088e-07,
+      "loss": 82.2038,
+      "step": 115450
+    },
+    {
+      "epoch": 0.9551226372171899,
+      "grad_norm": 744.4939575195312,
+      "learning_rate": 1.360797557749155e-07,
+      "loss": 81.3256,
+      "step": 115460
+    },
+    {
+      "epoch": 0.9552053604665591,
+      "grad_norm": 757.7814331054688,
+      "learning_rate": 1.3575314804176176e-07,
+      "loss": 68.7753,
+      "step": 115470
+    },
+    {
+      "epoch": 0.9552880837159283,
+      "grad_norm": 502.91387939453125,
+      "learning_rate": 1.3542692733440555e-07,
+      "loss": 86.8093,
+      "step": 115480
+    },
+    {
+      "epoch": 0.9553708069652976,
+      "grad_norm": 464.1839904785156,
+      "learning_rate": 1.3510109367880387e-07,
+      "loss": 91.0751,
+      "step": 115490
+    },
+    {
+      "epoch": 0.9554535302146668,
+      "grad_norm": 1076.8289794921875,
+      "learning_rate": 1.3477564710088097e-07,
+      "loss": 95.7027,
+      "step": 115500
+    },
+    {
+      "epoch": 0.955536253464036,
+      "grad_norm": 701.7645874023438,
+      "learning_rate": 1.3445058762653174e-07,
+      "loss": 91.7253,
+      "step": 115510
+    },
+    {
+      "epoch": 0.9556189767134053,
+      "grad_norm": 1526.50439453125,
+      "learning_rate": 1.3412591528161935e-07,
+      "loss": 109.7217,
+      "step": 115520
+    },
+    {
+      "epoch": 0.9557016999627745,
+      "grad_norm": 1529.469482421875,
+      "learning_rate": 1.338016300919759e-07,
+      "loss": 80.6137,
+      "step": 115530
+    },
+    {
+      "epoch": 0.9557844232121437,
+      "grad_norm": 649.20068359375,
+      "learning_rate": 1.3347773208340464e-07,
+      "loss": 82.6933,
+      "step": 115540
+    },
+    {
+      "epoch": 0.955867146461513,
+      "grad_norm": 1042.389404296875,
+      "learning_rate": 1.3315422128167555e-07,
+      "loss": 79.4586,
+      "step": 115550
+    },
+    {
+      "epoch": 0.9559498697108822,
+      "grad_norm": 921.5702514648438,
+      "learning_rate": 1.3283109771252966e-07,
+      "loss": 88.0491,
+      "step": 115560
+    },
+    {
+      "epoch": 0.9560325929602514,
+      "grad_norm": 533.2924194335938,
+      "learning_rate": 1.3250836140167588e-07,
+      "loss": 63.4962,
+      "step": 115570
+    },
+    {
+      "epoch": 0.9561153162096208,
+      "grad_norm": 578.7036743164062,
+      "learning_rate": 1.3218601237479255e-07,
+      "loss": 78.6079,
+      "step": 115580
+    },
+    {
+      "epoch": 0.95619803945899,
+      "grad_norm": 1517.4697265625,
+      "learning_rate": 1.3186405065752861e-07,
+      "loss": 79.1212,
+      "step": 115590
+    },
+    {
+      "epoch": 0.9562807627083592,
+      "grad_norm": 694.7293090820312,
+      "learning_rate": 1.315424762755002e-07,
+      "loss": 75.5039,
+      "step": 115600
+    },
+    {
+      "epoch": 0.9563634859577285,
+      "grad_norm": 866.1787719726562,
+      "learning_rate": 1.3122128925429356e-07,
+      "loss": 104.6923,
+      "step": 115610
+    },
+    {
+      "epoch": 0.9564462092070977,
+      "grad_norm": 821.7919921875,
+      "learning_rate": 1.3090048961946433e-07,
+      "loss": 111.5836,
+      "step": 115620
+    },
+    {
+      "epoch": 0.9565289324564669,
+      "grad_norm": 1143.0037841796875,
+      "learning_rate": 1.305800773965371e-07,
+      "loss": 107.7157,
+      "step": 115630
+    },
+    {
+      "epoch": 0.9566116557058362,
+      "grad_norm": 874.7697143554688,
+      "learning_rate": 1.3026005261100537e-07,
+      "loss": 58.3183,
+      "step": 115640
+    },
+    {
+      "epoch": 0.9566943789552054,
+      "grad_norm": 968.6356201171875,
+      "learning_rate": 1.2994041528833267e-07,
+      "loss": 73.539,
+      "step": 115650
+    },
+    {
+      "epoch": 0.9567771022045746,
+      "grad_norm": 657.7074584960938,
+      "learning_rate": 1.2962116545394977e-07,
+      "loss": 88.1406,
+      "step": 115660
+    },
+    {
+      "epoch": 0.9568598254539439,
+      "grad_norm": 964.78662109375,
+      "learning_rate": 1.2930230313325908e-07,
+      "loss": 92.0967,
+      "step": 115670
+    },
+    {
+      "epoch": 0.9569425487033131,
+      "grad_norm": 537.0193481445312,
+      "learning_rate": 1.2898382835163093e-07,
+      "loss": 71.7896,
+      "step": 115680
+    },
+    {
+      "epoch": 0.9570252719526823,
+      "grad_norm": 1111.5057373046875,
+      "learning_rate": 1.2866574113440444e-07,
+      "loss": 86.3393,
+      "step": 115690
+    },
+    {
+      "epoch": 0.9571079952020516,
+      "grad_norm": 862.1448364257812,
+      "learning_rate": 1.2834804150688828e-07,
+      "loss": 85.4919,
+      "step": 115700
+    },
+    {
+      "epoch": 0.9571907184514208,
+      "grad_norm": 816.7103271484375,
+      "learning_rate": 1.2803072949436058e-07,
+      "loss": 83.843,
+      "step": 115710
+    },
+    {
+      "epoch": 0.95727344170079,
+      "grad_norm": 776.513671875,
+      "learning_rate": 1.277138051220689e-07,
+      "loss": 98.6719,
+      "step": 115720
+    },
+    {
+      "epoch": 0.9573561649501593,
+      "grad_norm": 596.2342529296875,
+      "learning_rate": 1.2739726841522858e-07,
+      "loss": 82.0256,
+      "step": 115730
+    },
+    {
+      "epoch": 0.9574388881995285,
+      "grad_norm": 825.4178466796875,
+      "learning_rate": 1.270811193990257e-07,
+      "loss": 86.8246,
+      "step": 115740
+    },
+    {
+      "epoch": 0.9575216114488977,
+      "grad_norm": 740.2637329101562,
+      "learning_rate": 1.267653580986139e-07,
+      "loss": 74.5113,
+      "step": 115750
+    },
+    {
+      "epoch": 0.957604334698267,
+      "grad_norm": 432.4769592285156,
+      "learning_rate": 1.2644998453911762e-07,
+      "loss": 73.2089,
+      "step": 115760
+    },
+    {
+      "epoch": 0.9576870579476362,
+      "grad_norm": 2192.880126953125,
+      "learning_rate": 1.2613499874563006e-07,
+      "loss": 113.4138,
+      "step": 115770
+    },
+    {
+      "epoch": 0.9577697811970054,
+      "grad_norm": 771.948486328125,
+      "learning_rate": 1.2582040074321177e-07,
+      "loss": 85.1038,
+      "step": 115780
+    },
+    {
+      "epoch": 0.9578525044463747,
+      "grad_norm": 656.0718383789062,
+      "learning_rate": 1.255061905568955e-07,
+      "loss": 97.9296,
+      "step": 115790
+    },
+    {
+      "epoch": 0.9579352276957439,
+      "grad_norm": 875.1376953125,
+      "learning_rate": 1.251923682116807e-07,
+      "loss": 76.1461,
+      "step": 115800
+    },
+    {
+      "epoch": 0.9580179509451131,
+      "grad_norm": 622.1165771484375,
+      "learning_rate": 1.248789337325368e-07,
+      "loss": 81.3456,
+      "step": 115810
+    },
+    {
+      "epoch": 0.9581006741944823,
+      "grad_norm": 1635.0670166015625,
+      "learning_rate": 1.2456588714440167e-07,
+      "loss": 106.8132,
+      "step": 115820
+    },
+    {
+      "epoch": 0.9581833974438516,
+      "grad_norm": 610.9708251953125,
+      "learning_rate": 1.2425322847218368e-07,
+      "loss": 83.99,
+      "step": 115830
+    },
+    {
+      "epoch": 0.9582661206932208,
+      "grad_norm": 882.1947631835938,
+      "learning_rate": 1.239409577407602e-07,
+      "loss": 74.6342,
+      "step": 115840
+    },
+    {
+      "epoch": 0.95834884394259,
+      "grad_norm": 1295.30126953125,
+      "learning_rate": 1.2362907497497633e-07,
+      "loss": 88.346,
+      "step": 115850
+    },
+    {
+      "epoch": 0.9584315671919593,
+      "grad_norm": 1216.3671875,
+      "learning_rate": 1.233175801996478e-07,
+      "loss": 95.8267,
+      "step": 115860
+    },
+    {
+      "epoch": 0.9585142904413285,
+      "grad_norm": 1159.141845703125,
+      "learning_rate": 1.2300647343955807e-07,
+      "loss": 98.5586,
+      "step": 115870
+    },
+    {
+      "epoch": 0.9585970136906977,
+      "grad_norm": 1170.9654541015625,
+      "learning_rate": 1.2269575471946127e-07,
+      "loss": 77.9647,
+      "step": 115880
+    },
+    {
+      "epoch": 0.958679736940067,
+      "grad_norm": 526.090087890625,
+      "learning_rate": 1.2238542406407984e-07,
+      "loss": 86.4296,
+      "step": 115890
+    },
+    {
+      "epoch": 0.9587624601894362,
+      "grad_norm": 970.7127685546875,
+      "learning_rate": 1.22075481498104e-07,
+      "loss": 70.3665,
+      "step": 115900
+    },
+    {
+      "epoch": 0.9588451834388054,
+      "grad_norm": 747.5088500976562,
+      "learning_rate": 1.2176592704619628e-07,
+      "loss": 67.7107,
+      "step": 115910
+    },
+    {
+      "epoch": 0.9589279066881747,
+      "grad_norm": 1059.5880126953125,
+      "learning_rate": 1.2145676073298473e-07,
+      "loss": 61.7566,
+      "step": 115920
+    },
+    {
+      "epoch": 0.9590106299375439,
+      "grad_norm": 982.748046875,
+      "learning_rate": 1.211479825830697e-07,
+      "loss": 106.074,
+      "step": 115930
+    },
+    {
+      "epoch": 0.9590933531869131,
+      "grad_norm": 676.8903198242188,
+      "learning_rate": 1.2083959262101874e-07,
+      "loss": 96.3958,
+      "step": 115940
+    },
+    {
+      "epoch": 0.9591760764362824,
+      "grad_norm": 1083.4862060546875,
+      "learning_rate": 1.205315908713689e-07,
+      "loss": 84.409,
+      "step": 115950
+    },
+    {
+      "epoch": 0.9592587996856516,
+      "grad_norm": 1195.280517578125,
+      "learning_rate": 1.2022397735862724e-07,
+      "loss": 94.2063,
+      "step": 115960
+    },
+    {
+      "epoch": 0.9593415229350208,
+      "grad_norm": 891.7564086914062,
+      "learning_rate": 1.199167521072686e-07,
+      "loss": 121.5456,
+      "step": 115970
+    },
+    {
+      "epoch": 0.9594242461843902,
+      "grad_norm": 679.1651611328125,
+      "learning_rate": 1.196099151417368e-07,
+      "loss": 70.1369,
+      "step": 115980
+    },
+    {
+      "epoch": 0.9595069694337593,
+      "grad_norm": 1112.198486328125,
+      "learning_rate": 1.1930346648644675e-07,
+      "loss": 92.8188,
+      "step": 115990
+    },
+    {
+      "epoch": 0.9595896926831285,
+      "grad_norm": 540.4926147460938,
+      "learning_rate": 1.1899740616578004e-07,
+      "loss": 93.057,
+      "step": 116000
+    },
+    {
+      "epoch": 0.9596724159324979,
+      "grad_norm": 913.5923461914062,
+      "learning_rate": 1.1869173420408886e-07,
+      "loss": 83.6311,
+      "step": 116010
+    },
+    {
+      "epoch": 0.9597551391818671,
+      "grad_norm": 704.4148559570312,
+      "learning_rate": 1.1838645062569377e-07,
+      "loss": 82.1583,
+      "step": 116020
+    },
+    {
+      "epoch": 0.9598378624312363,
+      "grad_norm": 983.6731567382812,
+      "learning_rate": 1.1808155545488586e-07,
+      "loss": 71.0212,
+      "step": 116030
+    },
+    {
+      "epoch": 0.9599205856806056,
+      "grad_norm": 1065.49072265625,
+      "learning_rate": 1.1777704871592355e-07,
+      "loss": 97.0757,
+      "step": 116040
+    },
+    {
+      "epoch": 0.9600033089299748,
+      "grad_norm": 2115.354248046875,
+      "learning_rate": 1.174729304330352e-07,
+      "loss": 83.9105,
+      "step": 116050
+    },
+    {
+      "epoch": 0.960086032179344,
+      "grad_norm": 895.270751953125,
+      "learning_rate": 1.1716920063041815e-07,
+      "loss": 94.6967,
+      "step": 116060
+    },
+    {
+      "epoch": 0.9601687554287133,
+      "grad_norm": 776.3504638671875,
+      "learning_rate": 1.168658593322386e-07,
+      "loss": 75.9079,
+      "step": 116070
+    },
+    {
+      "epoch": 0.9602514786780825,
+      "grad_norm": 933.9031372070312,
+      "learning_rate": 1.165629065626317e-07,
+      "loss": 89.017,
+      "step": 116080
+    },
+    {
+      "epoch": 0.9603342019274517,
+      "grad_norm": 984.0797119140625,
+      "learning_rate": 1.1626034234570261e-07,
+      "loss": 112.102,
+      "step": 116090
+    },
+    {
+      "epoch": 0.960416925176821,
+      "grad_norm": 813.5068969726562,
+      "learning_rate": 1.1595816670552429e-07,
+      "loss": 51.8942,
+      "step": 116100
+    },
+    {
+      "epoch": 0.9604996484261902,
+      "grad_norm": 717.0478515625,
+      "learning_rate": 1.1565637966613974e-07,
+      "loss": 70.7065,
+      "step": 116110
+    },
+    {
+      "epoch": 0.9605823716755594,
+      "grad_norm": 879.2958984375,
+      "learning_rate": 1.1535498125156197e-07,
+      "loss": 81.3121,
+      "step": 116120
+    },
+    {
+      "epoch": 0.9606650949249287,
+      "grad_norm": 1321.8685302734375,
+      "learning_rate": 1.1505397148577013e-07,
+      "loss": 52.1673,
+      "step": 116130
+    },
+    {
+      "epoch": 0.9607478181742979,
+      "grad_norm": 987.5189208984375,
+      "learning_rate": 1.1475335039271507e-07,
+      "loss": 70.81,
+      "step": 116140
+    },
+    {
+      "epoch": 0.9608305414236671,
+      "grad_norm": 899.2925415039062,
+      "learning_rate": 1.1445311799631598e-07,
+      "loss": 92.3291,
+      "step": 116150
+    },
+    {
+      "epoch": 0.9609132646730364,
+      "grad_norm": 756.8441772460938,
+      "learning_rate": 1.1415327432046041e-07,
+      "loss": 112.0039,
+      "step": 116160
+    },
+    {
+      "epoch": 0.9609959879224056,
+      "grad_norm": 635.4735107421875,
+      "learning_rate": 1.1385381938900597e-07,
+      "loss": 71.928,
+      "step": 116170
+    },
+    {
+      "epoch": 0.9610787111717748,
+      "grad_norm": 4364.9423828125,
+      "learning_rate": 1.1355475322577858e-07,
+      "loss": 115.5651,
+      "step": 116180
+    },
+    {
+      "epoch": 0.9611614344211441,
+      "grad_norm": 619.9732055664062,
+      "learning_rate": 1.1325607585457366e-07,
+      "loss": 106.7126,
+      "step": 116190
+    },
+    {
+      "epoch": 0.9612441576705133,
+      "grad_norm": 897.743408203125,
+      "learning_rate": 1.1295778729915551e-07,
+      "loss": 70.6706,
+      "step": 116200
+    },
+    {
+      "epoch": 0.9613268809198825,
+      "grad_norm": 1066.9715576171875,
+      "learning_rate": 1.1265988758325742e-07,
+      "loss": 81.003,
+      "step": 116210
+    },
+    {
+      "epoch": 0.9614096041692518,
+      "grad_norm": 805.7496337890625,
+      "learning_rate": 1.1236237673058315e-07,
+      "loss": 107.8749,
+      "step": 116220
+    },
+    {
+      "epoch": 0.961492327418621,
+      "grad_norm": 593.794677734375,
+      "learning_rate": 1.1206525476480323e-07,
+      "loss": 103.9387,
+      "step": 116230
+    },
+    {
+      "epoch": 0.9615750506679902,
+      "grad_norm": 1031.0078125,
+      "learning_rate": 1.1176852170955821e-07,
+      "loss": 81.512,
+      "step": 116240
+    },
+    {
+      "epoch": 0.9616577739173595,
+      "grad_norm": 388.48583984375,
+      "learning_rate": 1.1147217758845752e-07,
+      "loss": 71.2489,
+      "step": 116250
+    },
+    {
+      "epoch": 0.9617404971667287,
+      "grad_norm": 1047.6055908203125,
+      "learning_rate": 1.1117622242508064e-07,
+      "loss": 84.0228,
+      "step": 116260
+    },
+    {
+      "epoch": 0.9618232204160979,
+      "grad_norm": 386.8448486328125,
+      "learning_rate": 1.1088065624297484e-07,
+      "loss": 51.314,
+      "step": 116270
+    },
+    {
+      "epoch": 0.9619059436654672,
+      "grad_norm": 1203.3616943359375,
+      "learning_rate": 1.1058547906565743e-07,
+      "loss": 67.1805,
+      "step": 116280
+    },
+    {
+      "epoch": 0.9619886669148364,
+      "grad_norm": 992.1272583007812,
+      "learning_rate": 1.1029069091661459e-07,
+      "loss": 103.1852,
+      "step": 116290
+    },
+    {
+      "epoch": 0.9620713901642056,
+      "grad_norm": 621.4592895507812,
+      "learning_rate": 1.0999629181929983e-07,
+      "loss": 73.4462,
+      "step": 116300
+    },
+    {
+      "epoch": 0.9621541134135749,
+      "grad_norm": 1026.2972412109375,
+      "learning_rate": 1.0970228179713827e-07,
+      "loss": 73.2935,
+      "step": 116310
+    },
+    {
+      "epoch": 0.9622368366629441,
+      "grad_norm": 725.5745239257812,
+      "learning_rate": 1.0940866087352287e-07,
+      "loss": 76.0764,
+      "step": 116320
+    },
+    {
+      "epoch": 0.9623195599123133,
+      "grad_norm": 780.3145141601562,
+      "learning_rate": 1.0911542907181605e-07,
+      "loss": 65.7862,
+      "step": 116330
+    },
+    {
+      "epoch": 0.9624022831616826,
+      "grad_norm": 584.7391357421875,
+      "learning_rate": 1.0882258641534749e-07,
+      "loss": 66.0809,
+      "step": 116340
+    },
+    {
+      "epoch": 0.9624850064110518,
+      "grad_norm": 1102.67919921875,
+      "learning_rate": 1.0853013292741854e-07,
+      "loss": 74.0244,
+      "step": 116350
+    },
+    {
+      "epoch": 0.962567729660421,
+      "grad_norm": 282.3551940917969,
+      "learning_rate": 1.0823806863129838e-07,
+      "loss": 74.1816,
+      "step": 116360
+    },
+    {
+      "epoch": 0.9626504529097903,
+      "grad_norm": 876.3226928710938,
+      "learning_rate": 1.0794639355022507e-07,
+      "loss": 88.0445,
+      "step": 116370
+    },
+    {
+      "epoch": 0.9627331761591595,
+      "grad_norm": 719.6063842773438,
+      "learning_rate": 1.0765510770740506e-07,
+      "loss": 75.5096,
+      "step": 116380
+    },
+    {
+      "epoch": 0.9628158994085287,
+      "grad_norm": 1058.7418212890625,
+      "learning_rate": 1.0736421112601592e-07,
+      "loss": 88.4987,
+      "step": 116390
+    },
+    {
+      "epoch": 0.962898622657898,
+      "grad_norm": 544.888671875,
+      "learning_rate": 1.070737038292019e-07,
+      "loss": 86.7714,
+      "step": 116400
+    },
+    {
+      "epoch": 0.9629813459072672,
+      "grad_norm": 1347.0355224609375,
+      "learning_rate": 1.0678358584007787e-07,
+      "loss": 100.632,
+      "step": 116410
+    },
+    {
+      "epoch": 0.9630640691566364,
+      "grad_norm": 557.1162719726562,
+      "learning_rate": 1.0649385718172756e-07,
+      "loss": 98.0301,
+      "step": 116420
+    },
+    {
+      "epoch": 0.9631467924060058,
+      "grad_norm": 601.667236328125,
+      "learning_rate": 1.0620451787720254e-07,
+      "loss": 55.4105,
+      "step": 116430
+    },
+    {
+      "epoch": 0.963229515655375,
+      "grad_norm": 627.3742065429688,
+      "learning_rate": 1.059155679495244e-07,
+      "loss": 89.8003,
+      "step": 116440
+    },
+    {
+      "epoch": 0.9633122389047442,
+      "grad_norm": 612.9856567382812,
+      "learning_rate": 1.0562700742168364e-07,
+      "loss": 79.46,
+      "step": 116450
+    },
+    {
+      "epoch": 0.9633949621541135,
+      "grad_norm": 969.4473876953125,
+      "learning_rate": 1.0533883631663966e-07,
+      "loss": 105.5287,
+      "step": 116460
+    },
+    {
+      "epoch": 0.9634776854034827,
+      "grad_norm": 1106.258544921875,
+      "learning_rate": 1.0505105465732135e-07,
+      "loss": 85.9468,
+      "step": 116470
+    },
+    {
+      "epoch": 0.9635604086528519,
+      "grad_norm": 1332.35546875,
+      "learning_rate": 1.0476366246662595e-07,
+      "loss": 84.5516,
+      "step": 116480
+    },
+    {
+      "epoch": 0.9636431319022212,
+      "grad_norm": 922.864013671875,
