diff --git "a/7b/trainer_state.json" "b/7b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/7b/trainer_state.json"
@@ -0,0 +1,15297 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 2181,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 51.12571867652777,
+      "learning_rate": 1.5151515151515153e-05,
+      "loss": 7.0351,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 28.813409095819612,
+      "learning_rate": 3.0303030303030306e-05,
+      "loss": 7.0554,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 138.96081014821542,
+      "learning_rate": 4.545454545454546e-05,
+      "loss": 6.705,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 34.89949129527658,
+      "learning_rate": 6.060606060606061e-05,
+      "loss": 5.8443,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 19.008875432326786,
+      "learning_rate": 7.575757575757576e-05,
+      "loss": 5.4393,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 12.915846879528685,
+      "learning_rate": 9.090909090909092e-05,
+      "loss": 5.2955,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 11.283133618838027,
+      "learning_rate": 0.00010606060606060606,
+      "loss": 5.0711,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 13.900496595962197,
+      "learning_rate": 0.00012121212121212122,
+      "loss": 4.8488,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 21.125193417860302,
+      "learning_rate": 0.00013636363636363637,
+      "loss": 4.7177,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 6.3242559995267555,
+      "learning_rate": 0.00015151515151515152,
+      "loss": 4.4701,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.992941356074019,
+      "learning_rate": 0.00016666666666666666,
+      "loss": 4.4175,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.329039711426427,
+      "learning_rate": 0.00018181818181818183,
+      "loss": 4.3316,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.582306902481923,
+      "learning_rate": 0.00019696969696969695,
+      "loss": 4.3429,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 11.3126789466573,
+      "learning_rate": 0.00021212121212121213,
+      "loss": 4.2997,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.811625911339918,
+      "learning_rate": 0.00022727272727272727,
+      "loss": 4.2564,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 9.5903363032382,
+      "learning_rate": 0.00024242424242424245,
+      "loss": 4.028,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 15.198906765954014,
+      "learning_rate": 0.00025757575757575756,
+      "loss": 4.2279,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.7821926363193925,
+      "learning_rate": 0.00027272727272727274,
+      "loss": 4.1822,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.335104053233697,
+      "learning_rate": 0.0002878787878787879,
+      "loss": 4.0787,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 7.95183385507652,
+      "learning_rate": 0.00030303030303030303,
+      "loss": 4.0708,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 10.676735057154325,
+      "learning_rate": 0.0003181818181818182,
+      "loss": 4.0348,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.050859112502977,
+      "learning_rate": 0.0003333333333333333,
+      "loss": 4.0048,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 39.96503009854195,
+      "learning_rate": 0.0003484848484848485,
+      "loss": 3.9933,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.691958802417633,
+      "learning_rate": 0.00036363636363636367,
+      "loss": 4.2225,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 5.186965211383013,
+      "learning_rate": 0.0003787878787878788,
+      "loss": 4.1611,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.745509305265689,
+      "learning_rate": 0.0003939393939393939,
+      "loss": 4.054,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 11.405571021724224,
+      "learning_rate": 0.00040909090909090913,
+      "loss": 4.3372,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 12.625192354128874,
+      "learning_rate": 0.00042424242424242425,
+      "loss": 4.5419,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.93796582549662,
+      "learning_rate": 0.0004393939393939394,
+      "loss": 4.1928,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 14.618199014304697,
+      "learning_rate": 0.00045454545454545455,
+      "loss": 4.3646,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 6.985974407171347,
+      "learning_rate": 0.0004696969696969697,
+      "loss": 4.1514,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.559480963775047,
+      "learning_rate": 0.0004848484848484849,
+      "loss": 4.3174,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.011913067388771,
+      "learning_rate": 0.0005,
+      "loss": 4.2602,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.043066793700539,
+      "learning_rate": 0.0005151515151515151,
+      "loss": 4.1961,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 7.7568859317560355,
+      "learning_rate": 0.0005303030303030302,
+      "loss": 4.0749,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 12.285541097470366,
+      "learning_rate": 0.0005454545454545455,
+      "loss": 3.9721,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 16.697919545949883,
+      "learning_rate": 0.0005606060606060606,
+      "loss": 4.1619,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 4.656675394707004,
+      "learning_rate": 0.0005757575757575758,
+      "loss": 4.1483,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 8.831613360385623,
+      "learning_rate": 0.0005909090909090909,
+      "loss": 4.4321,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 8.179115364215471,
+      "learning_rate": 0.0006060606060606061,
+      "loss": 4.3132,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7401151476935148,
+      "learning_rate": 0.0006212121212121212,
+      "loss": 3.9094,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.65893991280624,
+      "learning_rate": 0.0006363636363636364,
+      "loss": 4.1059,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8584144309503413,
+      "learning_rate": 0.0006515151515151515,
+      "loss": 3.9624,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.320833112399646,
+      "learning_rate": 0.0006666666666666666,
+      "loss": 3.8809,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.843624953525116,
+      "learning_rate": 0.0006818181818181818,
+      "loss": 3.9462,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6325004518892849,
+      "learning_rate": 0.000696969696969697,
+      "loss": 3.9654,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.40639583322992573,
+      "learning_rate": 0.0007121212121212122,
+      "loss": 3.9225,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6930264273586801,
+      "learning_rate": 0.0007272727272727273,
+      "loss": 3.9161,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.5798470320913216,
+      "learning_rate": 0.0007424242424242425,
+      "loss": 3.8952,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.3230931814670408,
+      "learning_rate": 0.0007575757575757576,
+      "loss": 3.9508,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.386291383284893,
+      "learning_rate": 0.0007727272727272727,
+      "loss": 3.9127,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.382895337095947,
+      "learning_rate": 0.0007878787878787878,
+      "loss": 3.8659,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.31364886791783014,
+      "learning_rate": 0.000803030303030303,
+      "loss": 3.8487,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5542382835227448,
+      "learning_rate": 0.0008181818181818183,
+      "loss": 3.9279,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5788611396910377,
+      "learning_rate": 0.0008333333333333334,
+      "loss": 3.9127,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4441160585274129,
+      "learning_rate": 0.0008484848484848485,
+      "loss": 3.9826,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.508910101035815,
+      "learning_rate": 0.0008636363636363636,
+      "loss": 3.8611,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.3256940313304638,
+      "learning_rate": 0.0008787878787878789,
+      "loss": 3.8742,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.286074043184616,
+      "learning_rate": 0.000893939393939394,
+      "loss": 3.8191,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.30181508508230925,
+      "learning_rate": 0.0009090909090909091,
+      "loss": 3.8443,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.3159673995741885,
+      "learning_rate": 0.0009242424242424242,
+      "loss": 3.8287,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.2587470234743116,
+      "learning_rate": 0.0009393939393939394,
+      "loss": 3.8267,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7429863309559954,
+      "learning_rate": 0.0009545454545454546,
+      "loss": 3.8076,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.3816735949652952,
+      "learning_rate": 0.0009696969696969698,
+      "loss": 3.7293,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4330448131178964,
+      "learning_rate": 0.000984848484848485,
+      "loss": 3.7833,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4069087284276679,
+      "learning_rate": 0.001,
+      "loss": 3.6887,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.7183262298980956,
+      "learning_rate": 0.0009999994484067654,
+      "loss": 3.6693,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.3625086216271086,
+      "learning_rate": 0.0009999977936282788,
+      "loss": 3.7058,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5512170891000137,
+      "learning_rate": 0.0009999950356681913,
+      "loss": 3.5869,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.371490876178538,
+      "learning_rate": 0.0009999911745325876,
+      "loss": 3.692,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5087160536993884,
+      "learning_rate": 0.0009999862102299873,
+      "loss": 3.5616,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.49903811342676846,
+      "learning_rate": 0.0009999801427713433,
+      "loss": 3.6001,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8747443774511781,
+      "learning_rate": 0.0009999729721700424,
+      "loss": 3.4767,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.49451731848225655,
+      "learning_rate": 0.000999964698441906,
+      "loss": 3.6041,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5081932058013201,
+      "learning_rate": 0.0009999553216051892,
+      "loss": 3.5172,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.4026058416129961,
+      "learning_rate": 0.00099994484168058,
+      "loss": 3.5635,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45493094735054684,
+      "learning_rate": 0.0009999332586912019,
+      "loss": 3.664,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5681937317412628,
+      "learning_rate": 0.0009999205726626108,
+      "loss": 3.3972,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4020018393165983,
+      "learning_rate": 0.000999906783622797,
+      "loss": 3.3568,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45139555254127406,
+      "learning_rate": 0.0009998918916021842,
+      "loss": 3.5435,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.556641444876475,
+      "learning_rate": 0.0009998758966336297,
+      "loss": 3.5986,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6156290098795689,
+      "learning_rate": 0.0009998587987524242,
+      "loss": 3.4441,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5964525975089114,
+      "learning_rate": 0.0009998405979962926,
+      "loss": 3.367,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.9562397599301302,
+      "learning_rate": 0.000999821294405392,
+      "loss": 3.3185,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6611414687886483,
+      "learning_rate": 0.0009998008880223134,
+      "loss": 3.4438,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45548265392455406,
+      "learning_rate": 0.000999779378892081,
+      "loss": 3.2941,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.6465772312147813,
+      "learning_rate": 0.0009997567670621522,
+      "loss": 3.3439,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5108347279316393,
+      "learning_rate": 0.0009997330525824165,
+      "loss": 3.275,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5553632932795113,
+      "learning_rate": 0.0009997082355051976,
+      "loss": 3.3266,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.47954678951427127,
+      "learning_rate": 0.000999682315885251,
+      "loss": 3.3273,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.46168685192650183,
+      "learning_rate": 0.0009996552937797645,
+      "loss": 3.2993,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.45082106108919384,
+      "learning_rate": 0.0009996271692483596,
+      "loss": 3.2391,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0811070147798267,
+      "learning_rate": 0.0009995979423530893,
+      "loss": 3.226,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.502311517620603,
+      "learning_rate": 0.000999567613158439,
+      "loss": 3.2044,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.3177623642321388,
+      "learning_rate": 0.0009995361817313263,
+      "loss": 3.2473,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.5633712470025015,
+      "learning_rate": 0.0009995036481411004,
+      "loss": 3.1807,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0080358353350487,
+      "learning_rate": 0.0009994700124595429,
+      "loss": 3.3016,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.49452280650380337,
+      "learning_rate": 0.0009994352747608663,
+      "loss": 3.121,
+      "step": 98
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5860064272456921,
+      "learning_rate": 0.0009993994351217151,
+      "loss": 3.2099,
+      "step": 99
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6159789853250235,
+      "learning_rate": 0.000999362493621165,
+      "loss": 3.1449,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4692676648908955,
+      "learning_rate": 0.0009993244503407226,
+      "loss": 3.2371,
+      "step": 101
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.4807447695444268,
+      "learning_rate": 0.0009992853053643258,
+      "loss": 3.0888,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5578125840378995,
+      "learning_rate": 0.0009992450587783426,
+      "loss": 3.1082,
+      "step": 103
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5332179225286429,
+      "learning_rate": 0.000999203710671572,
+      "loss": 3.0437,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.48034431029343894,
+      "learning_rate": 0.0009991612611352438,
+      "loss": 3.1209,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.7005951124464233,
+      "learning_rate": 0.0009991177102630173,
+      "loss": 3.0602,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.48447427066530524,
+      "learning_rate": 0.0009990730581509817,
+      "loss": 3.0245,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.49092046082914104,
+      "learning_rate": 0.0009990273048976566,
+      "loss": 3.0243,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5153251452124302,
+      "learning_rate": 0.0009989804506039905,
+      "loss": 3.0585,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5273653326132732,
+      "learning_rate": 0.0009989324953733614,
+      "loss": 3.0211,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5092364683930998,
+      "learning_rate": 0.0009988834393115766,
+      "loss": 3.0246,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5805276133713935,
+      "learning_rate": 0.000998833282526872,
+      "loss": 3.0737,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5120017507607832,
+      "learning_rate": 0.0009987820251299122,
+      "loss": 2.9752,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6318949876542042,
+      "learning_rate": 0.00099872966723379,
+      "loss": 3.0822,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5578730203934199,
+      "learning_rate": 0.0009986762089540266,
+      "loss": 2.9152,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5710999659839776,
+      "learning_rate": 0.0009986216504085709,
+      "loss": 2.9253,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.6855595744298109,
+      "learning_rate": 0.0009985659917177991,
+      "loss": 2.9513,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5446621470664448,
+      "learning_rate": 0.0009985092330045155,
+      "loss": 3.0201,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.5991703183974769,
+      "learning_rate": 0.0009984513743939508,
+      "loss": 3.0381,
+      "step": 119
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4828593709305156,
+      "learning_rate": 0.0009983924160137626,
+      "loss": 2.9488,
+      "step": 120
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5073978994584547,
+      "learning_rate": 0.000998332357994035,
+      "loss": 2.9386,
+      "step": 121
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.7189694716461682,
+      "learning_rate": 0.0009982712004672786,
+      "loss": 3.025,
+      "step": 122
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5083664294299417,
+      "learning_rate": 0.0009982089435684295,
+      "loss": 3.007,
+      "step": 123
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.44319312440375835,
+      "learning_rate": 0.0009981455874348499,
+      "loss": 2.9104,
+      "step": 124
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5010655691737599,
+      "learning_rate": 0.0009980811322063269,
+      "loss": 2.9157,
+      "step": 125
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4771999357823196,
+      "learning_rate": 0.0009980155780250728,
+      "loss": 2.9217,
+      "step": 126
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5117236458582715,
+      "learning_rate": 0.0009979489250357243,
+      "loss": 2.9303,
+      "step": 127
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4994148572929066,
+      "learning_rate": 0.0009978811733853431,
+      "loss": 2.9632,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5029645548885143,
+      "learning_rate": 0.0009978123232234147,
+      "loss": 2.9382,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.545609502749393,
+      "learning_rate": 0.000997742374701848,
+      "loss": 2.8769,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4772794208725163,
+      "learning_rate": 0.0009976713279749754,
+      "loss": 2.9464,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5011121776492394,
+      "learning_rate": 0.0009975991831995528,
+      "loss": 2.8841,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9154266111716247,
+      "learning_rate": 0.0009975259405347581,
+      "loss": 2.9107,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4911729141091437,
+      "learning_rate": 0.0009974516001421926,
+      "loss": 2.8801,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5454904571359191,
+      "learning_rate": 0.000997376162185878,
+      "loss": 2.8968,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.48862686737804906,
+      "learning_rate": 0.0009972996268322594,
+      "loss": 2.9251,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4652130164684616,
+      "learning_rate": 0.0009972219942502017,
+      "loss": 2.9476,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4537931987119542,
+      "learning_rate": 0.0009971432646109918,
+      "loss": 2.8528,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.5624883365979697,
+      "learning_rate": 0.0009970634380883365,
+      "loss": 2.8629,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.4778290589587997,
+      "learning_rate": 0.0009969825148583627,
+      "loss": 2.8247,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.46125396055309287,
+      "learning_rate": 0.0009969004950996173,
+      "loss": 2.9263,
+      "step": 141
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4681442325909148,
+      "learning_rate": 0.0009968173789930668,
+      "loss": 2.9211,
+      "step": 142
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4695720439705318,
+      "learning_rate": 0.0009967331667220958,
+      "loss": 2.8432,
+      "step": 143
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5132202359159447,
+      "learning_rate": 0.0009966478584725086,
+      "loss": 2.9728,
+      "step": 144
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5299502010625903,
+      "learning_rate": 0.0009965614544325263,
+      "loss": 2.8344,
+      "step": 145
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5048985390999353,
+      "learning_rate": 0.000996473954792789,
+      "loss": 2.8667,
+      "step": 146
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.44591417738242695,
+      "learning_rate": 0.0009963853597463532,
+      "loss": 2.7384,
+      "step": 147
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.8862229376021098,
+      "learning_rate": 0.000996295669488693,
+      "loss": 2.9248,
+      "step": 148
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5031047286522878,
+      "learning_rate": 0.0009962048842176979,
+      "loss": 2.8393,
+      "step": 149
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.47649619595312775,
+      "learning_rate": 0.0009961130041336748,
+      "loss": 2.8376,
+      "step": 150
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.45917427797504334,
+      "learning_rate": 0.0009960200294393449,
+      "loss": 2.7629,
+      "step": 151
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.45285572406133606,
+      "learning_rate": 0.0009959259603398453,
+      "loss": 2.8133,
+      "step": 152
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4515794199510278,
+      "learning_rate": 0.0009958307970427275,
+      "loss": 2.8808,
+      "step": 153
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4761521184572242,
+      "learning_rate": 0.0009957345397579572,
+      "loss": 2.8041,
+      "step": 154
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4830477716573732,
+      "learning_rate": 0.0009956371886979138,
+      "loss": 2.8419,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5882272411843048,
+      "learning_rate": 0.00099553874407739,
+      "loss": 2.9059,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4955427713641379,
+      "learning_rate": 0.0009954392061135916,
+      "loss": 2.797,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.46264177487951824,
+      "learning_rate": 0.0009953385750261364,
+      "loss": 2.7265,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.48007938707590836,
+      "learning_rate": 0.0009952368510370538,
+      "loss": 2.7305,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5095607596132808,
+      "learning_rate": 0.0009951340343707852,
+      "loss": 2.7921,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5043654751203291,
+      "learning_rate": 0.0009950301252541823,
+      "loss": 2.6859,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.4757269791550311,
+      "learning_rate": 0.0009949251239165075,
+      "loss": 2.7209,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.5086296206252968,
+      "learning_rate": 0.000994819030589433,
+      "loss": 2.642,
+      "step": 163
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6552869555680885,
+      "learning_rate": 0.00099471184550704,
+      "loss": 2.7031,
+      "step": 164
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.48754563212176244,
+      "learning_rate": 0.0009946035689058189,
+      "loss": 2.6954,
+      "step": 165
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5312943049348269,
+      "learning_rate": 0.0009944942010246681,
+      "loss": 2.7596,
+      "step": 166
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6709397471263573,
+      "learning_rate": 0.0009943837421048942,
+      "loss": 2.8201,
+      "step": 167
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.09982639192658,
+      "learning_rate": 0.0009942721923902106,
+      "loss": 2.7781,
+      "step": 168
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.48033316354426764,
+      "learning_rate": 0.0009941595521267377,
+      "loss": 2.819,
+      "step": 169
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5168245134579614,
+      "learning_rate": 0.0009940458215630017,
+      "loss": 2.7196,
+      "step": 170
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5438756346494406,
+      "learning_rate": 0.0009939310009499348,
+      "loss": 2.7028,
+      "step": 171
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5626455616119523,
+      "learning_rate": 0.000993815090540874,
+      "loss": 2.77,
+      "step": 172
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5801948586932619,
+      "learning_rate": 0.000993698090591561,
+      "loss": 2.7414,
+      "step": 173
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6343455693041222,
+      "learning_rate": 0.0009935800013601416,
+      "loss": 2.6209,
+      "step": 174
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5519371614776889,
+      "learning_rate": 0.000993460823107164,
+      "loss": 2.7892,
+      "step": 175
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.564778322262144,
+      "learning_rate": 0.0009933405560955803,
+      "loss": 2.7551,
+      "step": 176
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5862059170928506,
+      "learning_rate": 0.0009932192005907446,
+      "loss": 2.67,
+      "step": 177
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5316621247831305,
+      "learning_rate": 0.0009930967568604118,
+      "loss": 2.7299,
+      "step": 178
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.6075680577129055,
+      "learning_rate": 0.000992973225174739,
+      "loss": 2.7183,
+      "step": 179
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.512930931443382,
+      "learning_rate": 0.0009928486058062827,
+      "loss": 2.7161,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5138501930879785,
+      "learning_rate": 0.0009927228990299999,
+      "loss": 2.6443,
+      "step": 181
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4806673997666021,
+      "learning_rate": 0.0009925961051232468,
+      "loss": 2.7193,
+      "step": 182
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4839084792939232,
+      "learning_rate": 0.000992468224365778,
+      "loss": 2.7739,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.5249802035536133,
+      "learning_rate": 0.000992339257039746,
+      "loss": 2.6544,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4891951270144479,
+      "learning_rate": 0.0009922092034297006,
+      "loss": 2.6677,
+      "step": 185
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5380931345968355,
+      "learning_rate": 0.0009920780638225891,
+      "loss": 2.564,
+      "step": 186
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4856722736469776,
+      "learning_rate": 0.0009919458385077538,
+      "loss": 2.6803,
+      "step": 187
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4747576171513781,
+      "learning_rate": 0.0009918125277769336,
+      "loss": 2.6244,
+      "step": 188
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5463826912406253,
+      "learning_rate": 0.0009916781319242614,
+      "loss": 2.7937,
+      "step": 189
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5171114325125519,
+      "learning_rate": 0.0009915426512462646,
+      "loss": 2.6246,
+      "step": 190
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4869297038404041,
+      "learning_rate": 0.0009914060860418644,
+      "loss": 2.5784,
+      "step": 191
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6143623965827758,
+      "learning_rate": 0.000991268436612374,
+      "loss": 2.5744,
+      "step": 192
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5329178503143228,
+      "learning_rate": 0.0009911297032614997,
+      "loss": 2.6378,
+      "step": 193
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5574396373177448,
+      "learning_rate": 0.000990989886295339,
+      "loss": 2.7499,
+      "step": 194
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.566033850826537,
+      "learning_rate": 0.0009908489860223804,
+      "loss": 2.5747,
+      "step": 195
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5137375245869844,
+      "learning_rate": 0.000990707002753502,
+      "loss": 2.616,
+      "step": 196
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5079090125381454,
+      "learning_rate": 0.0009905639368019724,
+      "loss": 2.5804,
+      "step": 197
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.46729352683852143,
+      "learning_rate": 0.0009904197884834482,
+      "loss": 2.5824,
+      "step": 198
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5720048557189943,
+      "learning_rate": 0.0009902745581159742,
+      "loss": 2.5292,
+      "step": 199
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5535533451584191,
+      "learning_rate": 0.0009901282460199829,
+      "loss": 2.6455,
+      "step": 200
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6816874029572049,
+      "learning_rate": 0.0009899808525182935,
+      "loss": 2.6556,
+      "step": 201
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5100421284025154,
+      "learning_rate": 0.0009898323779361107,
+      "loss": 2.5528,
+      "step": 202
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4716786027489898,
+      "learning_rate": 0.000989682822601025,
+      "loss": 2.5773,
+      "step": 203
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.4782298286120504,
+      "learning_rate": 0.0009895321868430113,
+      "loss": 2.4961,
+      "step": 204
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5383832702735599,
+      "learning_rate": 0.0009893804709944281,
+      "loss": 2.672,
+      "step": 205
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.495824943612125,
+      "learning_rate": 0.0009892276753900174,
+      "loss": 2.6244,
+      "step": 206
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.5418655288547687,
+      "learning_rate": 0.0009890738003669028,
+      "loss": 2.5523,
+      "step": 207
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5534971891190894,
+      "learning_rate": 0.0009889188462645904,
+      "loss": 2.6091,
+      "step": 208
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5480316959874822,
+      "learning_rate": 0.0009887628134249667,
+      "loss": 2.5028,
+      "step": 209
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5098710667818314,
+      "learning_rate": 0.0009886057021922983,
+      "loss": 2.6255,
+      "step": 210
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4598573341474209,
+      "learning_rate": 0.0009884475129132311,
+      "loss": 2.6378,
+      "step": 211
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4985996422687174,
+      "learning_rate": 0.0009882882459367897,
+      "loss": 2.5585,
+      "step": 212
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.47487442607952396,
+      "learning_rate": 0.0009881279016143766,
+      "loss": 2.6189,
+      "step": 213
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5043627386999875,
+      "learning_rate": 0.0009879664802997707,
+      "loss": 2.5582,
+      "step": 214
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4779630816566409,
+      "learning_rate": 0.000987803982349128,
+      "loss": 2.5625,
+      "step": 215
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4628645813495461,
+      "learning_rate": 0.0009876404081209796,
+      "loss": 2.5866,
+      "step": 216
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.45045036617084877,
+      "learning_rate": 0.000987475757976231,
+      "loss": 2.5456,
+      "step": 217
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.479211605658884,
+      "learning_rate": 0.000987310032278162,
+      "loss": 2.6742,
+      "step": 218
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4807181632489761,
+      "learning_rate": 0.0009871432313924254,
+      "loss": 2.6321,
+      "step": 219
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4595699374077853,
+      "learning_rate": 0.000986975355687046,
+      "loss": 2.6373,
+      "step": 220
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5566387212839229,
+      "learning_rate": 0.0009868064055324204,
+      "loss": 2.5691,
+      "step": 221
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4983070240270771,
+      "learning_rate": 0.0009866363813013153,
+      "loss": 2.5535,
+      "step": 222
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.591135114519752,
+      "learning_rate": 0.0009864652833688676,
+      "loss": 2.6183,
+      "step": 223
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.47010795320698034,
+      "learning_rate": 0.0009862931121125836,
+      "loss": 2.6267,
+      "step": 224
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.48542563896669305,
+      "learning_rate": 0.000986119867912337,
+      "loss": 2.5275,
+      "step": 225
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.47213970851796344,
+      "learning_rate": 0.000985945551150369,
+      "loss": 2.4792,
+      "step": 226
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4960767656376482,
+      "learning_rate": 0.0009857701622112876,
+      "loss": 2.6612,
+      "step": 227
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.5357011412122866,
+      "learning_rate": 0.000985593701482066,
+      "loss": 2.5663,
+      "step": 228
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.4399546755826771,
+      "learning_rate": 0.0009854161693520424,
+      "loss": 2.6056,
+      "step": 229
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.428684795379488,
+      "learning_rate": 0.0009852375662129194,
+      "loss": 2.5225,
+      "step": 230
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.512046521899137,
+      "learning_rate": 0.0009850578924587613,
+      "loss": 2.6418,
+      "step": 231
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.43121598062136435,
+      "learning_rate": 0.000984877148485996,
+      "loss": 2.596,
+      "step": 232
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4647466011176806,
+      "learning_rate": 0.000984695334693412,
+      "loss": 2.5266,
+      "step": 233
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4502032122840246,
+      "learning_rate": 0.000984512451482158,
+      "loss": 2.6136,
+      "step": 234
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.43155908092316847,
+      "learning_rate": 0.0009843284992557431,
+      "loss": 2.5082,
+      "step": 235
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4598637047178943,
+      "learning_rate": 0.000984143478420034,
+      "loss": 2.5224,
+      "step": 236
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.43698576405101475,
+      "learning_rate": 0.0009839573893832563,
+      "loss": 2.5614,
+      "step": 237
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.44892845454207997,
+      "learning_rate": 0.000983770232555991,
+      "loss": 2.5751,
+      "step": 238
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5047291286215303,
+      "learning_rate": 0.0009835820083511765,
+      "loss": 2.5489,
+      "step": 239
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.44448859296538046,
+      "learning_rate": 0.0009833927171841055,
+      "loss": 2.5151,
+      "step": 240
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.6335943426590055,
+      "learning_rate": 0.0009832023594724246,
+      "loss": 2.4577,
+      "step": 241
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4579277488138868,
+      "learning_rate": 0.0009830109356361344,
+      "loss": 2.5578,
+      "step": 242
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.45830081362711034,
+      "learning_rate": 0.0009828184460975867,
+      "loss": 2.5511,
+      "step": 243
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5094790648416039,
+      "learning_rate": 0.0009826248912814855,
+      "loss": 2.5876,
+      "step": 244
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5239269433173943,
+      "learning_rate": 0.0009824302716148847,
+      "loss": 2.6402,
+      "step": 245
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4859166667801006,
+      "learning_rate": 0.0009822345875271884,
+      "loss": 2.533,
+      "step": 246
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.4908466401467397,
+      "learning_rate": 0.0009820378394501481,
+      "loss": 2.6216,
+      "step": 247
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.5991081019060942,
+      "learning_rate": 0.0009818400278178636,
+      "loss": 2.5603,
+      "step": 248
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.49653519371975274,
+      "learning_rate": 0.0009816411530667814,
+      "loss": 2.5832,
+      "step": 249
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.42957958742283164,
+      "learning_rate": 0.000981441215635693,
+      "loss": 2.5134,
+      "step": 250
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.46497467748438576,
+      "learning_rate": 0.0009812402159657353,
+      "loss": 2.5156,
+      "step": 251
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4369822051102667,
+      "learning_rate": 0.000981038154500388,
+      "loss": 2.5065,
+      "step": 252
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3960850487336733,
+      "learning_rate": 0.0009808350316854746,
+      "loss": 2.4837,
+      "step": 253
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4469145922319819,
+      "learning_rate": 0.0009806308479691594,
+      "loss": 2.474,
+      "step": 254
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.450372730744465,
+      "learning_rate": 0.0009804256038019482,
+      "loss": 2.5954,
+      "step": 255
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.44540995347209894,
+      "learning_rate": 0.0009802192996366857,
+      "loss": 2.5342,
+      "step": 256
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4979638288973265,
+      "learning_rate": 0.0009800119359285563,
+      "loss": 2.5545,
+      "step": 257
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4438621276233336,
+      "learning_rate": 0.0009798035131350813,
+      "loss": 2.5172,
+      "step": 258
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4042461846531273,
+      "learning_rate": 0.0009795940317161194,
+      "loss": 2.5585,
+      "step": 259
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4659780856270375,
+      "learning_rate": 0.0009793834921338646,
+      "loss": 2.4591,
+      "step": 260
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.43238661655801186,
+      "learning_rate": 0.0009791718948528457,
+      "loss": 2.6084,
+      "step": 261
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4263141179617928,
+      "learning_rate": 0.0009789592403399252,
+      "loss": 2.5207,
+      "step": 262
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.43442190706196754,
+      "learning_rate": 0.0009787455290642985,
+      "loss": 2.5362,
+      "step": 263
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4262249218788271,
+      "learning_rate": 0.000978530761497492,
+      "loss": 2.6311,
+      "step": 264
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4778569536442741,
+      "learning_rate": 0.0009783149381133633,
+      "loss": 2.569,
+      "step": 265
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.3963233001681935,
+      "learning_rate": 0.0009780980593880992,
+      "loss": 2.5313,
+      "step": 266
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4138838651559631,
+      "learning_rate": 0.0009778801258002153,
+      "loss": 2.523,
+      "step": 267
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4082932305466624,
+      "learning_rate": 0.000977661137830554,
+      "loss": 2.5619,
+      "step": 268
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4392646819081226,
+      "learning_rate": 0.0009774410959622845,
+      "loss": 2.5031,
+      "step": 269
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.44079267850265896,
+      "learning_rate": 0.000977220000680901,
+      "loss": 2.4371,
+      "step": 270
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4335218680367955,
+      "learning_rate": 0.000976997852474223,
+      "loss": 2.4306,
+      "step": 271
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.4271982517354844,
+      "learning_rate": 0.0009767746518323914,
+      "loss": 2.5581,
+      "step": 272
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4693404428741995,
+      "learning_rate": 0.0009765503992478704,
+      "loss": 2.5258,
+      "step": 273
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.440890627286386,
+      "learning_rate": 0.0009763250952154449,
+      "loss": 2.5548,
+      "step": 274
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4000703361675459,
+      "learning_rate": 0.0009760987402322195,
+      "loss": 2.4684,
+      "step": 275
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4323153416070329,
+      "learning_rate": 0.0009758713347976178,
+      "loss": 2.5138,
+      "step": 276
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4543137186131346,
+      "learning_rate": 0.000975642879413381,
+      "loss": 2.5818,
+      "step": 277
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.37826312090719155,
+      "learning_rate": 0.0009754133745835665,
+      "loss": 2.5383,
+      "step": 278
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.42330612621223773,
+      "learning_rate": 0.0009751828208145482,
+      "loss": 2.4972,
+      "step": 279
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.43626241975813124,
+      "learning_rate": 0.0009749512186150131,
+      "loss": 2.5767,
+      "step": 280
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4074270941186926,
+      "learning_rate": 0.0009747185684959625,
+      "loss": 2.5278,
+      "step": 281
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4595732429494049,
+      "learning_rate": 0.000974484870970709,
+      "loss": 2.4925,
+      "step": 282
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.45935072710195923,
+      "learning_rate": 0.0009742501265548767,
+      "loss": 2.4955,
+      "step": 283
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4665117009986919,
+      "learning_rate": 0.0009740143357663993,
+      "loss": 2.4795,
+      "step": 284
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4310766195303373,
+      "learning_rate": 0.000973777499125519,
+      "loss": 2.4924,
+      "step": 285
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4550677833050526,
+      "learning_rate": 0.0009735396171547859,
+      "loss": 2.5221,
+      "step": 286
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.42223238083991366,
+      "learning_rate": 0.0009733006903790564,
+      "loss": 2.5664,
+      "step": 287
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4842468838550934,
+      "learning_rate": 0.0009730607193254922,
+      "loss": 2.5791,
+      "step": 288
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.420635820589976,
+      "learning_rate": 0.0009728197045235585,
