diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4417 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9998799759951991,
+  "eval_steps": 500,
+  "global_step": 6248,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00016003200640128025,
+      "grad_norm": 1.1826485395431519,
+      "learning_rate": 8e-09,
+      "loss": 1.8238,
+      "step": 1
+    },
+    {
+      "epoch": 0.0016003200640128026,
+      "grad_norm": 1.2571839094161987,
+      "learning_rate": 8e-08,
+      "loss": 1.8529,
+      "step": 10
+    },
+    {
+      "epoch": 0.003200640128025605,
+      "grad_norm": 1.3447990417480469,
+      "learning_rate": 1.6e-07,
+      "loss": 1.8769,
+      "step": 20
+    },
+    {
+      "epoch": 0.004800960192038408,
+      "grad_norm": 1.0940663814544678,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 1.9161,
+      "step": 30
+    },
+    {
+      "epoch": 0.00640128025605121,
+      "grad_norm": 1.1993372440338135,
+      "learning_rate": 3.2e-07,
+      "loss": 1.8078,
+      "step": 40
+    },
+    {
+      "epoch": 0.008001600320064013,
+      "grad_norm": 1.06308913230896,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 1.9163,
+      "step": 50
+    },
+    {
+      "epoch": 0.009601920384076815,
+      "grad_norm": 1.1543391942977905,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 1.8377,
+      "step": 60
+    },
+    {
+      "epoch": 0.011202240448089618,
+      "grad_norm": 0.8527859449386597,
+      "learning_rate": 5.6e-07,
+      "loss": 1.8549,
+      "step": 70
+    },
+    {
+      "epoch": 0.01280256051210242,
+      "grad_norm": 0.9085479974746704,
+      "learning_rate": 6.4e-07,
+      "loss": 1.789,
+      "step": 80
+    },
+    {
+      "epoch": 0.014402880576115223,
+      "grad_norm": 0.7586134672164917,
+      "learning_rate": 7.2e-07,
+      "loss": 1.7819,
+      "step": 90
+    },
+    {
+      "epoch": 0.016003200640128026,
+      "grad_norm": 0.8043696284294128,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.7761,
+      "step": 100
+    },
+    {
+      "epoch": 0.017603520704140826,
+      "grad_norm": 0.7043126821517944,
+      "learning_rate": 8.8e-07,
+      "loss": 1.7702,
+      "step": 110
+    },
+    {
+      "epoch": 0.01920384076815363,
+      "grad_norm": 0.6750304698944092,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 1.7808,
+      "step": 120
+    },
+    {
+      "epoch": 0.02080416083216643,
+      "grad_norm": 0.6183897852897644,
+      "learning_rate": 1.04e-06,
+      "loss": 1.7655,
+      "step": 130
+    },
+    {
+      "epoch": 0.022404480896179236,
+      "grad_norm": 0.6176135540008545,
+      "learning_rate": 1.12e-06,
+      "loss": 1.7363,
+      "step": 140
+    },
+    {
+      "epoch": 0.024004800960192037,
+      "grad_norm": 0.5962868928909302,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 1.7132,
+      "step": 150
+    },
+    {
+      "epoch": 0.02560512102420484,
+      "grad_norm": 0.5577073693275452,
+      "learning_rate": 1.28e-06,
+      "loss": 1.7228,
+      "step": 160
+    },
+    {
+      "epoch": 0.027205441088217642,
+      "grad_norm": 0.5972305536270142,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 1.7134,
+      "step": 170
+    },
+    {
+      "epoch": 0.028805761152230446,
+      "grad_norm": 0.891796886920929,
+      "learning_rate": 1.44e-06,
+      "loss": 1.677,
+      "step": 180
+    },
+    {
+      "epoch": 0.030406081216243247,
+      "grad_norm": 0.500298023223877,
+      "learning_rate": 1.52e-06,
+      "loss": 1.6674,
+      "step": 190
+    },
+    {
+      "epoch": 0.03200640128025605,
+      "grad_norm": 0.4819110929965973,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.6432,
+      "step": 200
+    },
+    {
+      "epoch": 0.033606721344268856,
+      "grad_norm": 0.48815402388572693,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 1.6579,
+      "step": 210
+    },
+    {
+      "epoch": 0.03520704140828165,
+      "grad_norm": 0.43313154578208923,
+      "learning_rate": 1.76e-06,
+      "loss": 1.5858,
+      "step": 220
+    },
+    {
+      "epoch": 0.03680736147229446,
+      "grad_norm": 0.4705139100551605,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 1.6045,
+      "step": 230
+    },
+    {
+      "epoch": 0.03840768153630726,
+      "grad_norm": 0.46406659483909607,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 1.6324,
+      "step": 240
+    },
+    {
+      "epoch": 0.040008001600320066,
+      "grad_norm": 0.46592777967453003,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.601,
+      "step": 250
+    },
+    {
+      "epoch": 0.04160832166433286,
+      "grad_norm": 0.477504163980484,
+      "learning_rate": 2.08e-06,
+      "loss": 1.5639,
+      "step": 260
+    },
+    {
+      "epoch": 0.04320864172834567,
+      "grad_norm": 0.45321983098983765,
+      "learning_rate": 2.16e-06,
+      "loss": 1.5627,
+      "step": 270
+    },
+    {
+      "epoch": 0.04480896179235847,
+      "grad_norm": 0.463728129863739,
+      "learning_rate": 2.24e-06,
+      "loss": 1.5359,
+      "step": 280
+    },
+    {
+      "epoch": 0.046409281856371276,
+      "grad_norm": 0.43756529688835144,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 1.4996,
+      "step": 290
+    },
+    {
+      "epoch": 0.048009601920384073,
+      "grad_norm": 0.4395667314529419,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.51,
+      "step": 300
+    },
+    {
+      "epoch": 0.04960992198439688,
+      "grad_norm": 0.44691407680511475,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 1.4876,
+      "step": 310
+    },
+    {
+      "epoch": 0.05121024204840968,
+      "grad_norm": 0.4260852336883545,
+      "learning_rate": 2.56e-06,
+      "loss": 1.4463,
+      "step": 320
+    },
+    {
+      "epoch": 0.052810562112422486,
+      "grad_norm": 0.4296850264072418,
+      "learning_rate": 2.64e-06,
+      "loss": 1.4195,
+      "step": 330
+    },
+    {
+      "epoch": 0.054410882176435284,
+      "grad_norm": 0.3987746834754944,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 1.4248,
+      "step": 340
+    },
+    {
+      "epoch": 0.05601120224044809,
+      "grad_norm": 0.38749226927757263,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 1.4151,
+      "step": 350
+    },
+    {
+      "epoch": 0.05761152230446089,
+      "grad_norm": 0.3994350731372833,
+      "learning_rate": 2.88e-06,
+      "loss": 1.4625,
+      "step": 360
+    },
+    {
+      "epoch": 0.0592118423684737,
+      "grad_norm": 0.36386868357658386,
+      "learning_rate": 2.96e-06,
+      "loss": 1.4048,
+      "step": 370
+    },
+    {
+      "epoch": 0.060812162432486494,
+      "grad_norm": 0.3657371699810028,
+      "learning_rate": 3.04e-06,
+      "loss": 1.4143,
+      "step": 380
+    },
+    {
+      "epoch": 0.0624124824964993,
+      "grad_norm": 0.3797203302383423,
+      "learning_rate": 3.12e-06,
+      "loss": 1.4394,
+      "step": 390
+    },
+    {
+      "epoch": 0.0640128025605121,
+      "grad_norm": 0.36554431915283203,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 1.3839,
+      "step": 400
+    },
+    {
+      "epoch": 0.0656131226245249,
+      "grad_norm": 0.3783222734928131,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 1.4041,
+      "step": 410
+    },
+    {
+      "epoch": 0.06721344268853771,
+      "grad_norm": 0.36894097924232483,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 1.3962,
+      "step": 420
+    },
+    {
+      "epoch": 0.06881376275255051,
+      "grad_norm": 0.3558606207370758,
+      "learning_rate": 3.44e-06,
+      "loss": 1.4449,
+      "step": 430
+    },
+    {
+      "epoch": 0.0704140828165633,
+      "grad_norm": 0.5871737003326416,
+      "learning_rate": 3.52e-06,
+      "loss": 1.4062,
+      "step": 440
+    },
+    {
+      "epoch": 0.07201440288057612,
+      "grad_norm": 0.3803611099720001,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 1.3951,
+      "step": 450
+    },
+    {
+      "epoch": 0.07361472294458891,
+      "grad_norm": 0.37580356001853943,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 1.3606,
+      "step": 460
+    },
+    {
+      "epoch": 0.07521504300860173,
+      "grad_norm": 0.3693610429763794,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 1.4153,
+      "step": 470
+    },
+    {
+      "epoch": 0.07681536307261452,
+      "grad_norm": 0.3637755811214447,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 1.4622,
+      "step": 480
+    },
+    {
+      "epoch": 0.07841568313662732,
+      "grad_norm": 0.3501175343990326,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 1.4107,
+      "step": 490
+    },
+    {
+      "epoch": 0.08001600320064013,
+      "grad_norm": 2.7416932582855225,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.3932,
+      "step": 500
+    },
+    {
+      "epoch": 0.08161632326465293,
+      "grad_norm": 0.7613168358802795,
+      "learning_rate": 4.08e-06,
+      "loss": 1.3947,
+      "step": 510
+    },
+    {
+      "epoch": 0.08321664332866573,
+      "grad_norm": 0.34814804792404175,
+      "learning_rate": 4.16e-06,
+      "loss": 1.4168,
+      "step": 520
+    },
+    {
+      "epoch": 0.08481696339267854,
+      "grad_norm": 0.3658451735973358,
+      "learning_rate": 4.24e-06,
+      "loss": 1.383,
+      "step": 530
+    },
+    {
+      "epoch": 0.08641728345669134,
+      "grad_norm": 0.3520263135433197,
+      "learning_rate": 4.32e-06,
+      "loss": 1.3419,
+      "step": 540
+    },
+    {
+      "epoch": 0.08801760352070415,
+      "grad_norm": 0.37928885221481323,
+      "learning_rate": 4.4e-06,
+      "loss": 1.3982,
+      "step": 550
+    },
+    {
+      "epoch": 0.08961792358471694,
+      "grad_norm": 1.466685175895691,
+      "learning_rate": 4.48e-06,
+      "loss": 1.3914,
+      "step": 560
+    },
+    {
+      "epoch": 0.09121824364872974,
+      "grad_norm": 0.37202754616737366,
+      "learning_rate": 4.56e-06,
+      "loss": 1.3421,
+      "step": 570
+    },
+    {
+      "epoch": 0.09281856371274255,
+      "grad_norm": 0.37455129623413086,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 1.3564,
+      "step": 580
+    },
+    {
+      "epoch": 0.09441888377675535,
+      "grad_norm": 0.3516671359539032,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 1.3219,
+      "step": 590
+    },
+    {
+      "epoch": 0.09601920384076815,
+      "grad_norm": 0.37462764978408813,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.3639,
+      "step": 600
+    },
+    {
+      "epoch": 0.09761952390478096,
+      "grad_norm": 0.3746024966239929,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 1.3789,
+      "step": 610
+    },
+    {
+      "epoch": 0.09921984396879376,
+      "grad_norm": 0.3724232316017151,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 1.3499,
+      "step": 620
+    },
+    {
+      "epoch": 0.10082016403280657,
+      "grad_norm": 0.38169583678245544,
+      "learning_rate": 4.999990245313474e-06,
+      "loss": 1.3086,
+      "step": 630
+    },
+    {
+      "epoch": 0.10242048409681936,
+      "grad_norm": 0.3765912353992462,
+      "learning_rate": 4.999912208277997e-06,
+      "loss": 1.3307,
+      "step": 640
+    },
+    {
+      "epoch": 0.10402080416083216,
+      "grad_norm": 0.38425111770629883,
+      "learning_rate": 4.999756136642964e-06,
+      "loss": 1.3168,
+      "step": 650
+    },
+    {
+      "epoch": 0.10562112422484497,
+      "grad_norm": 0.38071689009666443,
+      "learning_rate": 4.999522035280142e-06,
+      "loss": 1.3717,
+      "step": 660
+    },
+    {
+      "epoch": 0.10722144428885777,
+      "grad_norm": 0.3890415132045746,
+      "learning_rate": 4.999209911496989e-06,
+      "loss": 1.362,
+      "step": 670
+    },
+    {
+      "epoch": 0.10882176435287057,
+      "grad_norm": 0.3932015001773834,
+      "learning_rate": 4.99881977503643e-06,
+      "loss": 1.3367,
+      "step": 680
+    },
+    {
+      "epoch": 0.11042208441688338,
+      "grad_norm": 0.39485597610473633,
+      "learning_rate": 4.998351638076548e-06,
+      "loss": 1.3263,
+      "step": 690
+    },
+    {
+      "epoch": 0.11202240448089618,
+      "grad_norm": 0.40244272351264954,
+      "learning_rate": 4.99780551523021e-06,
+      "loss": 1.3483,
+      "step": 700
+    },
+    {
+      "epoch": 0.11362272454490899,
+      "grad_norm": 0.38777467608451843,
+      "learning_rate": 4.9971814235446035e-06,
+      "loss": 1.4173,
+      "step": 710
+    },
+    {
+      "epoch": 0.11522304460892178,
+      "grad_norm": 0.3693300485610962,
+      "learning_rate": 4.996479382500714e-06,
+      "loss": 1.3711,
+      "step": 720
+    },
+    {
+      "epoch": 0.11682336467293458,
+      "grad_norm": 0.39682939648628235,
+      "learning_rate": 4.995699414012704e-06,
+      "loss": 1.2961,
+      "step": 730
+    },
+    {
+      "epoch": 0.1184236847369474,
+      "grad_norm": 0.4018958806991577,
+      "learning_rate": 4.994841542427242e-06,
+      "loss": 1.3109,
+      "step": 740
+    },
+    {
+      "epoch": 0.12002400480096019,
+      "grad_norm": 0.3853909969329834,
+      "learning_rate": 4.993905794522734e-06,
+      "loss": 1.3114,
+      "step": 750
+    },
+    {
+      "epoch": 0.12162432486497299,
+      "grad_norm": 0.4010235071182251,
+      "learning_rate": 4.992892199508492e-06,
+      "loss": 1.3348,
+      "step": 760
+    },
+    {
+      "epoch": 0.1232246449289858,
+      "grad_norm": 0.3733123242855072,
+      "learning_rate": 4.991800789023818e-06,
+      "loss": 1.324,
+      "step": 770
+    },
+    {
+      "epoch": 0.1248249649929986,
+      "grad_norm": 0.4108160138130188,
+      "learning_rate": 4.990631597137021e-06,
+      "loss": 1.2966,
+      "step": 780
+    },
+    {
+      "epoch": 0.1264252850570114,
+      "grad_norm": 0.41119498014450073,
+      "learning_rate": 4.989384660344351e-06,
+      "loss": 1.3555,
+      "step": 790
+    },
+    {
+      "epoch": 0.1280256051210242,
+      "grad_norm": 0.39664438366889954,
+      "learning_rate": 4.9880600175688596e-06,
+      "loss": 1.326,
+      "step": 800
+    },
+    {