+      "learning_rate": 1.044766597674196e-07,
+      "loss": 80.9004,
+      "step": 116490
+    },
+    {
+      "epoch": 0.9637258551515904,
+      "grad_norm": 458.0718688964844,
+      "learning_rate": 1.0419004658253795e-07,
+      "loss": 81.3689,
+      "step": 116500
+    },
+    {
+      "epoch": 0.9638085784009596,
+      "grad_norm": 821.3236083984375,
+      "learning_rate": 1.0390382293478551e-07,
+      "loss": 77.9656,
+      "step": 116510
+    },
+    {
+      "epoch": 0.9638913016503289,
+      "grad_norm": 836.4533081054688,
+      "learning_rate": 1.036179888469363e-07,
+      "loss": 127.9579,
+      "step": 116520
+    },
+    {
+      "epoch": 0.9639740248996981,
+      "grad_norm": 1029.4652099609375,
+      "learning_rate": 1.0333254434173212e-07,
+      "loss": 84.6967,
+      "step": 116530
+    },
+    {
+      "epoch": 0.9640567481490673,
+      "grad_norm": 869.0244750976562,
+      "learning_rate": 1.0304748944188425e-07,
+      "loss": 76.1726,
+      "step": 116540
+    },
+    {
+      "epoch": 0.9641394713984365,
+      "grad_norm": 824.9259033203125,
+      "learning_rate": 1.0276282417007399e-07,
+      "loss": 85.9181,
+      "step": 116550
+    },
+    {
+      "epoch": 0.9642221946478058,
+      "grad_norm": 630.62109375,
+      "learning_rate": 1.02478548548951e-07,
+      "loss": 92.7414,
+      "step": 116560
+    },
+    {
+      "epoch": 0.964304917897175,
+      "grad_norm": 1001.7970581054688,
+      "learning_rate": 1.0219466260113276e-07,
+      "loss": 93.9342,
+      "step": 116570
+    },
+    {
+      "epoch": 0.9643876411465442,
+      "grad_norm": 711.27392578125,
+      "learning_rate": 1.0191116634920728e-07,
+      "loss": 91.1572,
+      "step": 116580
+    },
+    {
+      "epoch": 0.9644703643959135,
+      "grad_norm": 980.2251586914062,
+      "learning_rate": 1.0162805981573154e-07,
+      "loss": 112.6816,
+      "step": 116590
+    },
+    {
+      "epoch": 0.9645530876452827,
+      "grad_norm": 983.453857421875,
+      "learning_rate": 1.0134534302323029e-07,
+      "loss": 99.6782,
+      "step": 116600
+    },
+    {
+      "epoch": 0.9646358108946519,
+      "grad_norm": 362.87335205078125,
+      "learning_rate": 1.0106301599419832e-07,
+      "loss": 80.6802,
+      "step": 116610
+    },
+    {
+      "epoch": 0.9647185341440212,
+      "grad_norm": 757.7172241210938,
+      "learning_rate": 1.0078107875109878e-07,
+      "loss": 75.8858,
+      "step": 116620
+    },
+    {
+      "epoch": 0.9648012573933904,
+      "grad_norm": 863.5123901367188,
+      "learning_rate": 1.0049953131636481e-07,
+      "loss": 93.2976,
+      "step": 116630
+    },
+    {
+      "epoch": 0.9648839806427596,
+      "grad_norm": 642.6458740234375,
+      "learning_rate": 1.002183737123974e-07,
+      "loss": 81.1781,
+      "step": 116640
+    },
+    {
+      "epoch": 0.9649667038921289,
+      "grad_norm": 626.2711791992188,
+      "learning_rate": 9.993760596156698e-08,
+      "loss": 75.3195,
+      "step": 116650
+    },
+    {
+      "epoch": 0.9650494271414981,
+      "grad_norm": 792.793212890625,
+      "learning_rate": 9.965722808621403e-08,
+      "loss": 73.6435,
+      "step": 116660
+    },
+    {
+      "epoch": 0.9651321503908673,
+      "grad_norm": 803.5825805664062,
+      "learning_rate": 9.937724010864402e-08,
+      "loss": 71.446,
+      "step": 116670
+    },
+    {
+      "epoch": 0.9652148736402366,
+      "grad_norm": 952.573486328125,
+      "learning_rate": 9.909764205113747e-08,
+      "loss": 82.1058,
+      "step": 116680
+    },
+    {
+      "epoch": 0.9652975968896058,
+      "grad_norm": 1239.050048828125,
+      "learning_rate": 9.881843393593882e-08,
+      "loss": 83.7428,
+      "step": 116690
+    },
+    {
+      "epoch": 0.965380320138975,
+      "grad_norm": 1195.249267578125,
+      "learning_rate": 9.853961578526417e-08,
+      "loss": 87.2989,
+      "step": 116700
+    },
+    {
+      "epoch": 0.9654630433883443,
+      "grad_norm": 518.0953369140625,
+      "learning_rate": 9.826118762129799e-08,
+      "loss": 81.8561,
+      "step": 116710
+    },
+    {
+      "epoch": 0.9655457666377135,
+      "grad_norm": 741.3461303710938,
+      "learning_rate": 9.798314946619258e-08,
+      "loss": 93.3796,
+      "step": 116720
+    },
+    {
+      "epoch": 0.9656284898870827,
+      "grad_norm": 578.9274291992188,
+      "learning_rate": 9.770550134207135e-08,
+      "loss": 83.2529,
+      "step": 116730
+    },
+    {
+      "epoch": 0.965711213136452,
+      "grad_norm": 700.8192138671875,
+      "learning_rate": 9.74282432710244e-08,
+      "loss": 67.5045,
+      "step": 116740
+    },
+    {
+      "epoch": 0.9657939363858212,
+      "grad_norm": 965.154541015625,
+      "learning_rate": 9.715137527511298e-08,
+      "loss": 94.1983,
+      "step": 116750
+    },
+    {
+      "epoch": 0.9658766596351904,
+      "grad_norm": 2566.7998046875,
+      "learning_rate": 9.687489737636502e-08,
+      "loss": 91.5708,
+      "step": 116760
+    },
+    {
+      "epoch": 0.9659593828845597,
+      "grad_norm": 775.1097412109375,
+      "learning_rate": 9.659880959677903e-08,
+      "loss": 90.2111,
+      "step": 116770
+    },
+    {
+      "epoch": 0.9660421061339289,
+      "grad_norm": 1181.168212890625,
+      "learning_rate": 9.632311195832245e-08,
+      "loss": 104.988,
+      "step": 116780
+    },
+    {
+      "epoch": 0.9661248293832981,
+      "grad_norm": 812.8969116210938,
+      "learning_rate": 9.604780448293105e-08,
+      "loss": 74.2418,
+      "step": 116790
+    },
+    {
+      "epoch": 0.9662075526326674,
+      "grad_norm": 762.244384765625,
+      "learning_rate": 9.57728871925101e-08,
+      "loss": 75.0032,
+      "step": 116800
+    },
+    {
+      "epoch": 0.9662902758820366,
+      "grad_norm": 1320.201171875,
+      "learning_rate": 9.549836010893265e-08,
+      "loss": 107.8506,
+      "step": 116810
+    },
+    {
+      "epoch": 0.9663729991314058,
+      "grad_norm": 821.8568725585938,
+      "learning_rate": 9.522422325404234e-08,
+      "loss": 91.6939,
+      "step": 116820
+    },
+    {
+      "epoch": 0.9664557223807752,
+      "grad_norm": 1307.53466796875,
+      "learning_rate": 9.495047664965063e-08,
+      "loss": 97.0378,
+      "step": 116830
+    },
+    {
+      "epoch": 0.9665384456301443,
+      "grad_norm": 891.2249145507812,
+      "learning_rate": 9.467712031753839e-08,
+      "loss": 80.264,
+      "step": 116840
+    },
+    {
+      "epoch": 0.9666211688795135,
+      "grad_norm": 5029.44189453125,
+      "learning_rate": 9.440415427945548e-08,
+      "loss": 183.4319,
+      "step": 116850
+    },
+    {
+      "epoch": 0.9667038921288829,
+      "grad_norm": 1075.81103515625,
+      "learning_rate": 9.413157855712007e-08,
+      "loss": 81.115,
+      "step": 116860
+    },
+    {
+      "epoch": 0.9667866153782521,
+      "grad_norm": 3139.591796875,
+      "learning_rate": 9.385939317221926e-08,
+      "loss": 154.7868,
+      "step": 116870
+    },
+    {
+      "epoch": 0.9668693386276213,
+      "grad_norm": 621.9048461914062,
+      "learning_rate": 9.358759814641127e-08,
+      "loss": 72.0611,
+      "step": 116880
+    },
+    {
+      "epoch": 0.9669520618769906,
+      "grad_norm": 1159.2205810546875,
+      "learning_rate": 9.331619350132049e-08,
+      "loss": 98.6789,
+      "step": 116890
+    },
+    {
+      "epoch": 0.9670347851263598,
+      "grad_norm": 1112.920654296875,
+      "learning_rate": 9.304517925854184e-08,
+      "loss": 83.0609,
+      "step": 116900
+    },
+    {
+      "epoch": 0.967117508375729,
+      "grad_norm": 585.3701171875,
+      "learning_rate": 9.277455543963809e-08,
+      "loss": 100.5077,
+      "step": 116910
+    },
+    {
+      "epoch": 0.9672002316250983,
+      "grad_norm": 488.8531494140625,
+      "learning_rate": 9.250432206614258e-08,
+      "loss": 84.3317,
+      "step": 116920
+    },
+    {
+      "epoch": 0.9672829548744675,
+      "grad_norm": 479.1993103027344,
+      "learning_rate": 9.22344791595553e-08,
+      "loss": 71.9188,
+      "step": 116930
+    },
+    {
+      "epoch": 0.9673656781238367,
+      "grad_norm": 1235.1121826171875,
+      "learning_rate": 9.196502674134689e-08,
+      "loss": 94.2771,
+      "step": 116940
+    },
+    {
+      "epoch": 0.967448401373206,
+      "grad_norm": 991.1707153320312,
+      "learning_rate": 9.169596483295628e-08,
+      "loss": 105.2712,
+      "step": 116950
+    },
+    {
+      "epoch": 0.9675311246225752,
+      "grad_norm": 1200.013427734375,
+      "learning_rate": 9.142729345579193e-08,
+      "loss": 92.9962,
+      "step": 116960
+    },
+    {
+      "epoch": 0.9676138478719444,
+      "grad_norm": 757.19580078125,
+      "learning_rate": 9.115901263123006e-08,
+      "loss": 83.3054,
+      "step": 116970
+    },
+    {
+      "epoch": 0.9676965711213137,
+      "grad_norm": 1006.8023071289062,
+      "learning_rate": 9.089112238061692e-08,
+      "loss": 100.9199,
+      "step": 116980
+    },
+    {
+      "epoch": 0.9677792943706829,
+      "grad_norm": 668.6526489257812,
+      "learning_rate": 9.062362272526825e-08,
+      "loss": 91.3445,
+      "step": 116990
+    },
+    {
+      "epoch": 0.9678620176200521,
+      "grad_norm": 1176.7666015625,
+      "learning_rate": 9.035651368646647e-08,
+      "loss": 89.7577,
+      "step": 117000
+    },
+    {
+      "epoch": 0.9679447408694214,
+      "grad_norm": 1044.1488037109375,
+      "learning_rate": 9.008979528546513e-08,
+      "loss": 88.0212,
+      "step": 117010
+    },
+    {
+      "epoch": 0.9680274641187906,
+      "grad_norm": 831.52197265625,
+      "learning_rate": 8.982346754348503e-08,
+      "loss": 84.6709,
+      "step": 117020
+    },
+    {
+      "epoch": 0.9681101873681598,
+      "grad_norm": 926.8240356445312,
+      "learning_rate": 8.955753048171645e-08,
+      "loss": 73.5235,
+      "step": 117030
+    },
+    {
+      "epoch": 0.9681929106175291,
+      "grad_norm": 1101.0938720703125,
+      "learning_rate": 8.929198412131968e-08,
+      "loss": 92.8208,
+      "step": 117040
+    },
+    {
+      "epoch": 0.9682756338668983,
+      "grad_norm": 721.4671020507812,
+      "learning_rate": 8.902682848342282e-08,
+      "loss": 74.5471,
+      "step": 117050
+    },
+    {
+      "epoch": 0.9683583571162675,
+      "grad_norm": 922.5578002929688,
+      "learning_rate": 8.876206358912232e-08,
+      "loss": 87.3108,
+      "step": 117060
+    },
+    {
+      "epoch": 0.9684410803656368,
+      "grad_norm": 815.2578735351562,
+      "learning_rate": 8.849768945948522e-08,
+      "loss": 89.3768,
+      "step": 117070
+    },
+    {
+      "epoch": 0.968523803615006,
+      "grad_norm": 905.2554321289062,
+      "learning_rate": 8.823370611554638e-08,
+      "loss": 80.9791,
+      "step": 117080
+    },
+    {
+      "epoch": 0.9686065268643752,
+      "grad_norm": 972.8367919921875,
+      "learning_rate": 8.797011357830953e-08,
+      "loss": 137.1645,
+      "step": 117090
+    },
+    {
+      "epoch": 0.9686892501137445,
+      "grad_norm": 931.77392578125,
+      "learning_rate": 8.770691186874791e-08,
+      "loss": 84.7926,
+      "step": 117100
+    },
+    {
+      "epoch": 0.9687719733631137,
+      "grad_norm": 788.7830200195312,
+      "learning_rate": 8.744410100780254e-08,
+      "loss": 58.1295,
+      "step": 117110
+    },
+    {
+      "epoch": 0.9688546966124829,
+      "grad_norm": 2165.523681640625,
+      "learning_rate": 8.718168101638446e-08,
+      "loss": 107.7649,
+      "step": 117120
+    },
+    {
+      "epoch": 0.9689374198618522,
+      "grad_norm": 1699.0238037109375,
+      "learning_rate": 8.69196519153731e-08,
+      "loss": 119.7703,
+      "step": 117130
+    },
+    {
+      "epoch": 0.9690201431112214,
+      "grad_norm": 1005.6736450195312,
+      "learning_rate": 8.665801372561677e-08,
+      "loss": 91.6887,
+      "step": 117140
+    },
+    {
+      "epoch": 0.9691028663605906,
+      "grad_norm": 997.28857421875,
+      "learning_rate": 8.639676646793382e-08,
+      "loss": 78.077,
+      "step": 117150
+    },
+    {
+      "epoch": 0.9691855896099599,
+      "grad_norm": 868.7511596679688,
+      "learning_rate": 8.613591016310874e-08,
+      "loss": 78.0869,
+      "step": 117160
+    },
+    {
+      "epoch": 0.9692683128593291,
+      "grad_norm": 1064.0950927734375,
+      "learning_rate": 8.58754448318988e-08,
+      "loss": 82.3546,
+      "step": 117170
+    },
+    {
+      "epoch": 0.9693510361086983,
+      "grad_norm": 615.5275268554688,
+      "learning_rate": 8.561537049502688e-08,
+      "loss": 108.2288,
+      "step": 117180
+    },
+    {
+      "epoch": 0.9694337593580676,
+      "grad_norm": 714.61328125,
+      "learning_rate": 8.535568717318533e-08,
+      "loss": 79.6565,
+      "step": 117190
+    },
+    {
+      "epoch": 0.9695164826074368,
+      "grad_norm": 695.4380493164062,
+      "learning_rate": 8.509639488703703e-08,
+      "loss": 81.4569,
+      "step": 117200
+    },
+    {
+      "epoch": 0.969599205856806,
+      "grad_norm": 1146.906005859375,
+      "learning_rate": 8.483749365721217e-08,
+      "loss": 87.309,
+      "step": 117210
+    },
+    {
+      "epoch": 0.9696819291061753,
+      "grad_norm": 1346.7725830078125,
+      "learning_rate": 8.457898350430982e-08,
+      "loss": 99.9304,
+      "step": 117220
+    },
+    {
+      "epoch": 0.9697646523555445,
+      "grad_norm": 692.3303833007812,
+      "learning_rate": 8.432086444889964e-08,
+      "loss": 82.1651,
+      "step": 117230
+    },
+    {
+      "epoch": 0.9698473756049137,
+      "grad_norm": 876.026123046875,
+      "learning_rate": 8.406313651151799e-08,
+      "loss": 124.7427,
+      "step": 117240
+    },
+    {
+      "epoch": 0.969930098854283,
+      "grad_norm": 654.4212646484375,
+      "learning_rate": 8.380579971267178e-08,
+      "loss": 60.3869,
+      "step": 117250
+    },
+    {
+      "epoch": 0.9700128221036523,
+      "grad_norm": 741.6707763671875,
+      "learning_rate": 8.354885407283574e-08,
+      "loss": 86.2612,
+      "step": 117260
+    },
+    {
+      "epoch": 0.9700955453530214,
+      "grad_norm": 688.7070922851562,
+      "learning_rate": 8.329229961245355e-08,
+      "loss": 73.4818,
+      "step": 117270
+    },
+    {
+      "epoch": 0.9701782686023906,
+      "grad_norm": 632.3734130859375,
+      "learning_rate": 8.303613635193886e-08,
+      "loss": 71.4801,
+      "step": 117280
+    },
+    {
+      "epoch": 0.97026099185176,
+      "grad_norm": 1043.305419921875,
+      "learning_rate": 8.278036431167313e-08,
+      "loss": 108.113,
+      "step": 117290
+    },
+    {
+      "epoch": 0.9703437151011292,
+      "grad_norm": 680.9979248046875,
+      "learning_rate": 8.252498351200621e-08,
+      "loss": 82.1075,
+      "step": 117300
+    },
+    {
+      "epoch": 0.9704264383504984,
+      "grad_norm": 1170.890380859375,
+      "learning_rate": 8.226999397325852e-08,
+      "loss": 78.4652,
+      "step": 117310
+    },
+    {
+      "epoch": 0.9705091615998677,
+      "grad_norm": 801.5345458984375,
+      "learning_rate": 8.201539571571826e-08,
+      "loss": 93.103,
+      "step": 117320
+    },
+    {
+      "epoch": 0.9705918848492369,
+      "grad_norm": 694.1389770507812,
+      "learning_rate": 8.176118875964201e-08,
+      "loss": 105.483,
+      "step": 117330
+    },
+    {
+      "epoch": 0.9706746080986061,
+      "grad_norm": 702.7589111328125,
+      "learning_rate": 8.150737312525692e-08,
+      "loss": 80.5185,
+      "step": 117340
+    },
+    {
+      "epoch": 0.9707573313479754,
+      "grad_norm": 735.4678955078125,
+      "learning_rate": 8.125394883275683e-08,
+      "loss": 95.8936,
+      "step": 117350
+    },
+    {
+      "epoch": 0.9708400545973446,
+      "grad_norm": 704.5406494140625,
+      "learning_rate": 8.100091590230618e-08,
+      "loss": 103.3761,
+      "step": 117360
+    },
+    {
+      "epoch": 0.9709227778467138,
+      "grad_norm": 966.3074340820312,
+      "learning_rate": 8.07482743540372e-08,
+      "loss": 101.2361,
+      "step": 117370
+    },
+    {
+      "epoch": 0.9710055010960831,
+      "grad_norm": 1027.426513671875,
+      "learning_rate": 8.049602420805214e-08,
+      "loss": 84.4538,
+      "step": 117380
+    },
+    {
+      "epoch": 0.9710882243454523,
+      "grad_norm": 901.7965087890625,
+      "learning_rate": 8.024416548442104e-08,
+      "loss": 86.5814,
+      "step": 117390
+    },
+    {
+      "epoch": 0.9711709475948215,
+      "grad_norm": 1091.1844482421875,
+      "learning_rate": 7.99926982031829e-08,
+      "loss": 99.3349,
+      "step": 117400
+    },
+    {
+      "epoch": 0.9712536708441908,
+      "grad_norm": 709.89697265625,
+      "learning_rate": 7.974162238434557e-08,
+      "loss": 76.4967,
+      "step": 117410
+    },
+    {
+      "epoch": 0.97133639409356,
+      "grad_norm": 859.7228393554688,
+      "learning_rate": 7.949093804788699e-08,
+      "loss": 87.3181,
+      "step": 117420
+    },
+    {
+      "epoch": 0.9714191173429292,
+      "grad_norm": 662.0986328125,
+      "learning_rate": 7.924064521375174e-08,
+      "loss": 80.255,
+      "step": 117430
+    },
+    {
+      "epoch": 0.9715018405922985,
+      "grad_norm": 804.8336791992188,
+      "learning_rate": 7.899074390185557e-08,
+      "loss": 82.7631,
+      "step": 117440
+    },
+    {
+      "epoch": 0.9715845638416677,
+      "grad_norm": 800.1364135742188,
+      "learning_rate": 7.874123413208145e-08,
+      "loss": 74.2819,
+      "step": 117450
+    },
+    {
+      "epoch": 0.9716672870910369,
+      "grad_norm": 465.7279052734375,
+      "learning_rate": 7.849211592428186e-08,
+      "loss": 66.4609,
+      "step": 117460
+    },
+    {
+      "epoch": 0.9717500103404062,
+      "grad_norm": 444.983154296875,
+      "learning_rate": 7.824338929827813e-08,
+      "loss": 72.5292,
+      "step": 117470
+    },
+    {
+      "epoch": 0.9718327335897754,
+      "grad_norm": 253.49969482421875,
+      "learning_rate": 7.799505427386001e-08,
+      "loss": 67.4403,
+      "step": 117480
+    },
+    {
+      "epoch": 0.9719154568391446,
+      "grad_norm": 559.1021118164062,
+      "learning_rate": 7.774711087078612e-08,
+      "loss": 71.8914,
+      "step": 117490
+    },
+    {
+      "epoch": 0.9719981800885139,
+      "grad_norm": 1051.5477294921875,
+      "learning_rate": 7.749955910878459e-08,
+      "loss": 80.7977,
+      "step": 117500
+    },
+    {
+      "epoch": 0.9720809033378831,
+      "grad_norm": 739.0164794921875,
+      "learning_rate": 7.725239900755244e-08,
+      "loss": 62.2263,
+      "step": 117510
+    },
+    {
+      "epoch": 0.9721636265872523,
+      "grad_norm": 1051.24560546875,
+      "learning_rate": 7.700563058675448e-08,
+      "loss": 72.5346,
+      "step": 117520
+    },
+    {
+      "epoch": 0.9722463498366216,
+      "grad_norm": 876.2381591796875,
+      "learning_rate": 7.67592538660239e-08,