+      "loss": 2.4612,
+      "step": 289
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.429425037867332,
+      "learning_rate": 0.0009725776465050242,
+      "loss": 2.5166,
+      "step": 290
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4376358690471994,
+      "learning_rate": 0.0009723345458039594,
+      "loss": 2.469,
+      "step": 291
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4745961706839911,
+      "learning_rate": 0.000972090402956735,
+      "loss": 2.4743,
+      "step": 292
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.4616821031871291,
+      "learning_rate": 0.0009718452185020212,
+      "loss": 2.4828,
+      "step": 293
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.43304398692556784,
+      "learning_rate": 0.0009715989929807862,
+      "loss": 2.4767,
+      "step": 294
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.40114111485319986,
+      "learning_rate": 0.0009713517269362955,
+      "loss": 2.4736,
+      "step": 295
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.44599477680229743,
+      "learning_rate": 0.0009711034209141101,
+      "loss": 2.4895,
+      "step": 296
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.44914302665883704,
+      "learning_rate": 0.0009708540754620856,
+      "loss": 2.5459,
+      "step": 297
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.3882369373695868,
+      "learning_rate": 0.0009706036911303713,
+      "loss": 2.556,
+      "step": 298
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4139892342315258,
+      "learning_rate": 0.0009703522684714083,
+      "loss": 2.5325,
+      "step": 299
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.41793581190723866,
+      "learning_rate": 0.0009700998080399286,
+      "loss": 2.4785,
+      "step": 300
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4218496936488842,
+      "learning_rate": 0.0009698463103929542,
+      "loss": 2.451,
+      "step": 301
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.435748727413112,
+      "learning_rate": 0.0009695917760897954,
+      "loss": 2.5152,
+      "step": 302
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.41465200521942236,
+      "learning_rate": 0.0009693362056920501,
+      "loss": 2.4114,
+      "step": 303
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4518930463503598,
+      "learning_rate": 0.0009690795997636015,
+      "loss": 2.5383,
+      "step": 304
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.44290603318250665,
+      "learning_rate": 0.0009688219588706179,
+      "loss": 2.532,
+      "step": 305
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4265988466451303,
+      "learning_rate": 0.0009685632835815518,
+      "loss": 2.4607,
+      "step": 306
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.40717373847343824,
+      "learning_rate": 0.0009683035744671367,
+      "loss": 2.5388,
+      "step": 307
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.39756406153161367,
+      "learning_rate": 0.0009680428321003883,
+      "loss": 2.5197,
+      "step": 308
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.42076073752375714,
+      "learning_rate": 0.000967781057056601,
+      "loss": 2.4599,
+      "step": 309
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.3704642198020802,
+      "learning_rate": 0.0009675182499133485,
+      "loss": 2.5019,
+      "step": 310
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.39353666994999453,
+      "learning_rate": 0.0009672544112504813,
+      "loss": 2.4515,
+      "step": 311
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4138083846729717,
+      "learning_rate": 0.0009669895416501257,
+      "loss": 2.4661,
+      "step": 312
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.42287009921637847,
+      "learning_rate": 0.0009667236416966833,
+      "loss": 2.5224,
+      "step": 313
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.39188970602011836,
+      "learning_rate": 0.0009664567119768281,
+      "loss": 2.5011,
+      "step": 314
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.4304014281205347,
+      "learning_rate": 0.0009661887530795067,
+      "loss": 2.3857,
+      "step": 315
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.40445689125088674,
+      "learning_rate": 0.0009659197655959365,
+      "loss": 2.3541,
+      "step": 316
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.40035320771634453,
+      "learning_rate": 0.000965649750119604,
+      "loss": 2.4789,
+      "step": 317
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.5074461008357748,
+      "learning_rate": 0.0009653787072462643,
+      "loss": 2.5385,
+      "step": 318
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3917565080154814,
+      "learning_rate": 0.0009651066375739388,
+      "loss": 2.5229,
+      "step": 319
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3953772507086133,
+      "learning_rate": 0.000964833541702915,
+      "loss": 2.4534,
+      "step": 320
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.40026214635399876,
+      "learning_rate": 0.0009645594202357438,
+      "loss": 2.4399,
+      "step": 321
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.413534991044047,
+      "learning_rate": 0.0009642842737772397,
+      "loss": 2.4556,
+      "step": 322
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4337864642517213,
+      "learning_rate": 0.0009640081029344782,
+      "loss": 2.4791,
+      "step": 323
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.39887086846374,
+      "learning_rate": 0.0009637309083167956,
+      "loss": 2.4164,
+      "step": 324
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4490961789815051,
+      "learning_rate": 0.0009634526905357859,
+      "loss": 2.5727,
+      "step": 325
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3957139788119504,
+      "learning_rate": 0.000963173450205302,
+      "loss": 2.506,
+      "step": 326
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3757232367720781,
+      "learning_rate": 0.0009628931879414517,
+      "loss": 2.4675,
+      "step": 327
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4274985034122862,
+      "learning_rate": 0.0009626119043625983,
+      "loss": 2.5474,
+      "step": 328
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.42406549018941425,
+      "learning_rate": 0.0009623296000893582,
+      "loss": 2.5006,
+      "step": 329
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.4264984428099638,
+      "learning_rate": 0.0009620462757446,
+      "loss": 2.5068,
+      "step": 330
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.41090291232826676,
+      "learning_rate": 0.0009617619319534428,
+      "loss": 2.5488,
+      "step": 331
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.44629649303482105,
+      "learning_rate": 0.000961476569343255,
+      "loss": 2.5564,
+      "step": 332
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3909006132116697,
+      "learning_rate": 0.0009611901885436529,
+      "loss": 2.5363,
+      "step": 333
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.40050420853997754,
+      "learning_rate": 0.0009609027901864996,
+      "loss": 2.4824,
+      "step": 334
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.383345429947794,
+      "learning_rate": 0.0009606143749059029,
+      "loss": 2.4503,
+      "step": 335
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.3767383795867858,
+      "learning_rate": 0.0009603249433382144,
+      "loss": 2.4386,
+      "step": 336
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.38631124579409454,
+      "learning_rate": 0.0009600344961220282,
+      "loss": 2.4772,
+      "step": 337
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.39913350077567683,
+      "learning_rate": 0.0009597430338981791,
+      "loss": 2.4655,
+      "step": 338
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.42403804676073625,
+      "learning_rate": 0.0009594505573097414,
+      "loss": 2.4149,
+      "step": 339
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.45403286969357937,
+      "learning_rate": 0.0009591570670020277,
+      "loss": 2.5795,
+      "step": 340
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.38146057907394515,
+      "learning_rate": 0.0009588625636225871,
+      "loss": 2.4377,
+      "step": 341
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3723890096724888,
+      "learning_rate": 0.0009585670478212036,
+      "loss": 2.4936,
+      "step": 342
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.44628764922022546,
+      "learning_rate": 0.0009582705202498956,
+      "loss": 2.4831,
+      "step": 343
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3999699475114257,
+      "learning_rate": 0.0009579729815629133,
+      "loss": 2.3923,
+      "step": 344
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.40929758806601163,
+      "learning_rate": 0.0009576744324167379,
+      "loss": 2.5499,
+      "step": 345
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4126432789490919,
+      "learning_rate": 0.0009573748734700804,
+      "loss": 2.4937,
+      "step": 346
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.37981234145315096,
+      "learning_rate": 0.0009570743053838796,
+      "loss": 2.5255,
+      "step": 347
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.39381738472335304,
+      "learning_rate": 0.0009567727288213005,
+      "loss": 2.461,
+      "step": 348
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.35702645494347907,
+      "learning_rate": 0.0009564701444477337,
+      "loss": 2.49,
+      "step": 349
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3732060363957987,
+      "learning_rate": 0.000956166552930793,
+      "loss": 2.4724,
+      "step": 350
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3590401720749476,
+      "learning_rate": 0.0009558619549403147,
+      "loss": 2.4598,
+      "step": 351
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3702945929015598,
+      "learning_rate": 0.0009555563511483555,
+      "loss": 2.4771,
+      "step": 352
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3989929984523915,
+      "learning_rate": 0.0009552497422291912,
+      "loss": 2.4804,
+      "step": 353
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.43731791510508733,
+      "learning_rate": 0.0009549421288593157,
+      "loss": 2.4685,
+      "step": 354
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.40160660892550687,
+      "learning_rate": 0.0009546335117174385,
+      "loss": 2.5043,
+      "step": 355
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.4105510743325178,
+      "learning_rate": 0.0009543238914844843,
+      "loss": 2.4286,
+      "step": 356
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3970429455002816,
+      "learning_rate": 0.0009540132688435907,
+      "loss": 2.378,
+      "step": 357
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3736423596901369,
+      "learning_rate": 0.0009537016444801074,
+      "loss": 2.4888,
+      "step": 358
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.37036009882386794,
+      "learning_rate": 0.0009533890190815935,
+      "loss": 2.4535,
+      "step": 359
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.36924428427981987,
+      "learning_rate": 0.0009530753933378173,
+      "loss": 2.4676,
+      "step": 360
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.40899627722749016,
+      "learning_rate": 0.0009527607679407545,
+      "loss": 2.4341,
+      "step": 361
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.38988566728195095,
+      "learning_rate": 0.0009524451435845857,
+      "loss": 2.4528,
+      "step": 362
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.40378485418706783,
+      "learning_rate": 0.0009521285209656963,
+      "loss": 2.4262,
+      "step": 363
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3804301531357516,
+      "learning_rate": 0.0009518109007826734,
+      "loss": 2.4213,
+      "step": 364
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.41690663138802536,
+      "learning_rate": 0.0009514922837363059,
+      "loss": 2.4674,
+      "step": 365
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.39661828740777816,
+      "learning_rate": 0.0009511726705295817,
+      "loss": 2.4503,
+      "step": 366
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.39010926936539314,
+      "learning_rate": 0.000950852061867687,
+      "loss": 2.4992,
+      "step": 367
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3555581902822759,
+      "learning_rate": 0.0009505304584580038,
+      "loss": 2.4204,
+      "step": 368
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.38786986685644853,
+      "learning_rate": 0.0009502078610101092,
+      "loss": 2.5043,
+      "step": 369
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3852345949107587,
+      "learning_rate": 0.0009498842702357736,
+      "loss": 2.4741,
+      "step": 370
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.35333671362272506,
+      "learning_rate": 0.0009495596868489587,
+      "loss": 2.3958,
+      "step": 371
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.35522129536651414,
+      "learning_rate": 0.0009492341115658167,
+      "loss": 2.379,
+      "step": 372
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3919774687493462,
+      "learning_rate": 0.0009489075451046879,
+      "loss": 2.4096,
+      "step": 373
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.3702478212433338,
+      "learning_rate": 0.0009485799881861,
+      "loss": 2.4445,
+      "step": 374
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.37160839327436673,
+      "learning_rate": 0.0009482514415327654,
+      "loss": 2.5704,
+      "step": 375
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4148339272074171,
+      "learning_rate": 0.000947921905869581,
+      "loss": 2.3951,
+      "step": 376
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.37493215781760914,
+      "learning_rate": 0.0009475913819236248,
+      "loss": 2.4075,
+      "step": 377
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.37162368697523046,
+      "learning_rate": 0.0009472598704241561,
+      "loss": 2.4789,
+      "step": 378
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.37295764692478217,
+      "learning_rate": 0.0009469273721026131,
+      "loss": 2.4833,
+      "step": 379
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.4183007953458358,
+      "learning_rate": 0.0009465938876926111,
+      "loss": 2.4732,
+      "step": 380
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.40086190626187934,
+      "learning_rate": 0.0009462594179299406,
+      "loss": 2.3687,
+      "step": 381
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3737583805545021,
+      "learning_rate": 0.0009459239635525672,
+      "loss": 2.4618,
+      "step": 382
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.37378842405897056,
+      "learning_rate": 0.0009455875253006281,
+      "loss": 2.527,
+      "step": 383
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.395349392777238,
+      "learning_rate": 0.0009452501039164315,
+      "loss": 2.4851,
+      "step": 384
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4291573285690305,
+      "learning_rate": 0.0009449117001444549,
+      "loss": 2.5013,
+      "step": 385
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.38601270620147526,
+      "learning_rate": 0.0009445723147313433,
+      "loss": 2.4164,
+      "step": 386
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.39550024296080677,
+      "learning_rate": 0.0009442319484259074,
+      "loss": 2.3229,
+      "step": 387
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.0595266430145296,
+      "learning_rate": 0.0009438906019791222,
+      "loss": 2.4211,
+      "step": 388
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.35781846203476025,
+      "learning_rate": 0.0009435482761441251,
+      "loss": 2.5453,
+      "step": 389
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.41269605779093066,
+      "learning_rate": 0.000943204971676215,
+      "loss": 2.452,
+      "step": 390
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3839530292652149,
+      "learning_rate": 0.0009428606893328493,
+      "loss": 2.4535,
+      "step": 391
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.39803329204809473,
+      "learning_rate": 0.0009425154298736432,
+      "loss": 2.5313,
+      "step": 392
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.398120437395968,
+      "learning_rate": 0.0009421691940603678,
+      "loss": 2.5213,
+      "step": 393
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.37227021719704056,
+      "learning_rate": 0.0009418219826569488,
+      "loss": 2.4089,
+      "step": 394
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3982623907414935,
+      "learning_rate": 0.0009414737964294635,
+      "loss": 2.4893,
+      "step": 395
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.41376737254201473,
+      "learning_rate": 0.000941124636146141,
+      "loss": 2.4997,
+      "step": 396
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.42403723999354304,
+      "learning_rate": 0.0009407745025773589,
+      "loss": 2.3783,
+      "step": 397
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.39031523364174403,
+      "learning_rate": 0.0009404233964956423,
+      "loss": 2.4885,
+      "step": 398
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.4015455550914117,
+      "learning_rate": 0.0009400713186756625,
+      "loss": 2.4155,
+      "step": 399
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.40418679847618094,
+      "learning_rate": 0.0009397182698942342,
+      "loss": 2.526,
+      "step": 400
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.40336200558810364,
+      "learning_rate": 0.0009393642509303149,
+      "loss": 2.5708,
+      "step": 401
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3843633370130936,
+      "learning_rate": 0.0009390092625650023,
+      "loss": 2.4275,
+      "step": 402
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.3611794280457007,
+      "learning_rate": 0.0009386533055815332,
+      "loss": 2.4524,
+      "step": 403
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.37461900940858506,
+      "learning_rate": 0.0009382963807652813,
+      "loss": 2.3929,
+      "step": 404
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.35982109104010745,
+      "learning_rate": 0.000937938488903756,
+      "loss": 2.5144,
+      "step": 405
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.35181522738221177,
+      "learning_rate": 0.0009375796307866003,
+      "loss": 2.4397,
+      "step": 406
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.36778157223315916,
+      "learning_rate": 0.0009372198072055888,
+      "loss": 2.3871,
+      "step": 407
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.39367811857282714,
+      "learning_rate": 0.0009368590189546268,
+      "loss": 2.4778,
+      "step": 408
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3816213321426315,
+      "learning_rate": 0.0009364972668297474,
+      "loss": 2.5021,
+      "step": 409
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.37662979370687905,
+      "learning_rate": 0.0009361345516291111,
+      "loss": 2.4222,
+      "step": 410
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.39077824133550926,
+      "learning_rate": 0.0009357708741530024,
+      "loss": 2.4532,
+      "step": 411
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.37255281706535454,
+      "learning_rate": 0.00093540623520383,
+      "loss": 2.3809,
+      "step": 412
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.40116058837187796,
+      "learning_rate": 0.000935040635586123,
+      "loss": 2.4712,
+      "step": 413
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4043710796640211,
+      "learning_rate": 0.0009346740761065305,
+      "loss": 2.4552,
+      "step": 414
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.38398453292127427,
+      "learning_rate": 0.0009343065575738197,
+      "loss": 2.4131,
+      "step": 415
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.37162449542045184,
+      "learning_rate": 0.0009339380807988733,
+      "loss": 2.4611,
+      "step": 416
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.36257010817955365,
+      "learning_rate": 0.0009335686465946887,
+      "loss": 2.5856,
+      "step": 417
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.393118365931577,
+      "learning_rate": 0.0009331982557763754,
+      "loss": 2.3109,
+      "step": 418
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.37885382850274146,
+      "learning_rate": 0.0009328269091611537,
+      "loss": 2.4782,
+      "step": 419
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3708941860896317,
+      "learning_rate": 0.0009324546075683524,
+      "loss": 2.4532,
+      "step": 420
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.371741708903257,
+      "learning_rate": 0.0009320813518194083,
+      "loss": 2.3083,
+      "step": 421
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.3630797929574602,
+      "learning_rate": 0.0009317071427378624,
+      "loss": 2.3538,
+      "step": 422
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.39484973730709616,
+      "learning_rate": 0.0009313319811493594,
+      "loss": 2.4897,
+      "step": 423
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.38674602078654036,
+      "learning_rate": 0.000930955867881646,
+      "loss": 2.3452,
+      "step": 424
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.36289480662525003,
+      "learning_rate": 0.0009305788037645681,
+      "loss": 2.4871,
+      "step": 425
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3471586260464887,
+      "learning_rate": 0.0009302007896300697,
+      "loss": 2.4294,
+      "step": 426
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.36464382107275317,
+      "learning_rate": 0.0009298218263121911,
+      "loss": 2.4322,
+      "step": 427
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3638375565738279,
+      "learning_rate": 0.0009294419146470668,
+      "loss": 2.4997,
+      "step": 428
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3720183545504425,
+      "learning_rate": 0.0009290610554729234,
+      "loss": 2.4538,
+      "step": 429
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3483936797536827,
+      "learning_rate": 0.0009286792496300784,
+      "loss": 2.5349,
+      "step": 430
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3719695856179318,
+      "learning_rate": 0.0009282964979609379,
+      "loss": 2.5131,
+      "step": 431
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.34107138075952104,
+      "learning_rate": 0.0009279128013099947,
+      "loss": 2.441,
+      "step": 432
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3627031361823046,
+      "learning_rate": 0.0009275281605238268,
+      "loss": 2.4696,
+      "step": 433
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3770204945143011,
+      "learning_rate": 0.0009271425764510953,
+      "loss": 2.4682,
+      "step": 434
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.371710209274811,
+      "learning_rate": 0.0009267560499425423,
+      "loss": 2.4238,
+      "step": 435
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.38718492079486605,
+      "learning_rate": 0.0009263685818509895,
+      "loss": 2.4143,
+      "step": 436
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3804241769771776,
+      "learning_rate": 0.000925980173031336,
+      "loss": 2.4128,
+      "step": 437
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.40123863506021573,
+      "learning_rate": 0.0009255908243405567,
+      "loss": 2.3946,
+      "step": 438
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3905109063689614,
+      "learning_rate": 0.0009252005366376996,
+      "loss": 2.4616,
+      "step": 439
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.37538428093762,
+      "learning_rate": 0.0009248093107838852,
+      "loss": 2.3516,
+      "step": 440
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3428217324221212,
+      "learning_rate": 0.0009244171476423036,
+      "loss": 2.488,
+      "step": 441
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.338435205777112,
+      "learning_rate": 0.0009240240480782129,
+      "loss": 2.4207,
+      "step": 442
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.37021262728458976,
+      "learning_rate": 0.0009236300129589376,
+      "loss": 2.4742,
+      "step": 443
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.35647748336027935,
+      "learning_rate": 0.0009232350431538657,
+      "loss": 2.4201,
+      "step": 444
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.36108327861613226,
+      "learning_rate": 0.0009228391395344482,
+      "loss": 2.4396,
+      "step": 445
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.371086595622876,
+      "learning_rate": 0.000922442302974196,
+      "loss": 2.4271,
+      "step": 446
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3779338040723432,
+      "learning_rate": 0.0009220445343486785,
+      "loss": 2.4329,
+      "step": 447
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3592683043603127,
+      "learning_rate": 0.0009216458345355217,
+      "loss": 2.4307,
+      "step": 448
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3668290141678517,
+      "learning_rate": 0.0009212462044144061,
+      "loss": 2.3321,
+      "step": 449
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.398419273567059,
+      "learning_rate": 0.0009208456448670648,
+      "loss": 2.348,
+      "step": 450
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4065285050342439,
+      "learning_rate": 0.0009204441567772816,
+      "loss": 2.5216,
+      "step": 451
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.358905100961114,
+      "learning_rate": 0.0009200417410308888,
+      "loss": 2.473,
+      "step": 452
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3993688658837592,
+      "learning_rate": 0.0009196383985157656,
+      "loss": 2.4082,
+      "step": 453
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4872566641767802,
+      "learning_rate": 0.000919234130121836,
+      "loss": 2.4546,
+      "step": 454
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3754008408636971,
+      "learning_rate": 0.0009188289367410672,
+      "loss": 2.3386,
+      "step": 455
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.35425824630429414,
+      "learning_rate": 0.0009184228192674666,
+      "loss": 2.341,
+      "step": 456
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3912511904805784,
+      "learning_rate": 0.0009180157785970808,
+      "loss": 2.4504,
+      "step": 457
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3915019499297843,
+      "learning_rate": 0.0009176078156279932,
+      "loss": 2.4296,
+      "step": 458
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3703612448420685,
+      "learning_rate": 0.0009171989312603226,
+      "loss": 2.4694,
+      "step": 459
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.39710403013618367,
+      "learning_rate": 0.0009167891263962202,
+      "loss": 2.3571,
+      "step": 460
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.37388039904815223,
+      "learning_rate": 0.0009163784019398685,
+      "loss": 2.5297,
+      "step": 461
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.359653463572671,
+      "learning_rate": 0.0009159667587974785,
+      "loss": 2.3281,
+      "step": 462
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3699210293678319,
+      "learning_rate": 0.0009155541978772887,
+      "loss": 2.3415,
+      "step": 463
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3736984231998537,
+      "learning_rate": 0.0009151407200895625,
+      "loss": 2.4206,
+      "step": 464
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.38291063040685963,
+      "learning_rate": 0.000914726326346586,
+      "loss": 2.4499,
+      "step": 465
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.41550495962459777,
+      "learning_rate": 0.0009143110175626661,
+      "loss": 2.4431,
+      "step": 466
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.407403442326318,
+      "learning_rate": 0.0009138947946541291,
+      "loss": 2.3992,
+      "step": 467
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.3479829093712718,
+      "learning_rate": 0.0009134776585393181,
+      "loss": 2.3348,
+      "step": 468
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3457875951476438,
+      "learning_rate": 0.0009130596101385906,
+      "loss": 2.4944,
+      "step": 469
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.35541066899432283,
+      "learning_rate": 0.0009126406503743174,
+      "loss": 2.4605,
+      "step": 470
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3620956790728183,
+      "learning_rate": 0.0009122207801708802,
+      "loss": 2.3659,
+      "step": 471
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3744101674715765,
+      "learning_rate": 0.0009118000004546689,
+      "loss": 2.3552,
+      "step": 472
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3666043852022924,
+      "learning_rate": 0.0009113783121540807,
+      "loss": 2.4108,
+      "step": 473
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.39494267784347975,
+      "learning_rate": 0.0009109557161995172,
+      "loss": 2.5126,
+      "step": 474
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3887336017605293,
+      "learning_rate": 0.0009105322135233828,
+      "loss": 2.3928,
+      "step": 475
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3703710693094936,
+      "learning_rate": 0.0009101078050600821,
+      "loss": 2.4511,
+      "step": 476
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3500656290560058,
+      "learning_rate": 0.0009096824917460186,
+      "loss": 2.379,
+      "step": 477
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3744219545856563,
+      "learning_rate": 0.0009092562745195921,
+      "loss": 2.4634,
+      "step": 478
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.39939356591873393,
+      "learning_rate": 0.0009088291543211967,
+      "loss": 2.3885,
+      "step": 479
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3674026331187117,
+      "learning_rate": 0.0009084011320932188,
+      "loss": 2.3663,
+      "step": 480
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3480800658722959,
+      "learning_rate": 0.0009079722087800352,
+      "loss": 2.3244,
+      "step": 481
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4034091981602653,
+      "learning_rate": 0.0009075423853280106,
+      "loss": 2.388,
+      "step": 482
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3834668804519419,
+      "learning_rate": 0.0009071116626854958,
+      "loss": 2.4861,
+      "step": 483
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3670918791725497,
+      "learning_rate": 0.0009066800418028256,
+      "loss": 2.3542,
+      "step": 484
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.40835852835745506,
+      "learning_rate": 0.0009062475236323168,
+      "loss": 2.3963,
+      "step": 485
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.4096985830446128,
+      "learning_rate": 0.0009058141091282656,
+      "loss": 2.3552,
+      "step": 486
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3637494133682254,
+      "learning_rate": 0.0009053797992469461,
+      "loss": 2.5588,
+      "step": 487
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.3348169785948692,
+      "learning_rate": 0.0009049445949466078,
+      "loss": 2.4276,
+      "step": 488
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.33841783576946444,
+      "learning_rate": 0.0009045084971874737,
+      "loss": 2.3167,
+      "step": 489
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.35215273651683593,
+      "learning_rate": 0.0009040715069317382,
+      "loss": 2.349,
+      "step": 490
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3595873749146284,
+      "learning_rate": 0.0009036336251435648,
+      "loss": 2.4133,
+      "step": 491
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.387608622313695,
+      "learning_rate": 0.0009031948527890839,
+      "loss": 2.4718,
+      "step": 492
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.35556916625928325,
+      "learning_rate": 0.000902755190836391,
+      "loss": 2.3513,
+      "step": 493
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3887806403960189,
+      "learning_rate": 0.0009023146402555442,
+      "loss": 2.5419,
+      "step": 494
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3586130573705468,
+      "learning_rate": 0.0009018732020185624,
+      "loss": 2.414,
+      "step": 495
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.35828365665443174,
+      "learning_rate": 0.0009014308770994235,
+      "loss": 2.3505,
+      "step": 496
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3495220818310776,
+      "learning_rate": 0.0009009876664740605,
+      "loss": 2.395,
+      "step": 497
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.37042387113230457,
+      "learning_rate": 0.0009005435711203618,
+      "loss": 2.41,
+      "step": 498
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3546377917996666,
+      "learning_rate": 0.000900098592018167,
+      "loss": 2.366,
+      "step": 499
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3471797227925989,
+      "learning_rate": 0.0008996527301492663,
+      "loss": 2.452,
+      "step": 500
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3544082574006691,
+      "learning_rate": 0.0008992059864973972,
+      "loss": 2.3725,
+      "step": 501
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.37585899774981824,
+      "learning_rate": 0.0008987583620482427,
+      "loss": 2.3835,
+      "step": 502
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3447524996281995,
+      "learning_rate": 0.0008983098577894292,
+      "loss": 2.408,
+      "step": 503
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3548187393194642,
+      "learning_rate": 0.0008978604747105246,
+      "loss": 2.3487,
+      "step": 504
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3489008241188146,
+      "learning_rate": 0.0008974102138030354,
+      "loss": 2.416,
+      "step": 505
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.4439385786214722,
+      "learning_rate": 0.000896959076060405,
+      "loss": 2.4348,
+      "step": 506
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.35910840495268803,
+      "learning_rate": 0.0008965070624780116,
+      "loss": 2.3391,
+      "step": 507
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.354197649028919,
+      "learning_rate": 0.0008960541740531658,
+      "loss": 2.4962,
+      "step": 508
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.3459427695717891,
+      "learning_rate": 0.0008956004117851083,
+      "loss": 2.4213,
+      "step": 509
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.35648417565708607,
+      "learning_rate": 0.0008951457766750079,
+      "loss": 2.4099,
+      "step": 510
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.37800314654655265,
+      "learning_rate": 0.0008946902697259593,
+      "loss": 2.3935,
+      "step": 511
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.35011902913053244,
+      "learning_rate": 0.0008942338919429805,
+      "loss": 2.4111,
+      "step": 512
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4066525648057426,
+      "learning_rate": 0.0008937766443330113,
+      "loss": 2.376,
+      "step": 513
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3761940416479475,
+      "learning_rate": 0.0008933185279049103,
+      "loss": 2.434,
+      "step": 514
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3552103992655528,
+      "learning_rate": 0.0008928595436694532,
+      "loss": 2.4583,
+      "step": 515
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.37723465624551467,
+      "learning_rate": 0.0008923996926393305,
+      "loss": 2.2755,
+      "step": 516
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3585107619847347,
+      "learning_rate": 0.0008919389758291449,
+      "loss": 2.4367,
+      "step": 517
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.35458232383071636,
+      "learning_rate": 0.0008914773942554098,
+      "loss": 2.3832,
+      "step": 518
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.343281988716576,
+      "learning_rate": 0.000891014948936546,
+      "loss": 2.5348,
+      "step": 519
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.338506256263189,
+      "learning_rate": 0.0008905516408928804,
+      "loss": 2.4904,
+      "step": 520
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.35479429954790154,
+      "learning_rate": 0.0008900874711466434,
+      "loss": 2.37,
+      "step": 521
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3619253962476406,
+      "learning_rate": 0.0008896224407219666,
+      "loss": 2.4743,
+      "step": 522
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3432567843135978,
+      "learning_rate": 0.0008891565506448804,
+      "loss": 2.4571,
+      "step": 523
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.46643997209945,
+      "learning_rate": 0.0008886898019433122,
+      "loss": 2.285,
+      "step": 524
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3823740660708181,
+      "learning_rate": 0.0008882221956470836,
+      "loss": 2.4196,
+      "step": 525
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.35962073689176144,
+      "learning_rate": 0.0008877537327879086,
+      "loss": 2.3706,
+      "step": 526
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.4114055956840627,
+      "learning_rate": 0.0008872844143993908,
+      "loss": 2.3809,
+      "step": 527
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3702957320643523,
+      "learning_rate": 0.0008868142415170218,
+      "loss": 2.4336,
+      "step": 528
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3580563089026477,
+      "learning_rate": 0.0008863432151781781,
+      "loss": 2.4826,
+      "step": 529
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.34137222085936114,
+      "learning_rate": 0.0008858713364221195,
+      "loss": 2.3634,
+      "step": 530
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.34518741732092884,
+      "learning_rate": 0.0008853986062899868,
+      "loss": 2.3667,
+      "step": 531
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.34367304087774225,
+      "learning_rate": 0.0008849250258247986,
+      "loss": 2.3988,
+      "step": 532
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.35460345443889224,
+      "learning_rate": 0.0008844505960714503,
+      "loss": 2.3649,
+      "step": 533
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3415813021586036,
+      "learning_rate": 0.0008839753180767108,
+      "loss": 2.3406,
+      "step": 534
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.37469453238171674,
+      "learning_rate": 0.0008834991928892204,
+      "loss": 2.4613,
+      "step": 535
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3531467136685005,
+      "learning_rate": 0.000883022221559489,
+      "loss": 2.4198,
+      "step": 536
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3672679870930414,
+      "learning_rate": 0.0008825444051398934,
+      "loss": 2.3888,
+      "step": 537
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3854395825629921,
+      "learning_rate": 0.0008820657446846745,
+      "loss": 2.4188,
+      "step": 538
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3501285296932991,
+      "learning_rate": 0.000881586241249936,
+      "loss": 2.4085,
+      "step": 539
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.40121803634779146,
+      "learning_rate": 0.0008811058958936411,
+      "loss": 2.4557,
+      "step": 540
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3396291261487534,
+      "learning_rate": 0.000880624709675611,
+      "loss": 2.3969,
+      "step": 541
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3595675783081387,
+      "learning_rate": 0.000880142683657522,
+      "loss": 2.3978,
+      "step": 542
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3386025204395416,
+      "learning_rate": 0.0008796598189029029,
+      "loss": 2.3475,
+      "step": 543
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3478607803236338,
+      "learning_rate": 0.0008791761164771338,
+      "loss": 2.3844,
+      "step": 544
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3599597036835319,
+      "learning_rate": 0.0008786915774474424,
+      "loss": 2.3019,
+      "step": 545
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.37459045560773385,
+      "learning_rate": 0.0008782062028829027,
+      "loss": 2.3462,
+      "step": 546
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3177496510394425,
+      "learning_rate": 0.0008777199938544318,
+      "loss": 2.375,
+      "step": 547
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3301519941700513,
+      "learning_rate": 0.0008772329514347883,
+      "loss": 2.283,
+      "step": 548
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3674660028144134,
+      "learning_rate": 0.0008767450766985694,
+      "loss": 2.4041,
+      "step": 549
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.32874630891900475,
+      "learning_rate": 0.0008762563707222086,
+      "loss": 2.4359,
+      "step": 550
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3345600535761565,
+      "learning_rate": 0.0008757668345839738,
+      "loss": 2.4629,
+      "step": 551
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.37282003278931103,
+      "learning_rate": 0.0008752764693639638,
+      "loss": 2.4556,
+      "step": 552
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3415040134105036,
+      "learning_rate": 0.0008747852761441078,
+      "loss": 2.3361,
+      "step": 553
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.34825467159869367,
+      "learning_rate": 0.0008742932560081607,
+      "loss": 2.3435,
+      "step": 554
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3895602928864221,
+      "learning_rate": 0.0008738004100417025,
+      "loss": 2.3627,
+      "step": 555
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.35755672827585355,
+      "learning_rate": 0.0008733067393321355,
+      "loss": 2.4851,
+      "step": 556
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.36121397621413276,
+      "learning_rate": 0.000872812244968681,
+      "loss": 2.4251,
+      "step": 557
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3624359814048522,
+      "learning_rate": 0.0008723169280423783,
+      "loss": 2.4353,
+      "step": 558
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.33193263431561504,
+      "learning_rate": 0.0008718207896460811,
+      "loss": 2.4398,
+      "step": 559
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3677852099731477,
+      "learning_rate": 0.0008713238308744557,
+      "loss": 2.4627,
+      "step": 560
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3233436501805117,
+      "learning_rate": 0.0008708260528239789,
+      "loss": 2.3211,
+      "step": 561