+      "epoch": 0.12962592518503702,
+      "grad_norm": 0.40418845415115356,
+      "learning_rate": 4.986657710159187e-06,
+      "loss": 1.3018,
+      "step": 810
+    },
+    {
+      "epoch": 0.1312262452490498,
+      "grad_norm": 0.4098711609840393,
+      "learning_rate": 4.98517778188827e-06,
+      "loss": 1.3963,
+      "step": 820
+    },
+    {
+      "epoch": 0.1328265653130626,
+      "grad_norm": 0.43664970993995667,
+      "learning_rate": 4.983620278951973e-06,
+      "loss": 1.3343,
+      "step": 830
+    },
+    {
+      "epoch": 0.13442688537707542,
+      "grad_norm": 0.4087508022785187,
+      "learning_rate": 4.981985249967651e-06,
+      "loss": 1.2787,
+      "step": 840
+    },
+    {
+      "epoch": 0.1360272054410882,
+      "grad_norm": 0.38437142968177795,
+      "learning_rate": 4.980272745972629e-06,
+      "loss": 1.3563,
+      "step": 850
+    },
+    {
+      "epoch": 0.13762752550510102,
+      "grad_norm": 0.41200581192970276,
+      "learning_rate": 4.978482820422612e-06,
+      "loss": 1.3338,
+      "step": 860
+    },
+    {
+      "epoch": 0.13922784556911383,
+      "grad_norm": 0.40172791481018066,
+      "learning_rate": 4.976615529190008e-06,
+      "loss": 1.3087,
+      "step": 870
+    },
+    {
+      "epoch": 0.1408281656331266,
+      "grad_norm": 0.44374722242355347,
+      "learning_rate": 4.974670930562195e-06,
+      "loss": 1.3104,
+      "step": 880
+    },
+    {
+      "epoch": 0.14242848569713942,
+      "grad_norm": 0.4120323657989502,
+      "learning_rate": 4.972649085239694e-06,
+      "loss": 1.303,
+      "step": 890
+    },
+    {
+      "epoch": 0.14402880576115223,
+      "grad_norm": 0.42948517203330994,
+      "learning_rate": 4.970550056334277e-06,
+      "loss": 1.2743,
+      "step": 900
+    },
+    {
+      "epoch": 0.14562912582516505,
+      "grad_norm": 0.4419565498828888,
+      "learning_rate": 4.968373909366998e-06,
+      "loss": 1.3082,
+      "step": 910
+    },
+    {
+      "epoch": 0.14722944588917783,
+      "grad_norm": 0.411284476518631,
+      "learning_rate": 4.966120712266146e-06,
+      "loss": 1.3116,
+      "step": 920
+    },
+    {
+      "epoch": 0.14882976595319064,
+      "grad_norm": 0.4247114062309265,
+      "learning_rate": 4.963790535365122e-06,
+      "loss": 1.3486,
+      "step": 930
+    },
+    {
+      "epoch": 0.15043008601720345,
+      "grad_norm": 0.4018177092075348,
+      "learning_rate": 4.961383451400251e-06,
+      "loss": 1.3029,
+      "step": 940
+    },
+    {
+      "epoch": 0.15203040608121624,
+      "grad_norm": 0.40833476185798645,
+      "learning_rate": 4.958899535508503e-06,
+      "loss": 1.2706,
+      "step": 950
+    },
+    {
+      "epoch": 0.15363072614522905,
+      "grad_norm": 0.41154181957244873,
+      "learning_rate": 4.956338865225155e-06,
+      "loss": 1.3063,
+      "step": 960
+    },
+    {
+      "epoch": 0.15523104620924186,
+      "grad_norm": 0.42829248309135437,
+      "learning_rate": 4.953701520481365e-06,
+      "loss": 1.3119,
+      "step": 970
+    },
+    {
+      "epoch": 0.15683136627325464,
+      "grad_norm": 0.42823630571365356,
+      "learning_rate": 4.950987583601681e-06,
+      "loss": 1.276,
+      "step": 980
+    },
+    {
+      "epoch": 0.15843168633726745,
+      "grad_norm": 0.4209652841091156,
+      "learning_rate": 4.9481971393014686e-06,
+      "loss": 1.2947,
+      "step": 990
+    },
+    {
+      "epoch": 0.16003200640128026,
+      "grad_norm": 0.4024667739868164,
+      "learning_rate": 4.945330274684268e-06,
+      "loss": 1.2704,
+      "step": 1000
+    },
+    {
+      "epoch": 0.16163232646529305,
+      "grad_norm": 0.4125162363052368,
+      "learning_rate": 4.942387079239075e-06,
+      "loss": 1.2686,
+      "step": 1010
+    },
+    {
+      "epoch": 0.16323264652930586,
+      "grad_norm": 0.39515283703804016,
+      "learning_rate": 4.9393676448375475e-06,
+      "loss": 1.2875,
+      "step": 1020
+    },
+    {
+      "epoch": 0.16483296659331867,
+      "grad_norm": 0.42820730805397034,
+      "learning_rate": 4.936272065731135e-06,
+      "loss": 1.2633,
+      "step": 1030
+    },
+    {
+      "epoch": 0.16643328665733145,
+      "grad_norm": 0.43181583285331726,
+      "learning_rate": 4.933100438548144e-06,
+      "loss": 1.3641,
+      "step": 1040
+    },
+    {
+      "epoch": 0.16803360672134426,
+      "grad_norm": 0.4250759482383728,
+      "learning_rate": 4.929852862290711e-06,
+      "loss": 1.3123,
+      "step": 1050
+    },
+    {
+      "epoch": 0.16963392678535708,
+      "grad_norm": 0.4282747805118561,
+      "learning_rate": 4.926529438331724e-06,
+      "loss": 1.3036,
+      "step": 1060
+    },
+    {
+      "epoch": 0.1712342468493699,
+      "grad_norm": 0.5093165040016174,
+      "learning_rate": 4.923130270411648e-06,
+      "loss": 1.2802,
+      "step": 1070
+    },
+    {
+      "epoch": 0.17283456691338267,
+      "grad_norm": 0.4356524348258972,
+      "learning_rate": 4.919655464635291e-06,
+      "loss": 1.3567,
+      "step": 1080
+    },
+    {
+      "epoch": 0.17443488697739548,
+      "grad_norm": 0.4331202507019043,
+      "learning_rate": 4.916105129468493e-06,
+      "loss": 1.3056,
+      "step": 1090
+    },
+    {
+      "epoch": 0.1760352070414083,
+      "grad_norm": 0.437448650598526,
+      "learning_rate": 4.91247937573474e-06,
+      "loss": 1.3202,
+      "step": 1100
+    },
+    {
+      "epoch": 0.17763552710542108,
+      "grad_norm": 0.4280548095703125,
+      "learning_rate": 4.9087783166117035e-06,
+      "loss": 1.2888,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1792358471694339,
+      "grad_norm": 0.430753231048584,
+      "learning_rate": 4.905002067627704e-06,
+      "loss": 1.2891,
+      "step": 1120
+    },
+    {
+      "epoch": 0.1808361672334467,
+      "grad_norm": 0.42238640785217285,
+      "learning_rate": 4.9011507466581146e-06,
+      "loss": 1.2956,
+      "step": 1130
+    },
+    {
+      "epoch": 0.18243648729745948,
+      "grad_norm": 0.4425515830516815,
+      "learning_rate": 4.897224473921672e-06,
+      "loss": 1.2969,
+      "step": 1140
+    },
+    {
+      "epoch": 0.1840368073614723,
+      "grad_norm": 0.4294049143791199,
+      "learning_rate": 4.893223371976728e-06,
+      "loss": 1.2962,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1856371274254851,
+      "grad_norm": 0.44323623180389404,
+      "learning_rate": 4.889147565717425e-06,
+      "loss": 1.2914,
+      "step": 1160
+    },
+    {
+      "epoch": 0.1872374474894979,
+      "grad_norm": 1.6440542936325073,
+      "learning_rate": 4.8849971823697925e-06,
+      "loss": 1.2961,
+      "step": 1170
+    },
+    {
+      "epoch": 0.1888377675535107,
+      "grad_norm": 0.44181469082832336,
+      "learning_rate": 4.880772351487781e-06,
+      "loss": 1.2617,
+      "step": 1180
+    },
+    {
+      "epoch": 0.1904380876175235,
+      "grad_norm": 0.42513954639434814,
+      "learning_rate": 4.8764732049492194e-06,
+      "loss": 1.3279,
+      "step": 1190
+    },
+    {
+      "epoch": 0.1920384076815363,
+      "grad_norm": 0.4171765446662903,
+      "learning_rate": 4.8720998769516876e-06,
+      "loss": 1.2486,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1936387277455491,
+      "grad_norm": 0.43634432554244995,
+      "learning_rate": 4.867652504008341e-06,
+      "loss": 1.2848,
+      "step": 1210
+    },
+    {
+      "epoch": 0.19523904780956192,
+      "grad_norm": 0.4300241768360138,
+      "learning_rate": 4.863131224943641e-06,
+      "loss": 1.3023,
+      "step": 1220
+    },
+    {
+      "epoch": 0.19683936787357473,
+      "grad_norm": 0.41159215569496155,
+      "learning_rate": 4.858536180889025e-06,
+      "loss": 1.2471,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1984396879375875,
+      "grad_norm": 0.44021639227867126,
+      "learning_rate": 4.853867515278499e-06,
+      "loss": 1.3048,
+      "step": 1240
+    },
+    {
+      "epoch": 0.20004000800160032,
+      "grad_norm": 0.43915456533432007,
+      "learning_rate": 4.849125373844159e-06,
+      "loss": 1.2624,
+      "step": 1250
+    },
+    {
+      "epoch": 0.20164032806561313,
+      "grad_norm": 0.4190748929977417,
+      "learning_rate": 4.844309904611648e-06,
+      "loss": 1.261,
+      "step": 1260
+    },
+    {
+      "epoch": 0.20324064812962592,
+      "grad_norm": 0.443705677986145,
+      "learning_rate": 4.839421257895531e-06,
+      "loss": 1.2977,
+      "step": 1270
+    },
+    {
+      "epoch": 0.20484096819363873,
+      "grad_norm": 0.5231160521507263,
+      "learning_rate": 4.8344595862945995e-06,
+      "loss": 1.3784,
+      "step": 1280
+    },
+    {
+      "epoch": 0.20644128825765154,
+      "grad_norm": 0.44119977951049805,
+      "learning_rate": 4.829425044687117e-06,
+      "loss": 1.275,
+      "step": 1290
+    },
+    {
+      "epoch": 0.20804160832166432,
+      "grad_norm": 0.4395291209220886,
+      "learning_rate": 4.8243177902259775e-06,
+      "loss": 1.2103,
+      "step": 1300
+    },
+    {
+      "epoch": 0.20964192838567713,
+      "grad_norm": 0.4597066044807434,
+      "learning_rate": 4.819137982333801e-06,
+      "loss": 1.3092,
+      "step": 1310
+    },
+    {
+      "epoch": 0.21124224844968995,
+      "grad_norm": 0.4260525107383728,
+      "learning_rate": 4.81388578269796e-06,
+      "loss": 1.2564,
+      "step": 1320
+    },
+    {
+      "epoch": 0.21284256851370273,
+      "grad_norm": 0.419204980134964,
+      "learning_rate": 4.80856135526553e-06,
+      "loss": 1.3276,
+      "step": 1330
+    },
+    {
+      "epoch": 0.21444288857771554,
+      "grad_norm": 0.4245491027832031,
+      "learning_rate": 4.8031648662381715e-06,
+      "loss": 1.297,
+      "step": 1340
+    },
+    {
+      "epoch": 0.21604320864172835,
+      "grad_norm": 0.4363405704498291,
+      "learning_rate": 4.797696484066945e-06,
+      "loss": 1.2796,
+      "step": 1350
+    },
+    {
+      "epoch": 0.21764352870574113,
+      "grad_norm": 0.4211016297340393,
+      "learning_rate": 4.7921563794470495e-06,
+      "loss": 1.2903,
+      "step": 1360
+    },
+    {
+      "epoch": 0.21924384876975395,
+      "grad_norm": 0.44692331552505493,
+      "learning_rate": 4.786544725312496e-06,
+      "loss": 1.2893,
+      "step": 1370
+    },
+    {
+      "epoch": 0.22084416883376676,
+      "grad_norm": 0.4414706826210022,
+      "learning_rate": 4.78086169683071e-06,
+      "loss": 1.2812,
+      "step": 1380
+    },
+    {
+      "epoch": 0.22244448889777957,
+      "grad_norm": 0.4407265782356262,
+      "learning_rate": 4.77510747139706e-06,
+      "loss": 1.2196,
+      "step": 1390
+    },
+    {
+      "epoch": 0.22404480896179235,
+      "grad_norm": 0.43710222840309143,
+      "learning_rate": 4.7692822286293266e-06,
+      "loss": 1.2762,
+      "step": 1400
+    },
+    {
+      "epoch": 0.22564512902580516,
+      "grad_norm": 0.4352550506591797,
+      "learning_rate": 4.76338615036209e-06,
+      "loss": 1.2983,
+      "step": 1410
+    },
+    {
+      "epoch": 0.22724544908981797,
+      "grad_norm": 0.45670264959335327,
+      "learning_rate": 4.757419420641053e-06,
+      "loss": 1.2998,
+      "step": 1420
+    },
+    {
+      "epoch": 0.22884576915383076,
+      "grad_norm": 0.4622712731361389,
+      "learning_rate": 4.751382225717304e-06,
+      "loss": 1.2664,
+      "step": 1430
+    },
+    {
+      "epoch": 0.23044608921784357,
+      "grad_norm": 0.4442577064037323,
+      "learning_rate": 4.745274754041497e-06,
+      "loss": 1.3014,
+      "step": 1440
+    },
+    {
+      "epoch": 0.23204640928185638,
+      "grad_norm": 0.4509070813655853,
+      "learning_rate": 4.739097196257967e-06,
+      "loss": 1.3166,
+      "step": 1450
+    },
+    {
+      "epoch": 0.23364672934586916,
+      "grad_norm": 0.4595312476158142,
+      "learning_rate": 4.732849745198786e-06,
+      "loss": 1.2714,
+      "step": 1460
+    },
+    {
+      "epoch": 0.23524704940988198,
+      "grad_norm": 0.46154943108558655,
+      "learning_rate": 4.72653259587774e-06,
+      "loss": 1.2457,
+      "step": 1470
+    },
+    {
+      "epoch": 0.2368473694738948,
+      "grad_norm": 0.45601826906204224,
+      "learning_rate": 4.7201459454842395e-06,
+      "loss": 1.3244,
+      "step": 1480
+    },
+    {
+      "epoch": 0.23844768953790757,
+      "grad_norm": 0.4444175958633423,
+      "learning_rate": 4.713689993377167e-06,
+      "loss": 1.2842,
+      "step": 1490
+    },
+    {
+      "epoch": 0.24004800960192038,
+      "grad_norm": 0.4575673043727875,
+      "learning_rate": 4.707164941078656e-06,
+      "loss": 1.2829,
+      "step": 1500
+    },
+    {
+      "epoch": 0.2416483296659332,
+      "grad_norm": 0.44316327571868896,
+      "learning_rate": 4.700570992267795e-06,
+      "loss": 1.2571,
+      "step": 1510
+    },
+    {
+      "epoch": 0.24324864972994598,
+      "grad_norm": 0.4629514515399933,
+      "learning_rate": 4.6939083527742735e-06,
+      "loss": 1.3043,
+      "step": 1520
+    },
+    {
+      "epoch": 0.2448489697939588,
+      "grad_norm": 0.45411327481269836,
+      "learning_rate": 4.6871772305719586e-06,
+      "loss": 1.3144,
+      "step": 1530
+    },
+    {
+      "epoch": 0.2464492898579716,
+      "grad_norm": 0.44114208221435547,
+      "learning_rate": 4.680377835772399e-06,
+      "loss": 1.2906,
+      "step": 1540
+    },
+    {
+      "epoch": 0.2480496099219844,
+      "grad_norm": 0.4612123966217041,
+      "learning_rate": 4.673510380618266e-06,
+      "loss": 1.2272,
+      "step": 1550
+    },
+    {
+      "epoch": 0.2496499299859972,
+      "grad_norm": 0.5552570819854736,
+      "learning_rate": 4.666575079476733e-06,
+      "loss": 1.3235,
+      "step": 1560
+    },
+    {
+      "epoch": 0.25125025005001,
+      "grad_norm": 0.4699633717536926,
+      "learning_rate": 4.659572148832783e-06,
+      "loss": 1.2614,
+      "step": 1570
+    },
+    {
+      "epoch": 0.2528505701140228,
+      "grad_norm": 0.4418560564517975,
+      "learning_rate": 4.652501807282448e-06,
+      "loss": 1.2433,
+      "step": 1580