+      "loss": 93.1742,
+      "step": 117530
+    },
+    {
+      "epoch": 0.9723290730859908,
+      "grad_norm": 892.1935424804688,
+      "learning_rate": 7.651326886496613e-08,
+      "loss": 75.8199,
+      "step": 117540
+    },
+    {
+      "epoch": 0.97241179633536,
+      "grad_norm": 513.9732055664062,
+      "learning_rate": 7.626767560315107e-08,
+      "loss": 75.7576,
+      "step": 117550
+    },
+    {
+      "epoch": 0.9724945195847293,
+      "grad_norm": 822.49267578125,
+      "learning_rate": 7.602247410012032e-08,
+      "loss": 82.0519,
+      "step": 117560
+    },
+    {
+      "epoch": 0.9725772428340985,
+      "grad_norm": 927.3632202148438,
+      "learning_rate": 7.577766437538325e-08,
+      "loss": 98.2942,
+      "step": 117570
+    },
+    {
+      "epoch": 0.9726599660834677,
+      "grad_norm": 772.4122314453125,
+      "learning_rate": 7.553324644841875e-08,
+      "loss": 106.7597,
+      "step": 117580
+    },
+    {
+      "epoch": 0.972742689332837,
+      "grad_norm": 796.9185791015625,
+      "learning_rate": 7.528922033867347e-08,
+      "loss": 103.0361,
+      "step": 117590
+    },
+    {
+      "epoch": 0.9728254125822062,
+      "grad_norm": 881.1489868164062,
+      "learning_rate": 7.5045586065563e-08,
+      "loss": 84.0167,
+      "step": 117600
+    },
+    {
+      "epoch": 0.9729081358315754,
+      "grad_norm": 965.630615234375,
+      "learning_rate": 7.480234364847349e-08,
+      "loss": 93.8757,
+      "step": 117610
+    },
+    {
+      "epoch": 0.9729908590809447,
+      "grad_norm": 820.9173583984375,
+      "learning_rate": 7.455949310675725e-08,
+      "loss": 76.266,
+      "step": 117620
+    },
+    {
+      "epoch": 0.9730735823303139,
+      "grad_norm": 748.765625,
+      "learning_rate": 7.43170344597377e-08,
+      "loss": 102.896,
+      "step": 117630
+    },
+    {
+      "epoch": 0.9731563055796831,
+      "grad_norm": 604.3673095703125,
+      "learning_rate": 7.407496772670609e-08,
+      "loss": 62.3851,
+      "step": 117640
+    },
+    {
+      "epoch": 0.9732390288290524,
+      "grad_norm": 783.4528198242188,
+      "learning_rate": 7.383329292692198e-08,
+      "loss": 89.3814,
+      "step": 117650
+    },
+    {
+      "epoch": 0.9733217520784216,
+      "grad_norm": 1100.453857421875,
+      "learning_rate": 7.359201007961503e-08,
+      "loss": 75.2225,
+      "step": 117660
+    },
+    {
+      "epoch": 0.9734044753277908,
+      "grad_norm": 647.7412109375,
+      "learning_rate": 7.335111920398263e-08,
+      "loss": 84.3431,
+      "step": 117670
+    },
+    {
+      "epoch": 0.9734871985771602,
+      "grad_norm": 796.82421875,
+      "learning_rate": 7.311062031919114e-08,
+      "loss": 103.5371,
+      "step": 117680
+    },
+    {
+      "epoch": 0.9735699218265293,
+      "grad_norm": 915.2636108398438,
+      "learning_rate": 7.28705134443769e-08,
+      "loss": 82.3625,
+      "step": 117690
+    },
+    {
+      "epoch": 0.9736526450758985,
+      "grad_norm": 1407.7706298828125,
+      "learning_rate": 7.263079859864298e-08,
+      "loss": 99.5093,
+      "step": 117700
+    },
+    {
+      "epoch": 0.9737353683252679,
+      "grad_norm": 515.5316162109375,
+      "learning_rate": 7.239147580106242e-08,
+      "loss": 85.3771,
+      "step": 117710
+    },
+    {
+      "epoch": 0.9738180915746371,
+      "grad_norm": 635.7969970703125,
+      "learning_rate": 7.215254507067782e-08,
+      "loss": 83.8497,
+      "step": 117720
+    },
+    {
+      "epoch": 0.9739008148240063,
+      "grad_norm": 915.8248901367188,
+      "learning_rate": 7.191400642649893e-08,
+      "loss": 72.568,
+      "step": 117730
+    },
+    {
+      "epoch": 0.9739835380733756,
+      "grad_norm": 774.489013671875,
+      "learning_rate": 7.167585988750669e-08,
+      "loss": 101.6645,
+      "step": 117740
+    },
+    {
+      "epoch": 0.9740662613227448,
+      "grad_norm": 866.99560546875,
+      "learning_rate": 7.143810547264762e-08,
+      "loss": 61.4445,
+      "step": 117750
+    },
+    {
+      "epoch": 0.974148984572114,
+      "grad_norm": 642.11474609375,
+      "learning_rate": 7.120074320083991e-08,
+      "loss": 64.4058,
+      "step": 117760
+    },
+    {
+      "epoch": 0.9742317078214833,
+      "grad_norm": 1115.784423828125,
+      "learning_rate": 7.096377309096846e-08,
+      "loss": 76.0183,
+      "step": 117770
+    },
+    {
+      "epoch": 0.9743144310708525,
+      "grad_norm": 734.0995483398438,
+      "learning_rate": 7.072719516188875e-08,
+      "loss": 96.2746,
+      "step": 117780
+    },
+    {
+      "epoch": 0.9743971543202217,
+      "grad_norm": 525.3031616210938,
+      "learning_rate": 7.049100943242404e-08,
+      "loss": 76.8062,
+      "step": 117790
+    },
+    {
+      "epoch": 0.974479877569591,
+      "grad_norm": 831.6923217773438,
+      "learning_rate": 7.025521592136597e-08,
+      "loss": 73.2237,
+      "step": 117800
+    },
+    {
+      "epoch": 0.9745626008189602,
+      "grad_norm": 1876.879150390625,
+      "learning_rate": 7.001981464747565e-08,
+      "loss": 87.2098,
+      "step": 117810
+    },
+    {
+      "epoch": 0.9746453240683294,
+      "grad_norm": 811.3916015625,
+      "learning_rate": 6.978480562948309e-08,
+      "loss": 94.3064,
+      "step": 117820
+    },
+    {
+      "epoch": 0.9747280473176987,
+      "grad_norm": 1149.4193115234375,
+      "learning_rate": 6.955018888608722e-08,
+      "loss": 100.6052,
+      "step": 117830
+    },
+    {
+      "epoch": 0.9748107705670679,
+      "grad_norm": 863.6624145507812,
+      "learning_rate": 6.931596443595478e-08,
+      "loss": 109.152,
+      "step": 117840
+    },
+    {
+      "epoch": 0.9748934938164371,
+      "grad_norm": 771.5863647460938,
+      "learning_rate": 6.908213229772254e-08,
+      "loss": 67.1083,
+      "step": 117850
+    },
+    {
+      "epoch": 0.9749762170658064,
+      "grad_norm": 581.7706298828125,
+      "learning_rate": 6.884869248999504e-08,
+      "loss": 92.5001,
+      "step": 117860
+    },
+    {
+      "epoch": 0.9750589403151756,
+      "grad_norm": 790.7437133789062,
+      "learning_rate": 6.861564503134688e-08,
+      "loss": 98.9197,
+      "step": 117870
+    },
+    {
+      "epoch": 0.9751416635645448,
+      "grad_norm": 710.2706909179688,
+      "learning_rate": 6.838298994031933e-08,
+      "loss": 100.5049,
+      "step": 117880
+    },
+    {
+      "epoch": 0.9752243868139141,
+      "grad_norm": 656.0684814453125,
+      "learning_rate": 6.815072723542426e-08,
+      "loss": 85.3911,
+      "step": 117890
+    },
+    {
+      "epoch": 0.9753071100632833,
+      "grad_norm": 482.3664245605469,
+      "learning_rate": 6.791885693514134e-08,
+      "loss": 58.3653,
+      "step": 117900
+    },
+    {
+      "epoch": 0.9753898333126525,
+      "grad_norm": 1170.74169921875,
+      "learning_rate": 6.768737905792022e-08,
+      "loss": 84.8319,
+      "step": 117910
+    },
+    {
+      "epoch": 0.9754725565620218,
+      "grad_norm": 755.68017578125,
+      "learning_rate": 6.745629362217731e-08,
+      "loss": 113.3002,
+      "step": 117920
+    },
+    {
+      "epoch": 0.975555279811391,
+      "grad_norm": 776.3069458007812,
+      "learning_rate": 6.722560064630013e-08,
+      "loss": 65.8851,
+      "step": 117930
+    },
+    {
+      "epoch": 0.9756380030607602,
+      "grad_norm": 1081.909912109375,
+      "learning_rate": 6.699530014864397e-08,
+      "loss": 92.803,
+      "step": 117940
+    },
+    {
+      "epoch": 0.9757207263101295,
+      "grad_norm": 552.3186645507812,
+      "learning_rate": 6.676539214753253e-08,
+      "loss": 92.8013,
+      "step": 117950
+    },
+    {
+      "epoch": 0.9758034495594987,
+      "grad_norm": 771.9279174804688,
+      "learning_rate": 6.653587666125782e-08,
+      "loss": 74.0391,
+      "step": 117960
+    },
+    {
+      "epoch": 0.9758861728088679,
+      "grad_norm": 670.3401489257812,
+      "learning_rate": 6.630675370808193e-08,
+      "loss": 77.2455,
+      "step": 117970
+    },
+    {
+      "epoch": 0.9759688960582372,
+      "grad_norm": 1167.330810546875,
+      "learning_rate": 6.607802330623525e-08,
+      "loss": 87.141,
+      "step": 117980
+    },
+    {
+      "epoch": 0.9760516193076064,
+      "grad_norm": 347.4068908691406,
+      "learning_rate": 6.584968547391657e-08,
+      "loss": 72.6703,
+      "step": 117990
+    },
+    {
+      "epoch": 0.9761343425569756,
+      "grad_norm": 1017.5067749023438,
+      "learning_rate": 6.562174022929358e-08,
+      "loss": 84.4651,
+      "step": 118000
+    },
+    {
+      "epoch": 0.9762170658063448,
+      "grad_norm": 818.614013671875,
+      "learning_rate": 6.539418759050286e-08,
+      "loss": 78.0972,
+      "step": 118010
+    },
+    {
+      "epoch": 0.9762997890557141,
+      "grad_norm": 870.0516357421875,
+      "learning_rate": 6.516702757564941e-08,
+      "loss": 60.4278,
+      "step": 118020
+    },
+    {
+      "epoch": 0.9763825123050833,
+      "grad_norm": 1681.1678466796875,
+      "learning_rate": 6.494026020280875e-08,
+      "loss": 94.2777,
+      "step": 118030
+    },
+    {
+      "epoch": 0.9764652355544525,
+      "grad_norm": 767.0552978515625,
+      "learning_rate": 6.471388549002255e-08,
+      "loss": 96.2769,
+      "step": 118040
+    },
+    {
+      "epoch": 0.9765479588038218,
+      "grad_norm": 904.4722900390625,
+      "learning_rate": 6.448790345530253e-08,
+      "loss": 73.3652,
+      "step": 118050
+    },
+    {
+      "epoch": 0.976630682053191,
+      "grad_norm": 916.374267578125,
+      "learning_rate": 6.426231411662876e-08,
+      "loss": 75.4504,
+      "step": 118060
+    },
+    {
+      "epoch": 0.9767134053025602,
+      "grad_norm": 327.85504150390625,
+      "learning_rate": 6.403711749195073e-08,
+      "loss": 55.0961,
+      "step": 118070
+    },
+    {
+      "epoch": 0.9767961285519295,
+      "grad_norm": 794.025146484375,
+      "learning_rate": 6.381231359918638e-08,
+      "loss": 82.1348,
+      "step": 118080
+    },
+    {
+      "epoch": 0.9768788518012987,
+      "grad_norm": 971.2992553710938,
+      "learning_rate": 6.358790245622193e-08,
+      "loss": 65.286,
+      "step": 118090
+    },
+    {
+      "epoch": 0.9769615750506679,
+      "grad_norm": 382.24456787109375,
+      "learning_rate": 6.336388408091366e-08,
+      "loss": 81.6362,
+      "step": 118100
+    },
+    {
+      "epoch": 0.9770442983000373,
+      "grad_norm": 700.8916625976562,
+      "learning_rate": 6.314025849108397e-08,
+      "loss": 66.6585,
+      "step": 118110
+    },
+    {
+      "epoch": 0.9771270215494064,
+      "grad_norm": 598.8179321289062,
+      "learning_rate": 6.291702570452806e-08,
+      "loss": 87.4747,
+      "step": 118120
+    },
+    {
+      "epoch": 0.9772097447987756,
+      "grad_norm": 832.1001586914062,
+      "learning_rate": 6.269418573900565e-08,
+      "loss": 78.9643,
+      "step": 118130
+    },
+    {
+      "epoch": 0.977292468048145,
+      "grad_norm": 1039.1939697265625,
+      "learning_rate": 6.247173861224753e-08,
+      "loss": 125.8844,
+      "step": 118140
+    },
+    {
+      "epoch": 0.9773751912975142,
+      "grad_norm": 537.9774169921875,
+      "learning_rate": 6.224968434195289e-08,
+      "loss": 60.1959,
+      "step": 118150
+    },
+    {
+      "epoch": 0.9774579145468834,
+      "grad_norm": 1239.988037109375,
+      "learning_rate": 6.202802294578981e-08,
+      "loss": 96.3588,
+      "step": 118160
+    },
+    {
+      "epoch": 0.9775406377962527,
+      "grad_norm": 629.1329345703125,
+      "learning_rate": 6.180675444139527e-08,
+      "loss": 78.8181,
+      "step": 118170
+    },
+    {
+      "epoch": 0.9776233610456219,
+      "grad_norm": 1456.6697998046875,
+      "learning_rate": 6.158587884637357e-08,
+      "loss": 83.5854,
+      "step": 118180
+    },
+    {
+      "epoch": 0.9777060842949911,
+      "grad_norm": 694.644775390625,
+      "learning_rate": 6.136539617829895e-08,
+      "loss": 75.1875,
+      "step": 118190
+    },
+    {
+      "epoch": 0.9777888075443604,
+      "grad_norm": 1134.1533203125,
+      "learning_rate": 6.114530645471461e-08,
+      "loss": 95.2716,
+      "step": 118200
+    },
+    {
+      "epoch": 0.9778715307937296,
+      "grad_norm": 819.7823486328125,
+      "learning_rate": 6.09256096931321e-08,
+      "loss": 93.0044,
+      "step": 118210
+    },
+    {
+      "epoch": 0.9779542540430988,
+      "grad_norm": 874.4256591796875,
+      "learning_rate": 6.070630591103188e-08,
+      "loss": 77.5878,
+      "step": 118220
+    },
+    {
+      "epoch": 0.9780369772924681,
+      "grad_norm": 941.7909545898438,
+      "learning_rate": 6.048739512586221e-08,
+      "loss": 106.6546,
+      "step": 118230
+    },
+    {
+      "epoch": 0.9781197005418373,
+      "grad_norm": 496.1390075683594,
+      "learning_rate": 6.026887735504083e-08,
+      "loss": 70.8411,
+      "step": 118240
+    },
+    {
+      "epoch": 0.9782024237912065,
+      "grad_norm": 1133.30517578125,
+      "learning_rate": 6.005075261595495e-08,
+      "loss": 62.7111,
+      "step": 118250
+    },
+    {
+      "epoch": 0.9782851470405758,
+      "grad_norm": 1455.4931640625,
+      "learning_rate": 5.983302092595955e-08,
+      "loss": 82.0333,
+      "step": 118260
+    },
+    {
+      "epoch": 0.978367870289945,
+      "grad_norm": 369.7554626464844,
+      "learning_rate": 5.961568230237858e-08,
+      "loss": 109.3388,
+      "step": 118270
+    },
+    {
+      "epoch": 0.9784505935393142,
+      "grad_norm": 983.8111572265625,
+      "learning_rate": 5.939873676250374e-08,
+      "loss": 74.7944,
+      "step": 118280
+    },
+    {
+      "epoch": 0.9785333167886835,
+      "grad_norm": 437.4408874511719,
+      "learning_rate": 5.91821843235979e-08,
+      "loss": 82.0863,
+      "step": 118290
+    },
+    {
+      "epoch": 0.9786160400380527,
+      "grad_norm": 1494.396484375,
+      "learning_rate": 5.8966025002889505e-08,
+      "loss": 103.6332,
+      "step": 118300
+    },
+    {
+      "epoch": 0.9786987632874219,
+      "grad_norm": 637.7553100585938,
+      "learning_rate": 5.8750258817578676e-08,
+      "loss": 59.2877,
+      "step": 118310
+    },
+    {
+      "epoch": 0.9787814865367912,
+      "grad_norm": 802.739990234375,
+      "learning_rate": 5.85348857848328e-08,
+      "loss": 55.1885,
+      "step": 118320
+    },
+    {
+      "epoch": 0.9788642097861604,
+      "grad_norm": 941.2757568359375,
+      "learning_rate": 5.8319905921787603e-08,
+      "loss": 69.7701,
+      "step": 118330
+    },
+    {
+      "epoch": 0.9789469330355296,
+      "grad_norm": 1327.10888671875,
+      "learning_rate": 5.810531924554774e-08,
+      "loss": 87.9876,
+      "step": 118340
+    },
+    {
+      "epoch": 0.9790296562848989,
+      "grad_norm": 1318.0863037109375,
+      "learning_rate": 5.7891125773187896e-08,
+      "loss": 139.0294,
+      "step": 118350
+    },
+    {
+      "epoch": 0.9791123795342681,
+      "grad_norm": 1360.8392333984375,
+      "learning_rate": 5.7677325521749983e-08,
+      "loss": 68.9617,
+      "step": 118360
+    },
+    {
+      "epoch": 0.9791951027836373,
+      "grad_norm": 1109.0506591796875,
+      "learning_rate": 5.746391850824484e-08,
+      "loss": 122.1048,
+      "step": 118370
+    },
+    {
+      "epoch": 0.9792778260330066,
+      "grad_norm": 1080.967041015625,
+      "learning_rate": 5.725090474965278e-08,
+      "loss": 67.3166,
+      "step": 118380
+    },
+    {
+      "epoch": 0.9793605492823758,
+      "grad_norm": 738.1102905273438,
+      "learning_rate": 5.703828426292191e-08,
+      "loss": 103.1861,
+      "step": 118390
+    },
+    {
+      "epoch": 0.979443272531745,
+      "grad_norm": 1060.9788818359375,
+      "learning_rate": 5.6826057064969244e-08,
+      "loss": 83.6084,
+      "step": 118400
+    },
+    {
+      "epoch": 0.9795259957811143,
+      "grad_norm": 731.0538330078125,
+      "learning_rate": 5.6614223172681836e-08,
+      "loss": 100.0388,
+      "step": 118410
+    },
+    {
+      "epoch": 0.9796087190304835,
+      "grad_norm": 1262.67578125,
+      "learning_rate": 5.640278260291287e-08,
+      "loss": 83.6562,
+      "step": 118420
+    },
+    {
+      "epoch": 0.9796914422798527,
+      "grad_norm": 726.7060546875,
+      "learning_rate": 5.6191735372487235e-08,
+      "loss": 89.0712,
+      "step": 118430
+    },
+    {
+      "epoch": 0.979774165529222,
+      "grad_norm": 708.3402099609375,
+      "learning_rate": 5.5981081498195365e-08,
+      "loss": 108.3079,
+      "step": 118440
+    },
+    {
+      "epoch": 0.9798568887785912,
+      "grad_norm": 731.2184448242188,
+      "learning_rate": 5.577082099679942e-08,
+      "loss": 64.9039,
+      "step": 118450
+    },
+    {
+      "epoch": 0.9799396120279604,
+      "grad_norm": 673.4403076171875,
+      "learning_rate": 5.556095388502824e-08,
+      "loss": 95.9862,
+      "step": 118460
+    },
+    {
+      "epoch": 0.9800223352773297,
+      "grad_norm": 1423.2120361328125,
+      "learning_rate": 5.535148017958014e-08,
+      "loss": 81.7384,
+      "step": 118470
+    },
+    {
+      "epoch": 0.9801050585266989,
+      "grad_norm": 706.9336547851562,
+      "learning_rate": 5.514239989712178e-08,
+      "loss": 82.0284,
+      "step": 118480
+    },
+    {
+      "epoch": 0.9801877817760681,
+      "grad_norm": 711.587646484375,
+      "learning_rate": 5.493371305428874e-08,
+      "loss": 101.2837,
+      "step": 118490
+    },
+    {
+      "epoch": 0.9802705050254374,
+      "grad_norm": 993.87353515625,
+      "learning_rate": 5.472541966768552e-08,
+      "loss": 74.9229,
+      "step": 118500
+    },
+    {
+      "epoch": 0.9803532282748066,
+      "grad_norm": 746.7269287109375,
+      "learning_rate": 5.451751975388442e-08,
+      "loss": 77.2581,
+      "step": 118510
+    },
+    {
+      "epoch": 0.9804359515241758,
+      "grad_norm": 1050.691162109375,
+      "learning_rate": 5.4310013329428314e-08,
+      "loss": 88.0851,
+      "step": 118520
+    },
+    {
+      "epoch": 0.9805186747735452,
+      "grad_norm": 496.2442321777344,
+      "learning_rate": 5.410290041082622e-08,
+      "loss": 141.8692,
+      "step": 118530
+    },
+    {
+      "epoch": 0.9806013980229144,
+      "grad_norm": 484.4472351074219,
+      "learning_rate": 5.3896181014557733e-08,
+      "loss": 62.8719,
+      "step": 118540
+    },
+    {
+      "epoch": 0.9806841212722835,
+      "grad_norm": 1243.587646484375,
+      "learning_rate": 5.368985515707137e-08,
+      "loss": 120.0687,
+      "step": 118550
+    },
+    {
+      "epoch": 0.9807668445216529,
+      "grad_norm": 533.1880493164062,
+      "learning_rate": 5.348392285478232e-08,
+      "loss": 66.4509,
+      "step": 118560
+    },
+    {
+      "epoch": 0.9808495677710221,
+      "grad_norm": 1625.4766845703125,
+      "learning_rate": 5.327838412407582e-08,
+      "loss": 110.9632,