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.37054951787773277,
+      "learning_rate": 0.000870327456592934,
+      "loss": 2.3932,
+      "step": 562
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3350860695309608,
+      "learning_rate": 0.0008698280432814107,
+      "loss": 2.4366,
+      "step": 563
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3611850174578661,
+      "learning_rate": 0.000869327813991301,
+      "loss": 2.4598,
+      "step": 564
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.31775654211358145,
+      "learning_rate": 0.0008688267698262971,
+      "loss": 2.4177,
+      "step": 565
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3478873529900364,
+      "learning_rate": 0.0008683249118918894,
+      "loss": 2.3688,
+      "step": 566
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.34812870559302367,
+      "learning_rate": 0.0008678222412953637,
+      "loss": 2.4241,
+      "step": 567
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.350306210626048,
+      "learning_rate": 0.0008673187591457987,
+      "loss": 2.5483,
+      "step": 568
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3350848444042627,
+      "learning_rate": 0.0008668144665540639,
+      "loss": 2.436,
+      "step": 569
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3419217501963111,
+      "learning_rate": 0.0008663093646328167,
+      "loss": 2.4478,
+      "step": 570
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3417861031002816,
+      "learning_rate": 0.0008658034544965003,
+      "loss": 2.4434,
+      "step": 571
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3326002201111991,
+      "learning_rate": 0.0008652967372613412,
+      "loss": 2.4115,
+      "step": 572
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.33725595110675155,
+      "learning_rate": 0.0008647892140453466,
+      "loss": 2.4718,
+      "step": 573
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3613279601499108,
+      "learning_rate": 0.0008642808859683021,
+      "loss": 2.3049,
+      "step": 574
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.36050101388687167,
+      "learning_rate": 0.0008637717541517689,
+      "loss": 2.4559,
+      "step": 575
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.35573833553000517,
+      "learning_rate": 0.0008632618197190816,
+      "loss": 2.3963,
+      "step": 576
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3297776623271599,
+      "learning_rate": 0.0008627510837953458,
+      "loss": 2.4268,
+      "step": 577
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3296333593576422,
+      "learning_rate": 0.0008622395475074355,
+      "loss": 2.3965,
+      "step": 578
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.33796404965507754,
+      "learning_rate": 0.0008617272119839903,
+      "loss": 2.3835,
+      "step": 579
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.348950909876335,
+      "learning_rate": 0.0008612140783554136,
+      "loss": 2.308,
+      "step": 580
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3566609822553432,
+      "learning_rate": 0.0008607001477538696,
+      "loss": 2.4077,
+      "step": 581
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.344060675569341,
+      "learning_rate": 0.0008601854213132807,
+      "loss": 2.5029,
+      "step": 582
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3365201740353835,
+      "learning_rate": 0.0008596699001693256,
+      "loss": 2.3498,
+      "step": 583
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.34173329851669504,
+      "learning_rate": 0.000859153585459436,
+      "loss": 2.3643,
+      "step": 584
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3439259051491479,
+      "learning_rate": 0.0008586364783227949,
+      "loss": 2.3222,
+      "step": 585
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3333280831796826,
+      "learning_rate": 0.0008581185799003332,
+      "loss": 2.4138,
+      "step": 586
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.364336789850431,
+      "learning_rate": 0.0008575998913347283,
+      "loss": 2.3764,
+      "step": 587
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.348046807727041,
+      "learning_rate": 0.0008570804137704004,
+      "loss": 2.3127,
+      "step": 588
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.33004584162530726,
+      "learning_rate": 0.0008565601483535108,
+      "loss": 2.306,
+      "step": 589
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.34714435197977966,
+      "learning_rate": 0.0008560390962319591,
+      "loss": 2.3334,
+      "step": 590
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3600486087267853,
+      "learning_rate": 0.0008555172585553804,
+      "loss": 2.3945,
+      "step": 591
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3528200545296786,
+      "learning_rate": 0.0008549946364751435,
+      "loss": 2.3209,
+      "step": 592
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.31862638893781464,
+      "learning_rate": 0.0008544712311443475,
+      "loss": 2.3953,
+      "step": 593
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.32852241992752873,
+      "learning_rate": 0.0008539470437178196,
+      "loss": 2.3739,
+      "step": 594
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.33330424786544,
+      "learning_rate": 0.000853422075352113,
+      "loss": 2.3792,
+      "step": 595
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3241334154639755,
+      "learning_rate": 0.0008528963272055035,
+      "loss": 2.3321,
+      "step": 596
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3374425344477458,
+      "learning_rate": 0.0008523698004379877,
+      "loss": 2.3918,
+      "step": 597
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3207254850140361,
+      "learning_rate": 0.00085184249621128,
+      "loss": 2.4575,
+      "step": 598
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.363975998346189,
+      "learning_rate": 0.0008513144156888101,
+      "loss": 2.4232,
+      "step": 599
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.34315932883703215,
+      "learning_rate": 0.0008507855600357207,
+      "loss": 2.4325,
+      "step": 600
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3544333014881256,
+      "learning_rate": 0.0008502559304188644,
+      "loss": 2.4128,
+      "step": 601
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.36448672844068714,
+      "learning_rate": 0.0008497255280068019,
+      "loss": 2.4458,
+      "step": 602
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3300989924507382,
+      "learning_rate": 0.0008491943539697986,
+      "loss": 2.269,
+      "step": 603
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.32846039082612916,
+      "learning_rate": 0.0008486624094798226,
+      "loss": 2.3327,
+      "step": 604
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3397549462110246,
+      "learning_rate": 0.0008481296957105417,
+      "loss": 2.4343,
+      "step": 605
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3447778247635745,
+      "learning_rate": 0.0008475962138373213,
+      "loss": 2.3368,
+      "step": 606
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.35246841012039,
+      "learning_rate": 0.0008470619650372211,
+      "loss": 2.4029,
+      "step": 607
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3581930326162995,
+      "learning_rate": 0.0008465269504889934,
+      "loss": 2.4421,
+      "step": 608
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.36170733347695855,
+      "learning_rate": 0.0008459911713730799,
+      "loss": 2.3796,
+      "step": 609
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.36904892427763614,
+      "learning_rate": 0.0008454546288716089,
+      "loss": 2.5082,
+      "step": 610
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.35011307421781696,
+      "learning_rate": 0.0008449173241683935,
+      "loss": 2.4066,
+      "step": 611
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.36829217223441835,
+      "learning_rate": 0.0008443792584489281,
+      "loss": 2.4389,
+      "step": 612
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.35072394026575726,
+      "learning_rate": 0.0008438404329003863,
+      "loss": 2.4144,
+      "step": 613
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.32954355370118715,
+      "learning_rate": 0.0008433008487116183,
+      "loss": 2.4401,
+      "step": 614
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.33031653759650337,
+      "learning_rate": 0.0008427605070731481,
+      "loss": 2.2946,
+      "step": 615
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3335321038476025,
+      "learning_rate": 0.0008422194091771708,
+      "loss": 2.4848,
+      "step": 616
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3395299799229233,
+      "learning_rate": 0.0008416775562175503,
+      "loss": 2.2874,
+      "step": 617
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.4251970893090745,
+      "learning_rate": 0.000841134949389816,
+      "loss": 2.3694,
+      "step": 618
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.34270649875396864,
+      "learning_rate": 0.0008405915898911611,
+      "loss": 2.37,
+      "step": 619
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3299036383545476,
+      "learning_rate": 0.0008400474789204396,
+      "loss": 2.4028,
+      "step": 620
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.33384828814912487,
+      "learning_rate": 0.0008395026176781626,
+      "loss": 2.422,
+      "step": 621
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3333120497426474,
+      "learning_rate": 0.0008389570073664976,
+      "loss": 2.4065,
+      "step": 622
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.340582775697048,
+      "learning_rate": 0.0008384106491892642,
+      "loss": 2.4036,
+      "step": 623
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3481401350948269,
+      "learning_rate": 0.0008378635443519327,
+      "loss": 2.4192,
+      "step": 624
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.33354400261870676,
+      "learning_rate": 0.0008373156940616199,
+      "loss": 2.4037,
+      "step": 625
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3574198746189019,
+      "learning_rate": 0.0008367670995270882,
+      "loss": 2.4281,
+      "step": 626
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.31395553705120366,
+      "learning_rate": 0.0008362177619587416,
+      "loss": 2.4082,
+      "step": 627
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.35963353914947493,
+      "learning_rate": 0.0008356676825686238,
+      "loss": 2.3745,
+      "step": 628
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.36051181678781635,
+      "learning_rate": 0.0008351168625704147,
+      "loss": 2.4021,
+      "step": 629
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.36406142960909904,
+      "learning_rate": 0.0008345653031794292,
+      "loss": 2.4722,
+      "step": 630
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.354687062651147,
+      "learning_rate": 0.0008340130056126125,
+      "loss": 2.4379,
+      "step": 631
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3753856113186296,
+      "learning_rate": 0.0008334599710885394,
+      "loss": 2.3469,
+      "step": 632
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.34037913416759547,
+      "learning_rate": 0.0008329062008274098,
+      "loss": 2.3679,
+      "step": 633
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3471143070676203,
+      "learning_rate": 0.000832351696051048,
+      "loss": 2.4414,
+      "step": 634
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.35548144542476484,
+      "learning_rate": 0.000831796457982898,
+      "loss": 2.4178,
+      "step": 635
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3693674233399044,
+      "learning_rate": 0.0008312404878480222,
+      "loss": 2.4369,
+      "step": 636
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.34024703634579245,
+      "learning_rate": 0.0008306837868730979,
+      "loss": 2.4586,
+      "step": 637
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3361548785519444,
+      "learning_rate": 0.0008301263562864152,
+      "loss": 2.3598,
+      "step": 638
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.31742923168115034,
+      "learning_rate": 0.0008295681973178737,
+      "loss": 2.3537,
+      "step": 639
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3578766971677568,
+      "learning_rate": 0.0008290093111989804,
+      "loss": 2.416,
+      "step": 640
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3577184267899438,
+      "learning_rate": 0.0008284496991628465,
+      "loss": 2.2857,
+      "step": 641
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3335025026653883,
+      "learning_rate": 0.0008278893624441847,
+      "loss": 2.2667,
+      "step": 642
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.3251734291018318,
+      "learning_rate": 0.000827328302279307,
+      "loss": 2.392,
+      "step": 643
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.32106764470010385,
+      "learning_rate": 0.0008267665199061211,
+      "loss": 2.2018,
+      "step": 644
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3535795212508593,
+      "learning_rate": 0.0008262040165641288,
+      "loss": 2.4039,
+      "step": 645
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3395903590923721,
+      "learning_rate": 0.0008256407934944219,
+      "loss": 2.4098,
+      "step": 646
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.35630511589573777,
+      "learning_rate": 0.0008250768519396807,
+      "loss": 2.4448,
+      "step": 647
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.4108869754247813,
+      "learning_rate": 0.0008245121931441706,
+      "loss": 2.3619,
+      "step": 648
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3299018852229439,
+      "learning_rate": 0.0008239468183537393,
+      "loss": 2.3771,
+      "step": 649
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3372952360444146,
+      "learning_rate": 0.0008233807288158146,
+      "loss": 2.3603,
+      "step": 650
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.32472636269721267,
+      "learning_rate": 0.0008228139257794012,
+      "loss": 2.4001,
+      "step": 651
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3633883833353663,
+      "learning_rate": 0.0008222464104950778,
+      "loss": 2.3336,
+      "step": 652
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3679787776360372,
+      "learning_rate": 0.000821678184214995,
+      "loss": 2.3743,
+      "step": 653
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3544774834604186,
+      "learning_rate": 0.0008211092481928716,
+      "loss": 2.3033,
+      "step": 654
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.42753242386006657,
+      "learning_rate": 0.0008205396036839927,
+      "loss": 2.4219,
+      "step": 655
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.36820917976673434,
+      "learning_rate": 0.0008199692519452069,
+      "loss": 2.4425,
+      "step": 656
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.36939798815795943,
+      "learning_rate": 0.0008193981942349224,
+      "loss": 2.37,
+      "step": 657
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3497426580103069,
+      "learning_rate": 0.0008188264318131056,
+      "loss": 2.347,
+      "step": 658
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3304232176478236,
+      "learning_rate": 0.0008182539659412776,
+      "loss": 2.2925,
+      "step": 659
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3359544446183685,
+      "learning_rate": 0.0008176807978825118,
+      "loss": 2.2985,
+      "step": 660
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.35521236625154395,
+      "learning_rate": 0.0008171069289014306,
+      "loss": 2.4282,
+      "step": 661
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.35171936424974115,
+      "learning_rate": 0.0008165323602642028,
+      "loss": 2.4822,
+      "step": 662
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.36031360713036886,
+      "learning_rate": 0.0008159570932385414,
+      "loss": 2.3291,
+      "step": 663
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3438888491376375,
+      "learning_rate": 0.0008153811290936999,
+      "loss": 2.3381,
+      "step": 664
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.3646643394581867,
+      "learning_rate": 0.0008148044691004698,
+      "loss": 2.2986,
+      "step": 665
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3236736292815385,
+      "learning_rate": 0.0008142271145311783,
+      "loss": 2.3976,
+      "step": 666
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3418416134355794,
+      "learning_rate": 0.000813649066659685,
+      "loss": 2.45,
+      "step": 667
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.34502103741050577,
+      "learning_rate": 0.0008130703267613787,
+      "loss": 2.4003,
+      "step": 668
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.37128613640745056,
+      "learning_rate": 0.0008124908961131759,
+      "loss": 2.3802,
+      "step": 669
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.36294597284034913,
+      "learning_rate": 0.0008119107759935163,
+      "loss": 2.2046,
+      "step": 670
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3482420722818896,
+      "learning_rate": 0.0008113299676823615,
+      "loss": 2.3893,
+      "step": 671
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.33034788485510147,
+      "learning_rate": 0.0008107484724611911,
+      "loss": 2.4657,
+      "step": 672
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3637181690643305,
+      "learning_rate": 0.0008101662916130006,
+      "loss": 2.3862,
+      "step": 673
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.36984960611309436,
+      "learning_rate": 0.0008095834264222979,
+      "loss": 2.3542,
+      "step": 674
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.35723547614800993,
+      "learning_rate": 0.0008089998781751009,
+      "loss": 2.4434,
+      "step": 675
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3269736415181225,
+      "learning_rate": 0.0008084156481589349,
+      "loss": 2.3833,
+      "step": 676
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3483426399777747,
+      "learning_rate": 0.0008078307376628291,
+      "loss": 2.4253,
+      "step": 677
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3157500448962639,
+      "learning_rate": 0.0008072451479773143,
+      "loss": 2.2885,
+      "step": 678
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.33495673125924164,
+      "learning_rate": 0.0008066588803944195,
+      "loss": 2.489,
+      "step": 679
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3513130609885305,
+      "learning_rate": 0.0008060719362076697,
+      "loss": 2.3903,
+      "step": 680
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.32374267306355253,
+      "learning_rate": 0.0008054843167120826,
+      "loss": 2.4595,
+      "step": 681
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.32012815885948004,
+      "learning_rate": 0.0008048960232041663,
+      "loss": 2.4066,
+      "step": 682
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.33347841596710215,
+      "learning_rate": 0.0008043070569819153,
+      "loss": 2.3431,
+      "step": 683
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.3286309280278726,
+      "learning_rate": 0.0008037174193448089,
+      "loss": 2.4446,
+      "step": 684
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.32675909874653697,
+      "learning_rate": 0.0008031271115938077,
+      "loss": 2.2867,
+      "step": 685
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.31408562718612537,
+      "learning_rate": 0.0008025361350313505,
+      "loss": 2.2578,
+      "step": 686
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.35654520680010077,
+      "learning_rate": 0.0008019444909613523,
+      "loss": 2.4839,
+      "step": 687
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.35473321918634054,
+      "learning_rate": 0.0008013521806892003,
+      "loss": 2.4422,
+      "step": 688
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3263701243643535,
+      "learning_rate": 0.000800759205521752,
+      "loss": 2.3006,
+      "step": 689
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3335205404467294,
+      "learning_rate": 0.0008001655667673318,
+      "loss": 2.5055,
+      "step": 690
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3284057795082039,
+      "learning_rate": 0.0007995712657357279,
+      "loss": 2.4488,
+      "step": 691
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3194165714234728,
+      "learning_rate": 0.0007989763037381904,
+      "loss": 2.3433,
+      "step": 692
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3014100149043869,
+      "learning_rate": 0.0007983806820874271,
+      "loss": 2.3472,
+      "step": 693
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.33765785147403127,
+      "learning_rate": 0.0007977844020976016,
+      "loss": 2.4718,
+      "step": 694
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.349127244610169,
+      "learning_rate": 0.00079718746508433,
+      "loss": 2.4164,
+      "step": 695
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3464919513788448,
+      "learning_rate": 0.0007965898723646776,
+      "loss": 2.3933,
+      "step": 696
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3388903370950258,
+      "learning_rate": 0.0007959916252571573,
+      "loss": 2.2988,
+      "step": 697
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.32047550371362243,
+      "learning_rate": 0.000795392725081725,
+      "loss": 2.3598,
+      "step": 698
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.35706368082253603,
+      "learning_rate": 0.000794793173159778,
+      "loss": 2.3931,
+      "step": 699
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3449978355205175,
+      "learning_rate": 0.0007941929708141513,
+      "loss": 2.4238,
+      "step": 700
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3439612305596844,
+      "learning_rate": 0.0007935921193691153,
+      "loss": 2.3962,
+      "step": 701
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3158003507187177,
+      "learning_rate": 0.0007929906201503722,
+      "loss": 2.3851,
+      "step": 702
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3471683982961772,
+      "learning_rate": 0.0007923884744850536,
+      "loss": 2.2554,
+      "step": 703
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.34309218600658,
+      "learning_rate": 0.0007917856837017176,
+      "loss": 2.4266,
+      "step": 704
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3352398842316874,
+      "learning_rate": 0.0007911822491303452,
+      "loss": 2.2636,
+      "step": 705
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.32564174727665246,
+      "learning_rate": 0.0007905781721023382,
+      "loss": 2.2644,
+      "step": 706
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.319543712739072,
+      "learning_rate": 0.000789973453950516,
+      "loss": 2.4434,
+      "step": 707
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.35427787912194375,
+      "learning_rate": 0.000789368096009112,
+      "loss": 2.3739,
+      "step": 708
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.39328036926912086,
+      "learning_rate": 0.0007887620996137721,
+      "loss": 2.3889,
+      "step": 709
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.29777691113332083,
+      "learning_rate": 0.0007881554661015497,
+      "loss": 2.4579,
+      "step": 710
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.32497794311947137,
+      "learning_rate": 0.0007875481968109051,
+      "loss": 2.3838,
+      "step": 711
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.2972226362974069,
+      "learning_rate": 0.0007869402930817007,
+      "loss": 2.2896,
+      "step": 712
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.32452823741473374,
+      "learning_rate": 0.0007863317562551987,
+      "loss": 2.301,
+      "step": 713
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3403237469102676,
+      "learning_rate": 0.0007857225876740584,
+      "loss": 2.2593,
+      "step": 714
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3653598760919418,
+      "learning_rate": 0.0007851127886823327,
+      "loss": 2.4855,
+      "step": 715
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.384418018414736,
+      "learning_rate": 0.0007845023606254658,
+      "loss": 2.4282,
+      "step": 716
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3479796383470647,
+      "learning_rate": 0.0007838913048502894,
+      "loss": 2.3262,
+      "step": 717
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3486991249143128,
+      "learning_rate": 0.0007832796227050208,
+      "loss": 2.3778,
+      "step": 718
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3335684730033968,
+      "learning_rate": 0.0007826673155392587,
+      "loss": 2.3994,
+      "step": 719
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3266459255062122,
+      "learning_rate": 0.000782054384703981,
+      "loss": 2.3678,
+      "step": 720
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.34348212479352735,
+      "learning_rate": 0.0007814408315515418,
+      "loss": 2.2806,
+      "step": 721
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.32141125096666223,
+      "learning_rate": 0.0007808266574356683,
+      "loss": 2.4045,
+      "step": 722
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3669084207430843,
+      "learning_rate": 0.0007802118637114573,
+      "loss": 2.4069,
+      "step": 723
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3499394910282515,
+      "learning_rate": 0.0007795964517353734,
+      "loss": 2.3176,
+      "step": 724
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.36158659519915776,
+      "learning_rate": 0.0007789804228652449,
+      "loss": 2.332,
+      "step": 725
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3670003566149308,
+      "learning_rate": 0.0007783637784602609,
+      "loss": 2.3656,
+      "step": 726
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.37567327874450757,
+      "learning_rate": 0.0007777465198809692,
+      "loss": 2.3766,
+      "step": 727
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3250864378316282,
+      "learning_rate": 0.0007771286484892722,
+      "loss": 2.4215,
+      "step": 728
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.35594719072415604,
+      "learning_rate": 0.000776510165648425,
+      "loss": 2.4118,
+      "step": 729
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.3464834040188932,
+      "learning_rate": 0.0007758910727230311,
+      "loss": 2.3609,
+      "step": 730
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.35756931111779144,
+      "learning_rate": 0.0007752713710790404,
+      "loss": 2.4359,
+      "step": 731
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.33453732379143286,
+      "learning_rate": 0.0007746510620837459,
+      "loss": 2.2596,
+      "step": 732
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.34798063437213816,
+      "learning_rate": 0.0007740301471057807,
+      "loss": 2.3705,
+      "step": 733
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3506666207102381,
+      "learning_rate": 0.0007734086275151146,
+      "loss": 2.3233,
+      "step": 734
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3632470832529857,
+      "learning_rate": 0.0007727865046830517,
+      "loss": 2.4654,
+      "step": 735
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.34323767822236306,
+      "learning_rate": 0.0007721637799822269,
+      "loss": 2.2985,
+      "step": 736
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3231169169454792,
+      "learning_rate": 0.0007715404547866032,
+      "loss": 2.3138,
+      "step": 737
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3302053383847459,
+      "learning_rate": 0.0007709165304714685,
+      "loss": 2.3241,
+      "step": 738
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3323159407786392,
+      "learning_rate": 0.0007702920084134324,
+      "loss": 2.3697,
+      "step": 739
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.34098074897980196,
+      "learning_rate": 0.0007696668899904236,
+      "loss": 2.3899,
+      "step": 740
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3285075603307696,
+      "learning_rate": 0.0007690411765816864,
+      "loss": 2.4502,
+      "step": 741
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3341925697239362,
+      "learning_rate": 0.0007684148695677778,
+      "loss": 2.3561,
+      "step": 742
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.32577130595053777,
+      "learning_rate": 0.000767787970330565,
+      "loss": 2.41,
+      "step": 743
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.35662071332294765,
+      "learning_rate": 0.000767160480253221,
+      "loss": 2.4316,
+      "step": 744
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.34216792234642956,
+      "learning_rate": 0.0007665324007202235,
+      "loss": 2.416,
+      "step": 745
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.32933259008107096,
+      "learning_rate": 0.0007659037331173498,
+      "loss": 2.3617,
+      "step": 746
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.37718544844050067,
+      "learning_rate": 0.0007652744788316752,
+      "loss": 2.2923,
+      "step": 747
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3435911687393134,
+      "learning_rate": 0.0007646446392515692,
+      "loss": 2.3882,
+      "step": 748
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3656911436293673,
+      "learning_rate": 0.000764014215766693,
+      "loss": 2.3424,
+      "step": 749
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.33004490696673916,
+      "learning_rate": 0.0007633832097679958,
+      "loss": 2.3493,
+      "step": 750
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.33855949878039954,
+      "learning_rate": 0.0007627516226477122,
+      "loss": 2.2887,
+      "step": 751
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.3493362908678133,
+      "learning_rate": 0.0007621194557993589,
+      "loss": 2.4506,
+      "step": 752
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.332401242955665,
+      "learning_rate": 0.0007614867106177319,
+      "loss": 2.2656,
+      "step": 753
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.32015079011358244,
+      "learning_rate": 0.0007608533884989029,
+      "loss": 2.387,
+      "step": 754
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3313724714812314,
+      "learning_rate": 0.0007602194908402166,
+      "loss": 2.3817,
+      "step": 755
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.34239666718727796,
+      "learning_rate": 0.0007595850190402877,
+      "loss": 2.346,
+      "step": 756
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3065125572196097,
+      "learning_rate": 0.0007589499744989976,
+      "loss": 2.3622,
+      "step": 757
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.35943670789011195,
+      "learning_rate": 0.0007583143586174916,
+      "loss": 2.4642,
+      "step": 758
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.32927861552553284,
+      "learning_rate": 0.000757678172798175,
+      "loss": 2.362,
+      "step": 759
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3401703667878008,
+      "learning_rate": 0.0007570414184447112,
+      "loss": 2.3867,
+      "step": 760
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3479691108895905,
+      "learning_rate": 0.0007564040969620179,
+      "loss": 2.3668,
+      "step": 761
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3564602188976791,
+      "learning_rate": 0.0007557662097562636,
+      "loss": 2.4224,
+      "step": 762
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3325953341811067,
+      "learning_rate": 0.0007551277582348658,
+      "loss": 2.3037,
+      "step": 763
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3096306105339475,
+      "learning_rate": 0.0007544887438064862,
+      "loss": 2.4259,
+      "step": 764
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3199752304525459,
+      "learning_rate": 0.0007538491678810294,
+      "loss": 2.4022,
+      "step": 765
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3562341147569589,
+      "learning_rate": 0.0007532090318696381,
+      "loss": 2.3511,
+      "step": 766
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.32548268504492667,
+      "learning_rate": 0.0007525683371846913,
+      "loss": 2.368,
+      "step": 767
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3605478889027448,
+      "learning_rate": 0.0007519270852398001,
+      "loss": 2.3811,
+      "step": 768
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3307162978905981,
+      "learning_rate": 0.000751285277449806,
+      "loss": 2.3511,
+      "step": 769
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3558165245789483,
+      "learning_rate": 0.0007506429152307756,
+      "loss": 2.3421,
+      "step": 770
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3532622292291561,
+      "learning_rate": 0.00075,
+      "loss": 2.465,
+      "step": 771
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3152689932059991,
+      "learning_rate": 0.00074935653317599,
+      "loss": 2.3449,
+      "step": 772
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3337243670517438,
+      "learning_rate": 0.000748712516178473,
+      "loss": 2.3226,
+      "step": 773
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.3207861027824236,
+      "learning_rate": 0.0007480679504283911,
+      "loss": 2.4824,
+      "step": 774
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3567653797935491,
+      "learning_rate": 0.0007474228373478964,
+      "loss": 2.3919,
+      "step": 775
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3346894157208737,
+      "learning_rate": 0.0007467771783603492,
+      "loss": 2.4775,
+      "step": 776
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.30751808246041545,
+      "learning_rate": 0.0007461309748903138,
+      "loss": 2.3289,
+      "step": 777
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3601586750018068,
+      "learning_rate": 0.0007454842283635562,
+      "loss": 2.3276,
+      "step": 778
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3128173651133522,
+      "learning_rate": 0.0007448369402070404,
+      "loss": 2.2925,
+      "step": 779
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.33324251648916375,
+      "learning_rate": 0.0007441891118489254,
+      "loss": 2.3825,
+      "step": 780
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3288890015436769,
+      "learning_rate": 0.0007435407447185622,
+      "loss": 2.3454,
+      "step": 781
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3274723679731149,
+      "learning_rate": 0.0007428918402464908,
+      "loss": 2.4117,
+      "step": 782
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.33819039837528986,
+      "learning_rate": 0.0007422423998644359,
+      "loss": 2.3445,
+      "step": 783
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.36959614345019703,
+      "learning_rate": 0.0007415924250053055,
+      "loss": 2.3347,
+      "step": 784
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.341138445046014,
+      "learning_rate": 0.0007409419171031865,
+      "loss": 2.3504,
+      "step": 785
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.34304074965580467,
+      "learning_rate": 0.0007402908775933419,
+      "loss": 2.3711,
+      "step": 786
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3116472713881393,
+      "learning_rate": 0.0007396393079122077,
+      "loss": 2.2848,
+      "step": 787
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3258643444124823,
+      "learning_rate": 0.0007389872094973896,
+      "loss": 2.348,
+      "step": 788
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.33655108359841646,
+      "learning_rate": 0.00073833458378766,
+      "loss": 2.418,
+      "step": 789
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3222152341069475,
+      "learning_rate": 0.0007376814322229544,
+      "loss": 2.3291,
+      "step": 790
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.325148064727578,
+      "learning_rate": 0.0007370277562443688,
+      "loss": 2.3882,
+      "step": 791
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3196430017889735,
+      "learning_rate": 0.0007363735572941564,
+      "loss": 2.3547,
+      "step": 792
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3285420991100482,
+      "learning_rate": 0.0007357188368157236,
+      "loss": 2.3347,
+      "step": 793
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3409955410281048,
+      "learning_rate": 0.0007350635962536284,
+      "loss": 2.3711,
+      "step": 794
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3277477105267692,
+      "learning_rate": 0.0007344078370535756,
+      "loss": 2.3606,
+      "step": 795
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.34218088878059993,
+      "learning_rate": 0.0007337515606624148,
+      "loss": 2.3767,
+      "step": 796
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.32064321249836963,
+      "learning_rate": 0.0007330947685281362,
+      "loss": 2.3909,
+      "step": 797
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.31795485218518793,
+      "learning_rate": 0.0007324374620998682,
+      "loss": 2.267,
+      "step": 798
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.31925534603252737,
+      "learning_rate": 0.000731779642827874,
+      "loss": 2.3803,
+      "step": 799
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.34771459310249997,
+      "learning_rate": 0.0007311213121635483,
+      "loss": 2.3531,
+      "step": 800
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.31918941233647247,
+      "learning_rate": 0.0007304624715594139,
+      "loss": 2.2971,
+      "step": 801
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.32819466737442515,
+      "learning_rate": 0.0007298031224691193,
+      "loss": 2.4472,
+      "step": 802
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3189034433044076,
+      "learning_rate": 0.0007291432663474339,
+      "loss": 2.3649,
+      "step": 803
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.31390623693655273,
+      "learning_rate": 0.0007284829046502467,
+      "loss": 2.3675,
+      "step": 804
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3265516471951474,
+      "learning_rate": 0.0007278220388345619,
+      "loss": 2.3344,
+      "step": 805
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.33000932708476066,
+      "learning_rate": 0.0007271606703584958,
+      "loss": 2.2187,
+      "step": 806
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.343752969457638,
+      "learning_rate": 0.000726498800681274,
+      "loss": 2.2825,
+      "step": 807
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.33339702338608646,
+      "learning_rate": 0.0007258364312632279,
+      "loss": 2.3568,
+      "step": 808
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3372497422887109,
+      "learning_rate": 0.0007251735635657915,
+      "loss": 2.4011,
+      "step": 809
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.306011304699242,
+      "learning_rate": 0.000724510199051498,
+      "loss": 2.3828,
+      "step": 810
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.32328693824380006,
+      "learning_rate": 0.0007238463391839769,
+      "loss": 2.4709,
+      "step": 811
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3152127870085075,
+      "learning_rate": 0.0007231819854279508,
+      "loss": 2.3715,
+      "step": 812
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3314605093751092,
+      "learning_rate": 0.0007225171392492316,
+      "loss": 2.2937,
+      "step": 813
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3238634835613084,
+      "learning_rate": 0.0007218518021147182,
+      "loss": 2.3484,
+      "step": 814
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.328720926770742,
+      "learning_rate": 0.0007211859754923923,
+      "loss": 2.4277,
+      "step": 815
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3316419762832727,
+      "learning_rate": 0.0007205196608513158,
+      "loss": 2.3019,
+      "step": 816
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3126928477539827,
+      "learning_rate": 0.0007198528596616272,
+      "loss": 2.3936,
+      "step": 817
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.318707864315425,
+      "learning_rate": 0.0007191855733945387,
+      "loss": 2.2313,
+      "step": 818
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3158164927647338,
+      "learning_rate": 0.0007185178035223327,
+      "loss": 2.2362,
+      "step": 819
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3317838644113709,
+      "learning_rate": 0.0007178495515183583,
+      "loss": 2.2461,
+      "step": 820
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3287330874357882,
+      "learning_rate": 0.000717180818857029,
+      "loss": 2.2722,
+      "step": 821
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.35517647031775873,
+      "learning_rate": 0.0007165116070138182,
+      "loss": 2.3351,
+      "step": 822
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3479494764122502,
+      "learning_rate": 0.0007158419174652569,
+      "loss": 2.269,
+      "step": 823
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.33061499039494735,
+      "learning_rate": 0.00071517175168893,
+      "loss": 2.36,
+      "step": 824
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.32402206240049514,
+      "learning_rate": 0.0007145011111634732,
+      "loss": 2.3048,
+      "step": 825
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3023558045290974,
+      "learning_rate": 0.0007138299973685694,
+      "loss": 2.4055,
+      "step": 826
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.31143790848708747,
+      "learning_rate": 0.0007131584117849459,
+      "loss": 2.3805,
+      "step": 827
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3110530400511612,
+      "learning_rate": 0.0007124863558943713,
+      "loss": 2.4095,
+      "step": 828
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.30435410112625394,
+      "learning_rate": 0.0007118138311796514,
+      "loss": 2.3874,
+      "step": 829
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3115489311922659,
+      "learning_rate": 0.0007111408391246262,
+      "loss": 2.3527,
+      "step": 830
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3290504600184735,
+      "learning_rate": 0.0007104673812141675,
+      "loss": 2.3196,
+      "step": 831
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3213139337787899,
+      "learning_rate": 0.0007097934589341745,
+      "loss": 2.3092,
+      "step": 832
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3171486772954657,
+      "learning_rate": 0.0007091190737715711,
+      "loss": 2.3587,
+      "step": 833
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3361340563682881,