+    },
+    {
+      "epoch": 0.2544508901780356,
+      "grad_norm": 0.4806090295314789,
+      "learning_rate": 4.645364275525984e-06,
+      "loss": 1.2728,
+      "step": 1590
+    },
+    {
+      "epoch": 0.2560512102420484,
+      "grad_norm": 0.4657137393951416,
+      "learning_rate": 4.63815977636099e-06,
+      "loss": 1.2605,
+      "step": 1600
+    },
+    {
+      "epoch": 0.2576515303060612,
+      "grad_norm": 0.48473361134529114,
+      "learning_rate": 4.630888534675446e-06,
+      "loss": 1.3102,
+      "step": 1610
+    },
+    {
+      "epoch": 0.25925185037007403,
+      "grad_norm": 0.4496365785598755,
+      "learning_rate": 4.6235507774406966e-06,
+      "loss": 1.2934,
+      "step": 1620
+    },
+    {
+      "epoch": 0.2608521704340868,
+      "grad_norm": 0.4331839978694916,
+      "learning_rate": 4.616146733704362e-06,
+      "loss": 1.2459,
+      "step": 1630
+    },
+    {
+      "epoch": 0.2624524904980996,
+      "grad_norm": 0.46064773201942444,
+      "learning_rate": 4.608676634583194e-06,
+      "loss": 1.2778,
+      "step": 1640
+    },
+    {
+      "epoch": 0.26405281056211244,
+      "grad_norm": 0.4446019232273102,
+      "learning_rate": 4.60114071325586e-06,
+      "loss": 1.2879,
+      "step": 1650
+    },
+    {
+      "epoch": 0.2656531306261252,
+      "grad_norm": 0.4695674479007721,
+      "learning_rate": 4.593539204955661e-06,
+      "loss": 1.294,
+      "step": 1660
+    },
+    {
+      "epoch": 0.267253450690138,
+      "grad_norm": 0.4513859450817108,
+      "learning_rate": 4.585872346963192e-06,
+      "loss": 1.241,
+      "step": 1670
+    },
+    {
+      "epoch": 0.26885377075415084,
+      "grad_norm": 0.4444881081581116,
+      "learning_rate": 4.578140378598935e-06,
+      "loss": 1.2657,
+      "step": 1680
+    },
+    {
+      "epoch": 0.27045409081816363,
+      "grad_norm": 0.4365733563899994,
+      "learning_rate": 4.570343541215787e-06,
+      "loss": 1.2716,
+      "step": 1690
+    },
+    {
+      "epoch": 0.2720544108821764,
+      "grad_norm": 0.4717450737953186,
+      "learning_rate": 4.562482078191527e-06,
+      "loss": 1.2458,
+      "step": 1700
+    },
+    {
+      "epoch": 0.27365473094618925,
+      "grad_norm": 0.45678940415382385,
+      "learning_rate": 4.554556234921221e-06,
+      "loss": 1.2837,
+      "step": 1710
+    },
+    {
+      "epoch": 0.27525505101020203,
+      "grad_norm": 0.4608180522918701,
+      "learning_rate": 4.54656625880956e-06,
+      "loss": 1.2805,
+      "step": 1720
+    },
+    {
+      "epoch": 0.2768553710742148,
+      "grad_norm": 0.46855640411376953,
+      "learning_rate": 4.538512399263136e-06,
+      "loss": 1.2629,
+      "step": 1730
+    },
+    {
+      "epoch": 0.27845569113822766,
+      "grad_norm": 0.4587608575820923,
+      "learning_rate": 4.530394907682659e-06,
+      "loss": 1.2804,
+      "step": 1740
+    },
+    {
+      "epoch": 0.28005601120224044,
+      "grad_norm": 0.4596967101097107,
+      "learning_rate": 4.52221403745511e-06,
+      "loss": 1.2699,
+      "step": 1750
+    },
+    {
+      "epoch": 0.2816563312662532,
+      "grad_norm": 0.4687954783439636,
+      "learning_rate": 4.513970043945828e-06,
+      "loss": 1.2563,
+      "step": 1760
+    },
+    {
+      "epoch": 0.28325665133026606,
+      "grad_norm": 0.4607718884944916,
+      "learning_rate": 4.505663184490543e-06,
+      "loss": 1.2504,
+      "step": 1770
+    },
+    {
+      "epoch": 0.28485697139427885,
+      "grad_norm": 0.4863135814666748,
+      "learning_rate": 4.497293718387342e-06,
+      "loss": 1.2735,
+      "step": 1780
+    },
+    {
+      "epoch": 0.2864572914582917,
+      "grad_norm": 0.4534025192260742,
+      "learning_rate": 4.488861906888574e-06,
+      "loss": 1.2426,
+      "step": 1790
+    },
+    {
+      "epoch": 0.28805761152230447,
+      "grad_norm": 0.4449337422847748,
+      "learning_rate": 4.4803680131926935e-06,
+      "loss": 1.2569,
+      "step": 1800
+    },
+    {
+      "epoch": 0.28965793158631725,
+      "grad_norm": 0.4723162055015564,
+      "learning_rate": 4.47181230243605e-06,
+      "loss": 1.2444,
+      "step": 1810
+    },
+    {
+      "epoch": 0.2912582516503301,
+      "grad_norm": 0.45309850573539734,
+      "learning_rate": 4.463195041684607e-06,
+      "loss": 1.2356,
+      "step": 1820
+    },
+    {
+      "epoch": 0.2928585717143429,
+      "grad_norm": 0.4788224399089813,
+      "learning_rate": 4.454516499925606e-06,
+      "loss": 1.2534,
+      "step": 1830
+    },
+    {
+      "epoch": 0.29445889177835566,
+      "grad_norm": 0.4625834822654724,
+      "learning_rate": 4.445776948059174e-06,
+      "loss": 1.2631,
+      "step": 1840
+    },
+    {
+      "epoch": 0.2960592118423685,
+      "grad_norm": 0.47085145115852356,
+      "learning_rate": 4.436976658889862e-06,
+      "loss": 1.3068,
+      "step": 1850
+    },
+    {
+      "epoch": 0.2976595319063813,
+      "grad_norm": 0.4663547873497009,
+      "learning_rate": 4.428115907118134e-06,
+      "loss": 1.2664,
+      "step": 1860
+    },
+    {
+      "epoch": 0.29925985197039406,
+      "grad_norm": 0.45759037137031555,
+      "learning_rate": 4.4191949693317886e-06,
+      "loss": 1.3019,
+      "step": 1870
+    },
+    {
+      "epoch": 0.3008601720344069,
+      "grad_norm": 0.4738781154155731,
+      "learning_rate": 4.410214123997328e-06,
+      "loss": 1.2699,
+      "step": 1880
+    },
+    {
+      "epoch": 0.3024604920984197,
+      "grad_norm": 0.4601910710334778,
+      "learning_rate": 4.401173651451264e-06,
+      "loss": 1.2779,
+      "step": 1890
+    },
+    {
+      "epoch": 0.30406081216243247,
+      "grad_norm": 0.4631597399711609,
+      "learning_rate": 4.392073833891369e-06,
+      "loss": 1.2617,
+      "step": 1900
+    },
+    {
+      "epoch": 0.3056611322264453,
+      "grad_norm": 0.46061640977859497,
+      "learning_rate": 4.3829149553678666e-06,
+      "loss": 1.2971,
+      "step": 1910
+    },
+    {
+      "epoch": 0.3072614522904581,
+      "grad_norm": 0.4908485412597656,
+      "learning_rate": 4.373697301774561e-06,
+      "loss": 1.2483,
+      "step": 1920
+    },
+    {
+      "epoch": 0.3088617723544709,
+      "grad_norm": 0.5022984147071838,
+      "learning_rate": 4.364421160839921e-06,
+      "loss": 1.2754,
+      "step": 1930
+    },
+    {
+      "epoch": 0.3104620924184837,
+      "grad_norm": 0.46905580163002014,
+      "learning_rate": 4.355086822118092e-06,
+      "loss": 1.2467,
+      "step": 1940
+    },
+    {
+      "epoch": 0.3120624124824965,
+      "grad_norm": 0.44887256622314453,
+      "learning_rate": 4.345694576979858e-06,
+      "loss": 1.2803,
+      "step": 1950
+    },
+    {
+      "epoch": 0.3136627325465093,
+      "grad_norm": 0.4691268503665924,
+      "learning_rate": 4.336244718603553e-06,
+      "loss": 1.2655,
+      "step": 1960
+    },
+    {
+      "epoch": 0.3152630526105221,
+      "grad_norm": 0.4812462329864502,
+      "learning_rate": 4.326737541965899e-06,
+      "loss": 1.2945,
+      "step": 1970
+    },
+    {
+      "epoch": 0.3168633726745349,
+      "grad_norm": 0.47187143564224243,
+      "learning_rate": 4.3171733438328065e-06,
+      "loss": 1.2625,
+      "step": 1980
+    },
+    {
+      "epoch": 0.3184636927385477,
+      "grad_norm": 0.46066814661026,
+      "learning_rate": 4.3075524227501074e-06,
+      "loss": 1.3126,
+      "step": 1990
+    },
+    {
+      "epoch": 0.3200640128025605,
+      "grad_norm": 0.4603062570095062,
+      "learning_rate": 4.297875079034239e-06,
+      "loss": 1.234,
+      "step": 2000
+    },
+    {
+      "epoch": 0.3216643328665733,
+      "grad_norm": 0.4697874188423157,
+      "learning_rate": 4.288141614762864e-06,
+      "loss": 1.2536,
+      "step": 2010
+    },
+    {
+      "epoch": 0.3232646529305861,
+      "grad_norm": 0.48067206144332886,
+      "learning_rate": 4.278352333765449e-06,
+      "loss": 1.2311,
+      "step": 2020
+    },
+    {
+      "epoch": 0.32486497299459893,
+      "grad_norm": 0.4830579459667206,
+      "learning_rate": 4.26850754161377e-06,
+      "loss": 1.2606,
+      "step": 2030
+    },
+    {
+      "epoch": 0.3264652930586117,
+      "grad_norm": 0.47166574001312256,
+      "learning_rate": 4.258607545612389e-06,
+      "loss": 1.2638,
+      "step": 2040
+    },
+    {
+      "epoch": 0.3280656131226245,
+      "grad_norm": 0.47056472301483154,
+      "learning_rate": 4.248652654789043e-06,
+      "loss": 1.2242,
+      "step": 2050
+    },
+    {
+      "epoch": 0.32966593318663734,
+      "grad_norm": 0.4559212625026703,
+      "learning_rate": 4.238643179885013e-06,
+      "loss": 1.2496,
+      "step": 2060
+    },
+    {
+      "epoch": 0.3312662532506501,
+      "grad_norm": 0.48663806915283203,
+      "learning_rate": 4.228579433345418e-06,
+      "loss": 1.2716,
+      "step": 2070
+    },
+    {
+      "epoch": 0.3328665733146629,
+      "grad_norm": 0.47868451476097107,
+      "learning_rate": 4.2184617293094625e-06,
+      "loss": 1.235,
+      "step": 2080
+    },
+    {
+      "epoch": 0.33446689337867574,
+      "grad_norm": 0.4848126769065857,
+      "learning_rate": 4.20829038360063e-06,
+      "loss": 1.2852,
+      "step": 2090
+    },
+    {
+      "epoch": 0.33606721344268853,
+      "grad_norm": 0.46406739950180054,
+      "learning_rate": 4.198065713716824e-06,
+      "loss": 1.2625,
+      "step": 2100
+    },
+    {
+      "epoch": 0.33766753350670137,
+      "grad_norm": 0.4762243628501892,
+      "learning_rate": 4.187788038820462e-06,
+      "loss": 1.2587,
+      "step": 2110
+    },
+    {
+      "epoch": 0.33926785357071415,
+      "grad_norm": 0.48254549503326416,
+      "learning_rate": 4.177457679728508e-06,
+      "loss": 1.2214,
+      "step": 2120
+    },
+    {
+      "epoch": 0.34086817363472693,
+      "grad_norm": 0.5035943388938904,
+      "learning_rate": 4.1670749589024605e-06,
+      "loss": 1.2365,
+      "step": 2130
+    },
+    {
+      "epoch": 0.3424684936987398,
+      "grad_norm": 0.4957742393016815,
+      "learning_rate": 4.156640200438283e-06,
+      "loss": 1.2667,
+      "step": 2140
+    },
+    {
+      "epoch": 0.34406881376275256,
+      "grad_norm": 0.47412991523742676,
+      "learning_rate": 4.146153730056296e-06,
+      "loss": 1.2442,
+      "step": 2150
+    },
+    {
+      "epoch": 0.34566913382676534,
+      "grad_norm": 0.461995393037796,
+      "learning_rate": 4.135615875090998e-06,
+      "loss": 1.3429,
+      "step": 2160
+    },
+    {
+      "epoch": 0.3472694538907782,
+      "grad_norm": 0.4934525787830353,
+      "learning_rate": 4.125026964480861e-06,
+      "loss": 1.2672,
+      "step": 2170
+    },
+    {
+      "epoch": 0.34886977395479096,
+      "grad_norm": 0.4495723247528076,
+      "learning_rate": 4.1143873287580485e-06,
+      "loss": 1.3079,
+      "step": 2180
+    },
+    {
+      "epoch": 0.35047009401880375,
+      "grad_norm": 0.4944385886192322,
+      "learning_rate": 4.10369730003811e-06,
+      "loss": 1.2615,
+      "step": 2190
+    },
+    {
+      "epoch": 0.3520704140828166,
+      "grad_norm": 0.5282748341560364,
+      "learning_rate": 4.092957212009607e-06,
+      "loss": 1.26,
+      "step": 2200
+    },
+    {
+      "epoch": 0.35367073414682937,
+      "grad_norm": 0.4717164635658264,
+      "learning_rate": 4.082167399923701e-06,
+      "loss": 1.2839,
+      "step": 2210
+    },
+    {
+      "epoch": 0.35527105421084215,
+      "grad_norm": 0.4664457142353058,
+      "learning_rate": 4.071328200583684e-06,
+      "loss": 1.2246,
+      "step": 2220
+    },
+    {
+      "epoch": 0.356871374274855,
+      "grad_norm": 0.4893302917480469,
+      "learning_rate": 4.0604399523344715e-06,
+      "loss": 1.2885,
+      "step": 2230
+    },
+    {
+      "epoch": 0.3584716943388678,
+      "grad_norm": 0.4661487638950348,
+      "learning_rate": 4.049502995052034e-06,
+      "loss": 1.287,
+      "step": 2240
+    },
+    {
+      "epoch": 0.36007201440288056,
+      "grad_norm": 0.4612460732460022,
+      "learning_rate": 4.038517670132794e-06,
+      "loss": 1.2483,
+      "step": 2250
+    },
+    {
+      "epoch": 0.3616723344668934,
+      "grad_norm": 0.4863792657852173,
+      "learning_rate": 4.0274843204829645e-06,
+      "loss": 1.231,
+      "step": 2260
+    },
+    {
+      "epoch": 0.3632726545309062,
+      "grad_norm": 0.514241635799408,
+      "learning_rate": 4.01640329050785e-06,
+      "loss": 1.2583,
+      "step": 2270
+    },
+    {
+      "epoch": 0.36487297459491896,
+      "grad_norm": 0.5051952004432678,
+      "learning_rate": 4.00527492610109e-06,
+      "loss": 1.246,
+      "step": 2280
+    },
+    {
+      "epoch": 0.3664732946589318,
+      "grad_norm": 0.4585234224796295,
+      "learning_rate": 3.994099574633869e-06,
+      "loss": 1.2585,
+      "step": 2290
+    },
+    {
+      "epoch": 0.3680736147229446,
+      "grad_norm": 0.4963371753692627,
+      "learning_rate": 3.982877584944066e-06,
+      "loss": 1.2529,
+      "step": 2300
+    },
+    {
+      "epoch": 0.36967393478695737,
+      "grad_norm": 0.4735323488712311,
+      "learning_rate": 3.971609307325373e-06,
+      "loss": 1.2464,
+      "step": 2310
+    },
+    {
+      "epoch": 0.3712742548509702,
+      "grad_norm": 0.4705098271369934,
+      "learning_rate": 3.960295093516352e-06,
+      "loss": 1.2705,
+      "step": 2320
+    },
+    {
+      "epoch": 0.372874574914983,
+      "grad_norm": 0.5204774737358093,
+      "learning_rate": 3.948935296689464e-06,
+      "loss": 1.2127,
+      "step": 2330
+    },
+    {
+      "epoch": 0.3744748949789958,
+      "grad_norm": 0.46412527561187744,
+      "learning_rate": 3.9375302714400415e-06,
+      "loss": 1.3021,
+      "step": 2340
+    },
+    {
+      "epoch": 0.3760752150430086,
+      "grad_norm": 0.47340336441993713,
+      "learning_rate": 3.926080373775215e-06,
+      "loss": 1.2325,
+      "step": 2350
+    },
+    {
+      "epoch": 0.3776755351070214,
+      "grad_norm": 0.5133469700813293,
+      "learning_rate": 3.914585961102808e-06,
+      "loss": 1.3012,
+      "step": 2360
+    },