+      "step": 118570
+    },
+    {
+      "epoch": 0.9809322910203913,
+      "grad_norm": 521.4833374023438,
+      "learning_rate": 5.3073238981305455e-08,
+      "loss": 88.1412,
+      "step": 118580
+    },
+    {
+      "epoch": 0.9810150142697606,
+      "grad_norm": 671.5929565429688,
+      "learning_rate": 5.2868487442794825e-08,
+      "loss": 59.3454,
+      "step": 118590
+    },
+    {
+      "epoch": 0.9810977375191298,
+      "grad_norm": 717.1107177734375,
+      "learning_rate": 5.266412952483424e-08,
+      "loss": 101.9609,
+      "step": 118600
+    },
+    {
+      "epoch": 0.981180460768499,
+      "grad_norm": 657.0799560546875,
+      "learning_rate": 5.246016524368347e-08,
+      "loss": 88.4751,
+      "step": 118610
+    },
+    {
+      "epoch": 0.9812631840178683,
+      "grad_norm": 616.4298706054688,
+      "learning_rate": 5.225659461557176e-08,
+      "loss": 111.9666,
+      "step": 118620
+    },
+    {
+      "epoch": 0.9813459072672375,
+      "grad_norm": 525.951416015625,
+      "learning_rate": 5.205341765669503e-08,
+      "loss": 113.7892,
+      "step": 118630
+    },
+    {
+      "epoch": 0.9814286305166067,
+      "grad_norm": 701.6843872070312,
+      "learning_rate": 5.185063438322091e-08,
+      "loss": 59.8481,
+      "step": 118640
+    },
+    {
+      "epoch": 0.981511353765976,
+      "grad_norm": 768.6402587890625,
+      "learning_rate": 5.1648244811282054e-08,
+      "loss": 75.6018,
+      "step": 118650
+    },
+    {
+      "epoch": 0.9815940770153452,
+      "grad_norm": 1144.209716796875,
+      "learning_rate": 5.14462489569828e-08,
+      "loss": 80.7018,
+      "step": 118660
+    },
+    {
+      "epoch": 0.9816768002647144,
+      "grad_norm": 1187.2999267578125,
+      "learning_rate": 5.1244646836394187e-08,
+      "loss": 93.3386,
+      "step": 118670
+    },
+    {
+      "epoch": 0.9817595235140837,
+      "grad_norm": 1293.2803955078125,
+      "learning_rate": 5.104343846555726e-08,
+      "loss": 73.9802,
+      "step": 118680
+    },
+    {
+      "epoch": 0.9818422467634529,
+      "grad_norm": 918.84765625,
+      "learning_rate": 5.0842623860482e-08,
+      "loss": 94.4681,
+      "step": 118690
+    },
+    {
+      "epoch": 0.9819249700128221,
+      "grad_norm": 558.7251586914062,
+      "learning_rate": 5.064220303714507e-08,
+      "loss": 82.5352,
+      "step": 118700
+    },
+    {
+      "epoch": 0.9820076932621913,
+      "grad_norm": 1152.9447021484375,
+      "learning_rate": 5.044217601149371e-08,
+      "loss": 65.6448,
+      "step": 118710
+    },
+    {
+      "epoch": 0.9820904165115606,
+      "grad_norm": 535.541748046875,
+      "learning_rate": 5.024254279944296e-08,
+      "loss": 72.6428,
+      "step": 118720
+    },
+    {
+      "epoch": 0.9821731397609298,
+      "grad_norm": 2769.91845703125,
+      "learning_rate": 5.004330341687735e-08,
+      "loss": 102.4574,
+      "step": 118730
+    },
+    {
+      "epoch": 0.982255863010299,
+      "grad_norm": 680.4102172851562,
+      "learning_rate": 4.9844457879648086e-08,
+      "loss": 94.1751,
+      "step": 118740
+    },
+    {
+      "epoch": 0.9823385862596683,
+      "grad_norm": 941.7730712890625,
+      "learning_rate": 4.9646006203577515e-08,
+      "loss": 112.8078,
+      "step": 118750
+    },
+    {
+      "epoch": 0.9824213095090375,
+      "grad_norm": 1552.2708740234375,
+      "learning_rate": 4.944794840445521e-08,
+      "loss": 78.17,
+      "step": 118760
+    },
+    {
+      "epoch": 0.9825040327584067,
+      "grad_norm": 1019.7816772460938,
+      "learning_rate": 4.9250284498039146e-08,
+      "loss": 65.3781,
+      "step": 118770
+    },
+    {
+      "epoch": 0.982586756007776,
+      "grad_norm": 658.9552001953125,
+      "learning_rate": 4.905301450005784e-08,
+      "loss": 74.2217,
+      "step": 118780
+    },
+    {
+      "epoch": 0.9826694792571452,
+      "grad_norm": 731.14111328125,
+      "learning_rate": 4.885613842620596e-08,
+      "loss": 93.1678,
+      "step": 118790
+    },
+    {
+      "epoch": 0.9827522025065144,
+      "grad_norm": 1084.143310546875,
+      "learning_rate": 4.865965629214819e-08,
+      "loss": 86.729,
+      "step": 118800
+    },
+    {
+      "epoch": 0.9828349257558837,
+      "grad_norm": 621.4204711914062,
+      "learning_rate": 4.846356811351871e-08,
+      "loss": 87.1071,
+      "step": 118810
+    },
+    {
+      "epoch": 0.9829176490052529,
+      "grad_norm": 686.3203735351562,
+      "learning_rate": 4.826787390591836e-08,
+      "loss": 66.5393,
+      "step": 118820
+    },
+    {
+      "epoch": 0.9830003722546221,
+      "grad_norm": 380.79351806640625,
+      "learning_rate": 4.8072573684918024e-08,
+      "loss": 60.1358,
+      "step": 118830
+    },
+    {
+      "epoch": 0.9830830955039914,
+      "grad_norm": 1424.990966796875,
+      "learning_rate": 4.787766746605638e-08,
+      "loss": 97.6002,
+      "step": 118840
+    },
+    {
+      "epoch": 0.9831658187533606,
+      "grad_norm": 489.2065734863281,
+      "learning_rate": 4.768315526484158e-08,
+      "loss": 69.9402,
+      "step": 118850
+    },
+    {
+      "epoch": 0.9832485420027298,
+      "grad_norm": 868.9462280273438,
+      "learning_rate": 4.7489037096750126e-08,
+      "loss": 106.7428,
+      "step": 118860
+    },
+    {
+      "epoch": 0.9833312652520991,
+      "grad_norm": 2591.3154296875,
+      "learning_rate": 4.7295312977226895e-08,
+      "loss": 102.9899,
+      "step": 118870
+    },
+    {
+      "epoch": 0.9834139885014683,
+      "grad_norm": 1134.68212890625,
+      "learning_rate": 4.710198292168566e-08,
+      "loss": 78.4484,
+      "step": 118880
+    },
+    {
+      "epoch": 0.9834967117508375,
+      "grad_norm": 805.161376953125,
+      "learning_rate": 4.690904694550913e-08,
+      "loss": 88.5338,
+      "step": 118890
+    },
+    {
+      "epoch": 0.9835794350002068,
+      "grad_norm": 747.1693115234375,
+      "learning_rate": 4.671650506404835e-08,
+      "loss": 83.3935,
+      "step": 118900
+    },
+    {
+      "epoch": 0.983662158249576,
+      "grad_norm": 807.8037109375,
+      "learning_rate": 4.6524357292622724e-08,
+      "loss": 78.79,
+      "step": 118910
+    },
+    {
+      "epoch": 0.9837448814989452,
+      "grad_norm": 695.4989013671875,
+      "learning_rate": 4.633260364652059e-08,
+      "loss": 82.1724,
+      "step": 118920
+    },
+    {
+      "epoch": 0.9838276047483145,
+      "grad_norm": 840.2431640625,
+      "learning_rate": 4.6141244140998634e-08,
+      "loss": 73.6012,
+      "step": 118930
+    },
+    {
+      "epoch": 0.9839103279976837,
+      "grad_norm": 857.3699340820312,
+      "learning_rate": 4.5950278791283e-08,
+      "loss": 75.8136,
+      "step": 118940
+    },
+    {
+      "epoch": 0.9839930512470529,
+      "grad_norm": 1018.6058959960938,
+      "learning_rate": 4.575970761256765e-08,
+      "loss": 85.1594,
+      "step": 118950
+    },
+    {
+      "epoch": 0.9840757744964223,
+      "grad_norm": 1265.7125244140625,
+      "learning_rate": 4.556953062001546e-08,
+      "loss": 89.4971,
+      "step": 118960
+    },
+    {
+      "epoch": 0.9841584977457914,
+      "grad_norm": 735.5936279296875,
+      "learning_rate": 4.5379747828757095e-08,
+      "loss": 81.9222,
+      "step": 118970
+    },
+    {
+      "epoch": 0.9842412209951606,
+      "grad_norm": 881.5132446289062,
+      "learning_rate": 4.5190359253894925e-08,
+      "loss": 88.397,
+      "step": 118980
+    },
+    {
+      "epoch": 0.98432394424453,
+      "grad_norm": 353.5128479003906,
+      "learning_rate": 4.500136491049578e-08,
+      "loss": 83.2312,
+      "step": 118990
+    },
+    {
+      "epoch": 0.9844066674938992,
+      "grad_norm": 1424.359130859375,
+      "learning_rate": 4.481276481359764e-08,
+      "loss": 86.6353,
+      "step": 119000
+    },
+    {
+      "epoch": 0.9844893907432684,
+      "grad_norm": 887.9814453125,
+      "learning_rate": 4.462455897820628e-08,
+      "loss": 76.8451,
+      "step": 119010
+    },
+    {
+      "epoch": 0.9845721139926377,
+      "grad_norm": 806.23193359375,
+      "learning_rate": 4.443674741929693e-08,
+      "loss": 66.0883,
+      "step": 119020
+    },
+    {
+      "epoch": 0.9846548372420069,
+      "grad_norm": 573.1680297851562,
+      "learning_rate": 4.424933015181265e-08,
+      "loss": 72.9488,
+      "step": 119030
+    },
+    {
+      "epoch": 0.9847375604913761,
+      "grad_norm": 976.209716796875,
+      "learning_rate": 4.40623071906654e-08,
+      "loss": 84.5281,
+      "step": 119040
+    },
+    {
+      "epoch": 0.9848202837407454,
+      "grad_norm": 947.809814453125,
+      "learning_rate": 4.387567855073604e-08,
+      "loss": 89.569,
+      "step": 119050
+    },
+    {
+      "epoch": 0.9849030069901146,
+      "grad_norm": 1052.1142578125,
+      "learning_rate": 4.368944424687271e-08,
+      "loss": 69.5182,
+      "step": 119060
+    },
+    {
+      "epoch": 0.9849857302394838,
+      "grad_norm": 498.6680603027344,
+      "learning_rate": 4.350360429389411e-08,
+      "loss": 92.5021,
+      "step": 119070
+    },
+    {
+      "epoch": 0.9850684534888531,
+      "grad_norm": 878.2149047851562,
+      "learning_rate": 4.3318158706586734e-08,
+      "loss": 60.1033,
+      "step": 119080
+    },
+    {
+      "epoch": 0.9851511767382223,
+      "grad_norm": 618.4829711914062,
+      "learning_rate": 4.3133107499704894e-08,
+      "loss": 80.3329,
+      "step": 119090
+    },
+    {
+      "epoch": 0.9852338999875915,
+      "grad_norm": 544.2587280273438,
+      "learning_rate": 4.294845068797349e-08,
+      "loss": 79.2608,
+      "step": 119100
+    },
+    {
+      "epoch": 0.9853166232369608,
+      "grad_norm": 989.1095581054688,
+      "learning_rate": 4.276418828608353e-08,
+      "loss": 69.4792,
+      "step": 119110
+    },
+    {
+      "epoch": 0.98539934648633,
+      "grad_norm": 1151.3955078125,
+      "learning_rate": 4.258032030869608e-08,
+      "loss": 83.3888,
+      "step": 119120
+    },
+    {
+      "epoch": 0.9854820697356992,
+      "grad_norm": 628.578857421875,
+      "learning_rate": 4.2396846770441644e-08,
+      "loss": 103.2333,
+      "step": 119130
+    },
+    {
+      "epoch": 0.9855647929850685,
+      "grad_norm": 830.7070922851562,
+      "learning_rate": 4.221376768591801e-08,
+      "loss": 83.7114,
+      "step": 119140
+    },
+    {
+      "epoch": 0.9856475162344377,
+      "grad_norm": 343.4464111328125,
+      "learning_rate": 4.203108306969128e-08,
+      "loss": 86.0502,
+      "step": 119150
+    },
+    {
+      "epoch": 0.9857302394838069,
+      "grad_norm": 621.4478759765625,
+      "learning_rate": 4.1848792936297064e-08,
+      "loss": 83.2397,
+      "step": 119160
+    },
+    {
+      "epoch": 0.9858129627331762,
+      "grad_norm": 964.1356811523438,
+      "learning_rate": 4.166689730023987e-08,
+      "loss": 77.1152,
+      "step": 119170
+    },
+    {
+      "epoch": 0.9858956859825454,
+      "grad_norm": 818.7540893554688,
+      "learning_rate": 4.148539617599201e-08,
+      "loss": 92.1264,
+      "step": 119180
+    },
+    {
+      "epoch": 0.9859784092319146,
+      "grad_norm": 1049.2435302734375,
+      "learning_rate": 4.13042895779947e-08,
+      "loss": 92.8093,
+      "step": 119190
+    },
+    {
+      "epoch": 0.9860611324812839,
+      "grad_norm": 622.9645385742188,
+      "learning_rate": 4.112357752065754e-08,
+      "loss": 73.6431,
+      "step": 119200
+    },
+    {
+      "epoch": 0.9861438557306531,
+      "grad_norm": 934.980224609375,
+      "learning_rate": 4.0943260018359024e-08,
+      "loss": 102.3326,
+      "step": 119210
+    },
+    {
+      "epoch": 0.9862265789800223,
+      "grad_norm": 775.8395385742188,
+      "learning_rate": 4.076333708544655e-08,
+      "loss": 99.6095,
+      "step": 119220
+    },
+    {
+      "epoch": 0.9863093022293916,
+      "grad_norm": 561.11181640625,
+      "learning_rate": 4.058380873623591e-08,
+      "loss": 60.4265,
+      "step": 119230
+    },
+    {
+      "epoch": 0.9863920254787608,
+      "grad_norm": 1175.651123046875,
+      "learning_rate": 4.040467498501011e-08,
+      "loss": 102.5078,
+      "step": 119240
+    },
+    {
+      "epoch": 0.98647474872813,
+      "grad_norm": 1031.730712890625,
+      "learning_rate": 4.02259358460233e-08,
+      "loss": 107.7175,
+      "step": 119250
+    },
+    {
+      "epoch": 0.9865574719774993,
+      "grad_norm": 1173.889404296875,
+      "learning_rate": 4.00475913334969e-08,
+      "loss": 112.8835,
+      "step": 119260
+    },
+    {
+      "epoch": 0.9866401952268685,
+      "grad_norm": 747.98486328125,
+      "learning_rate": 3.986964146162009e-08,
+      "loss": 72.1821,
+      "step": 119270
+    },
+    {
+      "epoch": 0.9867229184762377,
+      "grad_norm": 904.0946044921875,
+      "learning_rate": 3.969208624455212e-08,
+      "loss": 78.2023,
+      "step": 119280
+    },
+    {
+      "epoch": 0.986805641725607,
+      "grad_norm": 761.3492431640625,
+      "learning_rate": 3.951492569642001e-08,
+      "loss": 83.0522,
+      "step": 119290
+    },
+    {
+      "epoch": 0.9868883649749762,
+      "grad_norm": 1574.1885986328125,
+      "learning_rate": 3.9338159831319724e-08,
+      "loss": 108.7734,
+      "step": 119300
+    },
+    {
+      "epoch": 0.9869710882243454,
+      "grad_norm": 774.9620971679688,
+      "learning_rate": 3.9161788663315546e-08,
+      "loss": 79.6705,
+      "step": 119310
+    },
+    {
+      "epoch": 0.9870538114737147,
+      "grad_norm": 1015.9965209960938,
+      "learning_rate": 3.898581220644071e-08,
+      "loss": 126.5208,
+      "step": 119320
+    },
+    {
+      "epoch": 0.9871365347230839,
+      "grad_norm": 1341.8759765625,
+      "learning_rate": 3.881023047469679e-08,
+      "loss": 90.6255,
+      "step": 119330
+    },
+    {
+      "epoch": 0.9872192579724531,
+      "grad_norm": 511.98223876953125,
+      "learning_rate": 3.863504348205427e-08,
+      "loss": 85.9824,
+      "step": 119340
+    },
+    {
+      "epoch": 0.9873019812218224,
+      "grad_norm": 1165.170654296875,
+      "learning_rate": 3.8460251242451454e-08,
+      "loss": 123.9411,
+      "step": 119350
+    },
+    {
+      "epoch": 0.9873847044711916,
+      "grad_norm": 1152.4227294921875,
+      "learning_rate": 3.828585376979666e-08,
+      "loss": 91.7785,
+      "step": 119360
+    },
+    {
+      "epoch": 0.9874674277205608,
+      "grad_norm": 684.0379638671875,
+      "learning_rate": 3.811185107796489e-08,
+      "loss": 69.7867,
+      "step": 119370
+    },
+    {
+      "epoch": 0.9875501509699302,
+      "grad_norm": 664.3668823242188,
+      "learning_rate": 3.793824318080064e-08,
+      "loss": 74.003,
+      "step": 119380
+    },
+    {
+      "epoch": 0.9876328742192994,
+      "grad_norm": 800.0819091796875,
+      "learning_rate": 3.7765030092118404e-08,
+      "loss": 138.9721,
+      "step": 119390
+    },
+    {
+      "epoch": 0.9877155974686685,
+      "grad_norm": 845.6797485351562,
+      "learning_rate": 3.7592211825698835e-08,
+      "loss": 104.952,
+      "step": 119400
+    },
+    {
+      "epoch": 0.9877983207180379,
+      "grad_norm": 942.4180297851562,
+      "learning_rate": 3.741978839529259e-08,
+      "loss": 112.8675,
+      "step": 119410
+    },
+    {
+      "epoch": 0.9878810439674071,
+      "grad_norm": 1062.473876953125,
+      "learning_rate": 3.72477598146187e-08,
+      "loss": 90.5084,
+      "step": 119420
+    },
+    {
+      "epoch": 0.9879637672167763,
+      "grad_norm": 636.2952270507812,
+      "learning_rate": 3.7076126097363997e-08,
+      "loss": 70.5935,
+      "step": 119430
+    },
+    {
+      "epoch": 0.9880464904661455,
+      "grad_norm": 446.5930480957031,
+      "learning_rate": 3.690488725718588e-08,
+      "loss": 109.0157,
+      "step": 119440
+    },
+    {
+      "epoch": 0.9881292137155148,
+      "grad_norm": 1208.2958984375,
+      "learning_rate": 3.673404330770847e-08,
+      "loss": 64.4161,
+      "step": 119450
+    },
+    {
+      "epoch": 0.988211936964884,
+      "grad_norm": 501.6045227050781,
+      "learning_rate": 3.65635942625242e-08,
+      "loss": 64.2057,
+      "step": 119460
+    },
+    {
+      "epoch": 0.9882946602142532,
+      "grad_norm": 1004.0560913085938,
+      "learning_rate": 3.639354013519614e-08,
+      "loss": 118.3625,
+      "step": 119470
+    },
+    {
+      "epoch": 0.9883773834636225,
+      "grad_norm": 1431.545166015625,
+      "learning_rate": 3.6223880939254e-08,
+      "loss": 111.376,
+      "step": 119480
+    },
+    {
+      "epoch": 0.9884601067129917,
+      "grad_norm": 717.1210327148438,
+      "learning_rate": 3.605461668819754e-08,
+      "loss": 81.6542,
+      "step": 119490
+    },
+    {
+      "epoch": 0.9885428299623609,
+      "grad_norm": 844.7835083007812,
+      "learning_rate": 3.588574739549322e-08,
+      "loss": 92.5829,
+      "step": 119500
+    },
+    {
+      "epoch": 0.9886255532117302,
+      "grad_norm": 910.1040649414062,
+      "learning_rate": 3.571727307457806e-08,
+      "loss": 111.628,
+      "step": 119510
+    },
+    {
+      "epoch": 0.9887082764610994,
+      "grad_norm": 649.8385620117188,
+      "learning_rate": 3.5549193738856346e-08,
+      "loss": 86.6366,
+      "step": 119520
+    },
+    {
+      "epoch": 0.9887909997104686,
+      "grad_norm": 784.6031494140625,
+      "learning_rate": 3.5381509401701264e-08,
+      "loss": 72.6148,
+      "step": 119530
+    },
+    {
+      "epoch": 0.9888737229598379,
+      "grad_norm": 1095.903564453125,
+      "learning_rate": 3.5214220076455474e-08,
+      "loss": 120.5914,
+      "step": 119540
+    },
+    {
+      "epoch": 0.9889564462092071,
+      "grad_norm": 954.7218017578125,
+      "learning_rate": 3.5047325776428884e-08,
+      "loss": 101.9791,
+      "step": 119550
+    },
+    {
+      "epoch": 0.9890391694585763,
+      "grad_norm": 1238.466796875,
+      "learning_rate": 3.488082651490032e-08,
+      "loss": 100.0169,
+      "step": 119560
+    },
+    {
+      "epoch": 0.9891218927079456,
+      "grad_norm": 688.223388671875,
+      "learning_rate": 3.471472230511752e-08,
+      "loss": 96.7697,
+      "step": 119570
+    },
+    {
+      "epoch": 0.9892046159573148,
+      "grad_norm": 788.05859375,
+      "learning_rate": 3.454901316029657e-08,
+      "loss": 100.5506,
+      "step": 119580
+    },
+    {
+      "epoch": 0.989287339206684,
+      "grad_norm": 1479.240234375,
+      "learning_rate": 3.438369909362249e-08,
+      "loss": 109.9174,
+      "step": 119590
+    },
+    {
+      "epoch": 0.9893700624560533,
+      "grad_norm": 831.3067016601562,
+      "learning_rate": 3.421878011824864e-08,
+      "loss": 110.4286,
+      "step": 119600
+    },
+    {
+      "epoch": 0.9894527857054225,
+      "grad_norm": 924.3592529296875,
+      "learning_rate": 3.405425624729619e-08,
+      "loss": 91.2705,
+      "step": 119610