+      "learning_rate": 0.0007084442272143026,
+      "loss": 2.3177,
+      "step": 834
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3724891080202217,
+      "learning_rate": 0.000707768920751332,
+      "loss": 2.4006,
+      "step": 835
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3279951456398761,
+      "learning_rate": 0.0007070931558726373,
+      "loss": 2.447,
+      "step": 836
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3548358136579193,
+      "learning_rate": 0.0007064169340692076,
+      "loss": 2.432,
+      "step": 837
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.32607610560231143,
+      "learning_rate": 0.0007057402568330407,
+      "loss": 2.3165,
+      "step": 838
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.3214809107736882,
+      "learning_rate": 0.0007050631256571389,
+      "loss": 2.373,
+      "step": 839
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.32057829459480947,
+      "learning_rate": 0.000704385542035506,
+      "loss": 2.3885,
+      "step": 840
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3361826405194171,
+      "learning_rate": 0.000703707507463144,
+      "loss": 2.4413,
+      "step": 841
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.32029784396884564,
+      "learning_rate": 0.0007030290234360505,
+      "loss": 2.313,
+      "step": 842
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3150046154746048,
+      "learning_rate": 0.0007023500914512139,
+      "loss": 2.3191,
+      "step": 843
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.32861309813319284,
+      "learning_rate": 0.0007016707130066116,
+      "loss": 2.4038,
+      "step": 844
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3315416844131154,
+      "learning_rate": 0.0007009908896012055,
+      "loss": 2.3616,
+      "step": 845
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.333852202582535,
+      "learning_rate": 0.0007003106227349399,
+      "loss": 2.4106,
+      "step": 846
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.33936881033873867,
+      "learning_rate": 0.000699629913908737,
+      "loss": 2.2868,
+      "step": 847
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.32635619172513075,
+      "learning_rate": 0.0006989487646244943,
+      "loss": 2.3597,
+      "step": 848
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.32815647129909364,
+      "learning_rate": 0.0006982671763850814,
+      "loss": 2.2629,
+      "step": 849
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3203810551459643,
+      "learning_rate": 0.0006975851506943359,
+      "loss": 2.3915,
+      "step": 850
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3426320906028205,
+      "learning_rate": 0.0006969026890570611,
+      "loss": 2.4276,
+      "step": 851
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3274299026367865,
+      "learning_rate": 0.0006962197929790216,
+      "loss": 2.301,
+      "step": 852
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.32982680661051134,
+      "learning_rate": 0.0006955364639669409,
+      "loss": 2.3361,
+      "step": 853
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3385142627838122,
+      "learning_rate": 0.0006948527035284978,
+      "loss": 2.2909,
+      "step": 854
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.31393478088988014,
+      "learning_rate": 0.0006941685131723225,
+      "loss": 2.3393,
+      "step": 855
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.35995436350497134,
+      "learning_rate": 0.0006934838944079943,
+      "loss": 2.3497,
+      "step": 856
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3440878609197891,
+      "learning_rate": 0.0006927988487460378,
+      "loss": 2.3338,
+      "step": 857
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.312494885287798,
+      "learning_rate": 0.0006921133776979186,
+      "loss": 2.3524,
+      "step": 858
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3222413381251988,
+      "learning_rate": 0.0006914274827760418,
+      "loss": 2.3054,
+      "step": 859
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.30996726479740305,
+      "learning_rate": 0.0006907411654937475,
+      "loss": 2.3099,
+      "step": 860
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3308000312343332,
+      "learning_rate": 0.0006900544273653075,
+      "loss": 2.3824,
+      "step": 861
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3051626030325834,
+      "learning_rate": 0.000689367269905922,
+      "loss": 2.1658,
+      "step": 862
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.320151493046798,
+      "learning_rate": 0.0006886796946317168,
+      "loss": 2.2579,
+      "step": 863
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.29894601697068673,
+      "learning_rate": 0.0006879917030597397,
+      "loss": 2.3118,
+      "step": 864
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.31483589499546816,
+      "learning_rate": 0.0006873032967079561,
+      "loss": 2.3796,
+      "step": 865
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3350968684624178,
+      "learning_rate": 0.0006866144770952474,
+      "loss": 2.2906,
+      "step": 866
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3143361829651158,
+      "learning_rate": 0.0006859252457414067,
+      "loss": 2.3499,
+      "step": 867
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.34697757717577965,
+      "learning_rate": 0.0006852356041671351,
+      "loss": 2.309,
+      "step": 868
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3404897111849101,
+      "learning_rate": 0.0006845455538940394,
+      "loss": 2.3518,
+      "step": 869
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.322855121025158,
+      "learning_rate": 0.0006838550964446276,
+      "loss": 2.3678,
+      "step": 870
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3297412672979436,
+      "learning_rate": 0.0006831642333423067,
+      "loss": 2.2469,
+      "step": 871
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.33796344298959347,
+      "learning_rate": 0.000682472966111378,
+      "loss": 2.2828,
+      "step": 872
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3284399733560785,
+      "learning_rate": 0.0006817812962770348,
+      "loss": 2.3011,
+      "step": 873
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.34055450022623,
+      "learning_rate": 0.0006810892253653589,
+      "loss": 2.4256,
+      "step": 874
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.35169789331527285,
+      "learning_rate": 0.0006803967549033167,
+      "loss": 2.4206,
+      "step": 875
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3424455716637631,
+      "learning_rate": 0.0006797038864187564,
+      "loss": 2.3056,
+      "step": 876
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3085269868412449,
+      "learning_rate": 0.0006790106214404043,
+      "loss": 2.4024,
+      "step": 877
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3222878508984173,
+      "learning_rate": 0.0006783169614978614,
+      "loss": 2.3362,
+      "step": 878
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.33600084978280775,
+      "learning_rate": 0.0006776229081216001,
+      "loss": 2.3361,
+      "step": 879
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.33453284138489914,
+      "learning_rate": 0.0006769284628429611,
+      "loss": 2.3668,
+      "step": 880
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.3358384966167377,
+      "learning_rate": 0.0006762336271941498,
+      "loss": 2.2834,
+      "step": 881
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.30916536109358267,
+      "learning_rate": 0.0006755384027082326,
+      "loss": 2.2233,
+      "step": 882
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.32834660001253274,
+      "learning_rate": 0.0006748427909191342,
+      "loss": 2.3116,
+      "step": 883
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3590030277767803,
+      "learning_rate": 0.0006741467933616335,
+      "loss": 2.3552,
+      "step": 884
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.33924549007373356,
+      "learning_rate": 0.0006734504115713604,
+      "loss": 2.316,
+      "step": 885
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3373822212651072,
+      "learning_rate": 0.0006727536470847932,
+      "loss": 2.3907,
+      "step": 886
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3156531210563873,
+      "learning_rate": 0.000672056501439254,
+      "loss": 2.3484,
+      "step": 887
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.30232577065108385,
+      "learning_rate": 0.0006713589761729063,
+      "loss": 2.3422,
+      "step": 888
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.32679357485715804,
+      "learning_rate": 0.0006706610728247508,
+      "loss": 2.2347,
+      "step": 889
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.32952982891053145,
+      "learning_rate": 0.0006699627929346227,
+      "loss": 2.3958,
+      "step": 890
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.31416006410439123,
+      "learning_rate": 0.0006692641380431879,
+      "loss": 2.3508,
+      "step": 891
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.31976691095931475,
+      "learning_rate": 0.0006685651096919393,
+      "loss": 2.3573,
+      "step": 892
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3432477790461921,
+      "learning_rate": 0.0006678657094231944,
+      "loss": 2.3775,
+      "step": 893
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3494070771178411,
+      "learning_rate": 0.0006671659387800909,
+      "loss": 2.3531,
+      "step": 894
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3467868781617165,
+      "learning_rate": 0.000666465799306584,
+      "loss": 2.4034,
+      "step": 895
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.33963115232354674,
+      "learning_rate": 0.0006657652925474423,
+      "loss": 2.4095,
+      "step": 896
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3052838030006342,
+      "learning_rate": 0.000665064420048245,
+      "loss": 2.3037,
+      "step": 897
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3276846435276858,
+      "learning_rate": 0.0006643631833553785,
+      "loss": 2.3607,
+      "step": 898
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3126343812103487,
+      "learning_rate": 0.000663661584016032,
+      "loss": 2.4007,
+      "step": 899
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3106830317210614,
+      "learning_rate": 0.0006629596235781957,
+      "loss": 2.3719,
+      "step": 900
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3293973656528261,
+      "learning_rate": 0.0006622573035906556,
+      "loss": 2.2525,
+      "step": 901
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.336755851987241,
+      "learning_rate": 0.0006615546256029921,
+      "loss": 2.3228,
+      "step": 902
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3377103994430235,
+      "learning_rate": 0.0006608515911655743,
+      "loss": 2.3424,
+      "step": 903
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.32895208285844124,
+      "learning_rate": 0.0006601482018295591,
+      "loss": 2.4037,
+      "step": 904
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.3134833696408548,
+      "learning_rate": 0.0006594444591468851,
+      "loss": 2.2728,
+      "step": 905
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.32846455730601254,
+      "learning_rate": 0.0006587403646702713,
+      "loss": 2.2563,
+      "step": 906
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.319323292357709,
+      "learning_rate": 0.0006580359199532126,
+      "loss": 2.3391,
+      "step": 907
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.31860860927062956,
+      "learning_rate": 0.000657331126549977,
+      "loss": 2.3104,
+      "step": 908
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3224701737309374,
+      "learning_rate": 0.0006566259860156014,
+      "loss": 2.3488,
+      "step": 909
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.30690320725954295,
+      "learning_rate": 0.0006559204999058888,
+      "loss": 2.3893,
+      "step": 910
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3313824571680554,
+      "learning_rate": 0.0006552146697774049,
+      "loss": 2.3841,
+      "step": 911
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3528150801673695,
+      "learning_rate": 0.0006545084971874737,
+      "loss": 2.1983,
+      "step": 912
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3124392853117316,
+      "learning_rate": 0.0006538019836941758,
+      "loss": 2.1749,
+      "step": 913
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.33741395483060893,
+      "learning_rate": 0.0006530951308563431,
+      "loss": 2.3473,
+      "step": 914
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.32661969396845036,
+      "learning_rate": 0.0006523879402335567,
+      "loss": 2.414,
+      "step": 915
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.33181225636024936,
+      "learning_rate": 0.0006516804133861429,
+      "loss": 2.3853,
+      "step": 916
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3458896365071438,
+      "learning_rate": 0.0006509725518751698,
+      "loss": 2.439,
+      "step": 917
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.31153549802735847,
+      "learning_rate": 0.0006502643572624438,
+      "loss": 2.502,
+      "step": 918
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.31216707009464034,
+      "learning_rate": 0.0006495558311105064,
+      "loss": 2.2763,
+      "step": 919
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3235119331261518,
+      "learning_rate": 0.0006488469749826305,
+      "loss": 2.3995,
+      "step": 920
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3164397643263149,
+      "learning_rate": 0.000648137790442817,
+      "loss": 2.3993,
+      "step": 921
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3436873610558131,
+      "learning_rate": 0.0006474282790557916,
+      "loss": 2.2999,
+      "step": 922
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.29810466638743033,
+      "learning_rate": 0.000646718442387001,
+      "loss": 2.3556,
+      "step": 923
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.325317958931904,
+      "learning_rate": 0.0006460082820026094,
+      "loss": 2.3412,
+      "step": 924
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.2958031841477819,
+      "learning_rate": 0.0006452977994694959,
+      "loss": 2.267,
+      "step": 925
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3544029409883216,
+      "learning_rate": 0.0006445869963552496,
+      "loss": 2.4752,
+      "step": 926
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.32534768059996794,
+      "learning_rate": 0.0006438758742281672,
+      "loss": 2.3587,
+      "step": 927
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.32105135745702257,
+      "learning_rate": 0.0006431644346572495,
+      "loss": 2.2995,
+      "step": 928
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.31843542371837313,
+      "learning_rate": 0.0006424526792121974,
+      "loss": 2.3129,
+      "step": 929
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3220739616061797,
+      "learning_rate": 0.0006417406094634089,
+      "loss": 2.2944,
+      "step": 930
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3126717403345713,
+      "learning_rate": 0.0006410282269819756,
+      "loss": 2.1902,
+      "step": 931
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3169345168027278,
+      "learning_rate": 0.0006403155333396787,
+      "loss": 2.3799,
+      "step": 932
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.32586517347715577,
+      "learning_rate": 0.0006396025301089863,
+      "loss": 2.3421,
+      "step": 933
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.33374910705115224,
+      "learning_rate": 0.0006388892188630493,
+      "loss": 2.3134,
+      "step": 934
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.33166506187391426,
+      "learning_rate": 0.0006381756011756982,
+      "loss": 2.4764,
+      "step": 935
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.32317286671002265,
+      "learning_rate": 0.0006374616786214403,
+      "loss": 2.3915,
+      "step": 936
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.32411880898065504,
+      "learning_rate": 0.0006367474527754544,
+      "loss": 2.2781,
+      "step": 937
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.31625286646174194,
+      "learning_rate": 0.0006360329252135894,
+      "loss": 2.2111,
+      "step": 938
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.32040467681416823,
+      "learning_rate": 0.0006353180975123595,
+      "loss": 2.2662,
+      "step": 939
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.32813882902378005,
+      "learning_rate": 0.0006346029712489413,
+      "loss": 2.4014,
+      "step": 940
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3198358948066891,
+      "learning_rate": 0.0006338875480011698,
+      "loss": 2.2986,
+      "step": 941
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.31744867577260105,
+      "learning_rate": 0.0006331718293475357,
+      "loss": 2.3923,
+      "step": 942
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.323674699129199,
+      "learning_rate": 0.0006324558168671811,
+      "loss": 2.3017,
+      "step": 943
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.31619422274424275,
+      "learning_rate": 0.0006317395121398968,
+      "loss": 2.4215,
+      "step": 944
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.32723841345192733,
+      "learning_rate": 0.0006310229167461179,
+      "loss": 2.2779,
+      "step": 945
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.2975435835281563,
+      "learning_rate": 0.0006303060322669214,
+      "loss": 2.2576,
+      "step": 946
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3041767138105239,
+      "learning_rate": 0.0006295888602840214,
+      "loss": 2.2826,
+      "step": 947
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.3052844818015639,
+      "learning_rate": 0.0006288714023797671,
+      "loss": 2.2531,
+      "step": 948
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3309270730179494,
+      "learning_rate": 0.000628153660137138,
+      "loss": 2.3025,
+      "step": 949
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.32925981846869906,
+      "learning_rate": 0.0006274356351397413,
+      "loss": 2.3913,
+      "step": 950
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.32259515375266107,
+      "learning_rate": 0.0006267173289718079,
+      "loss": 2.359,
+      "step": 951
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.31949244722454084,
+      "learning_rate": 0.000625998743218189,
+      "loss": 2.4343,
+      "step": 952
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.2902821894529481,
+      "learning_rate": 0.000625279879464353,
+      "loss": 2.3073,
+      "step": 953
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.31287767368432445,
+      "learning_rate": 0.000624560739296381,
+      "loss": 2.3286,
+      "step": 954
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3243635901191911,
+      "learning_rate": 0.0006238413243009648,
+      "loss": 2.257,
+      "step": 955
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3121251484095827,
+      "learning_rate": 0.000623121636065402,
+      "loss": 2.392,
+      "step": 956
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3081847881169902,
+      "learning_rate": 0.0006224016761775933,
+      "loss": 2.3289,
+      "step": 957
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3246347608053308,
+      "learning_rate": 0.0006216814462260386,
+      "loss": 2.417,
+      "step": 958
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3106100136955923,
+      "learning_rate": 0.0006209609477998338,
+      "loss": 2.324,
+      "step": 959
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3096246050235048,
+      "learning_rate": 0.0006202401824886674,
+      "loss": 2.3947,
+      "step": 960
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.34664769413129054,
+      "learning_rate": 0.0006195191518828162,
+      "loss": 2.3279,
+      "step": 961
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3389739702228862,
+      "learning_rate": 0.0006187978575731427,
+      "loss": 2.3928,
+      "step": 962
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.33475731404588416,
+      "learning_rate": 0.0006180763011510911,
+      "loss": 2.3862,
+      "step": 963
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.2927920078017773,
+      "learning_rate": 0.000617354484208684,
+      "loss": 2.1668,
+      "step": 964
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.33114518325058034,
+      "learning_rate": 0.0006166324083385189,
+      "loss": 2.4325,
+      "step": 965
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3438554816506677,
+      "learning_rate": 0.0006159100751337642,
+      "loss": 2.3216,
+      "step": 966
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.34212787153086666,
+      "learning_rate": 0.0006151874861881565,
+      "loss": 2.3966,
+      "step": 967
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.31286553192214783,
+      "learning_rate": 0.0006144646430959964,
+      "loss": 2.4101,
+      "step": 968
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.325400701213478,
+      "learning_rate": 0.0006137415474521454,
+      "loss": 2.3334,
+      "step": 969
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.3189248080026219,
+      "learning_rate": 0.0006130182008520222,
+      "loss": 2.4279,
+      "step": 970
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3008412778640715,
+      "learning_rate": 0.000612294604891599,
+      "loss": 2.2966,
+      "step": 971
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3133297579914418,
+      "learning_rate": 0.0006115707611673986,
+      "loss": 2.3041,
+      "step": 972
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3234325279220917,
+      "learning_rate": 0.0006108466712764902,
+      "loss": 2.4158,
+      "step": 973
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3562366327735097,
+      "learning_rate": 0.0006101223368164858,
+      "loss": 2.3862,
+      "step": 974
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.33398666965551016,
+      "learning_rate": 0.0006093977593855375,
+      "loss": 2.3205,
+      "step": 975
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.31812680274989064,
+      "learning_rate": 0.0006086729405823335,
+      "loss": 2.3099,
+      "step": 976
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.326738372402196,
+      "learning_rate": 0.0006079478820060943,
+      "loss": 2.327,
+      "step": 977
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3075247289100618,
+      "learning_rate": 0.0006072225852565695,
+      "loss": 2.3258,
+      "step": 978
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.4192277297759658,
+      "learning_rate": 0.0006064970519340341,
+      "loss": 2.2647,
+      "step": 979
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3275532176097454,
+      "learning_rate": 0.0006057712836392856,
+      "loss": 2.2397,
+      "step": 980
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.32418532954439516,
+      "learning_rate": 0.0006050452819736389,
+      "loss": 2.3444,
+      "step": 981
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.34887009363856963,
+      "learning_rate": 0.000604319048538925,
+      "loss": 2.3524,
+      "step": 982
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.330291290544246,
+      "learning_rate": 0.0006035925849374855,
+      "loss": 2.4325,
+      "step": 983
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.33125540077256327,
+      "learning_rate": 0.0006028658927721697,
+      "loss": 2.3683,
+      "step": 984
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3160399549232559,
+      "learning_rate": 0.0006021389736463321,
+      "loss": 2.2623,
+      "step": 985
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.30018389656896394,
+      "learning_rate": 0.0006014118291638271,
+      "loss": 2.2523,
+      "step": 986
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.31765307647142366,
+      "learning_rate": 0.0006006844609290065,
+      "loss": 2.3597,
+      "step": 987
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3120451181691426,
+      "learning_rate": 0.0005999568705467161,
+      "loss": 2.3003,
+      "step": 988
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3344358054521809,
+      "learning_rate": 0.0005992290596222915,
+      "loss": 2.4032,
+      "step": 989
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.34535295974999397,
+      "learning_rate": 0.0005985010297615551,
+      "loss": 2.4168,
+      "step": 990
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3281675611519965,
+      "learning_rate": 0.0005977727825708123,
+      "loss": 2.321,
+      "step": 991
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.3146507826087485,
+      "learning_rate": 0.0005970443196568478,
+      "loss": 2.3544,
+      "step": 992
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.333365249424523,
+      "learning_rate": 0.0005963156426269227,
+      "loss": 2.3055,
+      "step": 993
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3469917003340476,
+      "learning_rate": 0.0005955867530887702,
+      "loss": 2.2836,
+      "step": 994
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.31627759771800706,
+      "learning_rate": 0.0005948576526505923,
+      "loss": 2.3189,
+      "step": 995
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.32241774008556184,
+      "learning_rate": 0.0005941283429210568,
+      "loss": 2.287,
+      "step": 996
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3130945911522531,
+      "learning_rate": 0.0005933988255092926,
+      "loss": 2.3186,
+      "step": 997
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3012209971677374,
+      "learning_rate": 0.0005926691020248874,
+      "loss": 2.205,
+      "step": 998
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.32574117147046766,
+      "learning_rate": 0.0005919391740778833,
+      "loss": 2.3536,
+      "step": 999
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.30938651988472604,
+      "learning_rate": 0.0005912090432787736,
+      "loss": 2.2637,
+      "step": 1000
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.35359993209433627,
+      "learning_rate": 0.000590478711238499,
+      "loss": 2.3242,
+      "step": 1001
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.32565291456150264,
+      "learning_rate": 0.0005897481795684446,
+      "loss": 2.3331,
+      "step": 1002
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3352221159248403,
+      "learning_rate": 0.0005890174498804355,
+      "loss": 2.3721,
+      "step": 1003
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.35146139509371727,
+      "learning_rate": 0.0005882865237867339,
+      "loss": 2.2994,
+      "step": 1004
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3323625407004526,
+      "learning_rate": 0.0005875554029000353,
+      "loss": 2.3213,
+      "step": 1005
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.31769590364780514,
+      "learning_rate": 0.0005868240888334653,
+      "loss": 2.3694,
+      "step": 1006
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.319345650956919,
+      "learning_rate": 0.0005860925832005753,
+      "loss": 2.325,
+      "step": 1007
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3161263295176036,
+      "learning_rate": 0.0005853608876153395,
+      "loss": 2.2414,
+      "step": 1008
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.32429481908445323,
+      "learning_rate": 0.0005846290036921512,
+      "loss": 2.298,
+      "step": 1009
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.31678186459022584,
+      "learning_rate": 0.0005838969330458195,
+      "loss": 2.3764,
+      "step": 1010
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.31823261598976704,
+      "learning_rate": 0.0005831646772915651,
+      "loss": 2.3449,
+      "step": 1011
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3184484056130353,
+      "learning_rate": 0.0005824322380450173,
+      "loss": 2.4109,
+      "step": 1012
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.31489253724303534,
+      "learning_rate": 0.0005816996169222102,
+      "loss": 2.3841,
+      "step": 1013
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.3267873616332654,
+      "learning_rate": 0.0005809668155395793,
+      "loss": 2.281,
+      "step": 1014
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3181483046186482,
+      "learning_rate": 0.0005802338355139578,
+      "loss": 2.2865,
+      "step": 1015
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3079328793248607,
+      "learning_rate": 0.0005795006784625728,
+      "loss": 2.3615,
+      "step": 1016
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3163840033661192,
+      "learning_rate": 0.0005787673460030423,
+      "loss": 2.3315,
+      "step": 1017
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.30558426997313565,
+      "learning_rate": 0.000578033839753371,
+      "loss": 2.3185,
+      "step": 1018
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.32678249653208646,
+      "learning_rate": 0.0005773001613319476,
+      "loss": 2.3794,
+      "step": 1019
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3584925286618791,
+      "learning_rate": 0.00057656631235754,
+      "loss": 2.3214,
+      "step": 1020
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.32753028032662845,
+      "learning_rate": 0.0005758322944492929,
+      "loss": 2.3967,
+      "step": 1021
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3161410697648002,
+      "learning_rate": 0.0005750981092267237,
+      "loss": 2.3614,
+      "step": 1022
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3213418150512745,
+      "learning_rate": 0.0005743637583097183,
+      "loss": 2.3236,
+      "step": 1023
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.31846407910603836,
+      "learning_rate": 0.0005736292433185291,
+      "loss": 2.2839,
+      "step": 1024
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3210521553380509,
+      "learning_rate": 0.0005728945658737699,
+      "loss": 2.371,
+      "step": 1025
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.32266730564000773,
+      "learning_rate": 0.0005721597275964133,
+      "loss": 2.3691,
+      "step": 1026
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.31398321961117975,
+      "learning_rate": 0.0005714247301077865,
+      "loss": 2.3334,
+      "step": 1027
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3137262660576014,
+      "learning_rate": 0.0005706895750295682,
+      "loss": 2.4076,
+      "step": 1028
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.31419503347527183,
+      "learning_rate": 0.0005699542639837844,
+      "loss": 2.2157,
+      "step": 1029
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3253276939215174,
+      "learning_rate": 0.0005692187985928055,
+      "loss": 2.4094,
+      "step": 1030
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.30623433789542936,
+      "learning_rate": 0.0005684831804793427,
+      "loss": 2.2556,
+      "step": 1031
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3162619774226131,
+      "learning_rate": 0.0005677474112664438,
+      "loss": 2.3088,
+      "step": 1032
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.3377422696261182,
+      "learning_rate": 0.0005670114925774899,
+      "loss": 2.2649,
+      "step": 1033
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.34112267656635076,
+      "learning_rate": 0.0005662754260361924,
+      "loss": 2.3261,
+      "step": 1034
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.319727575825281,
+      "learning_rate": 0.0005655392132665884,
+      "loss": 2.3467,
+      "step": 1035
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.32563327458884317,
+      "learning_rate": 0.000564802855893038,
+      "loss": 2.3074,
+      "step": 1036
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3591408471089698,
+      "learning_rate": 0.0005640663555402198,
+      "loss": 2.303,
+      "step": 1037
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3195266152551932,
+      "learning_rate": 0.0005633297138331285,
+      "loss": 2.282,
+      "step": 1038
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3416995964317091,
+      "learning_rate": 0.0005625929323970705,
+      "loss": 2.3558,
+      "step": 1039
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3259139273582026,
+      "learning_rate": 0.0005618560128576603,
+      "loss": 2.2926,
+      "step": 1040
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.32215019466551603,
+      "learning_rate": 0.0005611189568408173,
+      "loss": 2.4006,
+      "step": 1041
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.32489676907460974,
+      "learning_rate": 0.0005603817659727619,
+      "loss": 2.3513,
+      "step": 1042
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.33204883199615237,
+      "learning_rate": 0.0005596444418800121,
+      "loss": 2.4087,
+      "step": 1043
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.32417190407423074,
+      "learning_rate": 0.0005589069861893798,
+      "loss": 2.2896,
+      "step": 1044
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.31469926129048226,
+      "learning_rate": 0.0005581694005279673,
+      "loss": 2.3555,
+      "step": 1045
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.299805876382933,
+      "learning_rate": 0.0005574316865231637,
+      "loss": 2.2767,
+      "step": 1046
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3012786842840355,
+      "learning_rate": 0.0005566938458026411,
+      "loss": 2.277,
+      "step": 1047
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.31019630570100054,
+      "learning_rate": 0.0005559558799943514,
+      "loss": 2.3606,
+      "step": 1048
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3158170994048566,
+      "learning_rate": 0.0005552177907265223,
+      "loss": 2.302,
+      "step": 1049
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3335927860855599,
+      "learning_rate": 0.000554479579627654,
+      "loss": 2.3531,
+      "step": 1050
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3074913599612058,
+      "learning_rate": 0.0005537412483265157,
+      "loss": 2.3023,
+      "step": 1051
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3124799912745835,
+      "learning_rate": 0.0005530027984521413,
+      "loss": 2.3258,
+      "step": 1052
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3776512796766397,
+      "learning_rate": 0.0005522642316338268,
+      "loss": 2.2796,
+      "step": 1053
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3200118864469617,
+      "learning_rate": 0.0005515255495011259,
+      "loss": 2.3378,
+      "step": 1054
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.33010937797604284,
+      "learning_rate": 0.0005507867536838472,
+      "loss": 2.382,
+      "step": 1055
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.3135227214076043,
+      "learning_rate": 0.0005500478458120492,
+      "loss": 2.3325,
+      "step": 1056
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.28636180945498957,
+      "learning_rate": 0.0005493088275160387,
+      "loss": 2.2964,
+      "step": 1057
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.2960964947686682,
+      "learning_rate": 0.0005485697004263657,
+      "loss": 2.3248,
+      "step": 1058
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.29771671306574543,
+      "learning_rate": 0.0005478304661738199,
+      "loss": 2.4053,
+      "step": 1059
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3074413405401914,
+      "learning_rate": 0.0005470911263894279,
+      "loss": 2.2835,
+      "step": 1060
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3070490049710694,
+      "learning_rate": 0.0005463516827044491,
+      "loss": 2.3342,
+      "step": 1061
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3110218766635616,
+      "learning_rate": 0.000545612136750372,
+      "loss": 2.3703,
+      "step": 1062
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.32413782284059656,
+      "learning_rate": 0.0005448724901589107,
+      "loss": 2.2494,
+      "step": 1063
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.33054195562096395,
+      "learning_rate": 0.0005441327445620014,
+      "loss": 2.3167,
+      "step": 1064
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.2995234195121158,
+      "learning_rate": 0.0005433929015917988,
+      "loss": 2.3855,
+      "step": 1065
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.31974067694268155,
+      "learning_rate": 0.0005426529628806724,
+      "loss": 2.3281,
+      "step": 1066
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3289762327630004,
+      "learning_rate": 0.0005419129300612029,
+      "loss": 2.2907,
+      "step": 1067
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3314545804406946,
+      "learning_rate": 0.000541172804766179,
+      "loss": 2.3463,
+      "step": 1068
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3190085761350208,
+      "learning_rate": 0.0005404325886285927,
+      "loss": 2.3599,
+      "step": 1069
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3165572692700196,
+      "learning_rate": 0.000539692283281637,
+      "loss": 2.2449,
+      "step": 1070
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3170907271609993,
+      "learning_rate": 0.0005389518903587017,
+      "loss": 2.3472,
+      "step": 1071
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.31577451427882536,
+      "learning_rate": 0.0005382114114933695,
+      "loss": 2.3394,
+      "step": 1072
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.30076819323798826,
+      "learning_rate": 0.0005374708483194132,
+      "loss": 2.2715,
+      "step": 1073
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.31334952271554345,
+      "learning_rate": 0.000536730202470791,
+      "loss": 2.3085,
+      "step": 1074
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3097103122091878,
+      "learning_rate": 0.0005359894755816443,
+      "loss": 2.3387,
+      "step": 1075
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3260000860393223,
+      "learning_rate": 0.0005352486692862926,
+      "loss": 2.3972,
+      "step": 1076
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.32715639310833505,
+      "learning_rate": 0.0005345077852192307,
+      "loss": 2.3215,
+      "step": 1077
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.29989329920519897,
+      "learning_rate": 0.0005337668250151254,
+      "loss": 2.2725,
+      "step": 1078
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.31529899852940885,
+      "learning_rate": 0.0005330257903088111,
+      "loss": 2.3498,
+      "step": 1079
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.32171363146352805,
+      "learning_rate": 0.000532284682735287,
+      "loss": 2.2882,
+      "step": 1080
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.32578612318454786,
+      "learning_rate": 0.0005315435039297124,
+      "loss": 2.2995,
+      "step": 1081
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.33955606647230546,
+      "learning_rate": 0.0005308022555274046,
+      "loss": 2.2643,
+      "step": 1082
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3179519733026686,
+      "learning_rate": 0.0005300609391638336,
+      "loss": 2.3178,
+      "step": 1083
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.34470171706794805,
+      "learning_rate": 0.0005293195564746201,
+      "loss": 2.3344,
+      "step": 1084
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3340954211651353,
+      "learning_rate": 0.0005285781090955304,
+      "loss": 2.2987,
+      "step": 1085
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.33167140195620887,
+      "learning_rate": 0.0005278365986624743,
+      "loss": 2.3194,
+      "step": 1086
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3305509463520751,
+      "learning_rate": 0.0005270950268115001,
+      "loss": 2.3397,
+      "step": 1087
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.32231793541979536,
+      "learning_rate": 0.0005263533951787919,
+      "loss": 2.2676,
+      "step": 1088
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.30776529580782774,
+      "learning_rate": 0.000525611705400666,
+      "loss": 2.2599,
+      "step": 1089
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4392445908329955,
+      "learning_rate": 0.0005248699591135664,
+      "loss": 2.4093,
+      "step": 1090
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3154194540907418,
+      "learning_rate": 0.0005241281579540618,
+      "loss": 2.3645,
+      "step": 1091
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.31555683189059786,
+      "learning_rate": 0.0005233863035588427,
+      "loss": 2.2748,
+      "step": 1092
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.31555936297959236,
+      "learning_rate": 0.0005226443975647161,
+      "loss": 2.2836,
+      "step": 1093
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3474921396079148,
+      "learning_rate": 0.0005219024416086036,
+      "loss": 2.286,
+      "step": 1094
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.31966951572339947,
+      "learning_rate": 0.0005211604373275366,
+      "loss": 2.2555,
+      "step": 1095
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.33947989259207534,
+      "learning_rate": 0.0005204183863586533,
+      "loss": 2.3319,
+      "step": 1096
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.4358573334238069,
+      "learning_rate": 0.0005196762903391951,
+      "loss": 2.4403,
+      "step": 1097
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.336132179555079,
+      "learning_rate": 0.0005189341509065023,
+      "loss": 2.2586,
+      "step": 1098
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3920056776909444,
+      "learning_rate": 0.0005181919696980112,
+      "loss": 2.3691,
+      "step": 1099
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.32322115294972753,
+      "learning_rate": 0.0005174497483512506,
+      "loss": 2.1954,
+      "step": 1100
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.3853542674027668,
+      "learning_rate": 0.0005167074885038374,
+      "loss": 2.3437,
+      "step": 1101
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.32608259883211516,
+      "learning_rate": 0.0005159651917934735,
+      "loss": 2.3084,
+      "step": 1102
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.2984591819402342,
+      "learning_rate": 0.0005152228598579428,
+      "loss": 2.309,
+      "step": 1103
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3295725339971468,
+      "learning_rate": 0.000514480494335106,
+      "loss": 2.342,
+      "step": 1104
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3245502728549732,
+      "learning_rate": 0.0005137380968628983,
+      "loss": 2.2524,
+      "step": 1105