+    {
+      "epoch": 0.3792758551710342,
+      "grad_norm": 0.5203788876533508,
+      "learning_rate": 3.903047392220176e-06,
+      "loss": 1.1977,
+      "step": 2370
+    },
+    {
+      "epoch": 0.380876175235047,
+      "grad_norm": 0.5032314658164978,
+      "learning_rate": 3.8914650273030085e-06,
+      "loss": 1.2758,
+      "step": 2380
+    },
+    {
+      "epoch": 0.3824764952990598,
+      "grad_norm": 0.4731563627719879,
+      "learning_rate": 3.8798392278940846e-06,
+      "loss": 1.2457,
+      "step": 2390
+    },
+    {
+      "epoch": 0.3840768153630726,
+      "grad_norm": 0.5058057308197021,
+      "learning_rate": 3.86817035689199e-06,
+      "loss": 1.2344,
+      "step": 2400
+    },
+    {
+      "epoch": 0.3856771354270854,
+      "grad_norm": 0.4878542125225067,
+      "learning_rate": 3.856458778539784e-06,
+      "loss": 1.2323,
+      "step": 2410
+    },
+    {
+      "epoch": 0.3872774554910982,
+      "grad_norm": 0.4845779836177826,
+      "learning_rate": 3.844704858413637e-06,
+      "loss": 1.2345,
+      "step": 2420
+    },
+    {
+      "epoch": 0.38887777555511105,
+      "grad_norm": 0.4679320454597473,
+      "learning_rate": 3.832908963411412e-06,
+      "loss": 1.22,
+      "step": 2430
+    },
+    {
+      "epoch": 0.39047809561912383,
+      "grad_norm": 0.4730748236179352,
+      "learning_rate": 3.821071461741217e-06,
+      "loss": 1.2737,
+      "step": 2440
+    },
+    {
+      "epoch": 0.3920784156831366,
+      "grad_norm": 0.4998788833618164,
+      "learning_rate": 3.8091927229099094e-06,
+      "loss": 1.2222,
+      "step": 2450
+    },
+    {
+      "epoch": 0.39367873574714946,
+      "grad_norm": 0.4714483618736267,
+      "learning_rate": 3.797273117711561e-06,
+      "loss": 1.2434,
+      "step": 2460
+    },
+    {
+      "epoch": 0.39527905581116224,
+      "grad_norm": 0.5025110840797424,
+      "learning_rate": 3.7853130182158837e-06,
+      "loss": 1.2855,
+      "step": 2470
+    },
+    {
+      "epoch": 0.396879375875175,
+      "grad_norm": 0.4985499978065491,
+      "learning_rate": 3.77331279775662e-06,
+      "loss": 1.2668,
+      "step": 2480
+    },
+    {
+      "epoch": 0.39847969593918786,
+      "grad_norm": 0.49666157364845276,
+      "learning_rate": 3.7612728309198822e-06,
+      "loss": 1.26,
+      "step": 2490
+    },
+    {
+      "epoch": 0.40008001600320064,
+      "grad_norm": 0.47888752818107605,
+      "learning_rate": 3.7491934935324636e-06,
+      "loss": 1.2683,
+      "step": 2500
+    },
+    {
+      "epoch": 0.40168033606721343,
+      "grad_norm": 0.49205294251441956,
+      "learning_rate": 3.737075162650109e-06,
+      "loss": 1.2587,
+      "step": 2510
+    },
+    {
+      "epoch": 0.40328065613122627,
+      "grad_norm": 0.47279250621795654,
+      "learning_rate": 3.7249182165457405e-06,
+      "loss": 1.2564,
+      "step": 2520
+    },
+    {
+      "epoch": 0.40488097619523905,
+      "grad_norm": 0.4705461263656616,
+      "learning_rate": 3.7127230346976527e-06,
+      "loss": 1.2814,
+      "step": 2530
+    },
+    {
+      "epoch": 0.40648129625925183,
+      "grad_norm": 0.48040762543678284,
+      "learning_rate": 3.700489997777666e-06,
+      "loss": 1.2257,
+      "step": 2540
+    },
+    {
+      "epoch": 0.4080816163232647,
+      "grad_norm": 0.48487865924835205,
+      "learning_rate": 3.6882194876392454e-06,
+      "loss": 1.268,
+      "step": 2550
+    },
+    {
+      "epoch": 0.40968193638727746,
+      "grad_norm": 0.49241048097610474,
+      "learning_rate": 3.675911887305579e-06,
+      "loss": 1.2189,
+      "step": 2560
+    },
+    {
+      "epoch": 0.41128225645129024,
+      "grad_norm": 0.5074785351753235,
+      "learning_rate": 3.6635675809576234e-06,
+      "loss": 1.2319,
+      "step": 2570
+    },
+    {
+      "epoch": 0.4128825765153031,
+      "grad_norm": 0.5021569728851318,
+      "learning_rate": 3.65118695392211e-06,
+      "loss": 1.241,
+      "step": 2580
+    },
+    {
+      "epoch": 0.41448289657931586,
+      "grad_norm": 0.527664303779602,
+      "learning_rate": 3.638770392659522e-06,
+      "loss": 1.2692,
+      "step": 2590
+    },
+    {
+      "epoch": 0.41608321664332865,
+      "grad_norm": 0.4816740155220032,
+      "learning_rate": 3.626318284752022e-06,
+      "loss": 1.2027,
+      "step": 2600
+    },
+    {
+      "epoch": 0.4176835367073415,
+      "grad_norm": 0.4745207130908966,
+      "learning_rate": 3.6138310188913617e-06,
+      "loss": 1.2242,
+      "step": 2610
+    },
+    {
+      "epoch": 0.41928385677135427,
+      "grad_norm": 0.4819655120372772,
+      "learning_rate": 3.601308984866746e-06,
+      "loss": 1.2833,
+      "step": 2620
+    },
+    {
+      "epoch": 0.42088417683536705,
+      "grad_norm": 0.47727903723716736,
+      "learning_rate": 3.5887525735526645e-06,
+      "loss": 1.2501,
+      "step": 2630
+    },
+    {
+      "epoch": 0.4224844968993799,
+      "grad_norm": 0.4700309932231903,
+      "learning_rate": 3.5761621768966947e-06,
+      "loss": 1.2608,
+      "step": 2640
+    },
+    {
+      "epoch": 0.4240848169633927,
+      "grad_norm": 0.5003221035003662,
+      "learning_rate": 3.56353818790726e-06,
+      "loss": 1.2303,
+      "step": 2650
+    },
+    {
+      "epoch": 0.42568513702740546,
+      "grad_norm": 0.47159165143966675,
+      "learning_rate": 3.55088100064137e-06,
+      "loss": 1.2353,
+      "step": 2660
+    },
+    {
+      "epoch": 0.4272854570914183,
+      "grad_norm": 0.48961141705513,
+      "learning_rate": 3.538191010192314e-06,
+      "loss": 1.2455,
+      "step": 2670
+    },
+    {
+      "epoch": 0.4288857771554311,
+      "grad_norm": 0.5218635201454163,
+      "learning_rate": 3.525468612677333e-06,
+      "loss": 1.2207,
+      "step": 2680
+    },
+    {
+      "epoch": 0.43048609721944386,
+      "grad_norm": 0.5075445771217346,
+      "learning_rate": 3.512714205225249e-06,
+      "loss": 1.2562,
+      "step": 2690
+    },
+    {
+      "epoch": 0.4320864172834567,
+      "grad_norm": 0.4947759509086609,
+      "learning_rate": 3.4999281859640753e-06,
+      "loss": 1.2768,
+      "step": 2700
+    },
+    {
+      "epoch": 0.4336867373474695,
+      "grad_norm": 0.47496354579925537,
+      "learning_rate": 3.4871109540085853e-06,
+      "loss": 1.2451,
+      "step": 2710
+    },
+    {
+      "epoch": 0.43528705741148227,
+      "grad_norm": 0.4932936429977417,
+      "learning_rate": 3.474262909447853e-06,
+      "loss": 1.2292,
+      "step": 2720
+    },
+    {
+      "epoch": 0.4368873774754951,
+      "grad_norm": 0.5209819078445435,
+      "learning_rate": 3.4613844533327677e-06,
+      "loss": 1.2483,
+      "step": 2730
+    },
+    {
+      "epoch": 0.4384876975395079,
+      "grad_norm": 0.49459630250930786,
+      "learning_rate": 3.448475987663511e-06,
+      "loss": 1.2562,
+      "step": 2740
+    },
+    {
+      "epoch": 0.44008801760352073,
+      "grad_norm": 0.49290767312049866,
+      "learning_rate": 3.435537915377015e-06,
+      "loss": 1.2563,
+      "step": 2750
+    },
+    {
+      "epoch": 0.4416883376675335,
+      "grad_norm": 0.46836143732070923,
+      "learning_rate": 3.4225706403343767e-06,
+      "loss": 1.2324,
+      "step": 2760
+    },
+    {
+      "epoch": 0.4432886577315463,
+      "grad_norm": 0.47990599274635315,
+      "learning_rate": 3.4095745673082557e-06,
+      "loss": 1.2031,
+      "step": 2770
+    },
+    {
+      "epoch": 0.44488897779555914,
+      "grad_norm": 0.4839983880519867,
+      "learning_rate": 3.3965501019702406e-06,
+      "loss": 1.2534,
+      "step": 2780
+    },
+    {
+      "epoch": 0.4464892978595719,
+      "grad_norm": 0.4967963695526123,
+      "learning_rate": 3.383497650878183e-06,
+      "loss": 1.2496,
+      "step": 2790
+    },
+    {
+      "epoch": 0.4480896179235847,
+      "grad_norm": 0.5163443088531494,
+      "learning_rate": 3.370417621463508e-06,
+      "loss": 1.2078,
+      "step": 2800
+    },
+    {
+      "epoch": 0.44968993798759754,
+      "grad_norm": 0.49203282594680786,
+      "learning_rate": 3.3573104220184966e-06,
+      "loss": 1.1888,
+      "step": 2810
+    },
+    {
+      "epoch": 0.4512902580516103,
+      "grad_norm": 0.5206897854804993,
+      "learning_rate": 3.3441764616835405e-06,
+      "loss": 1.2233,
+      "step": 2820
+    },
+    {
+      "epoch": 0.4528905781156231,
+      "grad_norm": 0.48824429512023926,
+      "learning_rate": 3.3310161504343715e-06,
+      "loss": 1.2682,
+      "step": 2830
+    },
+    {
+      "epoch": 0.45449089817963595,
+      "grad_norm": 0.4975123703479767,
+      "learning_rate": 3.3178298990692614e-06,
+      "loss": 1.2532,
+      "step": 2840
+    },
+    {
+      "epoch": 0.45609121824364873,
+      "grad_norm": 0.5045737028121948,
+      "learning_rate": 3.3046181191962028e-06,
+      "loss": 1.2544,
+      "step": 2850
+    },
+    {
+      "epoch": 0.4576915383076615,
+      "grad_norm": 0.47918882966041565,
+      "learning_rate": 3.2913812232200593e-06,
+      "loss": 1.2512,
+      "step": 2860
+    },
+    {
+      "epoch": 0.45929185837167436,
+      "grad_norm": 0.5058532953262329,
+      "learning_rate": 3.278119624329692e-06,
+      "loss": 1.245,
+      "step": 2870
+    },
+    {
+      "epoch": 0.46089217843568714,
+      "grad_norm": 1.186800241470337,
+      "learning_rate": 3.2648337364850602e-06,
+      "loss": 1.2404,
+      "step": 2880
+    },
+    {
+      "epoch": 0.4624924984996999,
+      "grad_norm": 0.5249418616294861,
+      "learning_rate": 3.2515239744043033e-06,
+      "loss": 1.245,
+      "step": 2890
+    },
+    {
+      "epoch": 0.46409281856371276,
+      "grad_norm": 0.5032217502593994,
+      "learning_rate": 3.2381907535507912e-06,
+      "loss": 1.2201,
+      "step": 2900
+    },
+    {
+      "epoch": 0.46569313862772554,
+      "grad_norm": 0.49591124057769775,
+      "learning_rate": 3.2248344901201618e-06,
+      "loss": 1.2129,
+      "step": 2910
+    },
+    {
+      "epoch": 0.46729345869173833,
+      "grad_norm": 0.5112331509590149,
+      "learning_rate": 3.2114556010273213e-06,
+      "loss": 1.2446,
+      "step": 2920
+    },
+    {
+      "epoch": 0.46889377875575117,
+      "grad_norm": 0.5048455595970154,
+      "learning_rate": 3.198054503893436e-06,
+      "loss": 1.2595,
+      "step": 2930
+    },
+    {
+      "epoch": 0.47049409881976395,
+      "grad_norm": 0.4961850643157959,
+      "learning_rate": 3.184631617032897e-06,
+      "loss": 1.2276,
+      "step": 2940
+    },
+    {
+      "epoch": 0.47209441888377673,
+      "grad_norm": 0.4945153594017029,
+      "learning_rate": 3.171187359440257e-06,
+      "loss": 1.2423,
+      "step": 2950
+    },
+    {
+      "epoch": 0.4736947389477896,
+      "grad_norm": 0.49535536766052246,
+      "learning_rate": 3.157722150777156e-06,
+      "loss": 1.2322,
+      "step": 2960
+    },
+    {
+      "epoch": 0.47529505901180236,
+      "grad_norm": 0.47433891892433167,
+      "learning_rate": 3.144236411359221e-06,
+      "loss": 1.2201,
+      "step": 2970
+    },
+    {
+      "epoch": 0.47689537907581514,
+      "grad_norm": 0.4769941568374634,
+      "learning_rate": 3.1307305621429453e-06,
+      "loss": 1.2169,
+      "step": 2980
+    },
+    {
+      "epoch": 0.478495699139828,
+      "grad_norm": 0.5070396065711975,
+      "learning_rate": 3.117205024712546e-06,
+      "loss": 1.2166,
+      "step": 2990
+    },
+    {
+      "epoch": 0.48009601920384076,
+      "grad_norm": 1.291608214378357,
+      "learning_rate": 3.10366022126681e-06,
+      "loss": 1.2227,
+      "step": 3000
+    },
+    {
+      "epoch": 0.48169633926785355,
+      "grad_norm": 0.4981076717376709,
+      "learning_rate": 3.090096574605908e-06,
+      "loss": 1.244,
+      "step": 3010
+    },
+    {
+      "epoch": 0.4832966593318664,
+      "grad_norm": 0.49318239092826843,
+      "learning_rate": 3.0765145081182045e-06,
+      "loss": 1.2612,
+      "step": 3020
+    },
+    {
+      "epoch": 0.48489697939587917,
+      "grad_norm": 0.48173773288726807,
+      "learning_rate": 3.0629144457670334e-06,
+      "loss": 1.2403,
+      "step": 3030
+    },
+    {
+      "epoch": 0.48649729945989195,
+      "grad_norm": 0.5013173222541809,
+      "learning_rate": 3.0492968120774714e-06,
+      "loss": 1.2635,
+      "step": 3040
+    },
+    {
+      "epoch": 0.4880976195239048,
+      "grad_norm": 0.5153574347496033,
+      "learning_rate": 3.0356620321230823e-06,
+      "loss": 1.2283,
+      "step": 3050
+    },
+    {
+      "epoch": 0.4896979395879176,
+      "grad_norm": 0.49809423089027405,
+      "learning_rate": 3.0220105315126496e-06,
+      "loss": 1.2643,
+      "step": 3060
+    },
+    {
+      "epoch": 0.4912982596519304,
+      "grad_norm": 0.4933251142501831,
+      "learning_rate": 3.0083427363768908e-06,
+      "loss": 1.2687,
+      "step": 3070
+    },
+    {
+      "epoch": 0.4928985797159432,
+      "grad_norm": 0.497218519449234,
+      "learning_rate": 2.9946590733551562e-06,
+      "loss": 1.2131,
+      "step": 3080
+    },
+    {
+      "epoch": 0.494498899779956,
+      "grad_norm": 0.49667471647262573,
+      "learning_rate": 2.9809599695821114e-06,
+      "loss": 1.258,
+      "step": 3090
+    },
+    {
+      "epoch": 0.4960992198439688,
+      "grad_norm": 0.46937358379364014,
+      "learning_rate": 2.9672458526744034e-06,
+      "loss": 1.2243,
+      "step": 3100
+    },
+    {
+      "epoch": 0.4976995399079816,
+      "grad_norm": 0.466854065656662,
+      "learning_rate": 2.9535171507173144e-06,
+      "loss": 1.2265,
+      "step": 3110
+    },
+    {
+      "epoch": 0.4992998599719944,
+      "grad_norm": 0.5092979669570923,
+      "learning_rate": 2.939774292251395e-06,
+      "loss": 1.2694,
+      "step": 3120
+    },
+    {
+      "epoch": 0.5009001800360072,
+      "grad_norm": 0.5075433850288391,
+      "learning_rate": 2.926017706259095e-06,
+      "loss": 1.2295,
+      "step": 3130
+    },
+    {
+      "epoch": 0.50250050010002,
+      "grad_norm": 0.4989834129810333,
+      "learning_rate": 2.912247822151365e-06,
+      "loss": 1.2688,