+    },
+    {
+      "epoch": 0.9895355089547917,
+      "grad_norm": 674.0458374023438,
+      "learning_rate": 3.389012749385578e-08,
+      "loss": 74.0681,
+      "step": 119620
+    },
+    {
+      "epoch": 0.989618232204161,
+      "grad_norm": 523.977294921875,
+      "learning_rate": 3.3726393870986976e-08,
+      "loss": 97.7908,
+      "step": 119630
+    },
+    {
+      "epoch": 0.9897009554535302,
+      "grad_norm": 619.9107666015625,
+      "learning_rate": 3.356305539171656e-08,
+      "loss": 99.3192,
+      "step": 119640
+    },
+    {
+      "epoch": 0.9897836787028994,
+      "grad_norm": 644.1179809570312,
+      "learning_rate": 3.340011206904137e-08,
+      "loss": 80.9895,
+      "step": 119650
+    },
+    {
+      "epoch": 0.9898664019522687,
+      "grad_norm": 646.9247436523438,
+      "learning_rate": 3.323756391592548e-08,
+      "loss": 75.6963,
+      "step": 119660
+    },
+    {
+      "epoch": 0.9899491252016379,
+      "grad_norm": 613.2579956054688,
+      "learning_rate": 3.307541094530242e-08,
+      "loss": 106.8036,
+      "step": 119670
+    },
+    {
+      "epoch": 0.9900318484510071,
+      "grad_norm": 903.300048828125,
+      "learning_rate": 3.291365317007355e-08,
+      "loss": 66.5273,
+      "step": 119680
+    },
+    {
+      "epoch": 0.9901145717003764,
+      "grad_norm": 749.8668212890625,
+      "learning_rate": 3.2752290603109694e-08,
+      "loss": 81.0457,
+      "step": 119690
+    },
+    {
+      "epoch": 0.9901972949497456,
+      "grad_norm": 855.69677734375,
+      "learning_rate": 3.25913232572489e-08,
+      "loss": 103.7646,
+      "step": 119700
+    },
+    {
+      "epoch": 0.9902800181991148,
+      "grad_norm": 640.4257202148438,
+      "learning_rate": 3.243075114529981e-08,
+      "loss": 88.8553,
+      "step": 119710
+    },
+    {
+      "epoch": 0.9903627414484841,
+      "grad_norm": 642.8728637695312,
+      "learning_rate": 3.2270574280037213e-08,
+      "loss": 89.717,
+      "step": 119720
+    },
+    {
+      "epoch": 0.9904454646978533,
+      "grad_norm": 1653.7088623046875,
+      "learning_rate": 3.211079267420647e-08,
+      "loss": 106.5534,
+      "step": 119730
+    },
+    {
+      "epoch": 0.9905281879472225,
+      "grad_norm": 1032.344970703125,
+      "learning_rate": 3.195140634052074e-08,
+      "loss": 83.4833,
+      "step": 119740
+    },
+    {
+      "epoch": 0.9906109111965918,
+      "grad_norm": 886.9328002929688,
+      "learning_rate": 3.179241529166099e-08,
+      "loss": 82.0412,
+      "step": 119750
+    },
+    {
+      "epoch": 0.990693634445961,
+      "grad_norm": 930.1018676757812,
+      "learning_rate": 3.163381954027822e-08,
+      "loss": 77.0731,
+      "step": 119760
+    },
+    {
+      "epoch": 0.9907763576953302,
+      "grad_norm": 1377.98388671875,
+      "learning_rate": 3.147561909899066e-08,
+      "loss": 89.5693,
+      "step": 119770
+    },
+    {
+      "epoch": 0.9908590809446995,
+      "grad_norm": 943.3343505859375,
+      "learning_rate": 3.131781398038547e-08,
+      "loss": 82.712,
+      "step": 119780
+    },
+    {
+      "epoch": 0.9909418041940687,
+      "grad_norm": 901.633056640625,
+      "learning_rate": 3.1160404197018155e-08,
+      "loss": 70.4035,
+      "step": 119790
+    },
+    {
+      "epoch": 0.9910245274434379,
+      "grad_norm": 1146.6190185546875,
+      "learning_rate": 3.100338976141426e-08,
+      "loss": 68.581,
+      "step": 119800
+    },
+    {
+      "epoch": 0.9911072506928073,
+      "grad_norm": 1056.26904296875,
+      "learning_rate": 3.084677068606545e-08,
+      "loss": 76.8607,
+      "step": 119810
+    },
+    {
+      "epoch": 0.9911899739421764,
+      "grad_norm": 531.5973510742188,
+      "learning_rate": 3.0690546983433986e-08,
+      "loss": 101.3226,
+      "step": 119820
+    },
+    {
+      "epoch": 0.9912726971915456,
+      "grad_norm": 1205.59521484375,
+      "learning_rate": 3.053471866594993e-08,
+      "loss": 120.3366,
+      "step": 119830
+    },
+    {
+      "epoch": 0.991355420440915,
+      "grad_norm": 904.2116088867188,
+      "learning_rate": 3.0379285746011125e-08,
+      "loss": 75.5825,
+      "step": 119840
+    },
+    {
+      "epoch": 0.9914381436902842,
+      "grad_norm": 891.4710083007812,
+      "learning_rate": 3.022424823598546e-08,
+      "loss": 89.1019,
+      "step": 119850
+    },
+    {
+      "epoch": 0.9915208669396534,
+      "grad_norm": 869.971923828125,
+      "learning_rate": 3.0069606148208085e-08,
+      "loss": 88.6072,
+      "step": 119860
+    },
+    {
+      "epoch": 0.9916035901890227,
+      "grad_norm": 634.763916015625,
+      "learning_rate": 2.991535949498303e-08,
+      "loss": 87.1923,
+      "step": 119870
+    },
+    {
+      "epoch": 0.9916863134383919,
+      "grad_norm": 789.7999877929688,
+      "learning_rate": 2.9761508288583262e-08,
+      "loss": 72.2758,
+      "step": 119880
+    },
+    {
+      "epoch": 0.9917690366877611,
+      "grad_norm": 535.9653930664062,
+      "learning_rate": 2.96080525412501e-08,
+      "loss": 92.1307,
+      "step": 119890
+    },
+    {
+      "epoch": 0.9918517599371304,
+      "grad_norm": 1039.1324462890625,
+      "learning_rate": 2.9454992265193216e-08,
+      "loss": 54.078,
+      "step": 119900
+    },
+    {
+      "epoch": 0.9919344831864996,
+      "grad_norm": 616.3204345703125,
+      "learning_rate": 2.9302327472590653e-08,
+      "loss": 85.4111,
+      "step": 119910
+    },
+    {
+      "epoch": 0.9920172064358688,
+      "grad_norm": 799.8880615234375,
+      "learning_rate": 2.9150058175589356e-08,
+      "loss": 82.8982,
+      "step": 119920
+    },
+    {
+      "epoch": 0.9920999296852381,
+      "grad_norm": 1155.7872314453125,
+      "learning_rate": 2.8998184386305196e-08,
+      "loss": 100.5487,
+      "step": 119930
+    },
+    {
+      "epoch": 0.9921826529346073,
+      "grad_norm": 1084.304443359375,
+      "learning_rate": 2.8846706116821834e-08,
+      "loss": 90.8323,
+      "step": 119940
+    },
+    {
+      "epoch": 0.9922653761839765,
+      "grad_norm": 1023.7007446289062,
+      "learning_rate": 2.8695623379191296e-08,
+      "loss": 71.3728,
+      "step": 119950
+    },
+    {
+      "epoch": 0.9923480994333458,
+      "grad_norm": 681.5350952148438,
+      "learning_rate": 2.8544936185434525e-08,
+      "loss": 76.9926,
+      "step": 119960
+    },
+    {
+      "epoch": 0.992430822682715,
+      "grad_norm": 1248.9072265625,
+      "learning_rate": 2.8394644547541373e-08,
+      "loss": 89.5176,
+      "step": 119970
+    },
+    {
+      "epoch": 0.9925135459320842,
+      "grad_norm": 609.6895751953125,
+      "learning_rate": 2.8244748477470052e-08,
+      "loss": 81.6998,
+      "step": 119980
+    },
+    {
+      "epoch": 0.9925962691814535,
+      "grad_norm": 660.6539306640625,
+      "learning_rate": 2.809524798714658e-08,
+      "loss": 99.3655,
+      "step": 119990
+    },
+    {
+      "epoch": 0.9926789924308227,
+      "grad_norm": 712.8468017578125,
+      "learning_rate": 2.7946143088466437e-08,
+      "loss": 69.3642,
+      "step": 120000
+    },
+    {
+      "epoch": 0.9927617156801919,
+      "grad_norm": 1156.4256591796875,
+      "learning_rate": 2.7797433793292915e-08,
+      "loss": 94.3238,
+      "step": 120010
+    },
+    {
+      "epoch": 0.9928444389295612,
+      "grad_norm": 956.7645874023438,
+      "learning_rate": 2.7649120113458217e-08,
+      "loss": 95.6845,
+      "step": 120020
+    },
+    {
+      "epoch": 0.9929271621789304,
+      "grad_norm": 726.518310546875,
+      "learning_rate": 2.7501202060763454e-08,
+      "loss": 83.2681,
+      "step": 120030
+    },
+    {
+      "epoch": 0.9930098854282996,
+      "grad_norm": 1263.5406494140625,
+      "learning_rate": 2.7353679646976995e-08,
+      "loss": 91.9831,
+      "step": 120040
+    },
+    {
+      "epoch": 0.9930926086776689,
+      "grad_norm": 1196.4813232421875,
+      "learning_rate": 2.7206552883836667e-08,
+      "loss": 97.7018,
+      "step": 120050
+    },
+    {
+      "epoch": 0.9931753319270381,
+      "grad_norm": 481.29241943359375,
+      "learning_rate": 2.705982178304922e-08,
+      "loss": 94.1881,
+      "step": 120060
+    },
+    {
+      "epoch": 0.9932580551764073,
+      "grad_norm": 627.6724853515625,
+      "learning_rate": 2.691348635628921e-08,
+      "loss": 86.9981,
+      "step": 120070
+    },
+    {
+      "epoch": 0.9933407784257766,
+      "grad_norm": 792.4179077148438,
+      "learning_rate": 2.676754661519898e-08,
+      "loss": 85.3965,
+      "step": 120080
+    },
+    {
+      "epoch": 0.9934235016751458,
+      "grad_norm": 1112.1466064453125,
+      "learning_rate": 2.662200257139147e-08,
+      "loss": 107.8607,
+      "step": 120090
+    },
+    {
+      "epoch": 0.993506224924515,
+      "grad_norm": 851.7528076171875,
+      "learning_rate": 2.6476854236446858e-08,
+      "loss": 81.0411,
+      "step": 120100
+    },
+    {
+      "epoch": 0.9935889481738843,
+      "grad_norm": 679.0064697265625,
+      "learning_rate": 2.6332101621913133e-08,
+      "loss": 100.3239,
+      "step": 120110
+    },
+    {
+      "epoch": 0.9936716714232535,
+      "grad_norm": 970.4327392578125,
+      "learning_rate": 2.6187744739308297e-08,
+      "loss": 93.8976,
+      "step": 120120
+    },
+    {
+      "epoch": 0.9937543946726227,
+      "grad_norm": 783.4990234375,
+      "learning_rate": 2.604378360011761e-08,
+      "loss": 81.577,
+      "step": 120130
+    },
+    {
+      "epoch": 0.993837117921992,
+      "grad_norm": 464.9545593261719,
+      "learning_rate": 2.5900218215795802e-08,
+      "loss": 64.9535,
+      "step": 120140
+    },
+    {
+      "epoch": 0.9939198411713612,
+      "grad_norm": 1092.5909423828125,
+      "learning_rate": 2.57570485977654e-08,
+      "loss": 81.3496,
+      "step": 120150
+    },
+    {
+      "epoch": 0.9940025644207304,
+      "grad_norm": 300.0126647949219,
+      "learning_rate": 2.5614274757417846e-08,
+      "loss": 73.0116,
+      "step": 120160
+    },
+    {
+      "epoch": 0.9940852876700996,
+      "grad_norm": 1496.8515625,
+      "learning_rate": 2.5471896706113497e-08,
+      "loss": 90.5071,
+      "step": 120170
+    },
+    {
+      "epoch": 0.9941680109194689,
+      "grad_norm": 827.8729248046875,
+      "learning_rate": 2.5329914455180516e-08,
+      "loss": 68.6131,
+      "step": 120180
+    },
+    {
+      "epoch": 0.9942507341688381,
+      "grad_norm": 1299.62744140625,
+      "learning_rate": 2.5188328015914865e-08,
+      "loss": 96.4743,
+      "step": 120190
+    },
+    {
+      "epoch": 0.9943334574182073,
+      "grad_norm": 1085.9984130859375,
+      "learning_rate": 2.5047137399583088e-08,
+      "loss": 90.0169,
+      "step": 120200
+    },
+    {
+      "epoch": 0.9944161806675766,
+      "grad_norm": 712.6924438476562,
+      "learning_rate": 2.4906342617418976e-08,
+      "loss": 122.7976,
+      "step": 120210
+    },
+    {
+      "epoch": 0.9944989039169458,
+      "grad_norm": 1067.037109375,
+      "learning_rate": 2.4765943680624126e-08,
+      "loss": 105.9706,
+      "step": 120220
+    },
+    {
+      "epoch": 0.994581627166315,
+      "grad_norm": 1210.1380615234375,
+      "learning_rate": 2.4625940600369603e-08,
+      "loss": 77.7456,
+      "step": 120230
+    },
+    {
+      "epoch": 0.9946643504156844,
+      "grad_norm": 626.2311401367188,
+      "learning_rate": 2.4486333387795935e-08,
+      "loss": 73.6353,
+      "step": 120240
+    },
+    {
+      "epoch": 0.9947470736650535,
+      "grad_norm": 1163.8887939453125,
+      "learning_rate": 2.434712205400924e-08,
+      "loss": 72.8398,
+      "step": 120250
+    },
+    {
+      "epoch": 0.9948297969144227,
+      "grad_norm": 703.2742919921875,
+      "learning_rate": 2.4208306610087884e-08,
+      "loss": 88.9687,
+      "step": 120260
+    },
+    {
+      "epoch": 0.9949125201637921,
+      "grad_norm": 1137.736572265625,
+      "learning_rate": 2.406988706707525e-08,
+      "loss": 121.851,
+      "step": 120270
+    },
+    {
+      "epoch": 0.9949952434131613,
+      "grad_norm": 653.0780029296875,
+      "learning_rate": 2.3931863435985303e-08,
+      "loss": 79.4951,
+      "step": 120280
+    },
+    {
+      "epoch": 0.9950779666625305,
+      "grad_norm": 1000.9141845703125,
+      "learning_rate": 2.379423572779982e-08,
+      "loss": 72.5086,
+      "step": 120290
+    },
+    {
+      "epoch": 0.9951606899118998,
+      "grad_norm": 633.1144409179688,
+      "learning_rate": 2.3657003953468926e-08,
+      "loss": 94.2369,
+      "step": 120300
+    },
+    {
+      "epoch": 0.995243413161269,
+      "grad_norm": 827.4677734375,
+      "learning_rate": 2.352016812391278e-08,
+      "loss": 79.3089,
+      "step": 120310
+    },
+    {
+      "epoch": 0.9953261364106382,
+      "grad_norm": 759.8338012695312,
+      "learning_rate": 2.3383728250017112e-08,
+      "loss": 57.8192,
+      "step": 120320
+    },
+    {
+      "epoch": 0.9954088596600075,
+      "grad_norm": 735.8635864257812,
+      "learning_rate": 2.3247684342639355e-08,
+      "loss": 92.8142,
+      "step": 120330
+    },
+    {
+      "epoch": 0.9954915829093767,
+      "grad_norm": 739.5372314453125,
+      "learning_rate": 2.311203641260251e-08,
+      "loss": 82.3384,
+      "step": 120340
+    },
+    {
+      "epoch": 0.9955743061587459,
+      "grad_norm": 755.5853881835938,
+      "learning_rate": 2.2976784470700174e-08,
+      "loss": 117.4208,
+      "step": 120350
+    },
+    {
+      "epoch": 0.9956570294081152,
+      "grad_norm": 1694.4178466796875,
+      "learning_rate": 2.284192852769429e-08,
+      "loss": 125.9178,
+      "step": 120360
+    },
+    {
+      "epoch": 0.9957397526574844,
+      "grad_norm": 635.702880859375,
+      "learning_rate": 2.2707468594313497e-08,
+      "loss": 72.2283,
+      "step": 120370
+    },
+    {
+      "epoch": 0.9958224759068536,
+      "grad_norm": 1230.975341796875,
+      "learning_rate": 2.2573404681256463e-08,
+      "loss": 123.6782,
+      "step": 120380
+    },
+    {
+      "epoch": 0.9959051991562229,
+      "grad_norm": 678.0013427734375,
+      "learning_rate": 2.243973679919076e-08,
+      "loss": 80.5772,
+      "step": 120390
+    },
+    {
+      "epoch": 0.9959879224055921,
+      "grad_norm": 758.3480224609375,
+      "learning_rate": 2.2306464958751217e-08,
+      "loss": 67.9236,
+      "step": 120400
+    },
+    {
+      "epoch": 0.9960706456549613,
+      "grad_norm": 788.4033203125,
+      "learning_rate": 2.2173589170541576e-08,
+      "loss": 79.5632,
+      "step": 120410
+    },
+    {
+      "epoch": 0.9961533689043306,
+      "grad_norm": 529.8211059570312,
+      "learning_rate": 2.204110944513449e-08,
+      "loss": 113.7574,
+      "step": 120420
+    },
+    {
+      "epoch": 0.9962360921536998,
+      "grad_norm": 686.330078125,
+      "learning_rate": 2.1909025793070416e-08,
+      "loss": 67.1674,
+      "step": 120430
+    },
+    {
+      "epoch": 0.996318815403069,
+      "grad_norm": 659.1135864257812,
+      "learning_rate": 2.1777338224859278e-08,
+      "loss": 113.0164,
+      "step": 120440
+    },
+    {
+      "epoch": 0.9964015386524383,
+      "grad_norm": 969.425048828125,
+      "learning_rate": 2.1646046750978255e-08,
+      "loss": 99.7389,
+      "step": 120450
+    },
+    {
+      "epoch": 0.9964842619018075,
+      "grad_norm": 861.7916870117188,
+      "learning_rate": 2.1515151381873435e-08,
+      "loss": 84.4122,
+      "step": 120460
+    },
+    {
+      "epoch": 0.9965669851511767,
+      "grad_norm": 389.3373718261719,
+      "learning_rate": 2.1384652127959816e-08,
+      "loss": 83.2005,
+      "step": 120470
+    },
+    {
+      "epoch": 0.996649708400546,
+      "grad_norm": 1106.470458984375,
+      "learning_rate": 2.125454899962076e-08,
+      "loss": 96.3944,
+      "step": 120480
+    },
+    {
+      "epoch": 0.9967324316499152,
+      "grad_norm": 1094.505859375,
+      "learning_rate": 2.112484200720799e-08,
+      "loss": 83.0741,
+      "step": 120490
+    },
+    {
+      "epoch": 0.9968151548992844,
+      "grad_norm": 439.3631896972656,
+      "learning_rate": 2.0995531161041028e-08,
+      "loss": 97.843,
+      "step": 120500
+    },
+    {
+      "epoch": 0.9968978781486537,
+      "grad_norm": 987.277099609375,
+      "learning_rate": 2.0866616471409974e-08,
+      "loss": 105.7594,
+      "step": 120510
+    },
+    {
+      "epoch": 0.9969806013980229,
+      "grad_norm": 647.2058715820312,
+      "learning_rate": 2.0738097948570514e-08,
+      "loss": 68.505,
+      "step": 120520
+    },
+    {
+      "epoch": 0.9970633246473921,
+      "grad_norm": 1555.4075927734375,
+      "learning_rate": 2.0609975602749465e-08,
+      "loss": 75.6344,
+      "step": 120530
+    },
+    {
+      "epoch": 0.9971460478967614,
+      "grad_norm": 1339.879638671875,
+      "learning_rate": 2.048224944413979e-08,
+      "loss": 79.3426,
+      "step": 120540
+    },
+    {
+      "epoch": 0.9972287711461306,
+      "grad_norm": 446.56878662109375,
+      "learning_rate": 2.035491948290502e-08,
+      "loss": 111.5274,
+      "step": 120550
+    },
+    {
+      "epoch": 0.9973114943954998,
+      "grad_norm": 896.1454467773438,
+      "learning_rate": 2.0227985729175393e-08,
+      "loss": 93.6553,
+      "step": 120560
+    },
+    {
+      "epoch": 0.9973942176448691,
+      "grad_norm": 909.8489990234375,
+      "learning_rate": 2.0101448193051153e-08,
+      "loss": 81.4754,
+      "step": 120570
+    },
+    {
+      "epoch": 0.9974769408942383,
+      "grad_norm": 680.5419921875,
+      "learning_rate": 1.997530688459981e-08,
+      "loss": 64.9978,
+      "step": 120580
+    },
+    {
+      "epoch": 0.9975596641436075,
+      "grad_norm": 970.2647705078125,
+      "learning_rate": 1.984956181385833e-08,
+      "loss": 98.7731,
+      "step": 120590
+    },
+    {
+      "epoch": 0.9976423873929768,
+      "grad_norm": 1299.3114013671875,
+      "learning_rate": 1.9724212990830938e-08,
+      "loss": 73.416,
+      "step": 120600
+    },
+    {
+      "epoch": 0.997725110642346,
+      "grad_norm": 1053.347900390625,
+      "learning_rate": 1.9599260425491873e-08,
+      "loss": 82.9828,
+      "step": 120610
+    },
+    {
+      "epoch": 0.9978078338917152,
+      "grad_norm": 1263.4722900390625,
+      "learning_rate": 1.9474704127783184e-08,
+      "loss": 76.9199,
+      "step": 120620
+    },
+    {
+      "epoch": 0.9978905571410845,
+      "grad_norm": 490.1828918457031,
+      "learning_rate": 1.9350544107614165e-08,
+      "loss": 77.5371,
+      "step": 120630
+    },
+    {
+      "epoch": 0.9979732803904537,
+      "grad_norm": 1117.427734375,
+      "learning_rate": 1.9226780374864695e-08,
+      "loss": 103.023,
+      "step": 120640
+    },
+    {
+      "epoch": 0.9980560036398229,
+      "grad_norm": 476.96221923828125,
+      "learning_rate": 1.9103412939381338e-08,