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3127053723443028,
+      "learning_rate": 0.0005129956690793255,
+      "loss": 2.2936,
+      "step": 1106
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.30920189679731763,
+      "learning_rate": 0.0005122532126224601,
+      "loss": 2.2986,
+      "step": 1107
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.30908295944872727,
+      "learning_rate": 0.0005115107291304378,
+      "loss": 2.2788,
+      "step": 1108
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.31876095394817217,
+      "learning_rate": 0.0005107682202414544,
+      "loss": 2.2556,
+      "step": 1109
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.31657870994780885,
+      "learning_rate": 0.0005100256875937613,
+      "loss": 2.2694,
+      "step": 1110
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3194637138053709,
+      "learning_rate": 0.0005092831328256625,
+      "loss": 2.2723,
+      "step": 1111
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3373244548422759,
+      "learning_rate": 0.0005085405575755105,
+      "loss": 2.296,
+      "step": 1112
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3241341644088187,
+      "learning_rate": 0.0005077979634817034,
+      "loss": 2.3972,
+      "step": 1113
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.32637185200103297,
+      "learning_rate": 0.0005070553521826808,
+      "loss": 2.3204,
+      "step": 1114
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3145037982817732,
+      "learning_rate": 0.00050631272531692,
+      "loss": 2.3179,
+      "step": 1115
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.32020181510474477,
+      "learning_rate": 0.0005055700845229327,
+      "loss": 2.2413,
+      "step": 1116
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.32581042640848923,
+      "learning_rate": 0.000504827431439262,
+      "loss": 2.2698,
+      "step": 1117
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.32521875878591894,
+      "learning_rate": 0.000504084767704477,
+      "loss": 2.2604,
+      "step": 1118
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3436934897085911,
+      "learning_rate": 0.0005033420949571712,
+      "loss": 2.3327,
+      "step": 1119
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3145772878776768,
+      "learning_rate": 0.0005025994148359574,
+      "loss": 2.2988,
+      "step": 1120
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.36029347741235895,
+      "learning_rate": 0.0005018567289794651,
+      "loss": 2.271,
+      "step": 1121
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.312303827529939,
+      "learning_rate": 0.0005011140390263362,
+      "loss": 2.2618,
+      "step": 1122
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.3331233263741695,
+      "learning_rate": 0.0005003713466152218,
+      "loss": 2.2737,
+      "step": 1123
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3394079360387327,
+      "learning_rate": 0.0004996286533847783,
+      "loss": 2.3598,
+      "step": 1124
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.32906088473712725,
+      "learning_rate": 0.000498885960973664,
+      "loss": 2.2496,
+      "step": 1125
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.32096221156151,
+      "learning_rate": 0.000498143271020535,
+      "loss": 2.3572,
+      "step": 1126
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.33035078890586345,
+      "learning_rate": 0.0004974005851640428,
+      "loss": 2.2943,
+      "step": 1127
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.33196834964200467,
+      "learning_rate": 0.000496657905042829,
+      "loss": 2.2563,
+      "step": 1128
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.33340198763117945,
+      "learning_rate": 0.0004959152322955232,
+      "loss": 2.3344,
+      "step": 1129
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3048295246909787,
+      "learning_rate": 0.0004951725685607382,
+      "loss": 2.2899,
+      "step": 1130
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3436517800670356,
+      "learning_rate": 0.0004944299154770673,
+      "loss": 2.3966,
+      "step": 1131
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3295357972147818,
+      "learning_rate": 0.0004936872746830802,
+      "loss": 2.3568,
+      "step": 1132
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.32366413992257076,
+      "learning_rate": 0.0004929446478173195,
+      "loss": 2.2979,
+      "step": 1133
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.38079019512381906,
+      "learning_rate": 0.0004922020365182968,
+      "loss": 2.2454,
+      "step": 1134
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3378385452343894,
+      "learning_rate": 0.0004914594424244897,
+      "loss": 2.3333,
+      "step": 1135
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.32322496097631237,
+      "learning_rate": 0.0004907168671743376,
+      "loss": 2.3301,
+      "step": 1136
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3511777615258607,
+      "learning_rate": 0.0004899743124062387,
+      "loss": 2.3338,
+      "step": 1137
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3488322226256202,
+      "learning_rate": 0.0004892317797585456,
+      "loss": 2.3758,
+      "step": 1138
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.32930252232942586,
+      "learning_rate": 0.0004884892708695623,
+      "loss": 2.3063,
+      "step": 1139
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3081844138169794,
+      "learning_rate": 0.0004877467873775402,
+      "loss": 2.3843,
+      "step": 1140
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.33026538125406957,
+      "learning_rate": 0.00048700433092067473,
+      "loss": 2.1583,
+      "step": 1141
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.329169643845512,
+      "learning_rate": 0.0004862619031371019,
+      "loss": 2.2956,
+      "step": 1142
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.33425871618205194,
+      "learning_rate": 0.0004855195056648942,
+      "loss": 2.3605,
+      "step": 1143
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.32100454258732236,
+      "learning_rate": 0.00048477714014205734,
+      "loss": 2.3359,
+      "step": 1144
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.3273395846224853,
+      "learning_rate": 0.00048403480820652644,
+      "loss": 2.3045,
+      "step": 1145
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.32162321181826836,
+      "learning_rate": 0.0004832925114961629,
+      "loss": 2.3341,
+      "step": 1146
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3098113481738721,
+      "learning_rate": 0.0004825502516487497,
+      "loss": 2.2856,
+      "step": 1147
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.32193505318810145,
+      "learning_rate": 0.00048180803030198896,
+      "loss": 2.3312,
+      "step": 1148
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3151052293347772,
+      "learning_rate": 0.0004810658490934979,
+      "loss": 2.2719,
+      "step": 1149
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.36319259050503877,
+      "learning_rate": 0.000480323709660805,
+      "loss": 2.349,
+      "step": 1150
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3062815432626108,
+      "learning_rate": 0.0004795816136413467,
+      "loss": 2.2602,
+      "step": 1151
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3394558287442001,
+      "learning_rate": 0.00047883956267246353,
+      "loss": 2.3693,
+      "step": 1152
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.33084088099819214,
+      "learning_rate": 0.00047809755839139657,
+      "loss": 2.3731,
+      "step": 1153
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3225457287901387,
+      "learning_rate": 0.0004773556024352841,
+      "loss": 2.3676,
+      "step": 1154
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.32095825540212325,
+      "learning_rate": 0.00047661369644115754,
+      "loss": 2.3587,
+      "step": 1155
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.32685870296917446,
+      "learning_rate": 0.0004758718420459383,
+      "loss": 2.3194,
+      "step": 1156
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.31635850706130747,
+      "learning_rate": 0.0004751300408864339,
+      "loss": 2.2106,
+      "step": 1157
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3348795183749739,
+      "learning_rate": 0.00047438829459933414,
+      "loss": 2.2664,
+      "step": 1158
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3208909536193403,
+      "learning_rate": 0.0004736466048212082,
+      "loss": 2.287,
+      "step": 1159
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.313311319058538,
+      "learning_rate": 0.0004729049731885002,
+      "loss": 2.283,
+      "step": 1160
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.30482319513631123,
+      "learning_rate": 0.000472163401337526,
+      "loss": 2.1986,
+      "step": 1161
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.3176750027748306,
+      "learning_rate": 0.00047142189090446985,
+      "loss": 2.2148,
+      "step": 1162
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.307915113584847,
+      "learning_rate": 0.0004706804435253802,
+      "loss": 2.2622,
+      "step": 1163
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.31034458646619356,
+      "learning_rate": 0.0004699390608361665,
+      "loss": 2.2727,
+      "step": 1164
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.33078457218439944,
+      "learning_rate": 0.0004691977444725955,
+      "loss": 2.3075,
+      "step": 1165
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.31440853386880796,
+      "learning_rate": 0.0004684564960702877,
+      "loss": 2.2575,
+      "step": 1166
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3106226883318881,
+      "learning_rate": 0.0004677153172647131,
+      "loss": 2.2305,
+      "step": 1167
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3122104268124436,
+      "learning_rate": 0.00046697420969118894,
+      "loss": 2.3136,
+      "step": 1168
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.31896108221407193,
+      "learning_rate": 0.00046623317498487466,
+      "loss": 2.2417,
+      "step": 1169
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3008215828196148,
+      "learning_rate": 0.0004654922147807694,
+      "loss": 2.3072,
+      "step": 1170
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.31253057794560224,
+      "learning_rate": 0.00046475133071370757,
+      "loss": 2.3912,
+      "step": 1171
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3159429776216832,
+      "learning_rate": 0.00046401052441835574,
+      "loss": 2.1996,
+      "step": 1172
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3052394447477366,
+      "learning_rate": 0.000463269797529209,
+      "loss": 2.1804,
+      "step": 1173
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3304258377578447,
+      "learning_rate": 0.00046252915168058697,
+      "loss": 2.3275,
+      "step": 1174
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.33706507328947943,
+      "learning_rate": 0.0004617885885066305,
+      "loss": 2.2476,
+      "step": 1175
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.31787577881819856,
+      "learning_rate": 0.0004610481096412984,
+      "loss": 2.2792,
+      "step": 1176
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3287582657117998,
+      "learning_rate": 0.000460307716718363,
+      "loss": 2.3384,
+      "step": 1177
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.32291538334433845,
+      "learning_rate": 0.0004595674113714074,
+      "loss": 2.3143,
+      "step": 1178
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.31359727162327017,
+      "learning_rate": 0.0004588271952338212,
+      "loss": 2.3018,
+      "step": 1179
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3139285353128909,
+      "learning_rate": 0.00045808706993879714,
+      "loss": 2.3346,
+      "step": 1180
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.29625116601695484,
+      "learning_rate": 0.00045734703711932767,
+      "loss": 2.2577,
+      "step": 1181
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.31771375379651506,
+      "learning_rate": 0.0004566070984082013,
+      "loss": 2.274,
+      "step": 1182
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.32272064063681155,
+      "learning_rate": 0.00045586725543799865,
+      "loss": 2.2257,
+      "step": 1183
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.30905036117341317,
+      "learning_rate": 0.00045512750984108937,
+      "loss": 2.2575,
+      "step": 1184
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.3161223703411306,
+      "learning_rate": 0.000454387863249628,
+      "loss": 2.344,
+      "step": 1185
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.31695439116598845,
+      "learning_rate": 0.00045364831729555096,
+      "loss": 2.2991,
+      "step": 1186
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.33151404299392623,
+      "learning_rate": 0.0004529088736105721,
+      "loss": 2.2968,
+      "step": 1187
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.35185580114449155,
+      "learning_rate": 0.0004521695338261802,
+      "loss": 2.3887,
+      "step": 1188
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.307530438988675,
+      "learning_rate": 0.0004514302995736344,
+      "loss": 2.2682,
+      "step": 1189
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3138322889011514,
+      "learning_rate": 0.0004506911724839613,
+      "loss": 2.3285,
+      "step": 1190
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.34266176485986594,
+      "learning_rate": 0.0004499521541879508,
+      "loss": 2.3022,
+      "step": 1191
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.33462634499155575,
+      "learning_rate": 0.00044921324631615303,
+      "loss": 2.2574,
+      "step": 1192
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.307232303221464,
+      "learning_rate": 0.0004484744504988742,
+      "loss": 2.3582,
+      "step": 1193
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3233105840878531,
+      "learning_rate": 0.00044773576836617336,
+      "loss": 2.2642,
+      "step": 1194
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.33723384625183184,
+      "learning_rate": 0.0004469972015478588,
+      "loss": 2.3108,
+      "step": 1195
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.30611237576240674,
+      "learning_rate": 0.0004462587516734844,
+      "loss": 2.3781,
+      "step": 1196
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3177019445415829,
+      "learning_rate": 0.00044552042037234596,
+      "loss": 2.25,
+      "step": 1197
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.31865099279724857,
+      "learning_rate": 0.00044478220927347774,
+      "loss": 2.2986,
+      "step": 1198
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3224861824746387,
+      "learning_rate": 0.00044404412000564875,
+      "loss": 2.2725,
+      "step": 1199
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3273035202365556,
+      "learning_rate": 0.000443306154197359,
+      "loss": 2.2867,
+      "step": 1200
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.32702713314732407,
+      "learning_rate": 0.00044256831347683646,
+      "loss": 2.3126,
+      "step": 1201
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3053328412405154,
+      "learning_rate": 0.0004418305994720328,
+      "loss": 2.2677,
+      "step": 1202
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3320157596489124,
+      "learning_rate": 0.0004410930138106203,
+      "loss": 2.3218,
+      "step": 1203
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3089999230287238,
+      "learning_rate": 0.000440355558119988,
+      "loss": 2.2946,
+      "step": 1204
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.2981324835161221,
+      "learning_rate": 0.00043961823402723814,
+      "loss": 2.3123,
+      "step": 1205
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3067948196709865,
+      "learning_rate": 0.0004388810431591829,
+      "loss": 2.2271,
+      "step": 1206
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.29269484105047505,
+      "learning_rate": 0.0004381439871423398,
+      "loss": 2.246,
+      "step": 1207
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3035226589580223,
+      "learning_rate": 0.00043740706760292966,
+      "loss": 2.3617,
+      "step": 1208
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.32095300099389745,
+      "learning_rate": 0.0004366702861668716,
+      "loss": 2.2907,
+      "step": 1209
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.3059692381404876,
+      "learning_rate": 0.00043593364445978036,
+      "loss": 2.2269,
+      "step": 1210
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.32315488160311495,
+      "learning_rate": 0.0004351971441069622,
+      "loss": 2.3025,
+      "step": 1211
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.32908660267349993,
+      "learning_rate": 0.0004344607867334116,
+      "loss": 2.31,
+      "step": 1212
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3144196825198507,
+      "learning_rate": 0.00043372457396380766,
+      "loss": 2.2419,
+      "step": 1213
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.34335169093837686,
+      "learning_rate": 0.00043298850742251013,
+      "loss": 2.2115,
+      "step": 1214
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3311047498285028,
+      "learning_rate": 0.0004322525887335563,
+      "loss": 2.352,
+      "step": 1215
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.31034848967576917,
+      "learning_rate": 0.00043151681952065734,
+      "loss": 2.297,
+      "step": 1216
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3341265999777463,
+      "learning_rate": 0.00043078120140719456,
+      "loss": 2.3854,
+      "step": 1217
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3136674464422948,
+      "learning_rate": 0.0004300457360162158,
+      "loss": 2.3192,
+      "step": 1218
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.32208731312800476,
+      "learning_rate": 0.0004293104249704319,
+      "loss": 2.2999,
+      "step": 1219
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.2994642121531166,
+      "learning_rate": 0.00042857526989221355,
+      "loss": 2.2821,
+      "step": 1220
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.307909519657065,
+      "learning_rate": 0.00042784027240358674,
+      "loss": 2.2494,
+      "step": 1221
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3105608741410129,
+      "learning_rate": 0.0004271054341262301,
+      "loss": 2.1973,
+      "step": 1222
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.32785522607423534,
+      "learning_rate": 0.000426370756681471,
+      "loss": 2.3562,
+      "step": 1223
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3773624091928025,
+      "learning_rate": 0.0004256362416902817,
+      "loss": 2.2718,
+      "step": 1224
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.34685349422218437,
+      "learning_rate": 0.00042490189077327637,
+      "loss": 2.2445,
+      "step": 1225
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3288623695942945,
+      "learning_rate": 0.00042416770555070703,
+      "loss": 2.275,
+      "step": 1226
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.31274948479979264,
+      "learning_rate": 0.00042343368764246,
+      "loss": 2.3815,
+      "step": 1227
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.34555462271425086,
+      "learning_rate": 0.0004226998386680524,
+      "loss": 2.2952,
+      "step": 1228
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.3210744713936831,
+      "learning_rate": 0.000421966160246629,
+      "loss": 2.3543,
+      "step": 1229
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.31355387362226816,
+      "learning_rate": 0.00042123265399695783,
+      "loss": 2.2765,
+      "step": 1230
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.30051195086572935,
+      "learning_rate": 0.0004204993215374273,
+      "loss": 2.2616,
+      "step": 1231
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.2935332746803293,
+      "learning_rate": 0.00041976616448604226,
+      "loss": 2.3919,
+      "step": 1232
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.28595824555536214,
+      "learning_rate": 0.00041903318446042076,
+      "loss": 2.3432,
+      "step": 1233
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3062361593948023,
+      "learning_rate": 0.00041830038307778984,
+      "loss": 2.3717,
+      "step": 1234
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.29579254666948573,
+      "learning_rate": 0.0004175677619549828,
+      "loss": 2.2568,
+      "step": 1235
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3145180874905641,
+      "learning_rate": 0.000416835322708435,
+      "loss": 2.3309,
+      "step": 1236
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.32231730264949193,
+      "learning_rate": 0.00041610306695418056,
+      "loss": 2.3626,
+      "step": 1237
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.31107888110432763,
+      "learning_rate": 0.0004153709963078488,
+      "loss": 2.2593,
+      "step": 1238
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.30252894634513444,
+      "learning_rate": 0.0004146391123846606,
+      "loss": 2.1968,
+      "step": 1239
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3208762561451757,
+      "learning_rate": 0.0004139074167994249,
+      "loss": 2.4103,
+      "step": 1240
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3059236115734242,
+      "learning_rate": 0.00041317591116653486,
+      "loss": 2.3171,
+      "step": 1241
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3146631584468138,
+      "learning_rate": 0.0004124445970999648,
+      "loss": 2.2495,
+      "step": 1242
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2959881575116307,
+      "learning_rate": 0.00041171347621326627,
+      "loss": 2.2746,
+      "step": 1243
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3494362891089333,
+      "learning_rate": 0.00041098255011956465,
+      "loss": 2.4492,
+      "step": 1244
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3389971498024274,
+      "learning_rate": 0.00041025182043155547,
+      "loss": 2.2376,
+      "step": 1245
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.30948583146636194,
+      "learning_rate": 0.000409521288761501,
+      "loss": 2.2269,
+      "step": 1246
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3143409677892268,
+      "learning_rate": 0.00040879095672122646,
+      "loss": 2.3444,
+      "step": 1247
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.31610115750571316,
+      "learning_rate": 0.0004080608259221167,
+      "loss": 2.3312,
+      "step": 1248
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3211197620356205,
+      "learning_rate": 0.0004073308979751126,
+      "loss": 2.2843,
+      "step": 1249
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3126612679842174,
+      "learning_rate": 0.0004066011744907074,
+      "loss": 2.3301,
+      "step": 1250
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3346660506114923,
+      "learning_rate": 0.00040587165707894326,
+      "loss": 2.3543,
+      "step": 1251
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.30432221395588865,
+      "learning_rate": 0.0004051423473494076,
+      "loss": 2.2372,
+      "step": 1252
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.32435282058247383,
+      "learning_rate": 0.0004044132469112299,
+      "loss": 2.2444,
+      "step": 1253
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3162781811669345,
+      "learning_rate": 0.00040368435737307733,
+      "loss": 2.2109,
+      "step": 1254
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3244294520658478,
+      "learning_rate": 0.00040295568034315224,
+      "loss": 2.3601,
+      "step": 1255
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3406155952152521,
+      "learning_rate": 0.0004022272174291878,
+      "loss": 2.2964,
+      "step": 1256
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.318154454154696,
+      "learning_rate": 0.0004014989702384449,
+      "loss": 2.2791,
+      "step": 1257
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3296815949717671,
+      "learning_rate": 0.00040077094037770843,
+      "loss": 2.2102,
+      "step": 1258
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3615006632503655,
+      "learning_rate": 0.0004000431294532838,
+      "loss": 2.3375,
+      "step": 1259
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.33539642800980257,
+      "learning_rate": 0.0003993155390709935,
+      "loss": 2.2565,
+      "step": 1260
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3215530621009784,
+      "learning_rate": 0.0003985881708361729,
+      "loss": 2.3481,
+      "step": 1261
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3188197192917083,
+      "learning_rate": 0.00039786102635366784,
+      "loss": 2.2125,
+      "step": 1262
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3086111678806038,
+      "learning_rate": 0.0003971341072278302,
+      "loss": 2.2066,
+      "step": 1263
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.30550294195495925,
+      "learning_rate": 0.00039640741506251457,
+      "loss": 2.204,
+      "step": 1264
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3163408227740959,
+      "learning_rate": 0.00039568095146107495,
+      "loss": 2.3018,
+      "step": 1265
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.31315610447151193,
+      "learning_rate": 0.00039495471802636096,
+      "loss": 2.2717,
+      "step": 1266
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3137621319271193,
+      "learning_rate": 0.0003942287163607145,
+      "loss": 2.3737,
+      "step": 1267
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.32388954351587956,
+      "learning_rate": 0.0003935029480659658,
+      "loss": 2.3188,
+      "step": 1268
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3185112051272327,
+      "learning_rate": 0.00039277741474343054,
+      "loss": 2.2886,
+      "step": 1269
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.31664122225994035,
+      "learning_rate": 0.0003920521179939057,
+      "loss": 2.3105,
+      "step": 1270
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.32011133950657905,
+      "learning_rate": 0.00039132705941766644,
+      "loss": 2.3417,
+      "step": 1271
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.32260077325639697,
+      "learning_rate": 0.0003906022406144624,
+      "loss": 2.3131,
+      "step": 1272
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3329153069468472,
+      "learning_rate": 0.0003898776631835143,
+      "loss": 2.2848,
+      "step": 1273
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.2959226628294338,
+      "learning_rate": 0.00038915332872350994,
+      "loss": 2.2851,
+      "step": 1274
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3426716560616369,
+      "learning_rate": 0.00038842923883260135,
+      "loss": 2.2727,
+      "step": 1275
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.32118781029738236,
+      "learning_rate": 0.00038770539510840093,
+      "loss": 2.3346,
+      "step": 1276
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3211013594020553,
+      "learning_rate": 0.00038698179914797783,
+      "loss": 2.2304,
+      "step": 1277
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3355351268632028,
+      "learning_rate": 0.0003862584525478545,
+      "loss": 2.451,
+      "step": 1278
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.30133061116299487,
+      "learning_rate": 0.00038553535690400353,
+      "loss": 2.3331,
+      "step": 1279
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3133860704167152,
+      "learning_rate": 0.00038481251381184355,
+      "loss": 2.357,
+      "step": 1280
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3052569203989218,
+      "learning_rate": 0.00038408992486623584,
+      "loss": 2.2284,
+      "step": 1281
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3118911178406095,
+      "learning_rate": 0.00038336759166148117,
+      "loss": 2.1972,
+      "step": 1282
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.30256411193228433,
+      "learning_rate": 0.0003826455157913159,
+      "loss": 2.3375,
+      "step": 1283
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.33462025633839576,
+      "learning_rate": 0.00038192369884890886,
+      "loss": 2.2784,
+      "step": 1284
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.33425857052204405,
+      "learning_rate": 0.00038120214242685723,
+      "loss": 2.1834,
+      "step": 1285
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3235245999988734,
+      "learning_rate": 0.00038048084811718373,
+      "loss": 2.3219,
+      "step": 1286
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3373551535285787,
+      "learning_rate": 0.0003797598175113327,
+      "loss": 2.24,
+      "step": 1287
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.33446051848242564,
+      "learning_rate": 0.0003790390522001662,
+      "loss": 2.4137,
+      "step": 1288
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3018510305231225,
+      "learning_rate": 0.0003783185537739615,
+      "loss": 2.2011,
+      "step": 1289
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3112519429049824,
+      "learning_rate": 0.00037759832382240697,
+      "loss": 2.1858,
+      "step": 1290
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3199387625171417,
+      "learning_rate": 0.00037687836393459826,
+      "loss": 2.3541,
+      "step": 1291
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3253040490362159,
+      "learning_rate": 0.0003761586756990354,
+      "loss": 2.285,
+      "step": 1292
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3606840326403277,
+      "learning_rate": 0.0003754392607036191,
+      "loss": 2.3381,
+      "step": 1293
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3150336116130403,
+      "learning_rate": 0.0003747201205356472,
+      "loss": 2.1747,
+      "step": 1294
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.3066328652472098,
+      "learning_rate": 0.0003740012567818111,
+      "loss": 2.3313,
+      "step": 1295
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.30183856975043666,
+      "learning_rate": 0.0003732826710281922,
+      "loss": 2.2353,
+      "step": 1296
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.305820483009359,
+      "learning_rate": 0.0003725643648602588,
+      "loss": 2.2337,
+      "step": 1297
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.318046384643626,
+      "learning_rate": 0.0003718463398628621,
+      "loss": 2.3367,
+      "step": 1298
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3048802839142257,
+      "learning_rate": 0.0003711285976202331,
+      "loss": 2.2565,
+      "step": 1299
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.32543527472118045,
+      "learning_rate": 0.0003704111397159787,
+      "loss": 2.2799,
+      "step": 1300
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3298775644219932,
+      "learning_rate": 0.0003696939677330788,
+      "loss": 2.2965,
+      "step": 1301
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3165043946033763,
+      "learning_rate": 0.00036897708325388213,
+      "loss": 2.3024,
+      "step": 1302
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6097870086219918,
+      "learning_rate": 0.0003682604878601034,
+      "loss": 2.2331,
+      "step": 1303
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.34267900929671186,
+      "learning_rate": 0.000367544183132819,
+      "loss": 2.3526,
+      "step": 1304
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.33443944314557045,
+      "learning_rate": 0.0003668281706524645,
+      "loss": 2.2582,
+      "step": 1305
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.2991886787684565,
+      "learning_rate": 0.0003661124519988304,
+      "loss": 2.2751,
+      "step": 1306
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3132626888225301,
+      "learning_rate": 0.00036539702875105893,
+      "loss": 2.374,
+      "step": 1307
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3078105846607079,
+      "learning_rate": 0.0003646819024876406,
+      "loss": 2.2607,
+      "step": 1308
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3026138105458148,
+      "learning_rate": 0.0003639670747864107,
+      "loss": 2.3211,
+      "step": 1309
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.36333055651976026,
+      "learning_rate": 0.00036325254722454584,
+      "loss": 2.2476,
+      "step": 1310
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3336284680362645,
+      "learning_rate": 0.00036253832137855997,
+      "loss": 2.317,
+      "step": 1311
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.35644745983020276,
+      "learning_rate": 0.00036182439882430183,
+      "loss": 2.34,
+      "step": 1312
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.31195714536615704,
+      "learning_rate": 0.00036111078113695096,
+      "loss": 2.2796,
+      "step": 1313
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.30390336276484775,
+      "learning_rate": 0.0003603974698910139,
+      "loss": 2.3105,
+      "step": 1314
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3250683007389042,
+      "learning_rate": 0.0003596844666603214,
+      "loss": 2.318,
+      "step": 1315
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.30894496217652573,
+      "learning_rate": 0.0003589717730180245,
+      "loss": 2.2956,
+      "step": 1316
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3200029152393744,
+      "learning_rate": 0.00035825939053659117,
+      "loss": 2.4017,
+      "step": 1317
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3180435778985396,
+      "learning_rate": 0.00035754732078780273,
+      "loss": 2.3471,
+      "step": 1318
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.33602784865485985,
+      "learning_rate": 0.00035683556534275076,
+      "loss": 2.2434,
+      "step": 1319
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3043782944580565,
+      "learning_rate": 0.00035612412577183303,
+      "loss": 2.2187,
+      "step": 1320
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3415169056707833,
+      "learning_rate": 0.00035541300364475063,
+      "loss": 2.3098,
+      "step": 1321
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3100488552293367,
+      "learning_rate": 0.0003547022005305043,
+      "loss": 2.1779,
+      "step": 1322
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3306622859797888,
+      "learning_rate": 0.0003539917179973907,
+      "loss": 2.317,
+      "step": 1323
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.31131729537147274,
+      "learning_rate": 0.00035328155761299917,
+      "loss": 2.2131,
+      "step": 1324
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.31760849780516814,
+      "learning_rate": 0.0003525717209442085,
+      "loss": 2.3283,
+      "step": 1325
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.323813967878436,
+      "learning_rate": 0.00035186220955718306,
+      "loss": 2.3012,
+      "step": 1326
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.31248513270468303,
+      "learning_rate": 0.0003511530250173696,
+      "loss": 2.4097,
+      "step": 1327
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.31764346316256503,
+      "learning_rate": 0.00035044416888949364,
+      "loss": 2.3051,
+      "step": 1328
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3155034572723287,
+      "learning_rate": 0.0003497356427375562,
+      "loss": 2.3257,
+      "step": 1329
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.30273850341476727,
+      "learning_rate": 0.00034902744812483034,
+      "loss": 2.3284,
+      "step": 1330
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.31425529466149305,
+      "learning_rate": 0.00034831958661385714,
+      "loss": 2.3358,
+      "step": 1331
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3250833654994443,
+      "learning_rate": 0.0003476120597664434,
+      "loss": 2.2445,
+      "step": 1332
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.36926006560156727,
+      "learning_rate": 0.00034690486914365704,
+      "loss": 2.3364,
+      "step": 1333
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3215928459119012,
+      "learning_rate": 0.00034619801630582435,
+      "loss": 2.2814,
+      "step": 1334
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.2967946450796327,
+      "learning_rate": 0.00034549150281252633,
+      "loss": 2.2191,
+      "step": 1335
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.32824894256934917,
+      "learning_rate": 0.0003447853302225952,
+      "loss": 2.3424,
+      "step": 1336
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3313946567086212,
+      "learning_rate": 0.00034407950009411126,
+      "loss": 2.1894,
+      "step": 1337
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.32564417184575134,
+      "learning_rate": 0.00034337401398439873,
+      "loss": 2.2653,
+      "step": 1338
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.3244300385168141,
+      "learning_rate": 0.00034266887345002305,
+      "loss": 2.2563,
+      "step": 1339
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.34024565806051243,
+      "learning_rate": 0.0003419640800467874,
+      "loss": 2.2828,
+      "step": 1340
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.32710948092713704,
+      "learning_rate": 0.0003412596353297288,
+      "loss": 2.219,
+      "step": 1341
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3222186266952433,
+      "learning_rate": 0.00034055554085311493,
+      "loss": 2.3054,
+      "step": 1342
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3180793937641424,
+      "learning_rate": 0.00033985179817044105,
+      "loss": 2.2555,
+      "step": 1343
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3354887002260146,
+      "learning_rate": 0.0003391484088344257,
+      "loss": 2.3397,
+      "step": 1344
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.2978356897497722,
+      "learning_rate": 0.00033844537439700807,
+      "loss": 2.2768,
+      "step": 1345
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3187089877221622,
+      "learning_rate": 0.00033774269640934445,
+      "loss": 2.2864,
+      "step": 1346
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3104877258181367,
+      "learning_rate": 0.0003370403764218045,
+      "loss": 2.2363,
+      "step": 1347
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.2945323489662352,
+      "learning_rate": 0.000336338415983968,
+      "loss": 2.2801,
+      "step": 1348
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.30812105426740827,
+      "learning_rate": 0.00033563681664462155,
+      "loss": 2.2254,
+      "step": 1349
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.299448760318697,
+      "learning_rate": 0.000334935579951755,
+      "loss": 2.2625,
+      "step": 1350
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3340245008429087,
+      "learning_rate": 0.0003342347074525578,
+      "loss": 2.3106,
+      "step": 1351
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.32998023485811,
+      "learning_rate": 0.0003335342006934161,
+      "loss": 2.3222,
+      "step": 1352
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.31874201308715794,
+      "learning_rate": 0.00033283406121990914,
+      "loss": 2.2463,
+      "step": 1353
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3103904949520432,
+      "learning_rate": 0.0003321342905768057,
+      "loss": 2.2869,
+      "step": 1354
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3246632258961078,
+      "learning_rate": 0.00033143489030806086,
+      "loss": 2.3296,
+      "step": 1355
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3108336120559299,
+      "learning_rate": 0.00033073586195681227,
+      "loss": 2.3815,
+      "step": 1356
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.32806276607565155,
+      "learning_rate": 0.00033003720706537736,
+      "loss": 2.185,
+      "step": 1357
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3301234170099205,
+      "learning_rate": 0.0003293389271752492,
+      "loss": 2.2374,
+      "step": 1358
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3244675744969301,
+      "learning_rate": 0.00032864102382709374,
+      "loss": 2.2673,
+      "step": 1359
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3276951367155237,
+      "learning_rate": 0.000327943498560746,
+      "loss": 2.2357,
+      "step": 1360
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3035399850574307,
+      "learning_rate": 0.00032724635291520694,
+      "loss": 2.2582,
+      "step": 1361
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3111107681841673,
+      "learning_rate": 0.00032654958842863967,
+      "loss": 2.3026,
+      "step": 1362
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3120075584703583,
+      "learning_rate": 0.0003258532066383667,
+      "loss": 2.3036,
+      "step": 1363
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3244057005324511,
+      "learning_rate": 0.000325157209080866,
+      "loss": 2.2802,
+      "step": 1364
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3071974906295936,
+      "learning_rate": 0.00032446159729176743,
+      "loss": 2.2329,
+      "step": 1365
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3504860387979673,
+      "learning_rate": 0.0003237663728058502,
+      "loss": 2.4544,
+      "step": 1366
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3099825053410974,
+      "learning_rate": 0.0003230715371570389,
+      "loss": 2.3156,
+      "step": 1367
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.32092821603359023,
+      "learning_rate": 0.00032237709187839996,
+      "loss": 2.3099,
+      "step": 1368
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3062181517299731,
+      "learning_rate": 0.0003216830385021388,
+      "loss": 2.2367,
+      "step": 1369
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3087276958571579,
+      "learning_rate": 0.0003209893785595959,
+      "loss": 2.2006,
+      "step": 1370
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.2974749575514414,
+      "learning_rate": 0.00032029611358124366,
+      "loss": 2.1538,
+      "step": 1371
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3098384388093435,
+      "learning_rate": 0.00031960324509668336,
+      "loss": 2.3195,
+      "step": 1372
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3226331023400114,
+      "learning_rate": 0.0003189107746346412,
+      "loss": 2.2969,
+      "step": 1373
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3346189525898143,
+      "learning_rate": 0.0003182187037229653,
+      "loss": 2.2553,
+      "step": 1374
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3063906816980932,
+      "learning_rate": 0.0003175270338886221,
+      "loss": 2.2462,