+      "step": 3140
+    },
+    {
+      "epoch": 0.5041008201640328,
+      "grad_norm": 0.49576446413993835,
+      "learning_rate": 2.898465069754255e-06,
+      "loss": 1.251,
+      "step": 3150
+    },
+    {
+      "epoch": 0.5057011402280456,
+      "grad_norm": 0.49369537830352783,
+      "learning_rate": 2.8846698792954995e-06,
+      "loss": 1.2792,
+      "step": 3160
+    },
+    {
+      "epoch": 0.5073014602920584,
+      "grad_norm": 0.48225724697113037,
+      "learning_rate": 2.8708626813910857e-06,
+      "loss": 1.2142,
+      "step": 3170
+    },
+    {
+      "epoch": 0.5089017803560713,
+      "grad_norm": 0.48354271054267883,
+      "learning_rate": 2.8570439070318116e-06,
+      "loss": 1.2099,
+      "step": 3180
+    },
+    {
+      "epoch": 0.510502100420084,
+      "grad_norm": 0.49547797441482544,
+      "learning_rate": 2.843213987569835e-06,
+      "loss": 1.2254,
+      "step": 3190
+    },
+    {
+      "epoch": 0.5121024204840968,
+      "grad_norm": 0.49686744809150696,
+      "learning_rate": 2.8293733547052067e-06,
+      "loss": 1.255,
+      "step": 3200
+    },
+    {
+      "epoch": 0.5137027405481096,
+      "grad_norm": 0.49578046798706055,
+      "learning_rate": 2.8155224404723946e-06,
+      "loss": 1.1932,
+      "step": 3210
+    },
+    {
+      "epoch": 0.5153030606121224,
+      "grad_norm": 0.48902279138565063,
+      "learning_rate": 2.801661677226801e-06,
+      "loss": 1.2343,
+      "step": 3220
+    },
+    {
+      "epoch": 0.5169033806761353,
+      "grad_norm": 0.50181645154953,
+      "learning_rate": 2.787791497631263e-06,
+      "loss": 1.1983,
+      "step": 3230
+    },
+    {
+      "epoch": 0.5185037007401481,
+      "grad_norm": 0.49499696493148804,
+      "learning_rate": 2.7739123346425485e-06,
+      "loss": 1.2036,
+      "step": 3240
+    },
+    {
+      "epoch": 0.5201040208041608,
+      "grad_norm": 0.5074306130409241,
+      "learning_rate": 2.760024621497843e-06,
+      "loss": 1.2691,
+      "step": 3250
+    },
+    {
+      "epoch": 0.5217043408681736,
+      "grad_norm": 0.5145530700683594,
+      "learning_rate": 2.7461287917012207e-06,
+      "loss": 1.2452,
+      "step": 3260
+    },
+    {
+      "epoch": 0.5233046609321864,
+      "grad_norm": 0.5158199071884155,
+      "learning_rate": 2.732225279010121e-06,
+      "loss": 1.2463,
+      "step": 3270
+    },
+    {
+      "epoch": 0.5249049809961992,
+      "grad_norm": 0.48611709475517273,
+      "learning_rate": 2.7183145174218024e-06,
+      "loss": 1.2657,
+      "step": 3280
+    },
+    {
+      "epoch": 0.5265053010602121,
+      "grad_norm": 0.516374945640564,
+      "learning_rate": 2.704396941159798e-06,
+      "loss": 1.2098,
+      "step": 3290
+    },
+    {
+      "epoch": 0.5281056211242249,
+      "grad_norm": 0.5058311223983765,
+      "learning_rate": 2.690472984660359e-06,
+      "loss": 1.2323,
+      "step": 3300
+    },
+    {
+      "epoch": 0.5297059411882377,
+      "grad_norm": 0.5268154740333557,
+      "learning_rate": 2.676543082558896e-06,
+      "loss": 1.2465,
+      "step": 3310
+    },
+    {
+      "epoch": 0.5313062612522504,
+      "grad_norm": 0.5180853605270386,
+      "learning_rate": 2.6626076696764125e-06,
+      "loss": 1.2701,
+      "step": 3320
+    },
+    {
+      "epoch": 0.5329065813162632,
+      "grad_norm": 0.47433140873908997,
+      "learning_rate": 2.648667181005929e-06,
+      "loss": 1.2379,
+      "step": 3330
+    },
+    {
+      "epoch": 0.534506901380276,
+      "grad_norm": 0.5192251205444336,
+      "learning_rate": 2.6347220516989064e-06,
+      "loss": 1.2161,
+      "step": 3340
+    },
+    {
+      "epoch": 0.5361072214442889,
+      "grad_norm": 0.4724452495574951,
+      "learning_rate": 2.620772717051664e-06,
+      "loss": 1.2482,
+      "step": 3350
+    },
+    {
+      "epoch": 0.5377075415083017,
+      "grad_norm": 0.48347294330596924,
+      "learning_rate": 2.60681961249179e-06,
+      "loss": 1.2619,
+      "step": 3360
+    },
+    {
+      "epoch": 0.5393078615723145,
+      "grad_norm": 0.5179809331893921,
+      "learning_rate": 2.5928631735645503e-06,
+      "loss": 1.2228,
+      "step": 3370
+    },
+    {
+      "epoch": 0.5409081816363273,
+      "grad_norm": 0.5063818693161011,
+      "learning_rate": 2.5789038359192912e-06,
+      "loss": 1.1893,
+      "step": 3380
+    },
+    {
+      "epoch": 0.54250850170034,
+      "grad_norm": 0.5198962688446045,
+      "learning_rate": 2.5649420352958452e-06,
+      "loss": 1.2239,
+      "step": 3390
+    },
+    {
+      "epoch": 0.5441088217643528,
+      "grad_norm": 0.5000635981559753,
+      "learning_rate": 2.550978207510925e-06,
+      "loss": 1.2476,
+      "step": 3400
+    },
+    {
+      "epoch": 0.5457091418283657,
+      "grad_norm": 0.5031649470329285,
+      "learning_rate": 2.53701278844452e-06,
+      "loss": 1.2153,
+      "step": 3410
+    },
+    {
+      "epoch": 0.5473094618923785,
+      "grad_norm": 0.5141199231147766,
+      "learning_rate": 2.5230462140262936e-06,
+      "loss": 1.1893,
+      "step": 3420
+    },
+    {
+      "epoch": 0.5489097819563913,
+      "grad_norm": 0.499374657869339,
+      "learning_rate": 2.5090789202219705e-06,
+      "loss": 1.3086,
+      "step": 3430
+    },
+    {
+      "epoch": 0.5505101020204041,
+      "grad_norm": 0.5020270943641663,
+      "learning_rate": 2.495111343019735e-06,
+      "loss": 1.2257,
+      "step": 3440
+    },
+    {
+      "epoch": 0.5521104220844169,
+      "grad_norm": 0.5148738026618958,
+      "learning_rate": 2.4811439184166137e-06,
+      "loss": 1.2291,
+      "step": 3450
+    },
+    {
+      "epoch": 0.5537107421484296,
+      "grad_norm": 0.49779781699180603,
+      "learning_rate": 2.467177082404871e-06,
+      "loss": 1.2537,
+      "step": 3460
+    },
+    {
+      "epoch": 0.5553110622124425,
+      "grad_norm": 0.5102514028549194,
+      "learning_rate": 2.453211270958401e-06,
+      "loss": 1.2495,
+      "step": 3470
+    },
+    {
+      "epoch": 0.5569113822764553,
+      "grad_norm": 0.49807655811309814,
+      "learning_rate": 2.4392469200191135e-06,
+      "loss": 1.2416,
+      "step": 3480
+    },
+    {
+      "epoch": 0.5585117023404681,
+      "grad_norm": 0.5306846499443054,
+      "learning_rate": 2.4252844654833278e-06,
+      "loss": 1.2343,
+      "step": 3490
+    },
+    {
+      "epoch": 0.5601120224044809,
+      "grad_norm": 0.5187516212463379,
+      "learning_rate": 2.4113243431881696e-06,
+      "loss": 1.2483,
+      "step": 3500
+    },
+    {
+      "epoch": 0.5617123424684937,
+      "grad_norm": 0.48813924193382263,
+      "learning_rate": 2.3973669888979627e-06,
+      "loss": 1.2557,
+      "step": 3510
+    },
+    {
+      "epoch": 0.5633126625325064,
+      "grad_norm": 0.5215795040130615,
+      "learning_rate": 2.3834128382906277e-06,
+      "loss": 1.2194,
+      "step": 3520
+    },
+    {
+      "epoch": 0.5649129825965193,
+      "grad_norm": 0.515504777431488,
+      "learning_rate": 2.369462326944082e-06,
+      "loss": 1.246,
+      "step": 3530
+    },
+    {
+      "epoch": 0.5665133026605321,
+      "grad_norm": 0.5169162154197693,
+      "learning_rate": 2.3555158903226446e-06,
+      "loss": 1.2787,
+      "step": 3540
+    },
+    {
+      "epoch": 0.5681136227245449,
+      "grad_norm": 0.5044717788696289,
+      "learning_rate": 2.341573963763442e-06,
+      "loss": 1.2552,
+      "step": 3550
+    },
+    {
+      "epoch": 0.5697139427885577,
+      "grad_norm": 0.49591878056526184,
+      "learning_rate": 2.3276369824628193e-06,
+      "loss": 1.229,
+      "step": 3560
+    },
+    {
+      "epoch": 0.5713142628525705,
+      "grad_norm": 2.2468647956848145,
+      "learning_rate": 2.313705381462755e-06,
+      "loss": 1.3349,
+      "step": 3570
+    },
+    {
+      "epoch": 0.5729145829165834,
+      "grad_norm": 0.5160189867019653,
+      "learning_rate": 2.299779595637284e-06,
+      "loss": 1.2395,
+      "step": 3580
+    },
+    {
+      "epoch": 0.5745149029805962,
+      "grad_norm": 0.4940637946128845,
+      "learning_rate": 2.2858600596789186e-06,
+      "loss": 1.2176,
+      "step": 3590
+    },
+    {
+      "epoch": 0.5761152230446089,
+      "grad_norm": 0.5026009678840637,
+      "learning_rate": 2.2719472080850827e-06,
+      "loss": 1.2506,
+      "step": 3600
+    },
+    {
+      "epoch": 0.5777155431086217,
+      "grad_norm": 0.4862099587917328,
+      "learning_rate": 2.2580414751445483e-06,
+      "loss": 1.1798,
+      "step": 3610
+    },
+    {
+      "epoch": 0.5793158631726345,
+      "grad_norm": 0.4969867467880249,
+      "learning_rate": 2.2441432949238783e-06,
+      "loss": 1.2579,
+      "step": 3620
+    },
+    {
+      "epoch": 0.5809161832366473,
+      "grad_norm": 0.5029799938201904,
+      "learning_rate": 2.2302531012538796e-06,
+      "loss": 1.2203,
+      "step": 3630
+    },
+    {
+      "epoch": 0.5825165033006602,
+      "grad_norm": 0.5118884444236755,
+      "learning_rate": 2.2163713277160583e-06,
+      "loss": 1.2164,
+      "step": 3640
+    },
+    {
+      "epoch": 0.584116823364673,
+      "grad_norm": 0.4936463236808777,
+      "learning_rate": 2.202498407629086e-06,
+      "loss": 1.2404,
+      "step": 3650
+    },
+    {
+      "epoch": 0.5857171434286857,
+      "grad_norm": 0.5048061609268188,
+      "learning_rate": 2.188634774035276e-06,
+      "loss": 1.2116,
+      "step": 3660
+    },
+    {
+      "epoch": 0.5873174634926985,
+      "grad_norm": 0.5140055418014526,
+      "learning_rate": 2.1747808596870636e-06,
+      "loss": 1.187,
+      "step": 3670
+    },
+    {
+      "epoch": 0.5889177835567113,
+      "grad_norm": 0.4869968891143799,
+      "learning_rate": 2.160937097033497e-06,
+      "loss": 1.3085,
+      "step": 3680
+    },
+    {
+      "epoch": 0.5905181036207241,
+      "grad_norm": 0.4946134090423584,
+      "learning_rate": 2.1471039182067427e-06,
+      "loss": 1.2291,
+      "step": 3690
+    },
+    {
+      "epoch": 0.592118423684737,
+      "grad_norm": 0.5053882002830505,
+      "learning_rate": 2.133281755008592e-06,
+      "loss": 1.2178,
+      "step": 3700
+    },
+    {
+      "epoch": 0.5937187437487498,
+      "grad_norm": 0.5099985599517822,
+      "learning_rate": 2.1194710388969846e-06,
+      "loss": 1.2427,
+      "step": 3710
+    },
+    {
+      "epoch": 0.5953190638127626,
+      "grad_norm": 0.5032602548599243,
+      "learning_rate": 2.1056722009725386e-06,
+      "loss": 1.2182,
+      "step": 3720
+    },
+    {
+      "epoch": 0.5969193838767753,
+      "grad_norm": 0.9163219332695007,
+      "learning_rate": 2.0918856719650975e-06,
+      "loss": 1.249,
+      "step": 3730
+    },
+    {
+      "epoch": 0.5985197039407881,
+      "grad_norm": 0.5141546130180359,
+      "learning_rate": 2.078111882220282e-06,
+      "loss": 1.2313,
+      "step": 3740
+    },
+    {
+      "epoch": 0.6001200240048009,
+      "grad_norm": 0.5054259300231934,
+      "learning_rate": 2.0643512616860562e-06,
+      "loss": 1.2089,
+      "step": 3750
+    },
+    {
+      "epoch": 0.6017203440688138,
+      "grad_norm": 0.506846010684967,
+      "learning_rate": 2.050604239899311e-06,
+      "loss": 1.2399,
+      "step": 3760
+    },
+    {
+      "epoch": 0.6033206641328266,
+      "grad_norm": 0.49931642413139343,
+      "learning_rate": 2.03687124597245e-06,
+      "loss": 1.2006,
+      "step": 3770
+    },
+    {
+      "epoch": 0.6049209841968394,
+      "grad_norm": 0.5294950604438782,
+      "learning_rate": 2.0231527085799985e-06,
+      "loss": 1.2871,
+      "step": 3780
+    },
+    {
+      "epoch": 0.6065213042608522,
+      "grad_norm": 0.49745187163352966,
+      "learning_rate": 2.009449055945226e-06,
+      "loss": 1.2371,
+      "step": 3790
+    },
+    {
+      "epoch": 0.6081216243248649,
+      "grad_norm": 0.5167328119277954,
+      "learning_rate": 1.99576071582677e-06,
+      "loss": 1.2754,
+      "step": 3800
+    },
+    {
+      "epoch": 0.6097219443888777,
+      "grad_norm": 0.4940430819988251,
+      "learning_rate": 1.9820881155052922e-06,
+      "loss": 1.2542,
+      "step": 3810
+    },
+    {
+      "epoch": 0.6113222644528906,
+      "grad_norm": 0.500819206237793,
+      "learning_rate": 1.9684316817701365e-06,
+      "loss": 1.2487,
+      "step": 3820
+    },
+    {
+      "epoch": 0.6129225845169034,
+      "grad_norm": 0.5038882493972778,
+      "learning_rate": 1.9547918409060076e-06,
+      "loss": 1.2785,
+      "step": 3830
+    },
+    {
+      "epoch": 0.6145229045809162,
+      "grad_norm": 0.5073249936103821,
+      "learning_rate": 1.941169018679664e-06,
+      "loss": 1.2316,
+      "step": 3840
+    },
+    {
+      "epoch": 0.616123224644929,
+      "grad_norm": 0.5036085247993469,
+      "learning_rate": 1.9275636403266297e-06,
+      "loss": 1.2145,
+      "step": 3850
+    },
+    {
+      "epoch": 0.6177235447089418,
+      "grad_norm": 0.5362631678581238,
+      "learning_rate": 1.91397613053792e-06,
+      "loss": 1.2449,
+      "step": 3860
+    },
+    {
+      "epoch": 0.6193238647729545,
+      "grad_norm": 0.5268213748931885,
+      "learning_rate": 1.9004069134467834e-06,
+      "loss": 1.2062,
+      "step": 3870
+    },
+    {
+      "epoch": 0.6209241848369674,
+      "grad_norm": 0.4909802973270416,
+      "learning_rate": 1.8868564126154613e-06,
+      "loss": 1.2528,
+      "step": 3880
+    },
+    {
+      "epoch": 0.6225245049009802,
+      "grad_norm": 0.5266659259796143,
+      "learning_rate": 1.8733250510219693e-06,
+      "loss": 1.1999,
+      "step": 3890
+    },
+    {
+      "epoch": 0.624124824964993,
+      "grad_norm": 0.5033228397369385,
+      "learning_rate": 1.8598132510468967e-06,
+      "loss": 1.2249,
+      "step": 3900
+    },
+    {
+      "epoch": 0.6257251450290058,
+      "grad_norm": 0.510733425617218,
+      "learning_rate": 1.846321434460212e-06,
+      "loss": 1.2873,
+      "step": 3910
+    },
+    {
+      "epoch": 0.6273254650930186,
+      "grad_norm": 0.49860432744026184,
+      "learning_rate": 1.8328500224081069e-06,
+      "loss": 1.2465,