+      "loss": 69.9926,
+      "step": 120650
+    },
+    {
+      "epoch": 0.9981387268891923,
+      "grad_norm": 722.0250854492188,
+      "learning_rate": 1.898044181098013e-08,
+      "loss": 101.3328,
+      "step": 120660
+    },
+    {
+      "epoch": 0.9982214501385615,
+      "grad_norm": 736.2235107421875,
+      "learning_rate": 1.885786699944492e-08,
+      "loss": 79.6206,
+      "step": 120670
+    },
+    {
+      "epoch": 0.9983041733879306,
+      "grad_norm": 1309.5517578125,
+      "learning_rate": 1.873568851452956e-08,
+      "loss": 78.2351,
+      "step": 120680
+    },
+    {
+      "epoch": 0.9983868966373,
+      "grad_norm": 752.5504150390625,
+      "learning_rate": 1.8613906365954616e-08,
+      "loss": 87.0025,
+      "step": 120690
+    },
+    {
+      "epoch": 0.9984696198866692,
+      "grad_norm": 815.6478881835938,
+      "learning_rate": 1.8492520563409555e-08,
+      "loss": 64.3488,
+      "step": 120700
+    },
+    {
+      "epoch": 0.9985523431360384,
+      "grad_norm": 664.5618286132812,
+      "learning_rate": 1.837153111655221e-08,
+      "loss": 70.5869,
+      "step": 120710
+    },
+    {
+      "epoch": 0.9986350663854077,
+      "grad_norm": 1548.3511962890625,
+      "learning_rate": 1.825093803500988e-08,
+      "loss": 103.4076,
+      "step": 120720
+    },
+    {
+      "epoch": 0.9987177896347769,
+      "grad_norm": 667.9990234375,
+      "learning_rate": 1.813074132837711e-08,
+      "loss": 93.5051,
+      "step": 120730
+    },
+    {
+      "epoch": 0.9988005128841461,
+      "grad_norm": 587.8815307617188,
+      "learning_rate": 1.801094100621792e-08,
+      "loss": 84.6205,
+      "step": 120740
+    },
+    {
+      "epoch": 0.9988832361335154,
+      "grad_norm": 895.0107421875,
+      "learning_rate": 1.789153707806357e-08,
+      "loss": 76.1649,
+      "step": 120750
+    },
+    {
+      "epoch": 0.9989659593828846,
+      "grad_norm": 657.9972534179688,
+      "learning_rate": 1.7772529553414798e-08,
+      "loss": 74.9104,
+      "step": 120760
+    },
+    {
+      "epoch": 0.9990486826322538,
+      "grad_norm": 1787.9937744140625,
+      "learning_rate": 1.7653918441740693e-08,
+      "loss": 109.789,
+      "step": 120770
+    },
+    {
+      "epoch": 0.9991314058816231,
+      "grad_norm": 562.6156616210938,
+      "learning_rate": 1.753570375247815e-08,
+      "loss": 76.8233,
+      "step": 120780
+    },
+    {
+      "epoch": 0.9992141291309923,
+      "grad_norm": 759.29150390625,
+      "learning_rate": 1.7417885495033537e-08,
+      "loss": 86.3301,
+      "step": 120790
+    },
+    {
+      "epoch": 0.9992968523803615,
+      "grad_norm": 533.1914672851562,
+      "learning_rate": 1.730046367878102e-08,
+      "loss": 123.4786,
+      "step": 120800
+    },
+    {
+      "epoch": 0.9993795756297308,
+      "grad_norm": 1021.0844116210938,
+      "learning_rate": 1.7183438313062573e-08,
+      "loss": 79.5246,
+      "step": 120810
+    },
+    {
+      "epoch": 0.9994622988791,
+      "grad_norm": 579.9783935546875,
+      "learning_rate": 1.7066809407190187e-08,
+      "loss": 84.2175,
+      "step": 120820
+    },
+    {
+      "epoch": 0.9995450221284692,
+      "grad_norm": 669.8778076171875,
+      "learning_rate": 1.695057697044311e-08,
+      "loss": 100.9156,
+      "step": 120830
+    },
+    {
+      "epoch": 0.9996277453778385,
+      "grad_norm": 494.1153259277344,
+      "learning_rate": 1.68347410120695e-08,
+      "loss": 78.8109,
+      "step": 120840
+    },
+    {
+      "epoch": 0.9997104686272077,
+      "grad_norm": 1579.7835693359375,
+      "learning_rate": 1.671930154128587e-08,
+      "loss": 84.4901,
+      "step": 120850
+    },
+    {
+      "epoch": 0.9997931918765769,
+      "grad_norm": 1268.185546875,
+      "learning_rate": 1.6604258567277652e-08,
+      "loss": 77.1438,
+      "step": 120860
+    },
+    {
+      "epoch": 0.9998759151259462,
+      "grad_norm": 618.141845703125,
+      "learning_rate": 1.6489612099197527e-08,
+      "loss": 88.3295,
+      "step": 120870
+    },
+    {
+      "epoch": 0.9999586383753154,
+      "grad_norm": 1028.1591796875,
+      "learning_rate": 1.637536214616764e-08,
+      "loss": 69.9013,
+      "step": 120880
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 76.04621887207031,
+      "eval_runtime": 214.3291,
+      "eval_samples_per_second": 45.579,
+      "eval_steps_per_second": 5.702,
+      "step": 120885
+    },
+    {
+      "epoch": 1.0000413616246846,
+      "grad_norm": 835.1196899414062,
+      "learning_rate": 1.6261508717278497e-08,
+      "loss": 103.3933,
+      "step": 120890
+    },
+    {
+      "epoch": 1.0001240848740538,
+      "grad_norm": 625.5361938476562,
+      "learning_rate": 1.614805182158896e-08,
+      "loss": 81.46,
+      "step": 120900
+    },
+    {
+      "epoch": 1.000206808123423,
+      "grad_norm": 715.9172973632812,
+      "learning_rate": 1.603499146812626e-08,
+      "loss": 85.437,
+      "step": 120910
+    },
+    {
+      "epoch": 1.0002895313727924,
+      "grad_norm": 975.20361328125,
+      "learning_rate": 1.5922327665885416e-08,
+      "loss": 100.9134,
+      "step": 120920
+    },
+    {
+      "epoch": 1.0003722546221616,
+      "grad_norm": 1666.02734375,
+      "learning_rate": 1.5810060423831487e-08,
+      "loss": 88.4254,
+      "step": 120930
+    },
+    {
+      "epoch": 1.0004549778715308,
+      "grad_norm": 772.5983276367188,
+      "learning_rate": 1.5698189750896762e-08,
+      "loss": 97.352,
+      "step": 120940
+    },
+    {
+      "epoch": 1.0005377011209,
+      "grad_norm": 1283.052490234375,
+      "learning_rate": 1.5586715655982463e-08,
+      "loss": 107.4238,
+      "step": 120950
+    },
+    {
+      "epoch": 1.0006204243702692,
+      "grad_norm": 646.5191650390625,
+      "learning_rate": 1.5475638147957607e-08,
+      "loss": 72.1843,
+      "step": 120960
+    },
+    {
+      "epoch": 1.0007031476196384,
+      "grad_norm": 777.9674072265625,
+      "learning_rate": 1.5364957235660115e-08,
+      "loss": 90.0959,
+      "step": 120970
+    },
+    {
+      "epoch": 1.0007858708690078,
+      "grad_norm": 364.2453308105469,
+      "learning_rate": 1.525467292789684e-08,
+      "loss": 74.3597,
+      "step": 120980
+    },
+    {
+      "epoch": 1.000868594118377,
+      "grad_norm": 626.4849853515625,
+      "learning_rate": 1.5144785233442428e-08,
+      "loss": 70.714,
+      "step": 120990
+    },
+    {
+      "epoch": 1.0009513173677462,
+      "grad_norm": 418.33367919921875,
+      "learning_rate": 1.5035294161039882e-08,
+      "loss": 81.4252,
+      "step": 121000
+    },
+    {
+      "epoch": 1.0010340406171154,
+      "grad_norm": 1237.5550537109375,
+      "learning_rate": 1.4926199719401124e-08,
+      "loss": 93.9281,
+      "step": 121010
+    },
+    {
+      "epoch": 1.0011167638664846,
+      "grad_norm": 523.1364135742188,
+      "learning_rate": 1.4817501917205879e-08,
+      "loss": 67.6458,
+      "step": 121020
+    },
+    {
+      "epoch": 1.0011994871158538,
+      "grad_norm": 531.9727172851562,
+      "learning_rate": 1.4709200763103892e-08,
+      "loss": 67.4297,
+      "step": 121030
+    },
+    {
+      "epoch": 1.0012822103652232,
+      "grad_norm": 2390.410888671875,
+      "learning_rate": 1.460129626571105e-08,
+      "loss": 97.8159,
+      "step": 121040
+    },
+    {
+      "epoch": 1.0013649336145924,
+      "grad_norm": 923.2665405273438,
+      "learning_rate": 1.449378843361271e-08,
+      "loss": 96.3494,
+      "step": 121050
+    },
+    {
+      "epoch": 1.0014476568639616,
+      "grad_norm": 484.8533630371094,
+      "learning_rate": 1.4386677275363692e-08,
+      "loss": 62.1874,
+      "step": 121060
+    },
+    {
+      "epoch": 1.0015303801133308,
+      "grad_norm": 816.00146484375,
+      "learning_rate": 1.4279962799486069e-08,
+      "loss": 85.6988,
+      "step": 121070
+    },
+    {
+      "epoch": 1.0016131033627,
+      "grad_norm": 547.6641235351562,
+      "learning_rate": 1.4173645014470272e-08,
+      "loss": 101.0872,
+      "step": 121080
+    },
+    {
+      "epoch": 1.0016958266120692,
+      "grad_norm": 648.3076782226562,
+      "learning_rate": 1.4067723928775645e-08,
+      "loss": 74.9175,
+      "step": 121090
+    },
+    {
+      "epoch": 1.0017785498614387,
+      "grad_norm": 828.8029174804688,
+      "learning_rate": 1.3962199550829892e-08,
+      "loss": 83.7772,
+      "step": 121100
+    },
+    {
+      "epoch": 1.0018612731108079,
+      "grad_norm": 672.8508911132812,
+      "learning_rate": 1.3857071889029073e-08,
+      "loss": 99.7323,
+      "step": 121110
+    },
+    {
+      "epoch": 1.001943996360177,
+      "grad_norm": 847.6339111328125,
+      "learning_rate": 1.3752340951737609e-08,
+      "loss": 83.5709,
+      "step": 121120
+    },
+    {
+      "epoch": 1.0020267196095463,
+      "grad_norm": 1562.5223388671875,
+      "learning_rate": 1.3648006747288833e-08,
+      "loss": 85.9273,
+      "step": 121130
+    },
+    {
+      "epoch": 1.0021094428589155,
+      "grad_norm": 1429.5631103515625,
+      "learning_rate": 1.354406928398333e-08,
+      "loss": 100.3321,
+      "step": 121140
+    },
+    {
+      "epoch": 1.0021921661082847,
+      "grad_norm": 858.3600463867188,
+      "learning_rate": 1.3440528570092259e-08,
+      "loss": 88.2199,
+      "step": 121150
+    },
+    {
+      "epoch": 1.002274889357654,
+      "grad_norm": 965.1406860351562,
+      "learning_rate": 1.3337384613852922e-08,
+      "loss": 71.9796,
+      "step": 121160
+    },
+    {
+      "epoch": 1.0023576126070233,
+      "grad_norm": 1044.383056640625,
+      "learning_rate": 1.3234637423472085e-08,
+      "loss": 76.3436,
+      "step": 121170
+    },
+    {
+      "epoch": 1.0024403358563925,
+      "grad_norm": 696.2551879882812,
+      "learning_rate": 1.3132287007124877e-08,
+      "loss": 70.0724,
+      "step": 121180
+    },
+    {
+      "epoch": 1.0025230591057617,
+      "grad_norm": 814.8746948242188,
+      "learning_rate": 1.3030333372954784e-08,
+      "loss": 69.551,
+      "step": 121190
+    },
+    {
+      "epoch": 1.0026057823551309,
+      "grad_norm": 665.35009765625,
+      "learning_rate": 1.2928776529074205e-08,
+      "loss": 96.7717,
+      "step": 121200
+    },
+    {
+      "epoch": 1.0026885056045,
+      "grad_norm": 1074.0308837890625,
+      "learning_rate": 1.2827616483563343e-08,
+      "loss": 147.5443,
+      "step": 121210
+    },
+    {
+      "epoch": 1.0027712288538695,
+      "grad_norm": 541.9178466796875,
+      "learning_rate": 1.2726853244471316e-08,
+      "loss": 80.1292,
+      "step": 121220
+    },
+    {
+      "epoch": 1.0028539521032387,
+      "grad_norm": 1141.502685546875,
+      "learning_rate": 1.2626486819814488e-08,
+      "loss": 112.4385,
+      "step": 121230
+    },
+    {
+      "epoch": 1.0029366753526079,
+      "grad_norm": 396.3456115722656,
+      "learning_rate": 1.2526517217579248e-08,
+      "loss": 74.7738,
+      "step": 121240
+    },
+    {
+      "epoch": 1.003019398601977,
+      "grad_norm": 824.7189331054688,
+      "learning_rate": 1.2426944445719791e-08,
+      "loss": 81.8229,
+      "step": 121250
+    },
+    {
+      "epoch": 1.0031021218513463,
+      "grad_norm": 322.1405334472656,
+      "learning_rate": 1.2327768512158667e-08,
+      "loss": 62.011,
+      "step": 121260
+    },
+    {
+      "epoch": 1.0031848451007155,
+      "grad_norm": 780.4888916015625,
+      "learning_rate": 1.2228989424786786e-08,
+      "loss": 79.2352,
+      "step": 121270
+    },
+    {
+      "epoch": 1.0032675683500847,
+      "grad_norm": 664.0239868164062,
+      "learning_rate": 1.2130607191462863e-08,
+      "loss": 78.1969,
+      "step": 121280
+    },
+    {
+      "epoch": 1.003350291599454,
+      "grad_norm": 737.7160034179688,
+      "learning_rate": 1.2032621820015633e-08,
+      "loss": 89.1368,
+      "step": 121290
+    },
+    {
+      "epoch": 1.0034330148488233,
+      "grad_norm": 988.8171997070312,
+      "learning_rate": 1.1935033318241084e-08,
+      "loss": 84.0071,
+      "step": 121300
+    },
+    {
+      "epoch": 1.0035157380981925,
+      "grad_norm": 708.0661010742188,
+      "learning_rate": 1.1837841693904118e-08,
+      "loss": 72.5567,
+      "step": 121310
+    },
+    {
+      "epoch": 1.0035984613475617,
+      "grad_norm": 664.5126342773438,
+      "learning_rate": 1.174104695473688e-08,
+      "loss": 67.6821,
+      "step": 121320
+    },
+    {
+      "epoch": 1.003681184596931,
+      "grad_norm": 548.2943725585938,
+      "learning_rate": 1.1644649108441542e-08,
+      "loss": 95.8433,
+      "step": 121330
+    },
+    {
+      "epoch": 1.0037639078463,
+      "grad_norm": 988.5643310546875,
+      "learning_rate": 1.1548648162688081e-08,
+      "loss": 87.8513,
+      "step": 121340
+    },
+    {
+      "epoch": 1.0038466310956695,
+      "grad_norm": 895.0706176757812,
+      "learning_rate": 1.1453044125114832e-08,
+      "loss": 89.9198,
+      "step": 121350
+    },
+    {
+      "epoch": 1.0039293543450387,
+      "grad_norm": 858.029296875,
+      "learning_rate": 1.1357837003329042e-08,
+      "loss": 67.4552,
+      "step": 121360
+    },
+    {
+      "epoch": 1.004012077594408,
+      "grad_norm": 605.3368530273438,
+      "learning_rate": 1.1263026804904653e-08,
+      "loss": 54.8068,
+      "step": 121370
+    },
+    {
+      "epoch": 1.0040948008437771,
+      "grad_norm": 557.933349609375,
+      "learning_rate": 1.1168613537386186e-08,
+      "loss": 79.3044,
+      "step": 121380
+    },
+    {
+      "epoch": 1.0041775240931463,
+      "grad_norm": 797.0153198242188,
+      "learning_rate": 1.1074597208285965e-08,
+      "loss": 85.258,
+      "step": 121390
+    },
+    {
+      "epoch": 1.0042602473425155,
+      "grad_norm": 749.7955932617188,
+      "learning_rate": 1.0980977825083561e-08,
+      "loss": 67.815,
+      "step": 121400
+    },
+    {
+      "epoch": 1.004342970591885,
+      "grad_norm": 1231.0704345703125,
+      "learning_rate": 1.0887755395228018e-08,
+      "loss": 87.1247,
+      "step": 121410
+    },
+    {
+      "epoch": 1.0044256938412541,
+      "grad_norm": 730.1659545898438,
+      "learning_rate": 1.0794929926137287e-08,
+      "loss": 87.4435,
+      "step": 121420
+    },
+    {
+      "epoch": 1.0045084170906233,
+      "grad_norm": 1234.9593505859375,
+      "learning_rate": 1.0702501425196576e-08,
+      "loss": 78.5555,
+      "step": 121430
+    },
+    {
+      "epoch": 1.0045911403399925,
+      "grad_norm": 728.4467163085938,
+      "learning_rate": 1.0610469899760001e-08,
+      "loss": 91.6618,
+      "step": 121440
+    },
+    {
+      "epoch": 1.0046738635893617,
+      "grad_norm": 939.027099609375,
+      "learning_rate": 1.0518835357150036e-08,
+      "loss": 77.4593,
+      "step": 121450
+    },
+    {
+      "epoch": 1.004756586838731,
+      "grad_norm": 785.4007568359375,
+      "learning_rate": 1.0427597804657518e-08,
+      "loss": 100.9078,
+      "step": 121460
+    },
+    {
+      "epoch": 1.0048393100881003,
+      "grad_norm": 932.05224609375,
+      "learning_rate": 1.0336757249542195e-08,
+      "loss": 77.0113,
+      "step": 121470
+    },
+    {
+      "epoch": 1.0049220333374695,
+      "grad_norm": 836.0494995117188,
+      "learning_rate": 1.0246313699031618e-08,
+      "loss": 100.5597,
+      "step": 121480
+    },
+    {
+      "epoch": 1.0050047565868387,
+      "grad_norm": 1092.8907470703125,
+      "learning_rate": 1.0156267160322253e-08,
+      "loss": 95.8895,
+      "step": 121490
+    },
+    {
+      "epoch": 1.005087479836208,
+      "grad_norm": 1151.7713623046875,
+      "learning_rate": 1.006661764057837e-08,
+      "loss": 119.7996,
+      "step": 121500
+    },
+    {
+      "epoch": 1.0051702030855771,
+      "grad_norm": 1046.9149169921875,
+      "learning_rate": 9.977365146932595e-09,
+      "loss": 54.5647,
+      "step": 121510
+    },
+    {
+      "epoch": 1.0052529263349463,
+      "grad_norm": 482.7346496582031,
+      "learning_rate": 9.888509686487025e-09,
+      "loss": 66.3425,
+      "step": 121520
+    },
+    {
+      "epoch": 1.0053356495843158,
+      "grad_norm": 771.5325927734375,
+      "learning_rate": 9.800051266311006e-09,
+      "loss": 84.2417,
+      "step": 121530
+    },
+    {
+      "epoch": 1.005418372833685,
+      "grad_norm": 931.7367553710938,
+      "learning_rate": 9.711989893443353e-09,
+      "loss": 84.3377,
+      "step": 121540
+    },
+    {
+      "epoch": 1.0055010960830542,
+      "grad_norm": 843.2858276367188,
+      "learning_rate": 9.624325574890125e-09,
+      "loss": 74.5406,
+      "step": 121550
+    },
+    {
+      "epoch": 1.0055838193324234,
+      "grad_norm": 802.0359497070312,
+      "learning_rate": 9.537058317626857e-09,
+      "loss": 72.5852,
+      "step": 121560
+    },
+    {
+      "epoch": 1.0056665425817926,
+      "grad_norm": 631.546875,
+      "learning_rate": 9.450188128596328e-09,
+      "loss": 87.635,
+      "step": 121570
+    },
+    {
+      "epoch": 1.0057492658311618,
+      "grad_norm": 947.8718872070312,
+      "learning_rate": 9.363715014710784e-09,
+      "loss": 92.5211,
+      "step": 121580
+    },
+    {
+      "epoch": 1.0058319890805312,
+      "grad_norm": 304.4026184082031,
+      "learning_rate": 9.277638982850835e-09,
+      "loss": 99.9868,
+      "step": 121590
+    },
+    {
+      "epoch": 1.0059147123299004,
+      "grad_norm": 926.6283569335938,
+      "learning_rate": 9.191960039864334e-09,
+      "loss": 78.2546,
+      "step": 121600
+    },
+    {
+      "epoch": 1.0059974355792696,
+      "grad_norm": 812.9754028320312,
+      "learning_rate": 9.106678192569718e-09,
+      "loss": 71.0503,
+      "step": 121610
+    },
+    {
+      "epoch": 1.0060801588286388,
+      "grad_norm": 1050.507080078125,
+      "learning_rate": 9.021793447750448e-09,
+      "loss": 83.0579,
+      "step": 121620
+    },
+    {
+      "epoch": 1.006162882078008,
+      "grad_norm": 738.1742553710938,
+      "learning_rate": 8.937305812162234e-09,
+      "loss": 83.2435,
+      "step": 121630
+    },
+    {
+      "epoch": 1.0062456053273772,
+      "grad_norm": 946.2736206054688,
+      "learning_rate": 8.853215292526917e-09,
+      "loss": 77.3213,
+      "step": 121640
+    },
+    {
+      "epoch": 1.0063283285767466,
+      "grad_norm": 717.0030517578125,
+      "learning_rate": 8.769521895534705e-09,
+      "loss": 60.9286,
+      "step": 121650
+    },
+    {
+      "epoch": 1.0064110518261158,
+      "grad_norm": 722.735107421875,
+      "learning_rate": 8.686225627845268e-09,
+      "loss": 60.4515,
+      "step": 121660
+    },
+    {
+      "epoch": 1.006493775075485,
+      "grad_norm": 477.0111083984375,
+      "learning_rate": 8.603326496085529e-09,
+      "loss": 103.6363,
+      "step": 121670
+    },
+    {
+      "epoch": 1.0065764983248542,
+      "grad_norm": 554.1845092773438,