+      "step": 1375
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3297753388544789,
+      "learning_rate": 0.00031683576665769345,
+      "loss": 2.3277,
+      "step": 1376
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.31410174488797327,
+      "learning_rate": 0.0003161449035553724,
+      "loss": 2.3519,
+      "step": 1377
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3204751236829418,
+      "learning_rate": 0.00031545444610596077,
+      "loss": 2.2362,
+      "step": 1378
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.31186077531571926,
+      "learning_rate": 0.000314764395832865,
+      "loss": 2.2634,
+      "step": 1379
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3214301551126915,
+      "learning_rate": 0.0003140747542585934,
+      "loss": 2.3348,
+      "step": 1380
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3030402390856347,
+      "learning_rate": 0.00031338552290475266,
+      "loss": 2.3344,
+      "step": 1381
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.3183115769664388,
+      "learning_rate": 0.00031269670329204396,
+      "loss": 2.3412,
+      "step": 1382
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.2971604957151576,
+      "learning_rate": 0.0003120082969402604,
+      "loss": 2.2286,
+      "step": 1383
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.299108830655927,
+      "learning_rate": 0.00031132030536828314,
+      "loss": 2.2671,
+      "step": 1384
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.29727742104212845,
+      "learning_rate": 0.00031063273009407805,
+      "loss": 2.1779,
+      "step": 1385
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3016529491440415,
+      "learning_rate": 0.00030994557263469265,
+      "loss": 2.2461,
+      "step": 1386
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3357654332889309,
+      "learning_rate": 0.0003092588345062526,
+      "loss": 2.3903,
+      "step": 1387
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3221215419391709,
+      "learning_rate": 0.0003085725172239582,
+      "loss": 2.2807,
+      "step": 1388
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.302531843002309,
+      "learning_rate": 0.0003078866223020815,
+      "loss": 2.2306,
+      "step": 1389
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3207953355505163,
+      "learning_rate": 0.0003072011512539624,
+      "loss": 2.3695,
+      "step": 1390
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3267725232999686,
+      "learning_rate": 0.00030651610559200574,
+      "loss": 2.3157,
+      "step": 1391
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3205139600965797,
+      "learning_rate": 0.00030583148682767757,
+      "loss": 2.2864,
+      "step": 1392
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.314234231623635,
+      "learning_rate": 0.00030514729647150243,
+      "loss": 2.3091,
+      "step": 1393
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3207211692797031,
+      "learning_rate": 0.0003044635360330592,
+      "loss": 2.2833,
+      "step": 1394
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.32095034824399793,
+      "learning_rate": 0.00030378020702097845,
+      "loss": 2.3498,
+      "step": 1395
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3423758304295419,
+      "learning_rate": 0.000303097310942939,
+      "loss": 2.2389,
+      "step": 1396
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3226346032282319,
+      "learning_rate": 0.0003024148493056641,
+      "loss": 2.3132,
+      "step": 1397
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3414786792303411,
+      "learning_rate": 0.00030173282361491865,
+      "loss": 2.3048,
+      "step": 1398
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3062634793913699,
+      "learning_rate": 0.0003010512353755057,
+      "loss": 2.2612,
+      "step": 1399
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3101597923293207,
+      "learning_rate": 0.00030037008609126313,
+      "loss": 2.2848,
+      "step": 1400
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3320919928451226,
+      "learning_rate": 0.0002996893772650602,
+      "loss": 2.1852,
+      "step": 1401
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.31891161969175874,
+      "learning_rate": 0.0002990091103987945,
+      "loss": 2.2898,
+      "step": 1402
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.3145697224598552,
+      "learning_rate": 0.0002983292869933886,
+      "loss": 2.2565,
+      "step": 1403
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.31459152265408574,
+      "learning_rate": 0.0002976499085487862,
+      "loss": 2.2828,
+      "step": 1404
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.31828834938882405,
+      "learning_rate": 0.00029697097656394963,
+      "loss": 2.1628,
+      "step": 1405
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.32763647447452593,
+      "learning_rate": 0.00029629249253685595,
+      "loss": 2.3138,
+      "step": 1406
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3256145304428704,
+      "learning_rate": 0.00029561445796449416,
+      "loss": 2.2799,
+      "step": 1407
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.32189980938229407,
+      "learning_rate": 0.0002949368743428612,
+      "loss": 2.2741,
+      "step": 1408
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.32073957987186436,
+      "learning_rate": 0.0002942597431669593,
+      "loss": 2.226,
+      "step": 1409
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.31383899555698386,
+      "learning_rate": 0.0002935830659307924,
+      "loss": 2.2137,
+      "step": 1410
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.30239829436983756,
+      "learning_rate": 0.0002929068441273629,
+      "loss": 2.3367,
+      "step": 1411
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.30458134995729763,
+      "learning_rate": 0.0002922310792486681,
+      "loss": 2.2752,
+      "step": 1412
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3193855509240868,
+      "learning_rate": 0.00029155577278569745,
+      "loss": 2.2112,
+      "step": 1413
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3243848553535439,
+      "learning_rate": 0.00029088092622842895,
+      "loss": 2.3594,
+      "step": 1414
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.32252104726820896,
+      "learning_rate": 0.00029020654106582544,
+      "loss": 2.3597,
+      "step": 1415
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.31464995228236714,
+      "learning_rate": 0.0002895326187858326,
+      "loss": 2.1776,
+      "step": 1416
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3169959139372564,
+      "learning_rate": 0.00028885916087537377,
+      "loss": 2.2908,
+      "step": 1417
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.30228687149880507,
+      "learning_rate": 0.00028818616882034877,
+      "loss": 2.2589,
+      "step": 1418
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3037946126765073,
+      "learning_rate": 0.0002875136441056286,
+      "loss": 2.1906,
+      "step": 1419
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.32649768900889653,
+      "learning_rate": 0.000286841588215054,
+      "loss": 2.28,
+      "step": 1420
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.31418514328960356,
+      "learning_rate": 0.0002861700026314308,
+      "loss": 2.2685,
+      "step": 1421
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.31110836503072564,
+      "learning_rate": 0.00028549888883652686,
+      "loss": 2.3908,
+      "step": 1422
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.32299852241209137,
+      "learning_rate": 0.00028482824831107,
+      "loss": 2.3459,
+      "step": 1423
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.32076837642358424,
+      "learning_rate": 0.000284158082534743,
+      "loss": 2.2111,
+      "step": 1424
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3223993360461135,
+      "learning_rate": 0.00028348839298618177,
+      "loss": 2.3065,
+      "step": 1425
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3435655681290865,
+      "learning_rate": 0.0002828191811429709,
+      "loss": 2.3283,
+      "step": 1426
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.318823993899082,
+      "learning_rate": 0.00028215044848164164,
+      "loss": 2.1881,
+      "step": 1427
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.3183943343046477,
+      "learning_rate": 0.00028148219647766747,
+      "loss": 2.266,
+      "step": 1428
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.31710100282983855,
+      "learning_rate": 0.00028081442660546124,
+      "loss": 2.2805,
+      "step": 1429
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3119858960589159,
+      "learning_rate": 0.0002801471403383728,
+      "loss": 2.1936,
+      "step": 1430
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3136350138253329,
+      "learning_rate": 0.00027948033914868415,
+      "loss": 2.2467,
+      "step": 1431
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3123854817114685,
+      "learning_rate": 0.00027881402450760775,
+      "loss": 2.1653,
+      "step": 1432
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3124487517751605,
+      "learning_rate": 0.00027814819788528165,
+      "loss": 2.2286,
+      "step": 1433
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.313484540255671,
+      "learning_rate": 0.00027748286075076836,
+      "loss": 2.25,
+      "step": 1434
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3006923294258501,
+      "learning_rate": 0.00027681801457204937,
+      "loss": 2.2131,
+      "step": 1435
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3474433715015837,
+      "learning_rate": 0.00027615366081602306,
+      "loss": 2.3066,
+      "step": 1436
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3038772216996293,
+      "learning_rate": 0.0002754898009485021,
+      "loss": 2.3749,
+      "step": 1437
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3299365513753331,
+      "learning_rate": 0.0002748264364342085,
+      "loss": 2.2362,
+      "step": 1438
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3144130032408834,
+      "learning_rate": 0.00027416356873677204,
+      "loss": 2.3104,
+      "step": 1439
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.33267218593477,
+      "learning_rate": 0.0002735011993187258,
+      "loss": 2.2602,
+      "step": 1440
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3157018388042273,
+      "learning_rate": 0.0002728393296415042,
+      "loss": 2.3431,
+      "step": 1441
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.31407775499981905,
+      "learning_rate": 0.00027217796116543817,
+      "loss": 2.2603,
+      "step": 1442
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3215542417224039,
+      "learning_rate": 0.0002715170953497532,
+      "loss": 2.254,
+      "step": 1443
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3096263794603709,
+      "learning_rate": 0.00027085673365256614,
+      "loss": 2.2818,
+      "step": 1444
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.33118949711767776,
+      "learning_rate": 0.00027019687753088075,
+      "loss": 2.2402,
+      "step": 1445
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.32690251603230314,
+      "learning_rate": 0.00026953752844058597,
+      "loss": 2.3522,
+      "step": 1446
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3039857786964638,
+      "learning_rate": 0.0002688786878364516,
+      "loss": 2.2861,
+      "step": 1447
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.30887570760118227,
+      "learning_rate": 0.00026822035717212597,
+      "loss": 2.3049,
+      "step": 1448
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3204614509202105,
+      "learning_rate": 0.00026756253790013193,
+      "loss": 2.291,
+      "step": 1449
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.30230330316386533,
+      "learning_rate": 0.0002669052314718641,
+      "loss": 2.3425,
+      "step": 1450
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.30742161952402874,
+      "learning_rate": 0.0002662484393375855,
+      "loss": 2.3069,
+      "step": 1451
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3142586698759257,
+      "learning_rate": 0.00026559216294642446,
+      "loss": 2.3131,
+      "step": 1452
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.29798767347459754,
+      "learning_rate": 0.0002649364037463718,
+      "loss": 2.346,
+      "step": 1453
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3045676682854831,
+      "learning_rate": 0.0002642811631842764,
+      "loss": 2.2672,
+      "step": 1454
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3262374325267991,
+      "learning_rate": 0.0002636264427058439,
+      "loss": 2.232,
+      "step": 1455
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3218971138651971,
+      "learning_rate": 0.00026297224375563123,
+      "loss": 2.3159,
+      "step": 1456
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.309196251678486,
+      "learning_rate": 0.00026231856777704575,
+      "loss": 2.1965,
+      "step": 1457
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3113683052559942,
+      "learning_rate": 0.00026166541621234026,
+      "loss": 2.2704,
+      "step": 1458
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3309277611262742,
+      "learning_rate": 0.00026101279050261045,
+      "loss": 2.2536,
+      "step": 1459
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3144579577678016,
+      "learning_rate": 0.00026036069208779247,
+      "loss": 2.3203,
+      "step": 1460
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3154073286768212,
+      "learning_rate": 0.0002597091224066581,
+      "loss": 2.2395,
+      "step": 1461
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3098085580860028,
+      "learning_rate": 0.00025905808289681365,
+      "loss": 2.1965,
+      "step": 1462
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.31143610211072054,
+      "learning_rate": 0.0002584075749946946,
+      "loss": 2.2492,
+      "step": 1463
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3248915151835927,
+      "learning_rate": 0.00025775760013556424,
+      "loss": 2.2703,
+      "step": 1464
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3195983307830321,
+      "learning_rate": 0.0002571081597535095,
+      "loss": 2.2805,
+      "step": 1465
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.33994374092510554,
+      "learning_rate": 0.00025645925528143776,
+      "loss": 2.2554,
+      "step": 1466
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.32855615370730856,
+      "learning_rate": 0.0002558108881510747,
+      "loss": 2.2179,
+      "step": 1467
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3263289406652224,
+      "learning_rate": 0.00025516305979295963,
+      "loss": 2.318,
+      "step": 1468
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3217111525213071,
+      "learning_rate": 0.0002545157716364439,
+      "loss": 2.2686,
+      "step": 1469
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.30495301733775587,
+      "learning_rate": 0.00025386902510968624,
+      "loss": 2.2383,
+      "step": 1470
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3042540508972044,
+      "learning_rate": 0.00025322282163965095,
+      "loss": 2.3271,
+      "step": 1471
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3027218660119525,
+      "learning_rate": 0.00025257716265210384,
+      "loss": 2.2553,
+      "step": 1472
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3170221601393989,
+      "learning_rate": 0.0002519320495716091,
+      "loss": 2.377,
+      "step": 1473
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3038436967084475,
+      "learning_rate": 0.00025128748382152716,
+      "loss": 2.2519,
+      "step": 1474
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.30826442543667115,
+      "learning_rate": 0.00025064346682401016,
+      "loss": 2.308,
+      "step": 1475
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.31438249979232114,
+      "learning_rate": 0.0002500000000000001,
+      "loss": 2.1692,
+      "step": 1476
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.32161574848417407,
+      "learning_rate": 0.0002493570847692246,
+      "loss": 2.3209,
+      "step": 1477
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3156613203154543,
+      "learning_rate": 0.00024871472255019424,
+      "loss": 2.2782,
+      "step": 1478
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.32430420227660595,
+      "learning_rate": 0.00024807291476019994,
+      "loss": 2.2969,
+      "step": 1479
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.30997454803963076,
+      "learning_rate": 0.00024743166281530877,
+      "loss": 2.1456,
+      "step": 1480
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.32547757492555834,
+      "learning_rate": 0.000246790968130362,
+      "loss": 2.1948,
+      "step": 1481
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.33374032966262057,
+      "learning_rate": 0.0002461508321189706,
+      "loss": 2.1667,
+      "step": 1482
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.34251898261238944,
+      "learning_rate": 0.00024551125619351385,
+      "loss": 2.2652,
+      "step": 1483
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3262393594854259,
+      "learning_rate": 0.00024487224176513453,
+      "loss": 2.2637,
+      "step": 1484
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.32137222146294314,
+      "learning_rate": 0.0002442337902437365,
+      "loss": 2.1744,
+      "step": 1485
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3316215057205316,
+      "learning_rate": 0.0002435959030379824,
+      "loss": 2.3753,
+      "step": 1486
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3234674078630194,
+      "learning_rate": 0.00024295858155528888,
+      "loss": 2.2589,
+      "step": 1487
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3330737571528488,
+      "learning_rate": 0.00024232182720182523,
+      "loss": 2.3283,
+      "step": 1488
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3043213085537192,
+      "learning_rate": 0.00024168564138250855,
+      "loss": 2.3356,
+      "step": 1489
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.31647656939172314,
+      "learning_rate": 0.00024105002550100246,
+      "loss": 2.2329,
+      "step": 1490
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.31418137781328687,
+      "learning_rate": 0.00024041498095971254,
+      "loss": 2.2653,
+      "step": 1491
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.3077775609058911,
+      "learning_rate": 0.0002397805091597835,
+      "loss": 2.2476,
+      "step": 1492
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.296725975813188,
+      "learning_rate": 0.0002391466115010973,
+      "loss": 2.3369,
+      "step": 1493
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3095096656930156,
+      "learning_rate": 0.00023851328938226808,
+      "loss": 2.2248,
+      "step": 1494
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.30844894314571314,
+      "learning_rate": 0.00023788054420064109,
+      "loss": 2.2955,
+      "step": 1495
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.325623543928534,
+      "learning_rate": 0.00023724837735228773,
+      "loss": 2.3004,
+      "step": 1496
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3078427124944613,
+      "learning_rate": 0.00023661679023200422,
+      "loss": 2.1525,
+      "step": 1497
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3263634397674122,
+      "learning_rate": 0.00023598578423330714,
+      "loss": 2.2127,
+      "step": 1498
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3275624592569635,
+      "learning_rate": 0.00023535536074843083,
+      "loss": 2.2309,
+      "step": 1499
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.33250226064796606,
+      "learning_rate": 0.00023472552116832502,
+      "loss": 2.2962,
+      "step": 1500
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3223550109927865,
+      "learning_rate": 0.0002340962668826503,
+      "loss": 2.2127,
+      "step": 1501
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3386172766482032,
+      "learning_rate": 0.00023346759927977663,
+      "loss": 2.2959,
+      "step": 1502
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3327667716258987,
+      "learning_rate": 0.0002328395197467789,
+      "loss": 2.2667,
+      "step": 1503
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.30290385431211764,
+      "learning_rate": 0.00023221202966943515,
+      "loss": 2.2052,
+      "step": 1504
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.32788356584433853,
+      "learning_rate": 0.0002315851304322223,
+      "loss": 2.3698,
+      "step": 1505
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.30962385055602015,
+      "learning_rate": 0.0002309588234183137,
+      "loss": 2.3099,
+      "step": 1506
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.31927248464306557,
+      "learning_rate": 0.00023033311000957653,
+      "loss": 2.221,
+      "step": 1507
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.30525521383461784,
+      "learning_rate": 0.00022970799158656758,
+      "loss": 2.2419,
+      "step": 1508
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.30097022556997105,
+      "learning_rate": 0.0002290834695285316,
+      "loss": 2.3481,
+      "step": 1509
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.32493661224644954,
+      "learning_rate": 0.00022845954521339678,
+      "loss": 2.2406,
+      "step": 1510
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.29960375054610433,
+      "learning_rate": 0.0002278362200177732,
+      "loss": 2.3008,
+      "step": 1511
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.30291708183062777,
+      "learning_rate": 0.00022721349531694852,
+      "loss": 2.2996,
+      "step": 1512
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3150855640062585,
+      "learning_rate": 0.0002265913724848855,
+      "loss": 2.3022,
+      "step": 1513
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.30602394872630145,
+      "learning_rate": 0.00022596985289421946,
+      "loss": 2.1644,
+      "step": 1514
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.30459575708504466,
+      "learning_rate": 0.00022534893791625405,
+      "loss": 2.1933,
+      "step": 1515
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.34340237448893607,
+      "learning_rate": 0.00022472862892095968,
+      "loss": 2.3441,
+      "step": 1516
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3269561529104272,
+      "learning_rate": 0.00022410892727696896,
+      "loss": 2.2657,
+      "step": 1517
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3263498822580946,
+      "learning_rate": 0.0002234898343515751,
+      "loss": 2.2476,
+      "step": 1518
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3373034414982997,
+      "learning_rate": 0.00022287135151072792,
+      "loss": 2.2051,
+      "step": 1519
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3357516012036515,
+      "learning_rate": 0.00022225348011903096,
+      "loss": 2.338,
+      "step": 1520
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3304935365493637,
+      "learning_rate": 0.0002216362215397393,
+      "loss": 2.2429,
+      "step": 1521
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3194519153705695,
+      "learning_rate": 0.00022101957713475522,
+      "loss": 2.2371,
+      "step": 1522
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3328605273549015,
+      "learning_rate": 0.00022040354826462666,
+      "loss": 2.2622,
+      "step": 1523
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3291371178324196,
+      "learning_rate": 0.0002197881362885426,
+      "loss": 2.3218,
+      "step": 1524
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.31337522896988945,
+      "learning_rate": 0.0002191733425643318,
+      "loss": 2.2288,
+      "step": 1525
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.316552027751662,
+      "learning_rate": 0.00021855916844845826,
+      "loss": 2.2778,
+      "step": 1526
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.31792449071719353,
+      "learning_rate": 0.00021794561529601898,
+      "loss": 2.1436,
+      "step": 1527
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3171514428021474,
+      "learning_rate": 0.00021733268446074138,
+      "loss": 2.2062,
+      "step": 1528
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.31423227972776435,
+      "learning_rate": 0.00021672037729497917,
+      "loss": 2.3348,
+      "step": 1529
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3229724075151397,
+      "learning_rate": 0.0002161086951497106,
+      "loss": 2.2296,
+      "step": 1530
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.2967974821232448,
+      "learning_rate": 0.00021549763937453442,
+      "loss": 2.2361,
+      "step": 1531
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3130144875093324,
+      "learning_rate": 0.00021488721131766736,
+      "loss": 2.3737,
+      "step": 1532
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3037765015969037,
+      "learning_rate": 0.00021427741232594183,
+      "loss": 2.311,
+      "step": 1533
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.3170118945235017,
+      "learning_rate": 0.0002136682437448013,
+      "loss": 2.357,
+      "step": 1534
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.29914971416914915,
+      "learning_rate": 0.0002130597069182994,
+      "loss": 2.3224,
+      "step": 1535
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.32141086017775006,
+      "learning_rate": 0.0002124518031890948,
+      "loss": 2.2434,
+      "step": 1536
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.2910136206378644,
+      "learning_rate": 0.0002118445338984502,
+      "loss": 2.0984,
+      "step": 1537
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3164499702794291,
+      "learning_rate": 0.00021123790038622808,
+      "loss": 2.2616,
+      "step": 1538
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3077569212552924,
+      "learning_rate": 0.0002106319039908879,
+      "loss": 2.2759,
+      "step": 1539
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.30668628039788604,
+      "learning_rate": 0.00021002654604948412,
+      "loss": 2.2678,
+      "step": 1540
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3111952019485813,
+      "learning_rate": 0.00020942182789766172,
+      "loss": 2.2714,
+      "step": 1541
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.32405217258137414,
+      "learning_rate": 0.00020881775086965492,
+      "loss": 2.2432,
+      "step": 1542
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3185363971069462,
+      "learning_rate": 0.00020821431629828246,
+      "loss": 2.3268,
+      "step": 1543
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3157615313366632,
+      "learning_rate": 0.00020761152551494643,
+      "loss": 2.2814,
+      "step": 1544
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3446947508969591,
+      "learning_rate": 0.00020700937984962798,
+      "loss": 2.3469,
+      "step": 1545
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3146281585841679,
+      "learning_rate": 0.0002064078806308848,
+      "loss": 2.2231,
+      "step": 1546
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.32653216960380754,
+      "learning_rate": 0.00020580702918584882,
+      "loss": 2.2619,
+      "step": 1547
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.32223597470632537,
+      "learning_rate": 0.000205206826840222,
+      "loss": 2.3053,
+      "step": 1548
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.325340675441311,
+      "learning_rate": 0.0002046072749182751,
+      "loss": 2.2634,
+      "step": 1549
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.31928163272312265,
+      "learning_rate": 0.00020400837474284273,
+      "loss": 2.2913,
+      "step": 1550
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3165394466985783,
+      "learning_rate": 0.0002034101276353224,
+      "loss": 2.343,
+      "step": 1551
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3107588127787961,
+      "learning_rate": 0.00020281253491567027,
+      "loss": 2.2904,
+      "step": 1552
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3045164065608385,
+      "learning_rate": 0.0002022155979023984,
+      "loss": 2.2546,
+      "step": 1553
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3308038158721351,
+      "learning_rate": 0.000201619317912573,
+      "loss": 2.3386,
+      "step": 1554
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.32929770774941114,
+      "learning_rate": 0.00020102369626180962,
+      "loss": 2.2754,
+      "step": 1555
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3111404710874792,
+      "learning_rate": 0.0002004287342642721,
+      "loss": 2.243,
+      "step": 1556
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.32056815237679465,
+      "learning_rate": 0.00019983443323266824,
+      "loss": 2.2571,
+      "step": 1557
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3049023283896207,
+      "learning_rate": 0.00019924079447824805,
+      "loss": 2.294,
+      "step": 1558
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.3187778477072764,
+      "learning_rate": 0.00019864781931079977,
+      "loss": 2.1963,
+      "step": 1559
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3174930955365824,
+      "learning_rate": 0.00019805550903864773,
+      "loss": 2.3356,
+      "step": 1560
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3007006938784742,
+      "learning_rate": 0.00019746386496864948,
+      "loss": 2.2056,
+      "step": 1561
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.31856534672453013,
+      "learning_rate": 0.00019687288840619226,
+      "loss": 2.2424,
+      "step": 1562
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3183868240492882,
+      "learning_rate": 0.0001962825806551911,
+      "loss": 2.2789,
+      "step": 1563
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3321530989117453,
+      "learning_rate": 0.0001956929430180846,
+      "loss": 2.251,
+      "step": 1564
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3051115502707693,
+      "learning_rate": 0.00019510397679583374,
+      "loss": 2.2902,
+      "step": 1565
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.32868102452101816,
+      "learning_rate": 0.0001945156832879174,
+      "loss": 2.2956,
+      "step": 1566
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.31531333295694863,
+      "learning_rate": 0.00019392806379233036,
+      "loss": 2.1489,
+      "step": 1567
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.32548730419193705,
+      "learning_rate": 0.00019334111960558065,
+      "loss": 2.3115,
+      "step": 1568
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.30662577036752847,
+      "learning_rate": 0.00019275485202268573,
+      "loss": 2.3189,
+      "step": 1569
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.32455738603293144,
+      "learning_rate": 0.00019216926233717085,
+      "loss": 2.2153,
+      "step": 1570
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3272253394513705,
+      "learning_rate": 0.00019158435184106498,
+      "loss": 2.339,
+      "step": 1571
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3272809390041462,
+      "learning_rate": 0.00019100012182489905,
+      "loss": 2.2757,
+      "step": 1572
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.2974883740969384,
+      "learning_rate": 0.00019041657357770226,
+      "loss": 2.25,
+      "step": 1573
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3339951163314029,
+      "learning_rate": 0.00018983370838699943,
+      "loss": 2.2499,
+      "step": 1574
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3490077454225252,
+      "learning_rate": 0.00018925152753880892,
+      "loss": 2.3098,
+      "step": 1575
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3206051759737299,
+      "learning_rate": 0.00018867003231763847,
+      "loss": 2.3515,
+      "step": 1576
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3238320849013584,
+      "learning_rate": 0.00018808922400648375,
+      "loss": 2.3855,
+      "step": 1577
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3264984222862411,
+      "learning_rate": 0.00018750910388682428,
+      "loss": 2.3047,
+      "step": 1578
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.31058863971221135,
+      "learning_rate": 0.00018692967323862125,
+      "loss": 2.2117,
+      "step": 1579
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.30728926846879623,
+      "learning_rate": 0.00018635093334031517,
+      "loss": 2.279,
+      "step": 1580
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.32734950847030675,
+      "learning_rate": 0.00018577288546882165,
+      "loss": 2.2273,
+      "step": 1581
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3243926645108166,
+      "learning_rate": 0.00018519553089953023,
+      "loss": 2.2619,
+      "step": 1582
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.30708342729342064,
+      "learning_rate": 0.0001846188709063001,
+      "loss": 2.2028,
+      "step": 1583
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.2998605792941661,
+      "learning_rate": 0.00018404290676145857,
+      "loss": 2.2498,
+      "step": 1584
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.31118219960896554,
+      "learning_rate": 0.00018346763973579722,
+      "loss": 2.2321,
+      "step": 1585
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.31264879014433195,
+      "learning_rate": 0.00018289307109856939,
+      "loss": 2.363,
+      "step": 1586
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.34035101325077277,
+      "learning_rate": 0.0001823192021174882,
+      "loss": 2.2243,
+      "step": 1587
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3148209377436567,
+      "learning_rate": 0.0001817460340587223,
+      "loss": 2.2788,
+      "step": 1588
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.30149105205532356,
+      "learning_rate": 0.00018117356818689445,
+      "loss": 2.3275,
+      "step": 1589
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.2950239800012248,
+      "learning_rate": 0.00018060180576507756,
+      "loss": 2.245,
+      "step": 1590
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.31997500893616193,
+      "learning_rate": 0.00018003074805479313,
+      "loss": 2.1281,
+      "step": 1591
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3146976632247854,
+      "learning_rate": 0.00017946039631600724,
+      "loss": 2.2608,
+      "step": 1592
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.30592639164700486,
+      "learning_rate": 0.00017889075180712837,
+      "loss": 2.2826,
+      "step": 1593
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.30511813026171863,
+      "learning_rate": 0.00017832181578500512,
+      "loss": 2.2356,
+      "step": 1594
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.30692875369050715,
+      "learning_rate": 0.0001777535895049221,
+      "loss": 2.2126,
+      "step": 1595
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.30823537986317406,
+      "learning_rate": 0.0001771860742205988,
+      "loss": 2.3009,
+      "step": 1596
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.31420735838428265,
+      "learning_rate": 0.00017661927118418525,
+      "loss": 2.3224,
+      "step": 1597
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3237875199056948,
+      "learning_rate": 0.00017605318164626066,
+      "loss": 2.3463,
+      "step": 1598
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3177139636822562,
+      "learning_rate": 0.00017548780685582949,
+      "loss": 2.2704,
+      "step": 1599
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.30555634088849937,
+      "learning_rate": 0.00017492314806031922,
+      "loss": 2.3577,
+      "step": 1600
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3083044240096357,
+      "learning_rate": 0.00017435920650557806,
+      "loss": 2.2723,
+      "step": 1601
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3303154988824857,
+      "learning_rate": 0.00017379598343587112,
+      "loss": 2.2455,
+      "step": 1602
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.3097884684737045,
+      "learning_rate": 0.00017323348009387878,
+      "loss": 2.2153,
+      "step": 1603
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.32680301910728954,
+      "learning_rate": 0.0001726716977206929,
+      "loss": 2.2429,
+      "step": 1604
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.31441904784649993,
+      "learning_rate": 0.00017211063755581525,
+      "loss": 2.231,
+      "step": 1605
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.32696737766293227,
+      "learning_rate": 0.0001715503008371536,
+      "loss": 2.2858,
+      "step": 1606
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3284636575477364,
+      "learning_rate": 0.0001709906888010196,
+      "loss": 2.1937,
+      "step": 1607
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3176006946151791,
+      "learning_rate": 0.00017043180268212638,
+      "loss": 2.3301,
+      "step": 1608
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3134245518833369,
+      "learning_rate": 0.00016987364371358481,
+      "loss": 2.2586,
+      "step": 1609
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.2959733667812727,
+      "learning_rate": 0.00016931621312690214,
+      "loss": 2.1949,
+      "step": 1610
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3364369260767532,
+      "learning_rate": 0.00016875951215197777,
+      "loss": 2.2547,
+      "step": 1611
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.32455827789106156,
+      "learning_rate": 0.00016820354201710214,
+      "loss": 2.3008,
+      "step": 1612
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.2998057088340452,
+      "learning_rate": 0.00016764830394895203,
+      "loss": 2.1275,
+      "step": 1613
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.31299200150232076,
+      "learning_rate": 0.00016709379917259027,
+      "loss": 2.2223,
+      "step": 1614
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3086778296748282,
+      "learning_rate": 0.00016654002891146091,
+      "loss": 2.3161,
+      "step": 1615
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3249319719972869,
+      "learning_rate": 0.00016598699438738764,
+      "loss": 2.2854,
+      "step": 1616
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.30065804230231885,
+      "learning_rate": 0.00016543469682057105,
+      "loss": 2.349,
+      "step": 1617
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.30929366589583834,
+      "learning_rate": 0.00016488313742958526,
+      "loss": 2.3386,
+      "step": 1618
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3080511575841249,
+      "learning_rate": 0.00016433231743137646,
+      "loss": 2.1632,
+      "step": 1619
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3140558970083249,
+      "learning_rate": 0.0001637822380412584,
+      "loss": 2.2358,
+      "step": 1620
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.30785831633729277,
+      "learning_rate": 0.00016323290047291195,
+      "loss": 2.2281,
+      "step": 1621
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.30940737482105435,
+      "learning_rate": 0.0001626843059383803,
+      "loss": 2.3297,
+      "step": 1622
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.3070384968080322,
+      "learning_rate": 0.00016213645564806752,
+      "loss": 2.2622,
+      "step": 1623
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.31238566216893177,
+      "learning_rate": 0.0001615893508107359,
+      "loss": 2.2125,
+      "step": 1624
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.30360459765486364,
+      "learning_rate": 0.00016104299263350252,
+      "loss": 2.2919,
+      "step": 1625
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.31606600863954526,
+      "learning_rate": 0.00016049738232183758,
+      "loss": 2.2436,
+      "step": 1626
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3322041192754495,
+      "learning_rate": 0.0001599525210795606,
+      "loss": 2.3209,
+      "step": 1627
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3146734170542244,
+      "learning_rate": 0.00015940841010883889,
+      "loss": 2.1897,
+      "step": 1628
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.29769180295751935,
+      "learning_rate": 0.00015886505061018413,
+      "loss": 2.2126,
+      "step": 1629
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3167242213836951,
+      "learning_rate": 0.0001583224437824498,
+      "loss": 2.2358,
+      "step": 1630
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3221204361308331,
+      "learning_rate": 0.0001577805908228293,
+      "loss": 2.1689,
+      "step": 1631
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3144596917718178,
+      "learning_rate": 0.00015723949292685191,
+      "loss": 2.2016,
+      "step": 1632
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.2983653255304679,
+      "learning_rate": 0.0001566991512883818,
+      "loss": 2.1894,
+      "step": 1633
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3176141785205311,
+      "learning_rate": 0.00015615956709961378,
+      "loss": 2.1747,
+      "step": 1634
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3299891838974438,
+      "learning_rate": 0.00015562074155107215,
+      "loss": 2.2381,
+      "step": 1635
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3347380280748284,
+      "learning_rate": 0.0001550826758316068,
+      "loss": 2.2741,
+      "step": 1636
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3026249804876062,
+      "learning_rate": 0.00015454537112839122,
+      "loss": 2.227,
+      "step": 1637
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.30039031318444476,
+      "learning_rate": 0.00015400882862692033,
+      "loss": 2.2408,
+      "step": 1638
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.30650303765128767,
+      "learning_rate": 0.00015347304951100665,
+      "loss": 2.2503,
+      "step": 1639
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.2994776000354664,
+      "learning_rate": 0.00015293803496277907,
+      "loss": 2.277,
+      "step": 1640
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3076826799364287,
+      "learning_rate": 0.00015240378616267886,
+      "loss": 2.2187,
+      "step": 1641
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3116747316804314,
+      "learning_rate": 0.00015187030428945843,
+      "loss": 2.1668,
+      "step": 1642
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3211011298847836,
+      "learning_rate": 0.0001513375905201776,
+      "loss": 2.3209,
+      "step": 1643
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.325558446374991,
+      "learning_rate": 0.00015080564603020142,
+      "loss": 2.2992,
+      "step": 1644
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.32216880052414154,