+      "step": 3920
+    },
+    {
+      "epoch": 0.6289257851570315,
+      "grad_norm": 0.5228532552719116,
+      "learning_rate": 1.8193994353998462e-06,
+      "loss": 1.2396,
+      "step": 3930
+    },
+    {
+      "epoch": 0.6305261052210442,
+      "grad_norm": 0.507424533367157,
+      "learning_rate": 1.8059700932946434e-06,
+      "loss": 1.2233,
+      "step": 3940
+    },
+    {
+      "epoch": 0.632126425285057,
+      "grad_norm": 0.52374267578125,
+      "learning_rate": 1.7925624152885512e-06,
+      "loss": 1.2586,
+      "step": 3950
+    },
+    {
+      "epoch": 0.6337267453490698,
+      "grad_norm": 0.5252387523651123,
+      "learning_rate": 1.7791768199013808e-06,
+      "loss": 1.2898,
+      "step": 3960
+    },
+    {
+      "epoch": 0.6353270654130826,
+      "grad_norm": 0.496242880821228,
+      "learning_rate": 1.7658137249636358e-06,
+      "loss": 1.2213,
+      "step": 3970
+    },
+    {
+      "epoch": 0.6369273854770954,
+      "grad_norm": 0.49199914932250977,
+      "learning_rate": 1.7524735476034689e-06,
+      "loss": 1.2509,
+      "step": 3980
+    },
+    {
+      "epoch": 0.6385277055411083,
+      "grad_norm": 0.5169931650161743,
+      "learning_rate": 1.7391567042336626e-06,
+      "loss": 1.2099,
+      "step": 3990
+    },
+    {
+      "epoch": 0.640128025605121,
+      "grad_norm": 0.5024374723434448,
+      "learning_rate": 1.7258636105386289e-06,
+      "loss": 1.2134,
+      "step": 4000
+    },
+    {
+      "epoch": 0.6417283456691338,
+      "grad_norm": 0.4861580431461334,
+      "learning_rate": 1.7125946814614396e-06,
+      "loss": 1.2247,
+      "step": 4010
+    },
+    {
+      "epoch": 0.6433286657331466,
+      "grad_norm": 0.5099141001701355,
+      "learning_rate": 1.6993503311908638e-06,
+      "loss": 1.235,
+      "step": 4020
+    },
+    {
+      "epoch": 0.6449289857971594,
+      "grad_norm": 0.5228403210639954,
+      "learning_rate": 1.6861309731484487e-06,
+      "loss": 1.2156,
+      "step": 4030
+    },
+    {
+      "epoch": 0.6465293058611722,
+      "grad_norm": 0.5319278240203857,
+      "learning_rate": 1.6729370199756083e-06,
+      "loss": 1.2589,
+      "step": 4040
+    },
+    {
+      "epoch": 0.6481296259251851,
+      "grad_norm": 0.5031079649925232,
+      "learning_rate": 1.6597688835207477e-06,
+      "loss": 1.2299,
+      "step": 4050
+    },
+    {
+      "epoch": 0.6497299459891979,
+      "grad_norm": 0.4944058954715729,
+      "learning_rate": 1.646626974826403e-06,
+      "loss": 1.2086,
+      "step": 4060
+    },
+    {
+      "epoch": 0.6513302660532106,
+      "grad_norm": 0.5121133327484131,
+      "learning_rate": 1.633511704116412e-06,
+      "loss": 1.2477,
+      "step": 4070
+    },
+    {
+      "epoch": 0.6529305861172234,
+      "grad_norm": 0.5161622166633606,
+      "learning_rate": 1.620423480783111e-06,
+      "loss": 1.2717,
+      "step": 4080
+    },
+    {
+      "epoch": 0.6545309061812362,
+      "grad_norm": 0.4925840198993683,
+      "learning_rate": 1.6073627133745531e-06,
+      "loss": 1.2057,
+      "step": 4090
+    },
+    {
+      "epoch": 0.656131226245249,
+      "grad_norm": 0.5267453193664551,
+      "learning_rate": 1.5943298095817572e-06,
+      "loss": 1.2009,
+      "step": 4100
+    },
+    {
+      "epoch": 0.6577315463092619,
+      "grad_norm": 0.5138763785362244,
+      "learning_rate": 1.5813251762259813e-06,
+      "loss": 1.2124,
+      "step": 4110
+    },
+    {
+      "epoch": 0.6593318663732747,
+      "grad_norm": 0.5108798146247864,
+      "learning_rate": 1.5683492192460226e-06,
+      "loss": 1.2226,
+      "step": 4120
+    },
+    {
+      "epoch": 0.6609321864372875,
+      "grad_norm": 0.5101296901702881,
+      "learning_rate": 1.5554023436855488e-06,
+      "loss": 1.2533,
+      "step": 4130
+    },
+    {
+      "epoch": 0.6625325065013002,
+      "grad_norm": 0.4998767375946045,
+      "learning_rate": 1.5424849536804497e-06,
+      "loss": 1.195,
+      "step": 4140
+    },
+    {
+      "epoch": 0.664132826565313,
+      "grad_norm": 0.4969738721847534,
+      "learning_rate": 1.529597452446229e-06,
+      "loss": 1.2269,
+      "step": 4150
+    },
+    {
+      "epoch": 0.6657331466293258,
+      "grad_norm": 0.47680023312568665,
+      "learning_rate": 1.5167402422654122e-06,
+      "loss": 1.2024,
+      "step": 4160
+    },
+    {
+      "epoch": 0.6673334666933387,
+      "grad_norm": 0.49921032786369324,
+      "learning_rate": 1.503913724474992e-06,
+      "loss": 1.1991,
+      "step": 4170
+    },
+    {
+      "epoch": 0.6689337867573515,
+      "grad_norm": 0.5103694796562195,
+      "learning_rate": 1.491118299453901e-06,
+      "loss": 1.2349,
+      "step": 4180
+    },
+    {
+      "epoch": 0.6705341068213643,
+      "grad_norm": 0.5037977695465088,
+      "learning_rate": 1.478354366610511e-06,
+      "loss": 1.2537,
+      "step": 4190
+    },
+    {
+      "epoch": 0.6721344268853771,
+      "grad_norm": 0.5193389654159546,
+      "learning_rate": 1.4656223243701692e-06,
+      "loss": 1.2304,
+      "step": 4200
+    },
+    {
+      "epoch": 0.6737347469493898,
+      "grad_norm": 7.260618209838867,
+      "learning_rate": 1.4529225701627587e-06,
+      "loss": 1.2909,
+      "step": 4210
+    },
+    {
+      "epoch": 0.6753350670134027,
+      "grad_norm": 0.5033296942710876,
+      "learning_rate": 1.4402555004102946e-06,
+      "loss": 1.2202,
+      "step": 4220
+    },
+    {
+      "epoch": 0.6769353870774155,
+      "grad_norm": 0.49899816513061523,
+      "learning_rate": 1.4276215105145486e-06,
+      "loss": 1.1994,
+      "step": 4230
+    },
+    {
+      "epoch": 0.6785357071414283,
+      "grad_norm": 0.4914785325527191,
+      "learning_rate": 1.415020994844706e-06,
+      "loss": 1.2149,
+      "step": 4240
+    },
+    {
+      "epoch": 0.6801360272054411,
+      "grad_norm": 0.5077610015869141,
+      "learning_rate": 1.4024543467250573e-06,
+      "loss": 1.213,
+      "step": 4250
+    },
+    {
+      "epoch": 0.6817363472694539,
+      "grad_norm": 0.5179877281188965,
+      "learning_rate": 1.389921958422719e-06,
+      "loss": 1.2507,
+      "step": 4260
+    },
+    {
+      "epoch": 0.6833366673334667,
+      "grad_norm": 0.5066015124320984,
+      "learning_rate": 1.3774242211353906e-06,
+      "loss": 1.2518,
+      "step": 4270
+    },
+    {
+      "epoch": 0.6849369873974795,
+      "grad_norm": 0.5118170380592346,
+      "learning_rate": 1.3649615249791396e-06,
+      "loss": 1.2214,
+      "step": 4280
+    },
+    {
+      "epoch": 0.6865373074614923,
+      "grad_norm": 0.502426266670227,
+      "learning_rate": 1.3525342589762298e-06,
+      "loss": 1.2264,
+      "step": 4290
+    },
+    {
+      "epoch": 0.6881376275255051,
+      "grad_norm": 0.5098256468772888,
+      "learning_rate": 1.3401428110429734e-06,
+      "loss": 1.2319,
+      "step": 4300
+    },
+    {
+      "epoch": 0.6897379475895179,
+      "grad_norm": 0.49994170665740967,
+      "learning_rate": 1.3277875679776252e-06,
+      "loss": 1.2075,
+      "step": 4310
+    },
+    {
+      "epoch": 0.6913382676535307,
+      "grad_norm": 0.5266156196594238,
+      "learning_rate": 1.3154689154483055e-06,
+      "loss": 1.2298,
+      "step": 4320
+    },
+    {
+      "epoch": 0.6929385877175435,
+      "grad_norm": 0.5013164281845093,
+      "learning_rate": 1.3031872379809663e-06,
+      "loss": 1.1819,
+      "step": 4330
+    },
+    {
+      "epoch": 0.6945389077815564,
+      "grad_norm": 0.5253974199295044,
+      "learning_rate": 1.2909429189473833e-06,
+      "loss": 1.2661,
+      "step": 4340
+    },
+    {
+      "epoch": 0.6961392278455691,
+      "grad_norm": 0.4963892996311188,
+      "learning_rate": 1.27873634055319e-06,
+      "loss": 1.2064,
+      "step": 4350
+    },
+    {
+      "epoch": 0.6977395479095819,
+      "grad_norm": 0.5143953561782837,
+      "learning_rate": 1.2665678838259498e-06,
+      "loss": 1.2714,
+      "step": 4360
+    },
+    {
+      "epoch": 0.6993398679735947,
+      "grad_norm": 0.5049063563346863,
+      "learning_rate": 1.254437928603261e-06,
+      "loss": 1.1939,
+      "step": 4370
+    },
+    {
+      "epoch": 0.7009401880376075,
+      "grad_norm": 0.5228173136711121,
+      "learning_rate": 1.242346853520899e-06,
+      "loss": 1.2205,
+      "step": 4380
+    },
+    {
+      "epoch": 0.7025405081016203,
+      "grad_norm": 0.5170574188232422,
+      "learning_rate": 1.2302950360009994e-06,
+      "loss": 1.2494,
+      "step": 4390
+    },
+    {
+      "epoch": 0.7041408281656332,
+      "grad_norm": 0.5283452272415161,
+      "learning_rate": 1.2182828522402737e-06,
+      "loss": 1.2615,
+      "step": 4400
+    },
+    {
+      "epoch": 0.705741148229646,
+      "grad_norm": 0.5086333751678467,
+      "learning_rate": 1.2063106771982693e-06,
+      "loss": 1.2446,
+      "step": 4410
+    },
+    {
+      "epoch": 0.7073414682936587,
+      "grad_norm": 0.5221850872039795,
+      "learning_rate": 1.194378884585664e-06,
+      "loss": 1.2235,
+      "step": 4420
+    },
+    {
+      "epoch": 0.7089417883576715,
+      "grad_norm": 0.525809109210968,
+      "learning_rate": 1.1824878468526001e-06,
+      "loss": 1.2425,
+      "step": 4430
+    },
+    {
+      "epoch": 0.7105421084216843,
+      "grad_norm": 0.5064604878425598,
+      "learning_rate": 1.1706379351770597e-06,
+      "loss": 1.2386,
+      "step": 4440
+    },
+    {
+      "epoch": 0.7121424284856971,
+      "grad_norm": 0.50041264295578,
+      "learning_rate": 1.158829519453277e-06,
+      "loss": 1.2283,
+      "step": 4450
+    },
+    {
+      "epoch": 0.71374274854971,
+      "grad_norm": 0.5169569253921509,
+      "learning_rate": 1.1470629682801925e-06,
+      "loss": 1.2147,
+      "step": 4460
+    },
+    {
+      "epoch": 0.7153430686137228,
+      "grad_norm": 0.5069741010665894,
+      "learning_rate": 1.1353386489499487e-06,
+      "loss": 1.2122,
+      "step": 4470
+    },
+    {
+      "epoch": 0.7169433886777355,
+      "grad_norm": 0.5095933079719543,
+      "learning_rate": 1.1236569274364224e-06,
+      "loss": 1.2038,
+      "step": 4480
+    },
+    {
+      "epoch": 0.7185437087417483,
+      "grad_norm": 0.5300613641738892,
+      "learning_rate": 1.112018168383803e-06,
+      "loss": 1.2187,
+      "step": 4490
+    },
+    {
+      "epoch": 0.7201440288057611,
+      "grad_norm": 0.5320760011672974,
+      "learning_rate": 1.1004227350952102e-06,
+      "loss": 1.2461,
+      "step": 4500
+    },
+    {
+      "epoch": 0.7217443488697739,
+      "grad_norm": 0.5251277685165405,
+      "learning_rate": 1.0888709895213498e-06,
+      "loss": 1.2513,
+      "step": 4510
+    },
+    {
+      "epoch": 0.7233446689337868,
+      "grad_norm": 0.4987024962902069,
+      "learning_rate": 1.077363292249222e-06,
+      "loss": 1.2544,
+      "step": 4520
+    },
+    {
+      "epoch": 0.7249449889977996,
+      "grad_norm": 0.5069317817687988,
+      "learning_rate": 1.0659000024908588e-06,
+      "loss": 1.2288,
+      "step": 4530
+    },
+    {
+      "epoch": 0.7265453090618124,
+      "grad_norm": 0.5270032286643982,
+      "learning_rate": 1.0544814780721185e-06,
+      "loss": 1.2335,
+      "step": 4540
+    },
+    {
+      "epoch": 0.7281456291258251,
+      "grad_norm": 0.5044519305229187,
+      "learning_rate": 1.0431080754215092e-06,
+      "loss": 1.2408,
+      "step": 4550
+    },
+    {
+      "epoch": 0.7297459491898379,
+      "grad_norm": 0.5111990571022034,
+      "learning_rate": 1.031780149559066e-06,
+      "loss": 1.2371,
+      "step": 4560
+    },
+    {
+      "epoch": 0.7313462692538508,
+      "grad_norm": 0.520692765712738,
+      "learning_rate": 1.020498054085271e-06,
+      "loss": 1.2359,
+      "step": 4570
+    },
+    {
+      "epoch": 0.7329465893178636,
+      "grad_norm": 0.5196593403816223,
+      "learning_rate": 1.0092621411700101e-06,
+      "loss": 1.237,
+      "step": 4580
+    },
+    {
+      "epoch": 0.7345469093818764,
+      "grad_norm": 0.5122175216674805,
+      "learning_rate": 9.980727615415867e-07,
+      "loss": 1.2347,
+      "step": 4590
+    },
+    {
+      "epoch": 0.7361472294458892,
+      "grad_norm": 0.5103949308395386,
+      "learning_rate": 9.869302644757696e-07,
+      "loss": 1.1981,
+      "step": 4600
+    },
+    {
+      "epoch": 0.737747549509902,
+      "grad_norm": 0.49739348888397217,
+      "learning_rate": 9.758349977848917e-07,
+      "loss": 1.2809,
+      "step": 4610
+    },
+    {
+      "epoch": 0.7393478695739147,
+      "grad_norm": 0.5084463953971863,
+      "learning_rate": 9.647873078069923e-07,
+      "loss": 1.2486,
+      "step": 4620
+    },
+    {
+      "epoch": 0.7409481896379276,
+      "grad_norm": 0.5033512711524963,
+      "learning_rate": 9.537875393950067e-07,
+      "loss": 1.2317,
+      "step": 4630
+    },
+    {
+      "epoch": 0.7425485097019404,
+      "grad_norm": 0.5531705617904663,
+      "learning_rate": 9.428360359060021e-07,
+      "loss": 1.2196,
+      "step": 4640
+    },
+    {
+      "epoch": 0.7441488297659532,
+      "grad_norm": 0.5110276937484741,
+      "learning_rate": 9.319331391904587e-07,
+      "loss": 1.2432,
+      "step": 4650
+    },
+    {
+      "epoch": 0.745749149829966,
+      "grad_norm": 0.5248765349388123,
+      "learning_rate": 9.210791895815996e-07,
+      "loss": 1.2137,
+      "step": 4660
+    },
+    {
+      "epoch": 0.7473494698939788,
+      "grad_norm": 0.4994560778141022,
+      "learning_rate": 9.10274525884767e-07,
+      "loss": 1.196,
+      "step": 4670
+    },
+    {
+      "epoch": 0.7489497899579916,
+      "grad_norm": 0.49998924136161804,
+      "learning_rate": 8.995194853668466e-07,
+      "loss": 1.2414,
+      "step": 4680
+    },
+    {
+      "epoch": 0.7505501100220044,
+      "grad_norm": 0.5122482776641846,
+      "learning_rate": 8.888144037457397e-07,
+      "loss": 1.2107,
+      "step": 4690
+    },
+    {
+      "epoch": 0.7521504300860172,
+      "grad_norm": 0.5213466286659241,
+      "learning_rate": 8.781596151798838e-07,
+      "loss": 1.1997,
+      "step": 4700