+      "learning_rate": 8.520824506851877e-09,
+      "loss": 88.1589,
+      "step": 121680
+    },
+    {
+      "epoch": 1.0066592215742234,
+      "grad_norm": 795.9306640625,
+      "learning_rate": 8.438719666707951e-09,
+      "loss": 70.1685,
+      "step": 121690
+    },
+    {
+      "epoch": 1.0067419448235926,
+      "grad_norm": 959.230224609375,
+      "learning_rate": 8.357011982187412e-09,
+      "loss": 82.1214,
+      "step": 121700
+    },
+    {
+      "epoch": 1.006824668072962,
+      "grad_norm": 684.617919921875,
+      "learning_rate": 8.27570145979062e-09,
+      "loss": 75.703,
+      "step": 121710
+    },
+    {
+      "epoch": 1.0069073913223312,
+      "grad_norm": 424.90740966796875,
+      "learning_rate": 8.194788105987395e-09,
+      "loss": 74.4388,
+      "step": 121720
+    },
+    {
+      "epoch": 1.0069901145717004,
+      "grad_norm": 777.08251953125,
+      "learning_rate": 8.114271927215923e-09,
+      "loss": 97.2741,
+      "step": 121730
+    },
+    {
+      "epoch": 1.0070728378210696,
+      "grad_norm": 970.0604858398438,
+      "learning_rate": 8.034152929881633e-09,
+      "loss": 78.7951,
+      "step": 121740
+    },
+    {
+      "epoch": 1.0071555610704388,
+      "grad_norm": 800.564697265625,
+      "learning_rate": 7.954431120359985e-09,
+      "loss": 73.0719,
+      "step": 121750
+    },
+    {
+      "epoch": 1.007238284319808,
+      "grad_norm": 970.1133422851562,
+      "learning_rate": 7.875106504994234e-09,
+      "loss": 84.3109,
+      "step": 121760
+    },
+    {
+      "epoch": 1.0073210075691774,
+      "grad_norm": 535.0064086914062,
+      "learning_rate": 7.796179090094891e-09,
+      "loss": 114.3683,
+      "step": 121770
+    },
+    {
+      "epoch": 1.0074037308185466,
+      "grad_norm": 1022.1173095703125,
+      "learning_rate": 7.71764888194304e-09,
+      "loss": 74.4927,
+      "step": 121780
+    },
+    {
+      "epoch": 1.0074864540679158,
+      "grad_norm": 387.936767578125,
+      "learning_rate": 7.63951588678591e-09,
+      "loss": 93.5733,
+      "step": 121790
+    },
+    {
+      "epoch": 1.007569177317285,
+      "grad_norm": 637.2479248046875,
+      "learning_rate": 7.561780110840744e-09,
+      "loss": 64.9509,
+      "step": 121800
+    },
+    {
+      "epoch": 1.0076519005666542,
+      "grad_norm": 817.9259033203125,
+      "learning_rate": 7.484441560292599e-09,
+      "loss": 107.0969,
+      "step": 121810
+    },
+    {
+      "epoch": 1.0077346238160234,
+      "grad_norm": 1081.365234375,
+      "learning_rate": 7.407500241294885e-09,
+      "loss": 68.617,
+      "step": 121820
+    },
+    {
+      "epoch": 1.0078173470653928,
+      "grad_norm": 1211.7698974609375,
+      "learning_rate": 7.3309561599693715e-09,
+      "loss": 83.9199,
+      "step": 121830
+    },
+    {
+      "epoch": 1.007900070314762,
+      "grad_norm": 497.444091796875,
+      "learning_rate": 7.254809322406742e-09,
+      "loss": 69.8575,
+      "step": 121840
+    },
+    {
+      "epoch": 1.0079827935641312,
+      "grad_norm": 490.0628967285156,
+      "learning_rate": 7.1790597346649286e-09,
+      "loss": 70.6277,
+      "step": 121850
+    },
+    {
+      "epoch": 1.0080655168135004,
+      "grad_norm": 1028.51953125,
+      "learning_rate": 7.103707402771887e-09,
+      "loss": 88.6616,
+      "step": 121860
+    },
+    {
+      "epoch": 1.0081482400628696,
+      "grad_norm": 2735.880126953125,
+      "learning_rate": 7.028752332722266e-09,
+      "loss": 106.7273,
+      "step": 121870
+    },
+    {
+      "epoch": 1.0082309633122388,
+      "grad_norm": 659.5089111328125,
+      "learning_rate": 6.954194530480185e-09,
+      "loss": 65.3164,
+      "step": 121880
+    },
+    {
+      "epoch": 1.0083136865616082,
+      "grad_norm": 610.6782836914062,
+      "learning_rate": 6.880034001977565e-09,
+      "loss": 66.6304,
+      "step": 121890
+    },
+    {
+      "epoch": 1.0083964098109774,
+      "grad_norm": 908.3348999023438,
+      "learning_rate": 6.806270753115796e-09,
+      "loss": 91.2019,
+      "step": 121900
+    },
+    {
+      "epoch": 1.0084791330603466,
+      "grad_norm": 1129.9967041015625,
+      "learning_rate": 6.732904789762962e-09,
+      "loss": 87.364,
+      "step": 121910
+    },
+    {
+      "epoch": 1.0085618563097158,
+      "grad_norm": 1457.5179443359375,
+      "learning_rate": 6.659936117757171e-09,
+      "loss": 100.0551,
+      "step": 121920
+    },
+    {
+      "epoch": 1.008644579559085,
+      "grad_norm": 696.4898681640625,
+      "learning_rate": 6.587364742903779e-09,
+      "loss": 98.5237,
+      "step": 121930
+    },
+    {
+      "epoch": 1.0087273028084542,
+      "grad_norm": 702.8273315429688,
+      "learning_rate": 6.515190670977057e-09,
+      "loss": 54.427,
+      "step": 121940
+    },
+    {
+      "epoch": 1.0088100260578237,
+      "grad_norm": 1061.605224609375,
+      "learning_rate": 6.4434139077201865e-09,
+      "loss": 81.1102,
+      "step": 121950
+    },
+    {
+      "epoch": 1.0088927493071929,
+      "grad_norm": 623.2269287109375,
+      "learning_rate": 6.3720344588430464e-09,
+      "loss": 117.4438,
+      "step": 121960
+    },
+    {
+      "epoch": 1.008975472556562,
+      "grad_norm": 1200.9871826171875,
+      "learning_rate": 6.301052330025537e-09,
+      "loss": 97.576,
+      "step": 121970
+    },
+    {
+      "epoch": 1.0090581958059313,
+      "grad_norm": 605.901611328125,
+      "learning_rate": 6.230467526915362e-09,
+      "loss": 63.3802,
+      "step": 121980
+    },
+    {
+      "epoch": 1.0091409190553005,
+      "grad_norm": 609.5484619140625,
+      "learning_rate": 6.160280055128032e-09,
+      "loss": 70.7963,
+      "step": 121990
+    },
+    {
+      "epoch": 1.0092236423046697,
+      "grad_norm": 692.6685180664062,
+      "learning_rate": 6.090489920249076e-09,
+      "loss": 90.4518,
+      "step": 122000
+    },
+    {
+      "epoch": 1.0093063655540389,
+      "grad_norm": 900.5830688476562,
+      "learning_rate": 6.021097127831277e-09,
+      "loss": 80.2815,
+      "step": 122010
+    },
+    {
+      "epoch": 1.0093890888034083,
+      "grad_norm": 1926.3387451171875,
+      "learning_rate": 5.952101683394662e-09,
+      "loss": 118.7339,
+      "step": 122020
+    },
+    {
+      "epoch": 1.0094718120527775,
+      "grad_norm": 563.1376953125,
+      "learning_rate": 5.8835035924303955e-09,
+      "loss": 93.353,
+      "step": 122030
+    },
+    {
+      "epoch": 1.0095545353021467,
+      "grad_norm": 601.802978515625,
+      "learning_rate": 5.815302860395778e-09,
+      "loss": 94.3208,
+      "step": 122040
+    },
+    {
+      "epoch": 1.0096372585515159,
+      "grad_norm": 828.7846069335938,
+      "learning_rate": 5.7474994927170235e-09,
+      "loss": 67.016,
+      "step": 122050
+    },
+    {
+      "epoch": 1.009719981800885,
+      "grad_norm": 468.670654296875,
+      "learning_rate": 5.680093494789263e-09,
+      "loss": 67.1952,
+      "step": 122060
+    },
+    {
+      "epoch": 1.0098027050502543,
+      "grad_norm": 799.6488037109375,
+      "learning_rate": 5.613084871975982e-09,
+      "loss": 76.9322,
+      "step": 122070
+    },
+    {
+      "epoch": 1.0098854282996237,
+      "grad_norm": 481.51190185546875,
+      "learning_rate": 5.546473629607918e-09,
+      "loss": 68.0958,
+      "step": 122080
+    },
+    {
+      "epoch": 1.0099681515489929,
+      "grad_norm": 2192.124755859375,
+      "learning_rate": 5.48025977298583e-09,
+      "loss": 99.5286,
+      "step": 122090
+    },
+    {
+      "epoch": 1.010050874798362,
+      "grad_norm": 411.526123046875,
+      "learning_rate": 5.414443307377171e-09,
+      "loss": 96.2247,
+      "step": 122100
+    },
+    {
+      "epoch": 1.0101335980477313,
+      "grad_norm": 194.7018280029297,
+      "learning_rate": 5.349024238019973e-09,
+      "loss": 63.8754,
+      "step": 122110
+    },
+    {
+      "epoch": 1.0102163212971005,
+      "grad_norm": 1099.83154296875,
+      "learning_rate": 5.284002570117852e-09,
+      "loss": 108.386,
+      "step": 122120
+    },
+    {
+      "epoch": 1.0102990445464697,
+      "grad_norm": 622.5518798828125,
+      "learning_rate": 5.219378308845558e-09,
+      "loss": 81.0414,
+      "step": 122130
+    },
+    {
+      "epoch": 1.010381767795839,
+      "grad_norm": 436.0816955566406,
+      "learning_rate": 5.155151459343977e-09,
+      "loss": 105.7425,
+      "step": 122140
+    },
+    {
+      "epoch": 1.0104644910452083,
+      "grad_norm": 1281.2032470703125,
+      "learning_rate": 5.091322026724022e-09,
+      "loss": 83.9981,
+      "step": 122150
+    },
+    {
+      "epoch": 1.0105472142945775,
+      "grad_norm": 1101.347412109375,
+      "learning_rate": 5.027890016064408e-09,
+      "loss": 86.2431,
+      "step": 122160
+    },
+    {
+      "epoch": 1.0106299375439467,
+      "grad_norm": 490.4996032714844,
+      "learning_rate": 4.964855432411097e-09,
+      "loss": 70.88,
+      "step": 122170
+    },
+    {
+      "epoch": 1.010712660793316,
+      "grad_norm": 584.7899780273438,
+      "learning_rate": 4.90221828078008e-09,
+      "loss": 67.8227,
+      "step": 122180
+    },
+    {
+      "epoch": 1.010795384042685,
+      "grad_norm": 886.4894409179688,
+      "learning_rate": 4.8399785661557005e-09,
+      "loss": 79.1609,
+      "step": 122190
+    },
+    {
+      "epoch": 1.0108781072920545,
+      "grad_norm": 748.0830078125,
+      "learning_rate": 4.7781362934889995e-09,
+      "loss": 83.5357,
+      "step": 122200
+    },
+    {
+      "epoch": 1.0109608305414237,
+      "grad_norm": 582.5003662109375,
+      "learning_rate": 4.716691467701595e-09,
+      "loss": 70.7697,
+      "step": 122210
+    },
+    {
+      "epoch": 1.011043553790793,
+      "grad_norm": 1061.063720703125,
+      "learning_rate": 4.655644093681244e-09,
+      "loss": 86.0885,
+      "step": 122220
+    },
+    {
+      "epoch": 1.0111262770401621,
+      "grad_norm": 800.2383422851562,
+      "learning_rate": 4.5949941762862826e-09,
+      "loss": 67.8206,
+      "step": 122230
+    },
+    {
+      "epoch": 1.0112090002895313,
+      "grad_norm": 889.217041015625,
+      "learning_rate": 4.5347417203411845e-09,
+      "loss": 91.7229,
+      "step": 122240
+    },
+    {
+      "epoch": 1.0112917235389005,
+      "grad_norm": 465.7398681640625,
+      "learning_rate": 4.474886730641004e-09,
+      "loss": 76.3832,
+      "step": 122250
+    },
+    {
+      "epoch": 1.01137444678827,
+      "grad_norm": 638.1264038085938,
+      "learning_rate": 4.415429211948041e-09,
+      "loss": 64.9447,
+      "step": 122260
+    },
+    {
+      "epoch": 1.0114571700376391,
+      "grad_norm": 1501.3109130859375,
+      "learning_rate": 4.356369168992402e-09,
+      "loss": 76.8261,
+      "step": 122270
+    },
+    {
+      "epoch": 1.0115398932870083,
+      "grad_norm": 1041.6864013671875,
+      "learning_rate": 4.297706606473107e-09,
+      "loss": 101.5634,
+      "step": 122280
+    },
+    {
+      "epoch": 1.0116226165363775,
+      "grad_norm": 1001.2974243164062,
+      "learning_rate": 4.239441529058641e-09,
+      "loss": 85.1877,
+      "step": 122290
+    },
+    {
+      "epoch": 1.0117053397857467,
+      "grad_norm": 405.4379577636719,
+      "learning_rate": 4.181573941384187e-09,
+      "loss": 76.1107,
+      "step": 122300
+    },
+    {
+      "epoch": 1.011788063035116,
+      "grad_norm": 670.6312866210938,
+      "learning_rate": 4.1241038480543945e-09,
+      "loss": 63.3756,
+      "step": 122310
+    },
+    {
+      "epoch": 1.0118707862844853,
+      "grad_norm": 1032.002685546875,
+      "learning_rate": 4.067031253641162e-09,
+      "loss": 92.0514,
+      "step": 122320
+    },
+    {
+      "epoch": 1.0119535095338545,
+      "grad_norm": 894.6094970703125,
+      "learning_rate": 4.010356162686413e-09,
+      "loss": 166.9267,
+      "step": 122330
+    },
+    {
+      "epoch": 1.0120362327832237,
+      "grad_norm": 840.5252685546875,
+      "learning_rate": 3.9540785796993165e-09,
+      "loss": 72.8894,
+      "step": 122340
+    },
+    {
+      "epoch": 1.012118956032593,
+      "grad_norm": 867.8668823242188,
+      "learning_rate": 3.898198509157402e-09,
+      "loss": 55.9863,
+      "step": 122350
+    },
+    {
+      "epoch": 1.0122016792819621,
+      "grad_norm": 802.0833129882812,
+      "learning_rate": 3.842715955506559e-09,
+      "loss": 79.1176,
+      "step": 122360
+    },
+    {
+      "epoch": 1.0122844025313313,
+      "grad_norm": 1034.685302734375,
+      "learning_rate": 3.787630923161589e-09,
+      "loss": 77.4822,
+      "step": 122370
+    },
+    {
+      "epoch": 1.0123671257807008,
+      "grad_norm": 506.0362548828125,
+      "learning_rate": 3.7329434165050975e-09,
+      "loss": 66.0188,
+      "step": 122380
+    },
+    {
+      "epoch": 1.01244984903007,
+      "grad_norm": 1035.119384765625,
+      "learning_rate": 3.6786534398891573e-09,
+      "loss": 72.05,
+      "step": 122390
+    },
+    {
+      "epoch": 1.0125325722794392,
+      "grad_norm": 623.9898071289062,
+      "learning_rate": 3.6247609976319818e-09,
+      "loss": 56.7796,
+      "step": 122400
+    },
+    {
+      "epoch": 1.0126152955288084,
+      "grad_norm": 1366.751708984375,
+      "learning_rate": 3.5712660940229184e-09,
+      "loss": 81.9944,
+      "step": 122410
+    },
+    {
+      "epoch": 1.0126980187781776,
+      "grad_norm": 689.130615234375,
+      "learning_rate": 3.518168733317451e-09,
+      "loss": 130.3806,
+      "step": 122420
+    },
+    {
+      "epoch": 1.0127807420275468,
+      "grad_norm": 1009.6278686523438,
+      "learning_rate": 3.4654689197405335e-09,
+      "loss": 81.2598,
+      "step": 122430
+    },
+    {
+      "epoch": 1.0128634652769162,
+      "grad_norm": 437.2408142089844,
+      "learning_rate": 3.413166657485478e-09,
+      "loss": 80.2506,
+      "step": 122440
+    },
+    {
+      "epoch": 1.0129461885262854,
+      "grad_norm": 862.9107666015625,
+      "learning_rate": 3.3612619507134013e-09,
+      "loss": 128.401,
+      "step": 122450
+    },
+    {
+      "epoch": 1.0130289117756546,
+      "grad_norm": 486.2399597167969,
+      "learning_rate": 3.3097548035537776e-09,
+      "loss": 61.5472,
+      "step": 122460
+    },
+    {
+      "epoch": 1.0131116350250238,
+      "grad_norm": 731.587890625,
+      "learning_rate": 3.25864522010555e-09,
+      "loss": 79.2592,
+      "step": 122470
+    },
+    {
+      "epoch": 1.013194358274393,
+      "grad_norm": 1046.9647216796875,
+      "learning_rate": 3.207933204435465e-09,
+      "loss": 70.5895,
+      "step": 122480
+    },
+    {
+      "epoch": 1.0132770815237622,
+      "grad_norm": 621.4727172851562,
+      "learning_rate": 3.157618760577519e-09,
+      "loss": 69.9926,
+      "step": 122490
+    },
+    {
+      "epoch": 1.0133598047731316,
+      "grad_norm": 774.1251220703125,
+      "learning_rate": 3.1077018925351753e-09,
+      "loss": 124.7276,
+      "step": 122500
+    },
+    {
+      "epoch": 1.0134425280225008,
+      "grad_norm": 1431.8812255859375,
+      "learning_rate": 3.0581826042808126e-09,
+      "loss": 103.817,
+      "step": 122510
+    },
+    {
+      "epoch": 1.01352525127187,
+      "grad_norm": 441.38287353515625,
+      "learning_rate": 3.009060899754057e-09,
+      "loss": 70.7602,
+      "step": 122520
+    },
+    {
+      "epoch": 1.0136079745212392,
+      "grad_norm": 560.4646606445312,
+      "learning_rate": 2.960336782862894e-09,
+      "loss": 81.647,
+      "step": 122530
+    },
+    {
+      "epoch": 1.0136906977706084,
+      "grad_norm": 600.3382568359375,
+      "learning_rate": 2.9120102574842212e-09,
+      "loss": 81.0002,
+      "step": 122540
+    },
+    {
+      "epoch": 1.0137734210199776,
+      "grad_norm": 1430.2890625,
+      "learning_rate": 2.8640813274638525e-09,
+      "loss": 108.3468,
+      "step": 122550
+    },
+    {
+      "epoch": 1.013856144269347,
+      "grad_norm": 565.4375,
+      "learning_rate": 2.816549996614293e-09,
+      "loss": 111.723,
+      "step": 122560
+    },
+    {
+      "epoch": 1.0139388675187162,
+      "grad_norm": 706.692626953125,
+      "learning_rate": 2.7694162687186276e-09,
+      "loss": 75.623,
+      "step": 122570
+    },
+    {
+      "epoch": 1.0140215907680854,
+      "grad_norm": 2190.73681640625,
+      "learning_rate": 2.7226801475255248e-09,
+      "loss": 116.6049,
+      "step": 122580
+    },
+    {
+      "epoch": 1.0141043140174546,
+      "grad_norm": 573.954345703125,
+      "learning_rate": 2.6763416367547866e-09,
+      "loss": 77.4606,
+      "step": 122590
+    },
+    {
+      "epoch": 1.0141870372668238,
+      "grad_norm": 855.6474609375,
+      "learning_rate": 2.630400740092909e-09,
+      "loss": 81.674,
+      "step": 122600
+    },
+    {
+      "epoch": 1.014269760516193,
+      "grad_norm": 641.304443359375,
+      "learning_rate": 2.5848574611953003e-09,
+      "loss": 93.9946,
+      "step": 122610
+    },
+    {
+      "epoch": 1.0143524837655624,
+      "grad_norm": 899.0760498046875,
+      "learning_rate": 2.5397118036851743e-09,
+      "loss": 101.6424,
+      "step": 122620
+    },
+    {
+      "epoch": 1.0144352070149316,
+      "grad_norm": 1727.291259765625,
+      "learning_rate": 2.494963771155212e-09,
+      "loss": 74.2995,
+      "step": 122630
+    },
+    {
+      "epoch": 1.0145179302643008,
+      "grad_norm": 804.0676879882812,
+      "learning_rate": 2.4506133671653443e-09,
+      "loss": 93.2673,
+      "step": 122640
+    },
+    {
+      "epoch": 1.01460065351367,
+      "grad_norm": 1078.8939208984375,
+      "learning_rate": 2.4066605952444144e-09,
+      "loss": 78.2374,
+      "step": 122650
+    },
+    {
+      "epoch": 1.0146833767630392,
+      "grad_norm": 1028.9287109375,
+      "learning_rate": 2.3631054588901802e-09,
+      "loss": 89.6077,
+      "step": 122660
+    },
+    {
+      "epoch": 1.0147661000124084,
+      "grad_norm": 654.36328125,
+      "learning_rate": 2.319947961567093e-09,
+      "loss": 84.3641,
+      "step": 122670
+    },
+    {
+      "epoch": 1.0148488232617778,
+      "grad_norm": 532.8497314453125,
+      "learning_rate": 2.277188106709627e-09,
+      "loss": 120.6354,
+      "step": 122680
+    },
+    {
+      "epoch": 1.014931546511147,
+      "grad_norm": 970.8217163085938,
+      "learning_rate": 2.234825897720061e-09,
+      "loss": 66.7476,
+      "step": 122690
+    },
+    {
+      "epoch": 1.0150142697605162,
+      "grad_norm": 934.7479858398438,
+      "learning_rate": 2.192861337968477e-09,
+      "loss": 95.2176,
+      "step": 122700
+    },
+    {
+      "epoch": 1.0150969930098854,
+      "grad_norm": 740.1159057617188,
+      "learning_rate": 2.151294430794426e-09,
+      "loss": 86.6372,
+      "step": 122710
+    },
+    {
+      "epoch": 1.0151797162592546,
+      "grad_norm": 928.073974609375,
+      "learning_rate": 2.110125179504152e-09,