+      "learning_rate": 0.0001502744719931982,
+      "loss": 2.3176,
+      "step": 1645
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3172959266439997,
+      "learning_rate": 0.00014974406958113558,
+      "loss": 2.2218,
+      "step": 1646
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.31944076941376465,
+      "learning_rate": 0.00014921443996427947,
+      "loss": 2.2228,
+      "step": 1647
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3330060257204911,
+      "learning_rate": 0.0001486855843111901,
+      "loss": 2.2893,
+      "step": 1648
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.31453782973246314,
+      "learning_rate": 0.0001481575037887201,
+      "loss": 2.287,
+      "step": 1649
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3449707973389908,
+      "learning_rate": 0.00014763019956201253,
+      "loss": 2.2814,
+      "step": 1650
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.32228306085639596,
+      "learning_rate": 0.0001471036727944966,
+      "loss": 2.2339,
+      "step": 1651
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3440435847339428,
+      "learning_rate": 0.0001465779246478872,
+      "loss": 2.176,
+      "step": 1652
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.337857612421015,
+      "learning_rate": 0.00014605295628218045,
+      "loss": 2.2573,
+      "step": 1653
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3322291984508464,
+      "learning_rate": 0.0001455287688556527,
+      "loss": 2.2319,
+      "step": 1654
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3205468640419334,
+      "learning_rate": 0.00014500536352485673,
+      "loss": 2.21,
+      "step": 1655
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.31255310154492594,
+      "learning_rate": 0.00014448274144461965,
+      "loss": 2.2317,
+      "step": 1656
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3342788497202069,
+      "learning_rate": 0.00014396090376804112,
+      "loss": 2.3421,
+      "step": 1657
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3174051656174852,
+      "learning_rate": 0.00014343985164648926,
+      "loss": 2.2269,
+      "step": 1658
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.30557400245085997,
+      "learning_rate": 0.00014291958622959973,
+      "loss": 2.1738,
+      "step": 1659
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3224751487817808,
+      "learning_rate": 0.00014240010866527176,
+      "loss": 2.2649,
+      "step": 1660
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3104869613316413,
+      "learning_rate": 0.00014188142009966686,
+      "loss": 2.186,
+      "step": 1661
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.31311026255492463,
+      "learning_rate": 0.0001413635216772053,
+      "loss": 2.2233,
+      "step": 1662
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.31846116667735314,
+      "learning_rate": 0.000140846414540564,
+      "loss": 2.2081,
+      "step": 1663
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3274976249579299,
+      "learning_rate": 0.00014033009983067452,
+      "loss": 2.1767,
+      "step": 1664
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3047908310100857,
+      "learning_rate": 0.00013981457868671927,
+      "loss": 2.2504,
+      "step": 1665
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.347838645236411,
+      "learning_rate": 0.0001392998522461305,
+      "loss": 2.3933,
+      "step": 1666
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3260765027038489,
+      "learning_rate": 0.00013878592164458635,
+      "loss": 2.2056,
+      "step": 1667
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3225217892116252,
+      "learning_rate": 0.00013827278801600978,
+      "loss": 2.2962,
+      "step": 1668
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3492485392455155,
+      "learning_rate": 0.0001377604524925647,
+      "loss": 2.3701,
+      "step": 1669
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3054409460733161,
+      "learning_rate": 0.00013724891620465424,
+      "loss": 2.2335,
+      "step": 1670
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.30323467067586085,
+      "learning_rate": 0.0001367381802809185,
+      "loss": 2.1444,
+      "step": 1671
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3340911348792418,
+      "learning_rate": 0.00013622824584823113,
+      "loss": 2.2532,
+      "step": 1672
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.32633068638064927,
+      "learning_rate": 0.00013571911403169795,
+      "loss": 2.2369,
+      "step": 1673
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3129517235571715,
+      "learning_rate": 0.0001352107859546533,
+      "loss": 2.2052,
+      "step": 1674
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.31703472088743034,
+      "learning_rate": 0.00013470326273865886,
+      "loss": 2.2512,
+      "step": 1675
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3191218863184006,
+      "learning_rate": 0.00013419654550349985,
+      "loss": 2.3174,
+      "step": 1676
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.31594277354595923,
+      "learning_rate": 0.00013369063536718346,
+      "loss": 2.2418,
+      "step": 1677
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.32841404591772394,
+      "learning_rate": 0.00013318553344593632,
+      "loss": 2.2188,
+      "step": 1678
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3090409969477732,
+      "learning_rate": 0.00013268124085420136,
+      "loss": 2.285,
+      "step": 1679
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3168827598461631,
+      "learning_rate": 0.0001321777587046364,
+      "loss": 2.3282,
+      "step": 1680
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.31281387565146374,
+      "learning_rate": 0.00013167508810811059,
+      "loss": 2.2687,
+      "step": 1681
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.32566468736903503,
+      "learning_rate": 0.0001311732301737029,
+      "loss": 2.2834,
+      "step": 1682
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.33352532435539683,
+      "learning_rate": 0.0001306721860086991,
+      "loss": 2.3007,
+      "step": 1683
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3085301081429939,
+      "learning_rate": 0.00013017195671858928,
+      "loss": 2.1976,
+      "step": 1684
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.32672550584022636,
+      "learning_rate": 0.0001296725434070661,
+      "loss": 2.3333,
+      "step": 1685
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.32226570695499335,
+      "learning_rate": 0.00012917394717602121,
+      "loss": 2.3157,
+      "step": 1686
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.3132167691779886,
+      "learning_rate": 0.00012867616912554426,
+      "loss": 2.1473,
+      "step": 1687
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.30036232045659317,
+      "learning_rate": 0.00012817921035391882,
+      "loss": 2.2448,
+      "step": 1688
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.31211147385266036,
+      "learning_rate": 0.00012768307195762168,
+      "loss": 2.2657,
+      "step": 1689
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.30595694235121396,
+      "learning_rate": 0.00012718775503131908,
+      "loss": 2.2392,
+      "step": 1690
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3188481815202169,
+      "learning_rate": 0.0001266932606678646,
+      "loss": 2.2362,
+      "step": 1691
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.31397568431455575,
+      "learning_rate": 0.00012619958995829756,
+      "loss": 2.2798,
+      "step": 1692
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3043759868713739,
+      "learning_rate": 0.0001257067439918394,
+      "loss": 2.198,
+      "step": 1693
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3032893662807031,
+      "learning_rate": 0.00012521472385589234,
+      "loss": 2.2079,
+      "step": 1694
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3156347219675047,
+      "learning_rate": 0.00012472353063603626,
+      "loss": 2.2404,
+      "step": 1695
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.31993114812609014,
+      "learning_rate": 0.0001242331654160263,
+      "loss": 2.1959,
+      "step": 1696
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.31697249770798325,
+      "learning_rate": 0.0001237436292777914,
+      "loss": 2.2576,
+      "step": 1697
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3444105077242189,
+      "learning_rate": 0.00012325492330143061,
+      "loss": 2.2857,
+      "step": 1698
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3290896988094415,
+      "learning_rate": 0.00012276704856521175,
+      "loss": 2.1558,
+      "step": 1699
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3377593697696987,
+      "learning_rate": 0.00012228000614556816,
+      "loss": 2.3078,
+      "step": 1700
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3068322998885483,
+      "learning_rate": 0.00012179379711709738,
+      "loss": 2.2981,
+      "step": 1701
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3038427727144671,
+      "learning_rate": 0.0001213084225525577,
+      "loss": 2.3047,
+      "step": 1702
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.32689663560952803,
+      "learning_rate": 0.00012082388352286627,
+      "loss": 2.4007,
+      "step": 1703
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.31220668491706843,
+      "learning_rate": 0.00012034018109709716,
+      "loss": 2.2912,
+      "step": 1704
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.2855482916800673,
+      "learning_rate": 0.00011985731634247809,
+      "loss": 2.1791,
+      "step": 1705
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3145931828164151,
+      "learning_rate": 0.00011937529032438904,
+      "loss": 2.2555,
+      "step": 1706
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.31786685606327086,
+      "learning_rate": 0.00011889410410635887,
+      "loss": 2.1583,
+      "step": 1707
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3226422675997172,
+      "learning_rate": 0.0001184137587500641,
+      "loss": 2.169,
+      "step": 1708
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3193920242666578,
+      "learning_rate": 0.00011793425531532564,
+      "loss": 2.1636,
+      "step": 1709
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.29440309375779833,
+      "learning_rate": 0.00011745559486010671,
+      "loss": 2.2628,
+      "step": 1710
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.31514499829735276,
+      "learning_rate": 0.00011697777844051105,
+      "loss": 2.3082,
+      "step": 1711
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3151763981523586,
+      "learning_rate": 0.00011650080711077964,
+      "loss": 2.2509,
+      "step": 1712
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.31273403813395956,
+      "learning_rate": 0.00011602468192328936,
+      "loss": 2.3374,
+      "step": 1713
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3128527567961727,
+      "learning_rate": 0.00011554940392854973,
+      "loss": 2.2994,
+      "step": 1714
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3000417780300718,
+      "learning_rate": 0.00011507497417520146,
+      "loss": 2.3183,
+      "step": 1715
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3167672685087635,
+      "learning_rate": 0.00011460139371001339,
+      "loss": 2.2809,
+      "step": 1716
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.31908861602481253,
+      "learning_rate": 0.00011412866357788049,
+      "loss": 2.2273,
+      "step": 1717
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3039618076245166,
+      "learning_rate": 0.00011365678482182207,
+      "loss": 2.1142,
+      "step": 1718
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3168460287736233,
+      "learning_rate": 0.0001131857584829783,
+      "loss": 2.2188,
+      "step": 1719
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3356715769476872,
+      "learning_rate": 0.0001127155856006093,
+      "loss": 2.2877,
+      "step": 1720
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.33476817460742875,
+      "learning_rate": 0.00011224626721209141,
+      "loss": 2.1393,
+      "step": 1721
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.34987288163854874,
+      "learning_rate": 0.0001117778043529164,
+      "loss": 2.3361,
+      "step": 1722
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3170471374782586,
+      "learning_rate": 0.0001113101980566879,
+      "loss": 2.2086,
+      "step": 1723
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.30991314656489044,
+      "learning_rate": 0.00011084344935511958,
+      "loss": 2.1698,
+      "step": 1724
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3090426546763112,
+      "learning_rate": 0.00011037755927803345,
+      "loss": 2.266,
+      "step": 1725
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.29543876357326526,
+      "learning_rate": 0.00010991252885335651,
+      "loss": 2.2739,
+      "step": 1726
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3242579486614439,
+      "learning_rate": 0.00010944835910711958,
+      "loss": 2.2686,
+      "step": 1727
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.32403100107661637,
+      "learning_rate": 0.00010898505106345396,
+      "loss": 2.3085,
+      "step": 1728
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.31738690605784725,
+      "learning_rate": 0.00010852260574459022,
+      "loss": 2.2441,
+      "step": 1729
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.31300584001055415,
+      "learning_rate": 0.00010806102417085512,
+      "loss": 2.2013,
+      "step": 1730
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3203836925546152,
+      "learning_rate": 0.00010760030736066951,
+      "loss": 2.2235,
+      "step": 1731
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.3092022591777011,
+      "learning_rate": 0.00010714045633054687,
+      "loss": 2.2639,
+      "step": 1732
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.332828033355708,
+      "learning_rate": 0.00010668147209508971,
+      "loss": 2.2091,
+      "step": 1733
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3060500499107263,
+      "learning_rate": 0.00010622335566698877,
+      "loss": 2.314,
+      "step": 1734
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2977543542205654,
+      "learning_rate": 0.00010576610805701942,
+      "loss": 2.3034,
+      "step": 1735
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3236898822111785,
+      "learning_rate": 0.00010530973027404073,
+      "loss": 2.251,
+      "step": 1736
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3096162907082513,
+      "learning_rate": 0.00010485422332499212,
+      "loss": 2.2034,
+      "step": 1737
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.31353780768895506,
+      "learning_rate": 0.00010439958821489165,
+      "loss": 2.1945,
+      "step": 1738
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.32266983632387364,
+      "learning_rate": 0.00010394582594683428,
+      "loss": 2.1778,
+      "step": 1739
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2985288071951603,
+      "learning_rate": 0.0001034929375219884,
+      "loss": 2.178,
+      "step": 1740
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3149409058438918,
+      "learning_rate": 0.00010304092393959514,
+      "loss": 2.2314,
+      "step": 1741
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.31099146818252615,
+      "learning_rate": 0.00010258978619696468,
+      "loss": 2.159,
+      "step": 1742
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.31483410100142867,
+      "learning_rate": 0.00010213952528947551,
+      "loss": 2.295,
+      "step": 1743
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.29724956018069526,
+      "learning_rate": 0.00010169014221057089,
+      "loss": 2.156,
+      "step": 1744
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3205734136408605,
+      "learning_rate": 0.00010124163795175734,
+      "loss": 2.2614,
+      "step": 1745
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.301013799246263,
+      "learning_rate": 0.00010079401350260287,
+      "loss": 2.16,
+      "step": 1746
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3137031037985184,
+      "learning_rate": 0.00010034726985073362,
+      "loss": 2.3528,
+      "step": 1747
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3301956712522813,
+      "learning_rate": 9.9901407981833e-05,
+      "loss": 2.3237,
+      "step": 1748
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.31398791833080775,
+      "learning_rate": 9.94564288796384e-05,
+      "loss": 2.2972,
+      "step": 1749
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.32687230044382365,
+      "learning_rate": 9.901233352593953e-05,
+      "loss": 2.2298,
+      "step": 1750
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.325693673126282,
+      "learning_rate": 9.856912290057668e-05,
+      "loss": 2.1967,
+      "step": 1751
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.30901702992715496,
+      "learning_rate": 9.812679798143748e-05,
+      "loss": 2.2374,
+      "step": 1752
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3078137320037994,
+      "learning_rate": 9.768535974445586e-05,
+      "loss": 2.3064,
+      "step": 1753
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.3064120293441214,
+      "learning_rate": 9.724480916360906e-05,
+      "loss": 2.1792,
+      "step": 1754
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.31768892266888815,
+      "learning_rate": 9.68051472109162e-05,
+      "loss": 2.2269,
+      "step": 1755
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.31823184826387513,
+      "learning_rate": 9.636637485643529e-05,
+      "loss": 2.2924,
+      "step": 1756
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3170741782375582,
+      "learning_rate": 9.592849306826174e-05,
+      "loss": 2.3239,
+      "step": 1757
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3253330105946771,
+      "learning_rate": 9.549150281252633e-05,
+      "loss": 2.2895,
+      "step": 1758
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.32675020063179705,
+      "learning_rate": 9.505540505339223e-05,
+      "loss": 2.1653,
+      "step": 1759
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.31513070139834837,
+      "learning_rate": 9.4620200753054e-05,
+      "loss": 2.2655,
+      "step": 1760
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.31675595984702903,
+      "learning_rate": 9.418589087173441e-05,
+      "loss": 2.1528,
+      "step": 1761
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.30695923280172555,
+      "learning_rate": 9.375247636768325e-05,
+      "loss": 2.2337,
+      "step": 1762
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.31246235470306616,
+      "learning_rate": 9.331995819717443e-05,
+      "loss": 2.3252,
+      "step": 1763
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3584793915172355,
+      "learning_rate": 9.288833731450419e-05,
+      "loss": 2.2625,
+      "step": 1764
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3185772255870288,
+      "learning_rate": 9.245761467198948e-05,
+      "loss": 2.2462,
+      "step": 1765
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3305966875485118,
+      "learning_rate": 9.20277912199648e-05,
+      "loss": 2.1892,
+      "step": 1766
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.32721019281420566,
+      "learning_rate": 9.159886790678123e-05,
+      "loss": 2.3725,
+      "step": 1767
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.30633761116260794,
+      "learning_rate": 9.11708456788033e-05,
+      "loss": 2.3163,
+      "step": 1768
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.2994504690110083,
+      "learning_rate": 9.074372548040793e-05,
+      "loss": 2.2662,
+      "step": 1769
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.30254422824071325,
+      "learning_rate": 9.031750825398145e-05,
+      "loss": 2.2024,
+      "step": 1770
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3212994872420961,
+      "learning_rate": 8.98921949399179e-05,
+      "loss": 2.1552,
+      "step": 1771
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3223935467244081,
+      "learning_rate": 8.94677864766173e-05,
+      "loss": 2.2793,
+      "step": 1772
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.30540617567880507,
+      "learning_rate": 8.904428380048269e-05,
+      "loss": 2.2257,
+      "step": 1773
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.2955098699600702,
+      "learning_rate": 8.862168784591929e-05,
+      "loss": 2.3285,
+      "step": 1774
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.30692645081327785,
+      "learning_rate": 8.819999954533115e-05,
+      "loss": 2.3635,
+      "step": 1775
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.3171288984679785,
+      "learning_rate": 8.777921982911996e-05,
+      "loss": 2.2295,
+      "step": 1776
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.31154981631572237,
+      "learning_rate": 8.735934962568253e-05,
+      "loss": 2.1943,
+      "step": 1777
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.30511831683238677,
+      "learning_rate": 8.694038986140945e-05,
+      "loss": 2.1721,
+      "step": 1778
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.32353973290949894,
+      "learning_rate": 8.652234146068206e-05,
+      "loss": 2.2678,
+      "step": 1779
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.32019735866035526,
+      "learning_rate": 8.610520534587086e-05,
+      "loss": 2.251,
+      "step": 1780
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3103290934739414,
+      "learning_rate": 8.568898243733397e-05,
+      "loss": 2.321,
+      "step": 1781
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3137941565067693,
+      "learning_rate": 8.527367365341409e-05,
+      "loss": 2.2702,
+      "step": 1782
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.31751663449398687,
+      "learning_rate": 8.485927991043757e-05,
+      "loss": 2.0621,
+      "step": 1783
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.30370184961955443,
+      "learning_rate": 8.444580212271125e-05,
+      "loss": 2.2668,
+      "step": 1784
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3120171680918193,
+      "learning_rate": 8.403324120252159e-05,
+      "loss": 2.2663,
+      "step": 1785
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3174462254306441,
+      "learning_rate": 8.362159806013175e-05,
+      "loss": 2.2696,
+      "step": 1786
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.30823474192901007,
+      "learning_rate": 8.321087360377988e-05,
+      "loss": 2.2862,
+      "step": 1787
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3198614229481817,
+      "learning_rate": 8.280106873967752e-05,
+      "loss": 2.2369,
+      "step": 1788
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3339909330261886,
+      "learning_rate": 8.239218437200679e-05,
+      "loss": 2.2153,
+      "step": 1789
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3392772398184385,
+      "learning_rate": 8.198422140291939e-05,
+      "loss": 2.2867,
+      "step": 1790
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3046525384503458,
+      "learning_rate": 8.157718073253351e-05,
+      "loss": 2.2657,
+      "step": 1791
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3095529941193707,
+      "learning_rate": 8.117106325893287e-05,
+      "loss": 2.2097,
+      "step": 1792
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3080066332915626,
+      "learning_rate": 8.076586987816404e-05,
+      "loss": 2.225,
+      "step": 1793
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3115568659316559,
+      "learning_rate": 8.036160148423449e-05,
+      "loss": 2.3837,
+      "step": 1794
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.31469397124961523,
+      "learning_rate": 7.995825896911141e-05,
+      "loss": 2.1401,
+      "step": 1795
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.31076962783535556,
+      "learning_rate": 7.955584322271853e-05,
+      "loss": 2.218,
+      "step": 1796
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3103588135582236,
+      "learning_rate": 7.915435513293523e-05,
+      "loss": 2.3411,
+      "step": 1797
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3282146576900493,
+      "learning_rate": 7.875379558559387e-05,
+      "loss": 2.365,
+      "step": 1798
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.3161786236569608,
+      "learning_rate": 7.835416546447838e-05,
+      "loss": 2.2437,
+      "step": 1799
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.32616690245011287,
+      "learning_rate": 7.795546565132167e-05,
+      "loss": 2.1919,
+      "step": 1800
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.31426783358568094,
+      "learning_rate": 7.755769702580412e-05,
+      "loss": 2.2182,
+      "step": 1801
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3264835057971253,
+      "learning_rate": 7.716086046555193e-05,
+      "loss": 2.2255,
+      "step": 1802
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.32982014628506234,
+      "learning_rate": 7.676495684613432e-05,
+      "loss": 2.13,
+      "step": 1803
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.39976533421309224,
+      "learning_rate": 7.636998704106252e-05,
+      "loss": 2.3108,
+      "step": 1804
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.32255299227467493,
+      "learning_rate": 7.597595192178702e-05,
+      "loss": 2.2148,
+      "step": 1805
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3066612781135064,
+      "learning_rate": 7.558285235769646e-05,
+      "loss": 2.2599,
+      "step": 1806
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3268762491622282,
+      "learning_rate": 7.519068921611494e-05,
+      "loss": 2.2775,
+      "step": 1807
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3282194510406891,
+      "learning_rate": 7.479946336230047e-05,
+      "loss": 2.2483,
+      "step": 1808
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.34703301317824375,
+      "learning_rate": 7.440917565944349e-05,
+      "loss": 2.2595,
+      "step": 1809
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3332545366089718,
+      "learning_rate": 7.4019826968664e-05,
+      "loss": 2.3116,
+      "step": 1810
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3214847884861678,
+      "learning_rate": 7.363141814901053e-05,
+      "loss": 2.2031,
+      "step": 1811
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3112635548863918,
+      "learning_rate": 7.32439500574577e-05,
+      "loss": 2.2541,
+      "step": 1812
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.33810382807575573,
+      "learning_rate": 7.285742354890473e-05,
+      "loss": 2.2645,
+      "step": 1813
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.29976732100533193,
+      "learning_rate": 7.247183947617325e-05,
+      "loss": 2.2285,
+      "step": 1814
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.30948421132569537,
+      "learning_rate": 7.20871986900053e-05,
+      "loss": 2.338,
+      "step": 1815
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3313550457829786,
+      "learning_rate": 7.170350203906218e-05,
+      "loss": 2.2688,
+      "step": 1816
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.3164363342425958,
+      "learning_rate": 7.132075036992158e-05,
+      "loss": 2.2525,
+      "step": 1817
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.32056579534440555,
+      "learning_rate": 7.093894452707666e-05,
+      "loss": 2.2512,
+      "step": 1818
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.32548104257708815,
+      "learning_rate": 7.055808535293334e-05,
+      "loss": 2.203,
+      "step": 1819
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.30641353961566375,
+      "learning_rate": 7.017817368780888e-05,
+      "loss": 2.29,
+      "step": 1820
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.362463314150361,
+      "learning_rate": 6.979921036993042e-05,
+      "loss": 2.2474,
+      "step": 1821
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3321624893632081,
+      "learning_rate": 6.942119623543202e-05,
+      "loss": 2.2479,
+      "step": 1822
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.2999336026495958,
+      "learning_rate": 6.904413211835414e-05,
+      "loss": 2.2367,
+      "step": 1823
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3050442391131991,
+      "learning_rate": 6.866801885064056e-05,
+      "loss": 2.3242,
+      "step": 1824
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.31645412658132166,
+      "learning_rate": 6.829285726213769e-05,
+      "loss": 2.2427,
+      "step": 1825
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.32726901507087347,
+      "learning_rate": 6.79186481805918e-05,
+      "loss": 2.193,
+      "step": 1826
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3408092182952552,
+      "learning_rate": 6.754539243164754e-05,
+      "loss": 2.3532,
+      "step": 1827
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.2955826650000052,
+      "learning_rate": 6.717309083884654e-05,
+      "loss": 2.2148,
+      "step": 1828
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3387912034656634,
+      "learning_rate": 6.680174422362468e-05,
+      "loss": 2.2788,
+      "step": 1829
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3213661833725603,
+      "learning_rate": 6.643135340531136e-05,
+      "loss": 2.2467,
+      "step": 1830
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.30722568203334966,
+      "learning_rate": 6.606191920112664e-05,
+      "loss": 2.202,
+      "step": 1831
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.30847498008723884,
+      "learning_rate": 6.569344242618036e-05,
+      "loss": 2.2452,
+      "step": 1832
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.32003569251320063,
+      "learning_rate": 6.532592389346958e-05,
+      "loss": 2.3091,
+      "step": 1833
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.32238882824468296,
+      "learning_rate": 6.495936441387713e-05,
+      "loss": 2.3242,
+      "step": 1834
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3282209601301341,
+      "learning_rate": 6.459376479617013e-05,
+      "loss": 2.2512,
+      "step": 1835
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3126542032962482,
+      "learning_rate": 6.422912584699752e-05,
+      "loss": 2.1866,
+      "step": 1836
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3168662935439502,
+      "learning_rate": 6.386544837088904e-05,
+      "loss": 2.2848,
+      "step": 1837
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.31983979692327175,
+      "learning_rate": 6.350273317025251e-05,
+      "loss": 2.2874,
+      "step": 1838
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.30968836744025485,
+      "learning_rate": 6.314098104537324e-05,
+      "loss": 2.2979,
+      "step": 1839
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.31404184275029634,
+      "learning_rate": 6.278019279441122e-05,
+      "loss": 2.0821,
+      "step": 1840
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3225027686909786,
+      "learning_rate": 6.242036921339972e-05,
+      "loss": 2.3101,
+      "step": 1841
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3372849206223822,
+      "learning_rate": 6.206151109624402e-05,
+      "loss": 2.3221,
+      "step": 1842
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.32380823485971805,
+      "learning_rate": 6.170361923471868e-05,
+      "loss": 2.2876,
+      "step": 1843
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3172735172471084,
+      "learning_rate": 6.134669441846691e-05,
+      "loss": 2.3105,
+      "step": 1844
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.31864088396702034,
+      "learning_rate": 6.099073743499772e-05,
+      "loss": 2.2975,
+      "step": 1845
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.312923898413695,
+      "learning_rate": 6.063574906968511e-05,
+      "loss": 2.2928,
+      "step": 1846
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3264788136485936,
+      "learning_rate": 6.028173010576582e-05,
+      "loss": 2.2118,
+      "step": 1847
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.31215923822158126,
+      "learning_rate": 5.9928681324337544e-05,
+      "loss": 2.2252,
+      "step": 1848
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.316209645906311,
+      "learning_rate": 5.957660350435773e-05,
+      "loss": 2.1564,
+      "step": 1849
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.30819421145354925,
+      "learning_rate": 5.922549742264122e-05,
+      "loss": 2.3525,
+      "step": 1850
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.32139819619412496,
+      "learning_rate": 5.8875363853859166e-05,
+      "loss": 2.1874,
+      "step": 1851
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.310203303169172,
+      "learning_rate": 5.852620357053651e-05,
+      "loss": 2.317,
+      "step": 1852
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3175201052462213,
+      "learning_rate": 5.8178017343051336e-05,
+      "loss": 2.2136,
+      "step": 1853
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.321128762548206,
+      "learning_rate": 5.783080593963219e-05,
+      "loss": 2.2278,
+      "step": 1854
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3082971519922989,
+      "learning_rate": 5.748457012635683e-05,
+      "loss": 2.1284,
+      "step": 1855
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3183069936721064,
+      "learning_rate": 5.713931066715078e-05,
+      "loss": 2.2775,
+      "step": 1856
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3277775762233898,
+      "learning_rate": 5.679502832378497e-05,
+      "loss": 2.2926,
+      "step": 1857
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.2987900148617369,
+      "learning_rate": 5.645172385587482e-05,
+      "loss": 2.2107,
+      "step": 1858
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.29901913973560484,
+      "learning_rate": 5.6109398020877834e-05,
+      "loss": 2.2657,
+      "step": 1859
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3042595478044556,
+      "learning_rate": 5.576805157409265e-05,
+      "loss": 2.1314,
+      "step": 1860
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.321778268323102,
+      "learning_rate": 5.542768526865677e-05,
+      "loss": 2.179,
+      "step": 1861
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.31930667144974134,
+      "learning_rate": 5.508829985554509e-05,
+      "loss": 2.2152,
+      "step": 1862
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3289508231212693,
+      "learning_rate": 5.474989608356856e-05,
+      "loss": 2.2689,
+      "step": 1863
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.3307569474521261,
+      "learning_rate": 5.441247469937194e-05,
+      "loss": 2.1806,
+      "step": 1864
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.311022545890804,
+      "learning_rate": 5.407603644743286e-05,
+      "loss": 2.2526,
+      "step": 1865
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3095400736470584,
+      "learning_rate": 5.374058207005944e-05,
+      "loss": 2.2345,
+      "step": 1866
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.33407528689282473,
+      "learning_rate": 5.3406112307389066e-05,
+      "loss": 2.2554,
+      "step": 1867
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.31417730769176105,
+      "learning_rate": 5.3072627897386926e-05,
+      "loss": 2.1135,
+      "step": 1868
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3127679701441873,
+      "learning_rate": 5.27401295758439e-05,
+      "loss": 2.2648,
+      "step": 1869
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.30700398297678655,
+      "learning_rate": 5.2408618076375315e-05,
+      "loss": 2.2653,
+      "step": 1870
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.31760076341495674,
+      "learning_rate": 5.207809413041914e-05,
+      "loss": 2.1024,
+      "step": 1871
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3214106589683353,
+      "learning_rate": 5.174855846723459e-05,
+      "loss": 2.2931,
+      "step": 1872
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3212372569804571,
+      "learning_rate": 5.1420011813900104e-05,
+      "loss": 2.179,
+      "step": 1873
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3220910847977557,
+      "learning_rate": 5.109245489531211e-05,
+      "loss": 2.2622,
+      "step": 1874
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.32716512216166943,
+      "learning_rate": 5.0765888434183446e-05,
+      "loss": 2.2442,
+      "step": 1875
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.32492724531691003,
+      "learning_rate": 5.0440313151041364e-05,
+      "loss": 2.2029,
+      "step": 1876
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3299561656975357,
+      "learning_rate": 5.011572976422657e-05,
+      "loss": 2.2022,
+      "step": 1877
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.40331402323053916,
+      "learning_rate": 4.9792138989890825e-05,
+      "loss": 2.2499,
+      "step": 1878
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.2857414528625281,
+      "learning_rate": 4.9469541541996234e-05,
+      "loss": 2.1591,
+      "step": 1879
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3161630691936204,
+      "learning_rate": 4.914793813231305e-05,
+      "loss": 2.2651,
+      "step": 1880
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.34111866507407584,
+      "learning_rate": 4.882732947041818e-05,
+      "loss": 2.2679,
+      "step": 1881
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3078913979473448,
+      "learning_rate": 4.850771626369416e-05,
+      "loss": 2.1103,
+      "step": 1882
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.33663630984224596,
+      "learning_rate": 4.818909921732662e-05,
+      "loss": 2.3009,
+      "step": 1883
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3137946208529559,
+      "learning_rate": 4.787147903430383e-05,
+      "loss": 2.2817,
+      "step": 1884
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.32421915180074196,
+      "learning_rate": 4.755485641541424e-05,
+      "loss": 2.3727,
+      "step": 1885
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.31302679805679423,
+      "learning_rate": 4.723923205924557e-05,
+      "loss": 2.2944,
+      "step": 1886
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.31236456588628236,
+      "learning_rate": 4.6924606662182736e-05,
+      "loss": 2.233,
+      "step": 1887
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.31946041262284985,
+      "learning_rate": 4.6610980918406596e-05,
+      "loss": 2.2084,
+      "step": 1888
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.31724254518881956,
+      "learning_rate": 4.629835551989276e-05,
+      "loss": 2.2879,
+      "step": 1889
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.31303865084674903,
+      "learning_rate": 4.5986731156409224e-05,
+      "loss": 2.1431,
+      "step": 1890
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3477398596799482,
+      "learning_rate": 4.567610851551568e-05,
+      "loss": 2.2563,
+      "step": 1891
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3375926359393871,
+      "learning_rate": 4.536648828256146e-05,
+      "loss": 2.3131,
+      "step": 1892
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3142263883964917,
+      "learning_rate": 4.505787114068433e-05,
+      "loss": 2.3582,
+      "step": 1893
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3299994753661849,
+      "learning_rate": 4.4750257770808764e-05,
+      "loss": 2.234,
+      "step": 1894
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.320444647384152,
+      "learning_rate": 4.444364885164448e-05,
+      "loss": 2.2131,
+      "step": 1895
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.32164192898252364,
+      "learning_rate": 4.413804505968533e-05,
+      "loss": 2.1776,
+      "step": 1896
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.31648706112067465,
+      "learning_rate": 4.3833447069206944e-05,
+      "loss": 2.2543,
+      "step": 1897
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.31720693804715805,
+      "learning_rate": 4.352985555226635e-05,
+      "loss": 2.3413,
+      "step": 1898
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3214622975207692,
+      "learning_rate": 4.322727117869951e-05,
+      "loss": 2.313,
+      "step": 1899
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3001318082914308,
+      "learning_rate": 4.29256946161205e-05,
+      "loss": 2.2165,
+      "step": 1900
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.32940679465488437,
+      "learning_rate": 4.262512652991968e-05,
+      "loss": 2.1843,
+      "step": 1901
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.32502909317651635,
+      "learning_rate": 4.2325567583262113e-05,
+      "loss": 2.3588,
+      "step": 1902
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3295303079280436,
+      "learning_rate": 4.2027018437086895e-05,
+      "loss": 2.2769,
+      "step": 1903
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.32266387332638,
+      "learning_rate": 4.172947975010449e-05,
+      "loss": 2.1829,
+      "step": 1904
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3026464341408635,
+      "learning_rate": 4.143295217879645e-05,
+      "loss": 2.3357,
+      "step": 1905
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.319662589389544,
+      "learning_rate": 4.113743637741296e-05,
+      "loss": 2.3644,
+      "step": 1906
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3142695360660741,
+      "learning_rate": 4.084293299797226e-05,
+      "loss": 2.2429,
+      "step": 1907
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.3120909504441209,
+      "learning_rate": 4.054944269025862e-05,
+      "loss": 2.3265,
+      "step": 1908
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3025167720363361,
+      "learning_rate": 4.025696610182095e-05,
+      "loss": 2.278,
+      "step": 1909
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.31476567750107215,
+      "learning_rate": 3.996550387797187e-05,
+      "loss": 2.1811,
+      "step": 1910
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3573411005570571,
+      "learning_rate": 3.9675056661785556e-05,
+      "loss": 2.1188,
+      "step": 1911
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.32836048938085854,
+      "learning_rate": 3.9385625094097154e-05,
+      "loss": 2.2738,
+      "step": 1912
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3106362657459327,
+      "learning_rate": 3.909720981350034e-05,
+      "loss": 2.3658,
+      "step": 1913
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.32392562310372236,
+      "learning_rate": 3.880981145634704e-05,
+      "loss": 2.3069,
+      "step": 1914