+    },
+    {
+      "epoch": 0.75375075015003,
+      "grad_norm": 0.5351124405860901,
+      "learning_rate": 8.675554522578219e-07,
+      "loss": 1.2645,
+      "step": 4710
+    },
+    {
+      "epoch": 0.7553510702140428,
+      "grad_norm": 0.4969538450241089,
+      "learning_rate": 8.57002245987821e-07,
+      "loss": 1.2251,
+      "step": 4720
+    },
+    {
+      "epoch": 0.7569513902780556,
+      "grad_norm": 0.5164848566055298,
+      "learning_rate": 8.465003257875392e-07,
+      "loss": 1.2453,
+      "step": 4730
+    },
+    {
+      "epoch": 0.7585517103420684,
+      "grad_norm": 0.6240895986557007,
+      "learning_rate": 8.360500194737445e-07,
+      "loss": 1.2244,
+      "step": 4740
+    },
+    {
+      "epoch": 0.7601520304060813,
+      "grad_norm": 0.5174912214279175,
+      "learning_rate": 8.256516532520761e-07,
+      "loss": 1.2375,
+      "step": 4750
+    },
+    {
+      "epoch": 0.761752350470094,
+      "grad_norm": 0.5274839401245117,
+      "learning_rate": 8.153055517068733e-07,
+      "loss": 1.1858,
+      "step": 4760
+    },
+    {
+      "epoch": 0.7633526705341068,
+      "grad_norm": 0.5240426659584045,
+      "learning_rate": 8.050120377910326e-07,
+      "loss": 1.2411,
+      "step": 4770
+    },
+    {
+      "epoch": 0.7649529905981196,
+      "grad_norm": 0.5028566718101501,
+      "learning_rate": 7.947714328159314e-07,
+      "loss": 1.1977,
+      "step": 4780
+    },
+    {
+      "epoch": 0.7665533106621324,
+      "grad_norm": 0.5158740282058716,
+      "learning_rate": 7.845840564413992e-07,
+      "loss": 1.2006,
+      "step": 4790
+    },
+    {
+      "epoch": 0.7681536307261452,
+      "grad_norm": 0.5100390911102295,
+      "learning_rate": 7.744502266657372e-07,
+      "loss": 1.2377,
+      "step": 4800
+    },
+    {
+      "epoch": 0.7697539507901581,
+      "grad_norm": 0.5089078545570374,
+      "learning_rate": 7.643702598157937e-07,
+      "loss": 1.2058,
+      "step": 4810
+    },
+    {
+      "epoch": 0.7713542708541709,
+      "grad_norm": 0.5130503177642822,
+      "learning_rate": 7.543444705370873e-07,
+      "loss": 1.223,
+      "step": 4820
+    },
+    {
+      "epoch": 0.7729545909181836,
+      "grad_norm": 0.5297360420227051,
+      "learning_rate": 7.443731717839891e-07,
+      "loss": 1.2248,
+      "step": 4830
+    },
+    {
+      "epoch": 0.7745549109821964,
+      "grad_norm": 0.5005588531494141,
+      "learning_rate": 7.344566748099518e-07,
+      "loss": 1.2036,
+      "step": 4840
+    },
+    {
+      "epoch": 0.7761552310462092,
+      "grad_norm": 0.5383617877960205,
+      "learning_rate": 7.245952891577926e-07,
+      "loss": 1.2134,
+      "step": 4850
+    },
+    {
+      "epoch": 0.7777555511102221,
+      "grad_norm": 0.5036467909812927,
+      "learning_rate": 7.147893226500349e-07,
+      "loss": 1.2188,
+      "step": 4860
+    },
+    {
+      "epoch": 0.7793558711742349,
+      "grad_norm": 0.5479145646095276,
+      "learning_rate": 7.050390813792951e-07,
+      "loss": 1.2245,
+      "step": 4870
+    },
+    {
+      "epoch": 0.7809561912382477,
+      "grad_norm": 0.5276610851287842,
+      "learning_rate": 6.953448696987308e-07,
+      "loss": 1.2519,
+      "step": 4880
+    },
+    {
+      "epoch": 0.7825565113022604,
+      "grad_norm": 0.5226728916168213,
+      "learning_rate": 6.857069902125377e-07,
+      "loss": 1.166,
+      "step": 4890
+    },
+    {
+      "epoch": 0.7841568313662732,
+      "grad_norm": 0.5151464343070984,
+      "learning_rate": 6.761257437665075e-07,
+      "loss": 1.2345,
+      "step": 4900
+    },
+    {
+      "epoch": 0.785757151430286,
+      "grad_norm": 0.512322187423706,
+      "learning_rate": 6.666014294386347e-07,
+      "loss": 1.2238,
+      "step": 4910
+    },
+    {
+      "epoch": 0.7873574714942989,
+      "grad_norm": 0.5147780179977417,
+      "learning_rate": 6.571343445297817e-07,
+      "loss": 1.218,
+      "step": 4920
+    },
+    {
+      "epoch": 0.7889577915583117,
+      "grad_norm": 1.6754525899887085,
+      "learning_rate": 6.47724784554398e-07,
+      "loss": 1.2543,
+      "step": 4930
+    },
+    {
+      "epoch": 0.7905581116223245,
+      "grad_norm": 0.5027932524681091,
+      "learning_rate": 6.383730432312954e-07,
+      "loss": 1.2238,
+      "step": 4940
+    },
+    {
+      "epoch": 0.7921584316863373,
+      "grad_norm": 0.5050190687179565,
+      "learning_rate": 6.290794124744809e-07,
+      "loss": 1.2441,
+      "step": 4950
+    },
+    {
+      "epoch": 0.79375875175035,
+      "grad_norm": 0.5184482932090759,
+      "learning_rate": 6.198441823840439e-07,
+      "loss": 1.2078,
+      "step": 4960
+    },
+    {
+      "epoch": 0.7953590718143628,
+      "grad_norm": 0.5082205533981323,
+      "learning_rate": 6.106676412371002e-07,
+      "loss": 1.2144,
+      "step": 4970
+    },
+    {
+      "epoch": 0.7969593918783757,
+      "grad_norm": 0.5187720656394958,
+      "learning_rate": 6.015500754787942e-07,
+      "loss": 1.2269,
+      "step": 4980
+    },
+    {
+      "epoch": 0.7985597119423885,
+      "grad_norm": 0.5122787952423096,
+      "learning_rate": 5.924917697133579e-07,
+      "loss": 1.2332,
+      "step": 4990
+    },
+    {
+      "epoch": 0.8001600320064013,
+      "grad_norm": 0.5063095092773438,
+      "learning_rate": 5.834930066952254e-07,
+      "loss": 1.2522,
+      "step": 5000
+    },
+    {
+      "epoch": 0.8017603520704141,
+      "grad_norm": 0.5298603177070618,
+      "learning_rate": 5.745540673202088e-07,
+      "loss": 1.2776,
+      "step": 5010
+    },
+    {
+      "epoch": 0.8033606721344269,
+      "grad_norm": 0.532569944858551,
+      "learning_rate": 5.656752306167279e-07,
+      "loss": 1.2488,
+      "step": 5020
+    },
+    {
+      "epoch": 0.8049609921984396,
+      "grad_norm": 0.5093346834182739,
+      "learning_rate": 5.568567737371023e-07,
+      "loss": 1.2637,
+      "step": 5030
+    },
+    {
+      "epoch": 0.8065613122624525,
+      "grad_norm": 0.5202427506446838,
+      "learning_rate": 5.480989719489e-07,
+      "loss": 1.2482,
+      "step": 5040
+    },
+    {
+      "epoch": 0.8081616323264653,
+      "grad_norm": 0.5047549605369568,
+      "learning_rate": 5.394020986263412e-07,
+      "loss": 1.2504,
+      "step": 5050
+    },
+    {
+      "epoch": 0.8097619523904781,
+      "grad_norm": 0.5232532620429993,
+      "learning_rate": 5.30766425241771e-07,
+      "loss": 1.2636,
+      "step": 5060
+    },
+    {
+      "epoch": 0.8113622724544909,
+      "grad_norm": 0.49118658900260925,
+      "learning_rate": 5.221922213571809e-07,
+      "loss": 1.2292,
+      "step": 5070
+    },
+    {
+      "epoch": 0.8129625925185037,
+      "grad_norm": 0.5222775936126709,
+      "learning_rate": 5.136797546157974e-07,
+      "loss": 1.2117,
+      "step": 5080
+    },
+    {
+      "epoch": 0.8145629125825165,
+      "grad_norm": 0.49552589654922485,
+      "learning_rate": 5.052292907337239e-07,
+      "loss": 1.192,
+      "step": 5090
+    },
+    {
+      "epoch": 0.8161632326465293,
+      "grad_norm": 0.5257281064987183,
+      "learning_rate": 4.968410934916502e-07,
+      "loss": 1.2259,
+      "step": 5100
+    },
+    {
+      "epoch": 0.8177635527105421,
+      "grad_norm": 0.5237207412719727,
+      "learning_rate": 4.885154247266163e-07,
+      "loss": 1.2252,
+      "step": 5110
+    },
+    {
+      "epoch": 0.8193638727745549,
+      "grad_norm": 5.458524227142334,
+      "learning_rate": 4.80252544323839e-07,
+      "loss": 1.2544,
+      "step": 5120
+    },
+    {
+      "epoch": 0.8209641928385677,
+      "grad_norm": 0.4922438859939575,
+      "learning_rate": 4.7205271020860143e-07,
+      "loss": 1.2508,
+      "step": 5130
+    },
+    {
+      "epoch": 0.8225645129025805,
+      "grad_norm": 0.5238988995552063,
+      "learning_rate": 4.6391617833820066e-07,
+      "loss": 1.2277,
+      "step": 5140
+    },
+    {
+      "epoch": 0.8241648329665933,
+      "grad_norm": 0.5212239623069763,
+      "learning_rate": 4.5584320269395827e-07,
+      "loss": 1.256,
+      "step": 5150
+    },
+    {
+      "epoch": 0.8257651530306062,
+      "grad_norm": 0.4987280070781708,
+      "learning_rate": 4.478340352732924e-07,
+      "loss": 1.2181,
+      "step": 5160
+    },
+    {
+      "epoch": 0.8273654730946189,
+      "grad_norm": 0.519672155380249,
+      "learning_rate": 4.3988892608185035e-07,
+      "loss": 1.2403,
+      "step": 5170
+    },
+    {
+      "epoch": 0.8289657931586317,
+      "grad_norm": 0.5267203450202942,
+      "learning_rate": 4.3200812312570894e-07,
+      "loss": 1.1784,
+      "step": 5180
+    },
+    {
+      "epoch": 0.8305661132226445,
+      "grad_norm": 0.5164812207221985,
+      "learning_rate": 4.241918724036262e-07,
+      "loss": 1.2642,
+      "step": 5190
+    },
+    {
+      "epoch": 0.8321664332866573,
+      "grad_norm": 0.5407664775848389,
+      "learning_rate": 4.164404178993689e-07,
+      "loss": 1.2224,
+      "step": 5200
+    },
+    {
+      "epoch": 0.8337667533506702,
+      "grad_norm": 0.5061764717102051,
+      "learning_rate": 4.087540015740929e-07,
+      "loss": 1.2286,
+      "step": 5210
+    },
+    {
+      "epoch": 0.835367073414683,
+      "grad_norm": 0.8010200262069702,
+      "learning_rate": 4.011328633587919e-07,
+      "loss": 1.2238,
+      "step": 5220
+    },
+    {
+      "epoch": 0.8369673934786958,
+      "grad_norm": 0.5010237693786621,
+      "learning_rate": 3.935772411468075e-07,
+      "loss": 1.2296,
+      "step": 5230
+    },
+    {
+      "epoch": 0.8385677135427085,
+      "grad_norm": 0.5492453575134277,
+      "learning_rate": 3.8608737078640326e-07,
+      "loss": 1.2459,
+      "step": 5240
+    },
+    {
+      "epoch": 0.8401680336067213,
+      "grad_norm": 0.492929071187973,
+      "learning_rate": 3.7866348607340284e-07,
+      "loss": 1.1894,
+      "step": 5250
+    },
+    {
+      "epoch": 0.8417683536707341,
+      "grad_norm": 0.49885085225105286,
+      "learning_rate": 3.7130581874389285e-07,
+      "loss": 1.1877,
+      "step": 5260
+    },
+    {
+      "epoch": 0.843368673734747,
+      "grad_norm": 0.5160171389579773,
+      "learning_rate": 3.640145984669882e-07,
+      "loss": 1.1935,
+      "step": 5270
+    },
+    {
+      "epoch": 0.8449689937987598,
+      "grad_norm": 0.4957447052001953,
+      "learning_rate": 3.5679005283766136e-07,
+      "loss": 1.2315,
+      "step": 5280
+    },
+    {
+      "epoch": 0.8465693138627726,
+      "grad_norm": 0.4999656677246094,
+      "learning_rate": 3.496324073696439e-07,
+      "loss": 1.2198,
+      "step": 5290
+    },
+    {
+      "epoch": 0.8481696339267853,
+      "grad_norm": 0.5246298909187317,
+      "learning_rate": 3.4254188548837967e-07,
+      "loss": 1.1953,
+      "step": 5300
+    },
+    {
+      "epoch": 0.8497699539907981,
+      "grad_norm": 0.5125941634178162,
+      "learning_rate": 3.355187085240555e-07,
+      "loss": 1.193,
+      "step": 5310
+    },
+    {
+      "epoch": 0.8513702740548109,
+      "grad_norm": 0.5065252184867859,
+      "learning_rate": 3.285630957046912e-07,
+      "loss": 1.2423,
+      "step": 5320
+    },
+    {
+      "epoch": 0.8529705941188238,
+      "grad_norm": 0.48778149485588074,
+      "learning_rate": 3.216752641492951e-07,
+      "loss": 1.229,
+      "step": 5330
+    },
+    {
+      "epoch": 0.8545709141828366,
+      "grad_norm": 0.5513453483581543,
+      "learning_rate": 3.1485542886108945e-07,
+      "loss": 1.2244,
+      "step": 5340
+    },
+    {
+      "epoch": 0.8561712342468494,
+      "grad_norm": 0.5131115317344666,
+      "learning_rate": 3.081038027207944e-07,
+      "loss": 1.1894,
+      "step": 5350
+    },
+    {
+      "epoch": 0.8577715543108622,
+      "grad_norm": 0.49720299243927,
+      "learning_rate": 3.014205964799888e-07,
+      "loss": 1.2314,
+      "step": 5360
+    },
+    {
+      "epoch": 0.8593718743748749,
+      "grad_norm": 0.511868417263031,
+      "learning_rate": 2.948060187545276e-07,
+      "loss": 1.2065,
+      "step": 5370
+    },
+    {
+      "epoch": 0.8609721944388877,
+      "grad_norm": 0.5046089291572571,
+      "learning_rate": 2.8826027601803085e-07,
+      "loss": 1.2016,
+      "step": 5380
+    },
+    {
+      "epoch": 0.8625725145029006,
+      "grad_norm": 0.5005461573600769,
+      "learning_rate": 2.8178357259543876e-07,
+      "loss": 1.2073,
+      "step": 5390
+    },
+    {
+      "epoch": 0.8641728345669134,
+      "grad_norm": 0.5057045221328735,
+      "learning_rate": 2.7537611065663497e-07,
+      "loss": 1.2147,
+      "step": 5400
+    },
+    {
+      "epoch": 0.8657731546309262,
+      "grad_norm": 0.4953499436378479,
+      "learning_rate": 2.690380902101339e-07,
+      "loss": 1.2347,
+      "step": 5410
+    },
+    {
+      "epoch": 0.867373474694939,
+      "grad_norm": 0.5268727540969849,
+      "learning_rate": 2.6276970909683674e-07,
+      "loss": 1.2551,
+      "step": 5420
+    },
+    {
+      "epoch": 0.8689737947589518,
+      "grad_norm": 0.5176235437393188,
+      "learning_rate": 2.5657116298385964e-07,
+      "loss": 1.2194,
+      "step": 5430
+    },
+    {
+      "epoch": 0.8705741148229645,
+      "grad_norm": 0.5456222891807556,
+      "learning_rate": 2.5044264535842244e-07,
+      "loss": 1.2365,
+      "step": 5440
+    },
+    {
+      "epoch": 0.8721744348869774,
+      "grad_norm": 0.5065416693687439,
+      "learning_rate": 2.443843475218105e-07,
+      "loss": 1.2296,
+      "step": 5450
+    },
+    {
+      "epoch": 0.8737747549509902,
+      "grad_norm": 0.5186034440994263,
+      "learning_rate": 2.383964585834031e-07,
+      "loss": 1.2013,
+      "step": 5460
+    },
+    {
+      "epoch": 0.875375075015003,
+      "grad_norm": 0.5220500826835632,
+      "learning_rate": 2.3247916545477007e-07,
+      "loss": 1.2373,
+      "step": 5470
+    },
+    {
+      "epoch": 0.8769753950790158,
+      "grad_norm": 0.5157853960990906,
+      "learning_rate": 2.2663265284383757e-07,
+      "loss": 1.241,
+      "step": 5480
+    },
+    {
+      "epoch": 0.8785757151430286,