+      "loss": 88.6049,
+      "step": 122720
+    },
+    {
+      "epoch": 1.0152624395086238,
+      "grad_norm": 741.3151245117188,
+      "learning_rate": 2.0693535873744784e-09,
+      "loss": 87.0215,
+      "step": 122730
+    },
+    {
+      "epoch": 1.015345162757993,
+      "grad_norm": 805.8557739257812,
+      "learning_rate": 2.028979657648922e-09,
+      "loss": 51.4993,
+      "step": 122740
+    },
+    {
+      "epoch": 1.0154278860073624,
+      "grad_norm": 620.05224609375,
+      "learning_rate": 1.989003393539912e-09,
+      "loss": 97.4243,
+      "step": 122750
+    },
+    {
+      "epoch": 1.0155106092567316,
+      "grad_norm": 693.1014404296875,
+      "learning_rate": 1.9494247982282386e-09,
+      "loss": 131.4482,
+      "step": 122760
+    },
+    {
+      "epoch": 1.0155933325061008,
+      "grad_norm": 651.7940673828125,
+      "learning_rate": 1.9102438748624943e-09,
+      "loss": 74.7952,
+      "step": 122770
+    },
+    {
+      "epoch": 1.01567605575547,
+      "grad_norm": 810.8271484375,
+      "learning_rate": 1.8714606265607395e-09,
+      "loss": 77.6325,
+      "step": 122780
+    },
+    {
+      "epoch": 1.0157587790048392,
+      "grad_norm": 2490.756591796875,
+      "learning_rate": 1.8330750564088396e-09,
+      "loss": 81.8933,
+      "step": 122790
+    },
+    {
+      "epoch": 1.0158415022542084,
+      "grad_norm": 928.2571411132812,
+      "learning_rate": 1.795087167459908e-09,
+      "loss": 67.8226,
+      "step": 122800
+    },
+    {
+      "epoch": 1.0159242255035779,
+      "grad_norm": 970.4078979492188,
+      "learning_rate": 1.757496962738192e-09,
+      "loss": 105.3837,
+      "step": 122810
+    },
+    {
+      "epoch": 1.016006948752947,
+      "grad_norm": 399.0166320800781,
+      "learning_rate": 1.7203044452329676e-09,
+      "loss": 68.2508,
+      "step": 122820
+    },
+    {
+      "epoch": 1.0160896720023163,
+      "grad_norm": 846.7438354492188,
+      "learning_rate": 1.6835096179040888e-09,
+      "loss": 113.6935,
+      "step": 122830
+    },
+    {
+      "epoch": 1.0161723952516855,
+      "grad_norm": 1102.63623046875,
+      "learning_rate": 1.6471124836792142e-09,
+      "loss": 99.2099,
+      "step": 122840
+    },
+    {
+      "epoch": 1.0162551185010547,
+      "grad_norm": 1199.7076416015625,
+      "learning_rate": 1.61111304545436e-09,
+      "loss": 79.1822,
+      "step": 122850
+    },
+    {
+      "epoch": 1.0163378417504239,
+      "grad_norm": 708.465576171875,
+      "learning_rate": 1.575511306093902e-09,
+      "loss": 74.9537,
+      "step": 122860
+    },
+    {
+      "epoch": 1.0164205649997933,
+      "grad_norm": 665.5509643554688,
+      "learning_rate": 1.5403072684300191e-09,
+      "loss": 79.0717,
+      "step": 122870
+    },
+    {
+      "epoch": 1.0165032882491625,
+      "grad_norm": 616.6489868164062,
+      "learning_rate": 1.505500935264359e-09,
+      "loss": 127.0564,
+      "step": 122880
+    },
+    {
+      "epoch": 1.0165860114985317,
+      "grad_norm": 849.0372924804688,
+      "learning_rate": 1.471092309365818e-09,
+      "loss": 70.2291,
+      "step": 122890
+    },
+    {
+      "epoch": 1.0166687347479009,
+      "grad_norm": 669.7859497070312,
+      "learning_rate": 1.4370813934722062e-09,
+      "loss": 122.7297,
+      "step": 122900
+    },
+    {
+      "epoch": 1.01675145799727,
+      "grad_norm": 1538.0843505859375,
+      "learning_rate": 1.403468190290247e-09,
+      "loss": 75.6392,
+      "step": 122910
+    },
+    {
+      "epoch": 1.0168341812466393,
+      "grad_norm": 595.5289916992188,
+      "learning_rate": 1.3702527024933576e-09,
+      "loss": 91.8487,
+      "step": 122920
+    },
+    {
+      "epoch": 1.0169169044960087,
+      "grad_norm": 982.86083984375,
+      "learning_rate": 1.337434932724979e-09,
+      "loss": 80.7263,
+      "step": 122930
+    },
+    {
+      "epoch": 1.0169996277453779,
+      "grad_norm": 984.17333984375,
+      "learning_rate": 1.305014883595801e-09,
+      "loss": 79.123,
+      "step": 122940
+    },
+    {
+      "epoch": 1.017082350994747,
+      "grad_norm": 805.4910888671875,
+      "learning_rate": 1.2729925576859815e-09,
+      "loss": 83.601,
+      "step": 122950
+    },
+    {
+      "epoch": 1.0171650742441163,
+      "grad_norm": 1756.73193359375,
+      "learning_rate": 1.2413679575434823e-09,
+      "loss": 102.3782,
+      "step": 122960
+    },
+    {
+      "epoch": 1.0172477974934855,
+      "grad_norm": 719.05615234375,
+      "learning_rate": 1.210141085683514e-09,
+      "loss": 85.2048,
+      "step": 122970
+    },
+    {
+      "epoch": 1.0173305207428547,
+      "grad_norm": 581.4542846679688,
+      "learning_rate": 1.1793119445918656e-09,
+      "loss": 51.8009,
+      "step": 122980
+    },
+    {
+      "epoch": 1.017413243992224,
+      "grad_norm": 1233.638916015625,
+      "learning_rate": 1.1488805367204648e-09,
+      "loss": 70.137,
+      "step": 122990
+    },
+    {
+      "epoch": 1.0174959672415933,
+      "grad_norm": 595.7598876953125,
+      "learning_rate": 1.118846864490708e-09,
+      "loss": 119.1018,
+      "step": 123000
+    },
+    {
+      "epoch": 1.0175786904909625,
+      "grad_norm": 652.7130126953125,
+      "learning_rate": 1.0892109302929055e-09,
+      "loss": 81.6739,
+      "step": 123010
+    },
+    {
+      "epoch": 1.0176614137403317,
+      "grad_norm": 612.19482421875,
+      "learning_rate": 1.0599727364851708e-09,
+      "loss": 69.6915,
+      "step": 123020
+    },
+    {
+      "epoch": 1.017744136989701,
+      "grad_norm": 1230.201171875,
+      "learning_rate": 1.0311322853928662e-09,
+      "loss": 80.762,
+      "step": 123030
+    },
+    {
+      "epoch": 1.01782686023907,
+      "grad_norm": 1171.8270263671875,
+      "learning_rate": 1.0026895793108227e-09,
+      "loss": 116.272,
+      "step": 123040
+    },
+    {
+      "epoch": 1.0179095834884395,
+      "grad_norm": 644.4696044921875,
+      "learning_rate": 9.74644620502785e-10,
+      "loss": 89.4497,
+      "step": 123050
+    },
+    {
+      "epoch": 1.0179923067378087,
+      "grad_norm": 625.5791015625,
+      "learning_rate": 9.46997411200301e-10,
+      "loss": 84.8642,
+      "step": 123060
+    },
+    {
+      "epoch": 1.018075029987178,
+      "grad_norm": 844.577880859375,
+      "learning_rate": 9.197479536021681e-10,
+      "loss": 96.1589,
+      "step": 123070
+    },
+    {
+      "epoch": 1.0181577532365471,
+      "grad_norm": 844.6200561523438,
+      "learning_rate": 8.928962498772065e-10,
+      "loss": 93.7864,
+      "step": 123080
+    },
+    {
+      "epoch": 1.0182404764859163,
+      "grad_norm": 1154.9962158203125,
+      "learning_rate": 8.664423021614854e-10,
+      "loss": 72.7706,
+      "step": 123090
+    },
+    {
+      "epoch": 1.0183231997352855,
+      "grad_norm": 1171.821044921875,
+      "learning_rate": 8.403861125599877e-10,
+      "loss": 87.6492,
+      "step": 123100
+    },
+    {
+      "epoch": 1.018405922984655,
+      "grad_norm": 697.9021606445312,
+      "learning_rate": 8.147276831460548e-10,
+      "loss": 80.5435,
+      "step": 123110
+    },
+    {
+      "epoch": 1.0184886462340241,
+      "grad_norm": 1245.968017578125,
+      "learning_rate": 7.894670159613871e-10,
+      "loss": 83.4186,
+      "step": 123120
+    },
+    {
+      "epoch": 1.0185713694833933,
+      "grad_norm": 1226.178466796875,
+      "learning_rate": 7.646041130149329e-10,
+      "loss": 89.3538,
+      "step": 123130
+    },
+    {
+      "epoch": 1.0186540927327625,
+      "grad_norm": 1140.58984375,
+      "learning_rate": 7.401389762862199e-10,
+      "loss": 99.5532,
+      "step": 123140
+    },
+    {
+      "epoch": 1.0187368159821317,
+      "grad_norm": 732.798828125,
+      "learning_rate": 7.16071607720914e-10,
+      "loss": 95.5324,
+      "step": 123150
+    },
+    {
+      "epoch": 1.018819539231501,
+      "grad_norm": 850.7224731445312,
+      "learning_rate": 6.924020092335948e-10,
+      "loss": 65.989,
+      "step": 123160
+    },
+    {
+      "epoch": 1.0189022624808703,
+      "grad_norm": 660.1827392578125,
+      "learning_rate": 6.691301827088659e-10,
+      "loss": 83.6253,
+      "step": 123170
+    },
+    {
+      "epoch": 1.0189849857302395,
+      "grad_norm": 558.4231567382812,
+      "learning_rate": 6.46256129997469e-10,
+      "loss": 86.4292,
+      "step": 123180
+    },
+    {
+      "epoch": 1.0190677089796087,
+      "grad_norm": 1053.580078125,
+      "learning_rate": 6.237798529190597e-10,
+      "loss": 88.6869,
+      "step": 123190
+    },
+    {
+      "epoch": 1.019150432228978,
+      "grad_norm": 1924.813232421875,
+      "learning_rate": 6.017013532627625e-10,
+      "loss": 88.4377,
+      "step": 123200
+    },
+    {
+      "epoch": 1.0192331554783471,
+      "grad_norm": 1369.2806396484375,
+      "learning_rate": 5.800206327855051e-10,
+      "loss": 77.994,
+      "step": 123210
+    },
+    {
+      "epoch": 1.0193158787277163,
+      "grad_norm": 745.227294921875,
+      "learning_rate": 5.58737693210909e-10,
+      "loss": 89.5734,
+      "step": 123220
+    },
+    {
+      "epoch": 1.0193986019770858,
+      "grad_norm": 391.84759521484375,
+      "learning_rate": 5.378525362337294e-10,
+      "loss": 81.0701,
+      "step": 123230
+    },
+    {
+      "epoch": 1.019481325226455,
+      "grad_norm": 504.9894104003906,
+      "learning_rate": 5.173651635148602e-10,
+      "loss": 66.15,
+      "step": 123240
+    },
+    {
+      "epoch": 1.0195640484758242,
+      "grad_norm": 955.138671875,
+      "learning_rate": 4.972755766846637e-10,
+      "loss": 55.0407,
+      "step": 123250
+    },
+    {
+      "epoch": 1.0196467717251934,
+      "grad_norm": 1267.6429443359375,
+      "learning_rate": 4.775837773418612e-10,
+      "loss": 92.8494,
+      "step": 123260
+    },
+    {
+      "epoch": 1.0197294949745626,
+      "grad_norm": 495.1304016113281,
+      "learning_rate": 4.5828976705297735e-10,
+      "loss": 78.7691,
+      "step": 123270
+    },
+    {
+      "epoch": 1.0198122182239318,
+      "grad_norm": 913.0636596679688,
+      "learning_rate": 4.3939354735345053e-10,
+      "loss": 95.1808,
+      "step": 123280
+    },
+    {
+      "epoch": 1.0198949414733012,
+      "grad_norm": 1205.1334228515625,
+      "learning_rate": 4.2089511974596765e-10,
+      "loss": 88.0171,
+      "step": 123290
+    },
+    {
+      "epoch": 1.0199776647226704,
+      "grad_norm": 1519.8599853515625,
+      "learning_rate": 4.027944857032395e-10,
+      "loss": 74.2102,
+      "step": 123300
+    },
+    {
+      "epoch": 1.0200603879720396,
+      "grad_norm": 845.7083129882812,
+      "learning_rate": 3.8509164666522546e-10,
+      "loss": 81.2621,
+      "step": 123310
+    },
+    {
+      "epoch": 1.0201431112214088,
+      "grad_norm": 1035.19189453125,
+      "learning_rate": 3.677866040402434e-10,
+      "loss": 77.1908,
+      "step": 123320
+    },
+    {
+      "epoch": 1.020225834470778,
+      "grad_norm": 679.1898193359375,
+      "learning_rate": 3.5087935920496975e-10,
+      "loss": 78.4861,
+      "step": 123330
+    },
+    {
+      "epoch": 1.0203085577201472,
+      "grad_norm": 987.113525390625,
+      "learning_rate": 3.343699135049949e-10,
+      "loss": 106.0358,
+      "step": 123340
+    },
+    {
+      "epoch": 1.0203912809695166,
+      "grad_norm": 1130.323486328125,
+      "learning_rate": 3.182582682542679e-10,
+      "loss": 104.6267,
+      "step": 123350
+    },
+    {
+      "epoch": 1.0204740042188858,
+      "grad_norm": 952.3067016601562,
+      "learning_rate": 3.0254442473398594e-10,
+      "loss": 91.154,
+      "step": 123360
+    },
+    {
+      "epoch": 1.020556727468255,
+      "grad_norm": 777.0213012695312,
+      "learning_rate": 2.8722838419481535e-10,
+      "loss": 84.8183,
+      "step": 123370
+    },
+    {
+      "epoch": 1.0206394507176242,
+      "grad_norm": 873.121337890625,
+      "learning_rate": 2.723101478546708e-10,
+      "loss": 75.0147,
+      "step": 123380
+    },
+    {
+      "epoch": 1.0207221739669934,
+      "grad_norm": 1494.041015625,
+      "learning_rate": 2.57789716902046e-10,
+      "loss": 92.4943,
+      "step": 123390
+    },
+    {
+      "epoch": 1.0208048972163626,
+      "grad_norm": 729.3314819335938,
+      "learning_rate": 2.436670924910178e-10,
+      "loss": 89.5984,
+      "step": 123400
+    },
+    {
+      "epoch": 1.0208876204657318,
+      "grad_norm": 744.1867065429688,
+      "learning_rate": 2.2994227574568705e-10,
+      "loss": 103.0427,
+      "step": 123410
+    },
+    {
+      "epoch": 1.0209703437151012,
+      "grad_norm": 762.7804565429688,
+      "learning_rate": 2.1661526775795804e-10,
+      "loss": 75.6005,
+      "step": 123420
+    },
+    {
+      "epoch": 1.0210530669644704,
+      "grad_norm": 758.2909545898438,
+      "learning_rate": 2.0368606958809377e-10,
+      "loss": 85.8044,
+      "step": 123430
+    },
+    {
+      "epoch": 1.0211357902138396,
+      "grad_norm": 943.9282836914062,
+      "learning_rate": 1.9115468226527101e-10,
+      "loss": 90.3115,
+      "step": 123440
+    },
+    {
+      "epoch": 1.0212185134632088,
+      "grad_norm": 1385.7298583984375,
+      "learning_rate": 1.790211067859149e-10,
+      "loss": 82.1778,
+      "step": 123450
+    },
+    {
+      "epoch": 1.021301236712578,
+      "grad_norm": 875.9342041015625,
+      "learning_rate": 1.6728534411591946e-10,
+      "loss": 93.4806,
+      "step": 123460
+    },
+    {
+      "epoch": 1.0213839599619472,
+      "grad_norm": 732.7958374023438,
+      "learning_rate": 1.5594739518898227e-10,
+      "loss": 65.2388,
+      "step": 123470
+    },
+    {
+      "epoch": 1.0214666832113166,
+      "grad_norm": 978.7460327148438,
+      "learning_rate": 1.4500726090715956e-10,
+      "loss": 85.1013,
+      "step": 123480
+    },
+    {
+      "epoch": 1.0215494064606858,
+      "grad_norm": 1050.985595703125,
+      "learning_rate": 1.3446494214031102e-10,
+      "loss": 90.1326,
+      "step": 123490
+    },
+    {
+      "epoch": 1.021632129710055,
+      "grad_norm": 823.0833740234375,
+      "learning_rate": 1.2432043972832042e-10,
+      "loss": 71.4136,
+      "step": 123500
+    },
+    {
+      "epoch": 1.0217148529594242,
+      "grad_norm": 735.2103881835938,
+      "learning_rate": 1.1457375447776476e-10,
+      "loss": 74.5606,
+      "step": 123510
+    },
+    {
+      "epoch": 1.0217975762087934,
+      "grad_norm": 836.6810302734375,
+      "learning_rate": 1.0522488716413481e-10,
+      "loss": 59.5058,
+      "step": 123520
+    },
+    {
+      "epoch": 1.0218802994581626,
+      "grad_norm": 490.60614013671875,
+      "learning_rate": 9.627383853128003e-11,
+      "loss": 74.4532,
+      "step": 123530
+    },
+    {
+      "epoch": 1.021963022707532,
+      "grad_norm": 616.2906494140625,
+      "learning_rate": 8.772060929196357e-11,
+      "loss": 62.0144,
+      "step": 123540
+    },
+    {
+      "epoch": 1.0220457459569012,
+      "grad_norm": 775.4456787109375,
+      "learning_rate": 7.956520012564195e-11,
+      "loss": 84.0153,
+      "step": 123550
+    },
+    {
+      "epoch": 1.0221284692062704,
+      "grad_norm": 1081.5496826171875,
+      "learning_rate": 7.180761168179563e-11,
+      "loss": 79.5349,
+      "step": 123560
+    },
+    {
+      "epoch": 1.0222111924556396,
+      "grad_norm": 855.0829467773438,
+      "learning_rate": 6.444784457770858e-11,
+      "loss": 90.5716,
+      "step": 123570
+    },
+    {
+      "epoch": 1.0222939157050088,
+      "grad_norm": 1062.8492431640625,
+      "learning_rate": 5.748589939902349e-11,
+      "loss": 111.6922,
+      "step": 123580
+    },
+    {
+      "epoch": 1.022376638954378,
+      "grad_norm": 1228.608154296875,
+      "learning_rate": 5.0921776699741634e-11,
+      "loss": 90.2285,
+      "step": 123590
+    },
+    {
+      "epoch": 1.0224593622037474,
+      "grad_norm": 1081.4580078125,
+      "learning_rate": 4.4755477001667824e-11,
+      "loss": 100.085,
+      "step": 123600
+    },
+    {
+      "epoch": 1.0225420854531166,
+      "grad_norm": 735.9952392578125,
+      "learning_rate": 3.898700079607576e-11,
+      "loss": 85.3372,
+      "step": 123610
+    },
+    {
+      "epoch": 1.0226248087024858,
+      "grad_norm": 562.4883422851562,
+      "learning_rate": 3.361634854093243e-11,
+      "loss": 114.2243,
+      "step": 123620
+    },
+    {
+      "epoch": 1.022707531951855,
+      "grad_norm": 543.7153930664062,
+      "learning_rate": 2.864352066478393e-11,
+      "loss": 100.8226,
+      "step": 123630
+    },
+    {
+      "epoch": 1.0227902552012242,
+      "grad_norm": 463.9744873046875,
+      "learning_rate": 2.406851756231454e-11,
+      "loss": 62.1017,
+      "step": 123640
+    },
+    {
+      "epoch": 1.0228729784505934,
+      "grad_norm": 1143.583740234375,
+      "learning_rate": 1.9891339597677415e-11,
+      "loss": 97.4913,
+      "step": 123650
+    },
+    {
+      "epoch": 1.0229557016999629,
+      "grad_norm": 592.5479736328125,
+      "learning_rate": 1.6111987103939464e-11,
+      "loss": 69.8716,
+      "step": 123660
+    },
+    {
+      "epoch": 1.023038424949332,
+      "grad_norm": 738.0408935546875,
+      "learning_rate": 1.273046038141601e-11,
+      "loss": 81.3839,
+      "step": 123670
+    },
+    {
+      "epoch": 1.0231211481987013,
+      "grad_norm": 5198.33935546875,
+      "learning_rate": 9.746759698781027e-12,
+      "loss": 182.0562,
+      "step": 123680
+    },
+    {
+      "epoch": 1.0232038714480705,
+      "grad_norm": 724.8328857421875,
+      "learning_rate": 7.160885294177355e-12,
+      "loss": 113.7217,
+      "step": 123690
+    },
+    {
+      "epoch": 1.0232865946974397,
+      "grad_norm": 1005.2647705078125,
+      "learning_rate": 4.9728373724411416e-12,
+      "loss": 77.6447,
+      "step": 123700
+    },
+    {
+      "epoch": 1.0233693179468089,
+      "grad_norm": 1014.6140747070312,
+      "learning_rate": 3.1826161084325124e-12,
+      "loss": 78.813,
+      "step": 123710
+    },
+    {
+      "epoch": 1.0234520411961783,
+      "grad_norm": 925.9212036132812,
+      "learning_rate": 1.7902216442600152e-12,
+      "loss": 70.0796,
+      "step": 123720
+    },
+    {
+      "epoch": 1.0235347644455475,
+      "grad_norm": 570.4528198242188,
+      "learning_rate": 7.956540903908405e-13,
+      "loss": 79.1544,
+      "step": 123730
+    },
+    {
+      "epoch": 1.0236174876949167,
+      "grad_norm": 898.171142578125,
+      "learning_rate": 1.9891352676104647e-13,
+      "loss": 58.6368,
+      "step": 123740
+    },
+    {
+      "epoch": 1.0237002109442859,
+      "grad_norm": 748.3834838867188,
+      "learning_rate": 0.0,
+      "loss": 101.8211,
+      "step": 123750
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 123750,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 2000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}