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.9891597589221309,
+      "learning_rate": 3.852343065674507e-05,
+      "loss": 2.244,
+      "step": 1915
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3185142554199155,
+      "learning_rate": 3.8238068046557276e-05,
+      "loss": 2.2872,
+      "step": 1916
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.2940485898560435,
+      "learning_rate": 3.795372425540006e-05,
+      "loss": 2.3674,
+      "step": 1917
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.316657841116818,
+      "learning_rate": 3.76703999106418e-05,
+      "loss": 2.19,
+      "step": 1918
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.31997957143976186,
+      "learning_rate": 3.7388095637401754e-05,
+      "loss": 2.2733,
+      "step": 1919
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3231895741706535,
+      "learning_rate": 3.7106812058548376e-05,
+      "loss": 2.187,
+      "step": 1920
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.31889721804190413,
+      "learning_rate": 3.682654979469807e-05,
+      "loss": 2.25,
+      "step": 1921
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3114229904779001,
+      "learning_rate": 3.654730946421403e-05,
+      "loss": 2.2414,
+      "step": 1922
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.312065520038211,
+      "learning_rate": 3.6269091683204466e-05,
+      "loss": 2.2767,
+      "step": 1923
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.31085584423530194,
+      "learning_rate": 3.5991897065521693e-05,
+      "loss": 2.253,
+      "step": 1924
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.332385738256831,
+      "learning_rate": 3.571572622276026e-05,
+      "loss": 2.3497,
+      "step": 1925
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3248803540000162,
+      "learning_rate": 3.544057976425619e-05,
+      "loss": 2.219,
+      "step": 1926
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3051622766879877,
+      "learning_rate": 3.5166458297085146e-05,
+      "loss": 2.1663,
+      "step": 1927
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.31218004668897525,
+      "learning_rate": 3.489336242606111e-05,
+      "loss": 2.2055,
+      "step": 1928
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3225762830639944,
+      "learning_rate": 3.462129275373577e-05,
+      "loss": 2.2572,
+      "step": 1929
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3094881441358841,
+      "learning_rate": 3.4350249880395924e-05,
+      "loss": 2.3599,
+      "step": 1930
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.327367543052288,
+      "learning_rate": 3.408023440406355e-05,
+      "loss": 2.2543,
+      "step": 1931
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.314120294840051,
+      "learning_rate": 3.381124692049331e-05,
+      "loss": 2.24,
+      "step": 1932
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3224831473820762,
+      "learning_rate": 3.354328802317197e-05,
+      "loss": 2.2446,
+      "step": 1933
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3263660314020832,
+      "learning_rate": 3.327635830331677e-05,
+      "loss": 2.22,
+      "step": 1934
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3129345057770403,
+      "learning_rate": 3.3010458349874206e-05,
+      "loss": 2.1563,
+      "step": 1935
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3058634456794572,
+      "learning_rate": 3.2745588749518775e-05,
+      "loss": 2.2056,
+      "step": 1936
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.309155669346349,
+      "learning_rate": 3.248175008665161e-05,
+      "loss": 2.335,
+      "step": 1937
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3139197927286616,
+      "learning_rate": 3.221894294339911e-05,
+      "loss": 2.2877,
+      "step": 1938
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.34265780570541754,
+      "learning_rate": 3.1957167899611836e-05,
+      "loss": 2.2645,
+      "step": 1939
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3205774924465258,
+      "learning_rate": 3.169642553286334e-05,
+      "loss": 2.1879,
+      "step": 1940
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3225590845806455,
+      "learning_rate": 3.143671641844831e-05,
+      "loss": 2.1623,
+      "step": 1941
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.31017259058326524,
+      "learning_rate": 3.117804112938205e-05,
+      "loss": 2.2325,
+      "step": 1942
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3317502203796404,
+      "learning_rate": 3.092040023639869e-05,
+      "loss": 2.2786,
+      "step": 1943
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.32238626199914794,
+      "learning_rate": 3.066379430795002e-05,
+      "loss": 2.2188,
+      "step": 1944
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3227483833752387,
+      "learning_rate": 3.040822391020459e-05,
+      "loss": 2.2548,
+      "step": 1945
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3038532757056525,
+      "learning_rate": 3.0153689607045842e-05,
+      "loss": 2.2813,
+      "step": 1946
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.33450881089944023,
+      "learning_rate": 2.9900191960071545e-05,
+      "loss": 2.1338,
+      "step": 1947
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.33629174299326275,
+      "learning_rate": 2.9647731528591848e-05,
+      "loss": 2.2122,
+      "step": 1948
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.31111617910532036,
+      "learning_rate": 2.9396308869628795e-05,
+      "loss": 2.2006,
+      "step": 1949
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3260963199309813,
+      "learning_rate": 2.914592453791448e-05,
+      "loss": 2.2546,
+      "step": 1950
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3141810198877753,
+      "learning_rate": 2.8896579085889994e-05,
+      "loss": 2.3343,
+      "step": 1951
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.32373733145523986,
+      "learning_rate": 2.86482730637046e-05,
+      "loss": 2.2392,
+      "step": 1952
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.31919922192446865,
+      "learning_rate": 2.840100701921383e-05,
+      "loss": 2.1405,
+      "step": 1953
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3051412902584087,
+      "learning_rate": 2.8154781497978898e-05,
+      "loss": 2.2773,
+      "step": 1954
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.34552412442063296,
+      "learning_rate": 2.7909597043265013e-05,
+      "loss": 2.2104,
+      "step": 1955
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3160305863222974,
+      "learning_rate": 2.7665454196040662e-05,
+      "loss": 2.2449,
+      "step": 1956
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.31444209740917417,
+      "learning_rate": 2.7422353494975905e-05,
+      "loss": 2.2077,
+      "step": 1957
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3268875716978743,
+      "learning_rate": 2.7180295476441573e-05,
+      "loss": 2.3275,
+      "step": 1958
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3107647687187392,
+      "learning_rate": 2.6939280674508016e-05,
+      "loss": 2.2474,
+      "step": 1959
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.31771500230710253,
+      "learning_rate": 2.669930962094358e-05,
+      "loss": 2.2555,
+      "step": 1960
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.30949975422013837,
+      "learning_rate": 2.6460382845214126e-05,
+      "loss": 2.2227,
+      "step": 1961
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3211964644232933,
+      "learning_rate": 2.6222500874481025e-05,
+      "loss": 2.2659,
+      "step": 1962
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.32214580507038226,
+      "learning_rate": 2.5985664233600827e-05,
+      "loss": 2.2923,
+      "step": 1963
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3282929244006287,
+      "learning_rate": 2.574987344512336e-05,
+      "loss": 2.2552,
+      "step": 1964
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3138898861964472,
+      "learning_rate": 2.5515129029290984e-05,
+      "loss": 2.2754,
+      "step": 1965
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.30737811875015236,
+      "learning_rate": 2.5281431504037556e-05,
+      "loss": 2.241,
+      "step": 1966
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.29884685785211973,
+      "learning_rate": 2.504878138498684e-05,
+      "loss": 2.1524,
+      "step": 1967
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3017597827130586,
+      "learning_rate": 2.48171791854519e-05,
+      "loss": 2.2648,
+      "step": 1968
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3118240575412669,
+      "learning_rate": 2.4586625416433473e-05,
+      "loss": 2.1764,
+      "step": 1969
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.33047785489023573,
+      "learning_rate": 2.435712058661921e-05,
+      "loss": 2.2243,
+      "step": 1970
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3100945607027959,
+      "learning_rate": 2.4128665202382327e-05,
+      "loss": 2.2227,
+      "step": 1971
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3138534677253731,
+      "learning_rate": 2.3901259767780515e-05,
+      "loss": 2.1522,
+      "step": 1972
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3037209599345978,
+      "learning_rate": 2.367490478455514e-05,
+      "loss": 2.2304,
+      "step": 1973
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.32126091844411186,
+      "learning_rate": 2.3449600752129597e-05,
+      "loss": 2.3136,
+      "step": 1974
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.34294729201909907,
+      "learning_rate": 2.3225348167608685e-05,
+      "loss": 2.3449,
+      "step": 1975
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.31100697813964506,
+      "learning_rate": 2.3002147525777118e-05,
+      "loss": 2.2475,
+      "step": 1976
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.30824844080214525,
+      "learning_rate": 2.2779999319098856e-05,
+      "loss": 2.3317,
+      "step": 1977
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3209822371730949,
+      "learning_rate": 2.255890403771571e-05,
+      "loss": 2.2457,
+      "step": 1978
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3354159792150786,
+      "learning_rate": 2.233886216944614e-05,
+      "loss": 2.1116,
+      "step": 1979
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3048438333371173,
+      "learning_rate": 2.211987419978484e-05,
+      "loss": 2.1834,
+      "step": 1980
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.31085931351220314,
+      "learning_rate": 2.1901940611900705e-05,
+      "loss": 2.2019,
+      "step": 1981
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3099302444947614,
+      "learning_rate": 2.168506188663666e-05,
+      "loss": 2.255,
+      "step": 1982
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3234788585483872,
+      "learning_rate": 2.1469238502507925e-05,
+      "loss": 2.3201,
+      "step": 1983
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.32296494201254805,
+      "learning_rate": 2.125447093570154e-05,
+      "loss": 2.1713,
+      "step": 1984
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3206269408302948,
+      "learning_rate": 2.1040759660074793e-05,
+      "loss": 2.2361,
+      "step": 1985
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.312848343536363,
+      "learning_rate": 2.0828105147154273e-05,
+      "loss": 2.251,
+      "step": 1986
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.30674759485267006,
+      "learning_rate": 2.061650786613545e-05,
+      "loss": 2.2354,
+      "step": 1987
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.32223722795520493,
+      "learning_rate": 2.040596828388058e-05,
+      "loss": 2.2945,
+      "step": 1988
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3500025776972687,
+      "learning_rate": 2.019648686491865e-05,
+      "loss": 2.3285,
+      "step": 1989
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.31553027856579524,
+      "learning_rate": 1.9988064071443767e-05,
+      "loss": 2.1785,
+      "step": 1990
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.34158139197526743,
+      "learning_rate": 1.9780700363314253e-05,
+      "loss": 2.3272,
+      "step": 1991
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.3428626325995146,
+      "learning_rate": 1.957439619805196e-05,
+      "loss": 2.216,
+      "step": 1992
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.31064043317318907,
+      "learning_rate": 1.9369152030840554e-05,
+      "loss": 2.2877,
+      "step": 1993
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.31611754604002973,
+      "learning_rate": 1.916496831452552e-05,
+      "loss": 2.2615,
+      "step": 1994
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.31039788798749374,
+      "learning_rate": 1.8961845499611998e-05,
+      "loss": 2.2895,
+      "step": 1995
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.32343153351170323,
+      "learning_rate": 1.8759784034264925e-05,
+      "loss": 2.251,
+      "step": 1996
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.31221677080736565,
+      "learning_rate": 1.855878436430708e-05,
+      "loss": 2.1871,
+      "step": 1997
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3214843211056369,
+      "learning_rate": 1.835884693321871e-05,
+      "loss": 2.2292,
+      "step": 1998
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.31457289156961277,
+      "learning_rate": 1.8159972182136386e-05,
+      "loss": 2.2012,
+      "step": 1999
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.32303069708322907,
+      "learning_rate": 1.7962160549851945e-05,
+      "loss": 2.2464,
+      "step": 2000
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3340152262276117,
+      "learning_rate": 1.776541247281177e-05,
+      "loss": 2.1945,
+      "step": 2001
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.31118210693948284,
+      "learning_rate": 1.7569728385115224e-05,
+      "loss": 2.1223,
+      "step": 2002
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3177173331014299,
+      "learning_rate": 1.7375108718514665e-05,
+      "loss": 2.2184,
+      "step": 2003
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.32351937657339896,
+      "learning_rate": 1.7181553902413438e-05,
+      "loss": 2.2463,
+      "step": 2004
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.33628149650162775,
+      "learning_rate": 1.698906436386577e-05,
+      "loss": 2.2067,
+      "step": 2005
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3126577317563477,
+      "learning_rate": 1.679764052757532e-05,
+      "loss": 2.1888,
+      "step": 2006
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.30847298756297103,
+      "learning_rate": 1.6607282815894464e-05,
+      "loss": 2.1758,
+      "step": 2007
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3139387489497425,
+      "learning_rate": 1.6417991648823405e-05,
+      "loss": 2.2872,
+      "step": 2008
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.31997128517603357,
+      "learning_rate": 1.6229767444008835e-05,
+      "loss": 2.1523,
+      "step": 2009
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3054547787231991,
+      "learning_rate": 1.604261061674378e-05,
+      "loss": 2.2702,
+      "step": 2010
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3249392118675122,
+      "learning_rate": 1.5856521579965865e-05,
+      "loss": 2.2447,
+      "step": 2011
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3042312395021212,
+      "learning_rate": 1.5671500744256938e-05,
+      "loss": 2.0759,
+      "step": 2012
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.32059308775038403,
+      "learning_rate": 1.5487548517841953e-05,
+      "loss": 2.313,
+      "step": 2013
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.30284885519854204,
+      "learning_rate": 1.530466530658814e-05,
+      "loss": 2.1819,
+      "step": 2014
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3106157574588663,
+      "learning_rate": 1.5122851514004054e-05,
+      "loss": 2.2298,
+      "step": 2015
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.3146181358860623,
+      "learning_rate": 1.4942107541238703e-05,
+      "loss": 2.2036,
+      "step": 2016
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.30717999374904187,
+      "learning_rate": 1.4762433787080809e-05,
+      "loss": 2.2478,
+      "step": 2017
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.33577763104138225,
+      "learning_rate": 1.4583830647957541e-05,
+      "loss": 2.3279,
+      "step": 2018
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.34095926570130697,
+      "learning_rate": 1.4406298517934068e-05,
+      "loss": 2.2849,
+      "step": 2019
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3339419457407624,
+      "learning_rate": 1.4229837788712562e-05,
+      "loss": 2.2862,
+      "step": 2020
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.30919078783507115,
+      "learning_rate": 1.4054448849631085e-05,
+      "loss": 2.2597,
+      "step": 2021
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3136239617761759,
+      "learning_rate": 1.3880132087663145e-05,
+      "loss": 2.2415,
+      "step": 2022
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3199539627098768,
+      "learning_rate": 1.3706887887416419e-05,
+      "loss": 2.1574,
+      "step": 2023
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3084034941910532,
+      "learning_rate": 1.3534716631132316e-05,
+      "loss": 2.1876,
+      "step": 2024
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.31095689127085147,
+      "learning_rate": 1.3363618698684853e-05,
+      "loss": 2.1969,
+      "step": 2025
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3049617690823154,
+      "learning_rate": 1.3193594467579728e-05,
+      "loss": 2.1964,
+      "step": 2026
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3043023589850268,
+      "learning_rate": 1.3024644312954026e-05,
+      "loss": 2.213,
+      "step": 2027
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.319869159346228,
+      "learning_rate": 1.2856768607574564e-05,
+      "loss": 2.1914,
+      "step": 2028
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.31615878595964986,
+      "learning_rate": 1.2689967721837947e-05,
+      "loss": 2.1959,
+      "step": 2029
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3156628166896705,
+      "learning_rate": 1.2524242023769006e-05,
+      "loss": 2.1284,
+      "step": 2030
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.32977725958068355,
+      "learning_rate": 1.2359591879020526e-05,
+      "loss": 2.2547,
+      "step": 2031
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.32745177960168276,
+      "learning_rate": 1.2196017650872081e-05,
+      "loss": 2.2148,
+      "step": 2032
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3189935242294841,
+      "learning_rate": 1.2033519700229367e-05,
+      "loss": 2.3129,
+      "step": 2033
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.33131718917178943,
+      "learning_rate": 1.1872098385623586e-05,
+      "loss": 2.2112,
+      "step": 2034
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.30045788061388207,
+      "learning_rate": 1.1711754063210289e-05,
+      "loss": 2.2096,
+      "step": 2035
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.32262879338694744,
+      "learning_rate": 1.155248708676887e-05,
+      "loss": 2.2121,
+      "step": 2036
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3261223801152808,
+      "learning_rate": 1.1394297807701737e-05,
+      "loss": 2.2901,
+      "step": 2037
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.30643661703059033,
+      "learning_rate": 1.1237186575033254e-05,
+      "loss": 2.2018,
+      "step": 2038
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.320503925905042,
+      "learning_rate": 1.1081153735409522e-05,
+      "loss": 2.2637,
+      "step": 2039
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.31091693013901733,
+      "learning_rate": 1.0926199633097156e-05,
+      "loss": 2.2118,
+      "step": 2040
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.32092914178737075,
+      "learning_rate": 1.0772324609982787e-05,
+      "loss": 2.2099,
+      "step": 2041
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.32005000316584453,
+      "learning_rate": 1.0619529005571893e-05,
+      "loss": 2.2248,
+      "step": 2042
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.30957803522797567,
+      "learning_rate": 1.0467813156988748e-05,
+      "loss": 2.1983,
+      "step": 2043
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3195445142830627,
+      "learning_rate": 1.0317177398975031e-05,
+      "loss": 2.2769,
+      "step": 2044
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3155060199539104,
+      "learning_rate": 1.0167622063889326e-05,
+      "loss": 2.2636,
+      "step": 2045
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3244875876347678,
+      "learning_rate": 1.0019147481706625e-05,
+      "loss": 2.2083,
+      "step": 2046
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3224597220427775,
+      "learning_rate": 9.871753980017051e-06,
+      "loss": 2.2434,
+      "step": 2047
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.32556909632396436,
+      "learning_rate": 9.725441884025855e-06,
+      "loss": 2.156,
+      "step": 2048
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3218892691990981,
+      "learning_rate": 9.580211516551862e-06,
+      "loss": 2.3111,
+      "step": 2049
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.32043369714700276,
+      "learning_rate": 9.436063198027589e-06,
+      "loss": 2.1773,
+      "step": 2050
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.32409065340195553,
+      "learning_rate": 9.292997246497959e-06,
+      "loss": 2.1762,
+      "step": 2051
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.31447722570926706,
+      "learning_rate": 9.151013977619693e-06,
+      "loss": 2.3237,
+      "step": 2052
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3231771007597212,
+      "learning_rate": 9.010113704661038e-06,
+      "loss": 2.2285,
+      "step": 2053
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3173551760670162,
+      "learning_rate": 8.870296738500316e-06,
+      "loss": 2.2569,
+      "step": 2054
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.318396331597529,
+      "learning_rate": 8.731563387626095e-06,
+      "loss": 2.2161,
+      "step": 2055
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.32201636180909965,
+      "learning_rate": 8.59391395813569e-06,
+      "loss": 2.2408,
+      "step": 2056
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.32021124722181177,
+      "learning_rate": 8.457348753735328e-06,
+      "loss": 2.2675,
+      "step": 2057
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3138927679780405,
+      "learning_rate": 8.321868075738593e-06,
+      "loss": 2.2331,
+      "step": 2058
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.30578162010613025,
+      "learning_rate": 8.187472223066371e-06,
+      "loss": 2.302,
+      "step": 2059
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.30197164095670936,
+      "learning_rate": 8.054161492246136e-06,
+      "loss": 2.2116,
+      "step": 2060
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.3113115660409807,
+      "learning_rate": 7.921936177411049e-06,
+      "loss": 2.2854,
+      "step": 2061
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3261577830523105,
+      "learning_rate": 7.790796570299463e-06,
+      "loss": 2.1449,
+      "step": 2062
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3212864306763931,
+      "learning_rate": 7.660742960254207e-06,
+      "loss": 2.2524,
+      "step": 2063
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3176572888496612,
+      "learning_rate": 7.531775634222138e-06,
+      "loss": 2.1564,
+      "step": 2064
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.30274840392424374,
+      "learning_rate": 7.403894876753192e-06,
+      "loss": 2.1849,
+      "step": 2065
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.32033040840750504,
+      "learning_rate": 7.277100970000061e-06,
+      "loss": 2.2653,
+      "step": 2066
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3108015943315728,
+      "learning_rate": 7.151394193717408e-06,
+      "loss": 2.1749,
+      "step": 2067
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3143824321107337,
+      "learning_rate": 7.026774825261151e-06,
+      "loss": 2.2361,
+      "step": 2068
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.31032254448390345,
+      "learning_rate": 6.903243139588233e-06,
+      "loss": 2.1651,
+      "step": 2069
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.32993637550003807,
+      "learning_rate": 6.780799409255522e-06,
+      "loss": 2.181,
+      "step": 2070
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.33811005533427835,
+      "learning_rate": 6.659443904419637e-06,
+      "loss": 2.4152,
+      "step": 2071
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.30500154875741525,
+      "learning_rate": 6.539176892836008e-06,
+      "loss": 2.2822,
+      "step": 2072
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3114986063348592,
+      "learning_rate": 6.4199986398585375e-06,
+      "loss": 2.2892,
+      "step": 2073
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3250163736892647,
+      "learning_rate": 6.3019094084388884e-06,
+      "loss": 2.2678,
+      "step": 2074
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3027168597111556,
+      "learning_rate": 6.18490945912592e-06,
+      "loss": 2.2417,
+      "step": 2075
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3205615240673123,
+      "learning_rate": 6.068999050065249e-06,
+      "loss": 2.2375,
+      "step": 2076
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3172332972735916,
+      "learning_rate": 5.9541784369983586e-06,
+      "loss": 2.2532,
+      "step": 2077
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.33392367873929163,
+      "learning_rate": 5.840447873262433e-06,
+      "loss": 2.2937,
+      "step": 2078
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3291596450921899,
+      "learning_rate": 5.727807609789471e-06,
+      "loss": 2.1704,
+      "step": 2079
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.311349935559895,
+      "learning_rate": 5.616257895105892e-06,
+      "loss": 2.1911,
+      "step": 2080
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3194318238993267,
+      "learning_rate": 5.505798975331933e-06,
+      "loss": 2.329,
+      "step": 2081
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.31580796103551934,
+      "learning_rate": 5.396431094181198e-06,
+      "loss": 2.2727,
+      "step": 2082
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3064607572627899,
+      "learning_rate": 5.288154492960107e-06,
+      "loss": 2.3064,
+      "step": 2083
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3128813063163827,
+      "learning_rate": 5.1809694105671155e-06,
+      "loss": 2.3252,
+      "step": 2084
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.32654376050706097,
+      "learning_rate": 5.074876083492441e-06,
+      "loss": 2.2291,
+      "step": 2085
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.31969796299741277,
+      "learning_rate": 4.96987474581767e-06,
+      "loss": 2.2971,
+      "step": 2086
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.322987500857923,
+      "learning_rate": 4.865965629214819e-06,
+      "loss": 2.112,
+      "step": 2087
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.31139578775431415,
+      "learning_rate": 4.763148962946218e-06,
+      "loss": 2.2078,
+      "step": 2088
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.30397669679456607,
+      "learning_rate": 4.661424973863681e-06,
+      "loss": 2.1656,
+      "step": 2089
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3270013808225471,
+      "learning_rate": 4.560793886408398e-06,
+      "loss": 2.2866,
+      "step": 2090
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.330475096602533,
+      "learning_rate": 4.461255922609986e-06,
+      "loss": 2.3277,
+      "step": 2091
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.31002495651096096,
+      "learning_rate": 4.362811302086267e-06,
+      "loss": 2.3033,
+      "step": 2092
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.2925134936225451,
+      "learning_rate": 4.265460242042885e-06,
+      "loss": 2.1789,
+      "step": 2093
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3180284034439,
+      "learning_rate": 4.169202957272522e-06,
+      "loss": 2.2082,
+      "step": 2094
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.31256621401354195,
+      "learning_rate": 4.074039660154738e-06,
+      "loss": 2.2608,
+      "step": 2095
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.33236787819908853,
+      "learning_rate": 3.9799705606551325e-06,
+      "loss": 2.3458,
+      "step": 2096
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.33572092435659323,
+      "learning_rate": 3.886995866325294e-06,
+      "loss": 2.1531,
+      "step": 2097
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.31875114264540866,
+      "learning_rate": 3.795115782302072e-06,
+      "loss": 2.3272,
+      "step": 2098
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.31284485178114574,
+      "learning_rate": 3.704330511307197e-06,
+      "loss": 2.1486,
+      "step": 2099
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.313037364498853,
+      "learning_rate": 3.614640253646828e-06,
+      "loss": 2.2428,
+      "step": 2100
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.31544498496616524,
+      "learning_rate": 3.5260452072110594e-06,
+      "loss": 2.3567,
+      "step": 2101
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3077600185232213,
+      "learning_rate": 3.4385455674737498e-06,
+      "loss": 2.1791,
+      "step": 2102
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.32330294627886896,
+      "learning_rate": 3.3521415274915256e-06,
+      "loss": 2.2555,
+      "step": 2103
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.3169576815709317,
+      "learning_rate": 3.2668332779041133e-06,
+      "loss": 2.2057,
+      "step": 2104
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.309639478576868,
+      "learning_rate": 3.1826210069332838e-06,
+      "loss": 2.2059,
+      "step": 2105
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3197328066065118,
+      "learning_rate": 3.0995049003826324e-06,
+      "loss": 2.23,
+      "step": 2106
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3304261329569436,
+      "learning_rate": 3.017485141637355e-06,
+      "loss": 2.1826,
+      "step": 2107
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.32029442219468396,
+      "learning_rate": 2.9365619116636376e-06,
+      "loss": 2.2685,
+      "step": 2108
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.32591814608818637,
+      "learning_rate": 2.856735389008269e-06,
+      "loss": 2.1946,
+      "step": 2109
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.32507210688175103,
+      "learning_rate": 2.778005749798307e-06,
+      "loss": 2.1971,
+      "step": 2110
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3136440914288422,
+      "learning_rate": 2.700373167740744e-06,
+      "loss": 2.2098,
+      "step": 2111
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.31991962804562857,
+      "learning_rate": 2.62383781412201e-06,
+      "loss": 2.188,
+      "step": 2112
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.31710866751755606,
+      "learning_rate": 2.5483998578076373e-06,
+      "loss": 2.1538,
+      "step": 2113
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.305343809134597,
+      "learning_rate": 2.4740594652418736e-06,
+      "loss": 2.1269,
+      "step": 2114
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3310794969771493,
+      "learning_rate": 2.4008168004472917e-06,
+      "loss": 2.1947,
+      "step": 2115
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.31759218856634197,
+      "learning_rate": 2.3286720250246253e-06,
+      "loss": 2.2023,
+      "step": 2116
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3368691228400291,
+      "learning_rate": 2.2576252981520994e-06,
+      "loss": 2.1632,
+      "step": 2117
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3234596690388099,
+      "learning_rate": 2.1876767765853233e-06,
+      "loss": 2.1882,
+      "step": 2118
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3250733742657914,
+      "learning_rate": 2.118826614656788e-06,
+      "loss": 2.1735,
+      "step": 2119
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3179173661683933,
+      "learning_rate": 2.051074964275701e-06,
+      "loss": 2.2572,
+      "step": 2120
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.31347298559479575,
+      "learning_rate": 1.984421974927375e-06,
+      "loss": 2.1896,
+      "step": 2121
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3154363233216064,
+      "learning_rate": 1.9188677936731734e-06,
+      "loss": 2.2111,
+      "step": 2122
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3097785712198237,
+      "learning_rate": 1.8544125651501208e-06,
+      "loss": 2.2941,
+      "step": 2123
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.311618782334207,
+      "learning_rate": 1.7910564315704035e-06,
+      "loss": 2.2349,
+      "step": 2124
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.3159579437331596,
+      "learning_rate": 1.7287995327214257e-06,
+      "loss": 2.1823,
+      "step": 2125
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.321660716733451,
+      "learning_rate": 1.6676420059649754e-06,
+      "loss": 2.2085,
+      "step": 2126
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.32822950009057356,
+      "learning_rate": 1.6075839862374485e-06,
+      "loss": 2.2292,
+      "step": 2127
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.32055119087323825,
+      "learning_rate": 1.5486256060492366e-06,
+      "loss": 2.2553,
+      "step": 2128
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3145625396676682,
+      "learning_rate": 1.4907669954844495e-06,
+      "loss": 2.2398,
+      "step": 2129
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3133873045127362,
+      "learning_rate": 1.434008282200805e-06,
+      "loss": 2.1101,
+      "step": 2130
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3179571102960596,
+      "learning_rate": 1.3783495914291844e-06,
+      "loss": 2.173,
+      "step": 2131
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.33324180544761123,
+      "learning_rate": 1.3237910459734104e-06,
+      "loss": 2.285,
+      "step": 2132
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3292133648176681,
+      "learning_rate": 1.270332766210025e-06,
+      "loss": 2.2298,
+      "step": 2133
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.31864433404432807,
+      "learning_rate": 1.2179748700879012e-06,
+      "loss": 2.2384,
+      "step": 2134
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3207201356318713,
+      "learning_rate": 1.1667174731280205e-06,
+      "loss": 2.2882,
+      "step": 2135
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3159082132805986,
+      "learning_rate": 1.1165606884234182e-06,
+      "loss": 2.2756,
+      "step": 2136
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3018203588931066,
+      "learning_rate": 1.0675046266386268e-06,
+      "loss": 2.251,
+      "step": 2137
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.32808931587216655,
+      "learning_rate": 1.019549396009567e-06,
+      "loss": 2.2109,
+      "step": 2138
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.31546873541852566,
+      "learning_rate": 9.726951023434348e-07,
+      "loss": 2.2537,
+      "step": 2139
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.32627120555658057,
+      "learning_rate": 9.269418490182591e-07,
+      "loss": 2.2336,
+      "step": 2140
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3246167125271015,
+      "learning_rate": 8.822897369827332e-07,
+      "loss": 2.2997,
+      "step": 2141
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.32115802896635054,
+      "learning_rate": 8.387388647561611e-07,
+      "loss": 2.2001,
+      "step": 2142
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3707422678835187,
+      "learning_rate": 7.962893284279016e-07,
+      "loss": 2.242,
+      "step": 2143
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3107747729543837,
+      "learning_rate": 7.549412216574791e-07,
+      "loss": 2.2977,
+      "step": 2144
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3158879352472702,
+      "learning_rate": 7.146946356743067e-07,
+      "loss": 2.173,
+      "step": 2145
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3050989459572452,
+      "learning_rate": 6.755496592773524e-07,
+      "loss": 2.29,
+      "step": 2146
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.315183998736443,
+      "learning_rate": 6.375063788349733e-07,
+      "loss": 2.3428,
+      "step": 2147
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.3221225263454234,
+      "learning_rate": 6.005648782848594e-07,
+      "loss": 2.3475,
+      "step": 2148
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.32161040918778133,
+      "learning_rate": 5.647252391337565e-07,
+      "loss": 2.2356,
+      "step": 2149
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.315384436260205,
+      "learning_rate": 5.299875404572441e-07,
+      "loss": 2.1848,
+      "step": 2150
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.32417251664311475,
+      "learning_rate": 4.963518588996796e-07,
+      "loss": 2.1516,
+      "step": 2151
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3134746239933467,
+      "learning_rate": 4.638182686738657e-07,
+      "loss": 2.2578,
+      "step": 2152
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3136301229440725,
+      "learning_rate": 4.3238684156110543e-07,
+      "loss": 2.2513,
+      "step": 2153
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.38630163627889885,
+      "learning_rate": 4.020576469108139e-07,
+      "loss": 2.1517,
+      "step": 2154
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.30862174356322336,
+      "learning_rate": 3.7283075164046274e-07,
+      "loss": 2.2148,
+      "step": 2155
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.321734320190624,
+      "learning_rate": 3.4470622023557995e-07,
+      "loss": 2.2667,
+      "step": 2156
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3061375263655152,
+      "learning_rate": 3.176841147492504e-07,
+      "loss": 2.125,
+      "step": 2157
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3245929212613787,
+      "learning_rate": 2.9176449480244895e-07,
+      "loss": 2.1452,
+      "step": 2158
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3068686951967255,
+      "learning_rate": 2.6694741758342967e-07,
+      "loss": 2.2978,
+      "step": 2159
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3121218810065348,
+      "learning_rate": 2.432329378478926e-07,
+      "loss": 2.2672,
+      "step": 2160
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3535082785702842,
+      "learning_rate": 2.2062110791892798e-07,
+      "loss": 2.3504,
+      "step": 2161
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.31491345900740025,
+      "learning_rate": 1.9911197768662792e-07,
+      "loss": 2.2309,
+      "step": 2162
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.32463478156259307,
+      "learning_rate": 1.7870559460814173e-07,
+      "loss": 2.2175,
+      "step": 2163
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3204909422042002,
+      "learning_rate": 1.5940200370750947e-07,
+      "loss": 2.2743,
+      "step": 2164
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.33778789699551887,
+      "learning_rate": 1.4120124757577291e-07,
+      "loss": 2.2995,
+      "step": 2165
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3182301606140714,
+      "learning_rate": 1.2410336637047603e-07,
+      "loss": 2.2083,
+      "step": 2166
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3252525719983012,
+      "learning_rate": 1.081083978159425e-07,
+      "loss": 2.2841,
+      "step": 2167
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3159830015095795,
+      "learning_rate": 9.321637720310915e-08,
+      "loss": 2.3115,
+      "step": 2168
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.33621276138884454,
+      "learning_rate": 7.942733738924845e-08,
+      "loss": 2.1422,
+      "step": 2169
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.3545164848151948,
+      "learning_rate": 6.6741308798135e-08,
+      "loss": 2.1546,
+      "step": 2170
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3362881372061856,
+      "learning_rate": 5.5158319419934546e-08,
+      "loss": 2.2497,
+      "step": 2171
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.2985318778423691,
+      "learning_rate": 4.4678394810981906e-08,
+      "loss": 2.1508,
+      "step": 2172
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.2973742723920964,
+      "learning_rate": 3.5301558093947527e-08,
+      "loss": 2.1748,
+      "step": 2173
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3193964736424585,
+      "learning_rate": 2.7027829957559925e-08,
+      "loss": 2.1932,
+      "step": 2174
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.31686681390332955,
+      "learning_rate": 1.985722865682771e-08,
+      "loss": 2.3222,
+      "step": 2175
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.31680090814641054,
+      "learning_rate": 1.3789770012762048e-08,
+      "loss": 2.2008,
+      "step": 2176
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.32381143333686824,
+      "learning_rate": 8.825467412376665e-09,
+      "loss": 2.3238,
+      "step": 2177
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.31316442923675314,
+      "learning_rate": 4.96433180879885e-09,
+      "loss": 2.1443,
+      "step": 2178
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.31013294020063653,
+      "learning_rate": 2.206371721158451e-09,
+      "loss": 2.1727,
+      "step": 2179
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.3250368336472977,
+      "learning_rate": 5.515932345323549e-10,
+      "loss": 2.3571,
+      "step": 2180
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.32953541251446106,
+      "learning_rate": 0.0,
+      "loss": 2.2469,
+      "step": 2181
+    },
+    {
+      "epoch": 1.0,
+      "step": 2181,
+      "total_flos": 59888570243072.0,
+      "train_loss": 2.424219549385466,
+      "train_runtime": 8137.5064,
+      "train_samples_per_second": 68.587,
+      "train_steps_per_second": 0.268
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2181,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 50000.0,
+  "total_flos": 59888570243072.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}