+      "grad_norm": 0.48508715629577637,
+      "learning_rate": 2.2085710324912275e-07,
+      "loss": 1.2457,
+      "step": 5490
+    },
+    {
+      "epoch": 0.8801760352070415,
+      "grad_norm": 0.5108803510665894,
+      "learning_rate": 2.1515269695403656e-07,
+      "loss": 1.1906,
+      "step": 5500
+    },
+    {
+      "epoch": 0.8817763552710542,
+      "grad_norm": 0.5034008026123047,
+      "learning_rate": 2.0951961202125588e-07,
+      "loss": 1.1725,
+      "step": 5510
+    },
+    {
+      "epoch": 0.883376675335067,
+      "grad_norm": 0.5276970863342285,
+      "learning_rate": 2.0395802428716666e-07,
+      "loss": 1.2374,
+      "step": 5520
+    },
+    {
+      "epoch": 0.8849769953990798,
+      "grad_norm": 0.4890348017215729,
+      "learning_rate": 1.9846810735637378e-07,
+      "loss": 1.2215,
+      "step": 5530
+    },
+    {
+      "epoch": 0.8865773154630926,
+      "grad_norm": 0.5114810466766357,
+      "learning_rate": 1.9305003259628307e-07,
+      "loss": 1.2035,
+      "step": 5540
+    },
+    {
+      "epoch": 0.8881776355271054,
+      "grad_norm": 0.506430983543396,
+      "learning_rate": 1.8770396913175115e-07,
+      "loss": 1.2278,
+      "step": 5550
+    },
+    {
+      "epoch": 0.8897779555911183,
+      "grad_norm": 0.5225040912628174,
+      "learning_rate": 1.8243008383980716e-07,
+      "loss": 1.212,
+      "step": 5560
+    },
+    {
+      "epoch": 0.8913782756551311,
+      "grad_norm": 0.5081613659858704,
+      "learning_rate": 1.7722854134444246e-07,
+      "loss": 1.2222,
+      "step": 5570
+    },
+    {
+      "epoch": 0.8929785957191438,
+      "grad_norm": 0.5115835070610046,
+      "learning_rate": 1.7209950401147368e-07,
+      "loss": 1.2043,
+      "step": 5580
+    },
+    {
+      "epoch": 0.8945789157831566,
+      "grad_norm": 0.5055177211761475,
+      "learning_rate": 1.6704313194347204e-07,
+      "loss": 1.2989,
+      "step": 5590
+    },
+    {
+      "epoch": 0.8961792358471694,
+      "grad_norm": 0.49186971783638,
+      "learning_rate": 1.6205958297476787e-07,
+      "loss": 1.2083,
+      "step": 5600
+    },
+    {
+      "epoch": 0.8977795559111822,
+      "grad_norm": 0.5076245665550232,
+      "learning_rate": 1.5714901266652426e-07,
+      "loss": 1.2493,
+      "step": 5610
+    },
+    {
+      "epoch": 0.8993798759751951,
+      "grad_norm": 0.5159875154495239,
+      "learning_rate": 1.5231157430187853e-07,
+      "loss": 1.2123,
+      "step": 5620
+    },
+    {
+      "epoch": 0.9009801960392079,
+      "grad_norm": 0.48809802532196045,
+      "learning_rate": 1.4754741888115959e-07,
+      "loss": 1.2238,
+      "step": 5630
+    },
+    {
+      "epoch": 0.9025805161032207,
+      "grad_norm": 0.5183054208755493,
+      "learning_rate": 1.4285669511717364e-07,
+      "loss": 1.2591,
+      "step": 5640
+    },
+    {
+      "epoch": 0.9041808361672334,
+      "grad_norm": 0.5310025215148926,
+      "learning_rate": 1.3823954943056355e-07,
+      "loss": 1.215,
+      "step": 5650
+    },
+    {
+      "epoch": 0.9057811562312462,
+      "grad_norm": 0.4965580105781555,
+      "learning_rate": 1.3369612594523544e-07,
+      "loss": 1.2061,
+      "step": 5660
+    },
+    {
+      "epoch": 0.907381476295259,
+      "grad_norm": 0.5230700969696045,
+      "learning_rate": 1.2922656648386266e-07,
+      "loss": 1.2409,
+      "step": 5670
+    },
+    {
+      "epoch": 0.9089817963592719,
+      "grad_norm": 0.5221272706985474,
+      "learning_rate": 1.2483101056345815e-07,
+      "loss": 1.2387,
+      "step": 5680
+    },
+    {
+      "epoch": 0.9105821164232847,
+      "grad_norm": 0.5355955958366394,
+      "learning_rate": 1.2050959539101846e-07,
+      "loss": 1.2481,
+      "step": 5690
+    },
+    {
+      "epoch": 0.9121824364872975,
+      "grad_norm": 0.4908137619495392,
+      "learning_rate": 1.1626245585924123e-07,
+      "loss": 1.196,
+      "step": 5700
+    },
+    {
+      "epoch": 0.9137827565513102,
+      "grad_norm": 0.5292563438415527,
+      "learning_rate": 1.1208972454231526e-07,
+      "loss": 1.2779,
+      "step": 5710
+    },
+    {
+      "epoch": 0.915383076615323,
+      "grad_norm": 0.5044008493423462,
+      "learning_rate": 1.0799153169178206e-07,
+      "loss": 1.2197,
+      "step": 5720
+    },
+    {
+      "epoch": 0.9169833966793358,
+      "grad_norm": 1.3565585613250732,
+      "learning_rate": 1.0396800523246847e-07,
+      "loss": 1.2523,
+      "step": 5730
+    },
+    {
+      "epoch": 0.9185837167433487,
+      "grad_norm": 0.5061067938804626,
+      "learning_rate": 1.0001927075849543e-07,
+      "loss": 1.231,
+      "step": 5740
+    },
+    {
+      "epoch": 0.9201840368073615,
+      "grad_norm": 0.5245989561080933,
+      "learning_rate": 9.614545152935695e-08,
+      "loss": 1.2345,
+      "step": 5750
+    },
+    {
+      "epoch": 0.9217843568713743,
+      "grad_norm": 0.5027105808258057,
+      "learning_rate": 9.234666846607204e-08,
+      "loss": 1.3076,
+      "step": 5760
+    },
+    {
+      "epoch": 0.9233846769353871,
+      "grad_norm": 0.5443490743637085,
+      "learning_rate": 8.862304014741113e-08,
+      "loss": 1.2521,
+      "step": 5770
+    },
+    {
+      "epoch": 0.9249849969993998,
+      "grad_norm": 0.49271267652511597,
+      "learning_rate": 8.497468280619397e-08,
+      "loss": 1.2442,
+      "step": 5780
+    },
+    {
+      "epoch": 0.9265853170634126,
+      "grad_norm": 0.5291959047317505,
+      "learning_rate": 8.140171032566091e-08,
+      "loss": 1.2284,
+      "step": 5790
+    },
+    {
+      "epoch": 0.9281856371274255,
+      "grad_norm": 0.5186416506767273,
+      "learning_rate": 7.790423423591958e-08,
+      "loss": 1.2079,
+      "step": 5800
+    },
+    {
+      "epoch": 0.9297859571914383,
+      "grad_norm": 0.5180310010910034,
+      "learning_rate": 7.44823637104622e-08,
+      "loss": 1.2409,
+      "step": 5810
+    },
+    {
+      "epoch": 0.9313862772554511,
+      "grad_norm": 0.5114362835884094,
+      "learning_rate": 7.113620556275819e-08,
+      "loss": 1.2422,
+      "step": 5820
+    },
+    {
+      "epoch": 0.9329865973194639,
+      "grad_norm": 0.48198938369750977,
+      "learning_rate": 6.786586424291975e-08,
+      "loss": 1.2177,
+      "step": 5830
+    },
+    {
+      "epoch": 0.9345869173834767,
+      "grad_norm": 0.5073717832565308,
+      "learning_rate": 6.467144183444213e-08,
+      "loss": 1.2919,
+      "step": 5840
+    },
+    {
+      "epoch": 0.9361872374474896,
+      "grad_norm": 0.5134086608886719,
+      "learning_rate": 6.155303805101597e-08,
+      "loss": 1.2253,
+      "step": 5850
+    },
+    {
+      "epoch": 0.9377875575115023,
+      "grad_norm": 0.5133287310600281,
+      "learning_rate": 5.851075023341618e-08,
+      "loss": 1.2342,
+      "step": 5860
+    },
+    {
+      "epoch": 0.9393878775755151,
+      "grad_norm": 0.5021890997886658,
+      "learning_rate": 5.55446733464618e-08,
+      "loss": 1.2029,
+      "step": 5870
+    },
+    {
+      "epoch": 0.9409881976395279,
+      "grad_norm": 0.5119520425796509,
+      "learning_rate": 5.265489997605372e-08,
+      "loss": 1.2402,
+      "step": 5880
+    },
+    {
+      "epoch": 0.9425885177035407,
+      "grad_norm": 0.4851899743080139,
+      "learning_rate": 4.98415203262817e-08,
+      "loss": 1.1988,
+      "step": 5890
+    },
+    {
+      "epoch": 0.9441888377675535,
+      "grad_norm": 0.4942720830440521,
+      "learning_rate": 4.7104622216611884e-08,
+      "loss": 1.2398,
+      "step": 5900
+    },
+    {
+      "epoch": 0.9457891578315664,
+      "grad_norm": 3.8196513652801514,
+      "learning_rate": 4.444429107914372e-08,
+      "loss": 1.3265,
+      "step": 5910
+    },
+    {
+      "epoch": 0.9473894778955791,
+      "grad_norm": 0.5300382971763611,
+      "learning_rate": 4.186060995594293e-08,
+      "loss": 1.2989,
+      "step": 5920
+    },
+    {
+      "epoch": 0.9489897979595919,
+      "grad_norm": 0.5063347816467285,
+      "learning_rate": 3.935365949645109e-08,
+      "loss": 1.2128,
+      "step": 5930
+    },
+    {
+      "epoch": 0.9505901180236047,
+      "grad_norm": 0.5129570364952087,
+      "learning_rate": 3.692351795496568e-08,
+      "loss": 1.2242,
+      "step": 5940
+    },
+    {
+      "epoch": 0.9521904380876175,
+      "grad_norm": 0.4922707974910736,
+      "learning_rate": 3.457026118820012e-08,
+      "loss": 1.2326,
+      "step": 5950
+    },
+    {
+      "epoch": 0.9537907581516303,
+      "grad_norm": 0.48577573895454407,
+      "learning_rate": 3.229396265291285e-08,
+      "loss": 1.2291,
+      "step": 5960
+    },
+    {
+      "epoch": 0.9553910782156432,
+      "grad_norm": 0.491740882396698,
+      "learning_rate": 3.009469340361726e-08,
+      "loss": 1.228,
+      "step": 5970
+    },
+    {
+      "epoch": 0.956991398279656,
+      "grad_norm": 0.5267091393470764,
+      "learning_rate": 2.797252209036233e-08,
+      "loss": 1.2411,
+      "step": 5980
+    },
+    {
+      "epoch": 0.9585917183436687,
+      "grad_norm": 0.5358031988143921,
+      "learning_rate": 2.5927514956589628e-08,
+      "loss": 1.2506,
+      "step": 5990
+    },
+    {
+      "epoch": 0.9601920384076815,
+      "grad_norm": 0.50943922996521,
+      "learning_rate": 2.395973583706607e-08,
+      "loss": 1.2033,
+      "step": 6000
+    },
+    {
+      "epoch": 0.9617923584716943,
+      "grad_norm": 0.5052767992019653,
+      "learning_rate": 2.206924615589079e-08,
+      "loss": 1.1939,
+      "step": 6010
+    },
+    {
+      "epoch": 0.9633926785357071,
+      "grad_norm": 0.5414103269577026,
+      "learning_rate": 2.0256104924578357e-08,
+      "loss": 1.2267,
+      "step": 6020
+    },
+    {
+      "epoch": 0.96499299859972,
+      "grad_norm": 0.501817524433136,
+      "learning_rate": 1.8520368740215787e-08,
+      "loss": 1.2503,
+      "step": 6030
+    },
+    {
+      "epoch": 0.9665933186637328,
+      "grad_norm": 0.5033535957336426,
+      "learning_rate": 1.6862091783697576e-08,
+      "loss": 1.2043,
+      "step": 6040
+    },
+    {
+      "epoch": 0.9681936387277456,
+      "grad_norm": 0.5262925624847412,
+      "learning_rate": 1.5281325818032045e-08,
+      "loss": 1.2777,
+      "step": 6050
+    },
+    {
+      "epoch": 0.9697939587917583,
+      "grad_norm": 0.503886342048645,
+      "learning_rate": 1.3778120186728472e-08,
+      "loss": 1.2007,
+      "step": 6060
+    },
+    {
+      "epoch": 0.9713942788557711,
+      "grad_norm": 4.596221923828125,
+      "learning_rate": 1.2352521812253603e-08,
+      "loss": 1.2417,
+      "step": 6070
+    },
+    {
+      "epoch": 0.9729945989197839,
+      "grad_norm": 0.5384092330932617,
+      "learning_rate": 1.1004575194570044e-08,
+      "loss": 1.2333,
+      "step": 6080
+    },
+    {
+      "epoch": 0.9745949189837968,
+      "grad_norm": 0.4920881390571594,
+      "learning_rate": 9.734322409744867e-09,
+      "loss": 1.2,
+      "step": 6090
+    },
+    {
+      "epoch": 0.9761952390478096,
+      "grad_norm": 0.5168455839157104,
+      "learning_rate": 8.541803108637613e-09,
+      "loss": 1.1556,
+      "step": 6100
+    },
+    {
+      "epoch": 0.9777955591118224,
+      "grad_norm": 0.5123913884162903,
+      "learning_rate": 7.427054515661835e-09,
+      "loss": 1.2644,
+      "step": 6110
+    },
+    {
+      "epoch": 0.9793958791758351,
+      "grad_norm": 0.5197492837905884,
+      "learning_rate": 6.390111427623524e-09,
+      "loss": 1.2119,
+      "step": 6120
+    },
+    {
+      "epoch": 0.9809961992398479,
+      "grad_norm": 0.5286774039268494,
+      "learning_rate": 5.43100621263476e-09,
+      "loss": 1.208,
+      "step": 6130
+    },
+    {
+      "epoch": 0.9825965193038608,
+      "grad_norm": 0.5219496488571167,
+      "learning_rate": 4.549768809103406e-09,
+      "loss": 1.2614,
+      "step": 6140
+    },
+    {
+      "epoch": 0.9841968393678736,
+      "grad_norm": 0.4979791045188904,
+      "learning_rate": 3.74642672479858e-09,
+      "loss": 1.2224,
+      "step": 6150
+    },
+    {
+      "epoch": 0.9857971594318864,
+      "grad_norm": 0.5516430735588074,
+      "learning_rate": 3.021005035992175e-09,
+      "loss": 1.2278,
+      "step": 6160
+    },
+    {
+      "epoch": 0.9873974794958992,
+      "grad_norm": 0.5197909474372864,
+      "learning_rate": 2.373526386675873e-09,
+      "loss": 1.2345,
+      "step": 6170
+    },
+    {
+      "epoch": 0.988997799559912,
+      "grad_norm": 0.5151229500770569,
+      "learning_rate": 1.8040109878539326e-09,
+      "loss": 1.2125,
+      "step": 6180
+    },
+    {
+      "epoch": 0.9905981196239247,
+      "grad_norm": 0.5217447280883789,
+      "learning_rate": 1.3124766169131386e-09,
+      "loss": 1.2251,
+      "step": 6190
+    },
+    {
+      "epoch": 0.9921984396879376,
+      "grad_norm": 0.4913589060306549,
+      "learning_rate": 8.989386170674131e-10,
+      "loss": 1.232,
+      "step": 6200
+    },
+    {
+      "epoch": 0.9937987597519504,
+      "grad_norm": 0.5131626725196838,
+      "learning_rate": 5.634098968793078e-10,
+      "loss": 1.2282,
+      "step": 6210
+    },
+    {
+      "epoch": 0.9953990798159632,
+      "grad_norm": 0.48121556639671326,
+      "learning_rate": 3.059009298558846e-10,
+      "loss": 1.2193,
+      "step": 6220
+    },
+    {
+      "epoch": 0.996999399879976,
+      "grad_norm": 0.5183552503585815,
+      "learning_rate": 1.264197541234191e-10,
+      "loss": 1.2284,
+      "step": 6230
+    },
+    {
+      "epoch": 0.9985997199439888,
+      "grad_norm": 0.513854444026947,
+      "learning_rate": 2.4971972175102943e-11,
+      "loss": 1.177,
+      "step": 6240
+    },
+    {
+      "epoch": 0.9998799759951991,
+      "step": 6248,
+      "total_flos": 1.306193982790828e+18,
+      "train_loss": 1.279416140650665,
+      "train_runtime": 62601.4894,
+      "train_samples_per_second": 1.597,
+      "train_steps_per_second": 0.1
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 6248,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.306193982790828e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}