diff --git "a/checkpoint-3742/trainer_state.json" "b/checkpoint-3742/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-3742/trainer_state.json"
@@ -0,0 +1,5420 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 200,
+  "global_step": 3742,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0002672367717797969,
+      "grad_norm": 8.823081418058603,
+      "learning_rate": 5.3333333333333334e-08,
+      "loss": 1.906,
+      "step": 1
+    },
+    {
+      "epoch": 0.0013361838588989846,
+      "grad_norm": 9.375402635266639,
+      "learning_rate": 2.666666666666667e-07,
+      "loss": 1.8044,
+      "step": 5
+    },
+    {
+      "epoch": 0.002672367717797969,
+      "grad_norm": 8.65210459980887,
+      "learning_rate": 5.333333333333335e-07,
+      "loss": 1.8182,
+      "step": 10
+    },
+    {
+      "epoch": 0.004008551576696953,
+      "grad_norm": 6.51103320753397,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 1.7509,
+      "step": 15
+    },
+    {
+      "epoch": 0.005344735435595938,
+      "grad_norm": 4.268692783385918,
+      "learning_rate": 1.066666666666667e-06,
+      "loss": 1.767,
+      "step": 20
+    },
+    {
+      "epoch": 0.006680919294494923,
+      "grad_norm": 4.3863193537387355,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 1.7519,
+      "step": 25
+    },
+    {
+      "epoch": 0.008017103153393906,
+      "grad_norm": 3.465625691597637,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.7079,
+      "step": 30
+    },
+    {
+      "epoch": 0.00935328701229289,
+      "grad_norm": 2.6102827946343203,
+      "learning_rate": 1.8666666666666669e-06,
+      "loss": 1.6753,
+      "step": 35
+    },
+    {
+      "epoch": 0.010689470871191877,
+      "grad_norm": 2.3789490657646937,
+      "learning_rate": 2.133333333333334e-06,
+      "loss": 1.6337,
+      "step": 40
+    },
+    {
+      "epoch": 0.012025654730090861,
+      "grad_norm": 2.0835126121909235,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.6641,
+      "step": 45
+    },
+    {
+      "epoch": 0.013361838588989846,
+      "grad_norm": 1.8559179146822158,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 1.6296,
+      "step": 50
+    },
+    {
+      "epoch": 0.01469802244788883,
+      "grad_norm": 1.8106018636994041,
+      "learning_rate": 2.9333333333333338e-06,
+      "loss": 1.5509,
+      "step": 55
+    },
+    {
+      "epoch": 0.016034206306787813,
+      "grad_norm": 1.8755467245728914,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 1.5601,
+      "step": 60
+    },
+    {
+      "epoch": 0.017370390165686797,
+      "grad_norm": 1.7318124276404359,
+      "learning_rate": 3.4666666666666672e-06,
+      "loss": 1.5308,
+      "step": 65
+    },
+    {
+      "epoch": 0.01870657402458578,
+      "grad_norm": 1.6976770309274594,
+      "learning_rate": 3.7333333333333337e-06,
+      "loss": 1.5609,
+      "step": 70
+    },
+    {
+      "epoch": 0.020042757883484766,
+      "grad_norm": 1.7565922656582902,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.5082,
+      "step": 75
+    },
+    {
+      "epoch": 0.021378941742383754,
+      "grad_norm": 1.7393962369808933,
+      "learning_rate": 4.266666666666668e-06,
+      "loss": 1.5306,
+      "step": 80
+    },
+    {
+      "epoch": 0.022715125601282738,
+      "grad_norm": 1.720884117447266,
+      "learning_rate": 4.533333333333334e-06,
+      "loss": 1.5378,
+      "step": 85
+    },
+    {
+      "epoch": 0.024051309460181722,
+      "grad_norm": 1.7101397134036984,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.4813,
+      "step": 90
+    },
+    {
+      "epoch": 0.025387493319080707,
+      "grad_norm": 1.615617842911665,
+      "learning_rate": 5.0666666666666676e-06,
+      "loss": 1.4817,
+      "step": 95
+    },
+    {
+      "epoch": 0.02672367717797969,
+      "grad_norm": 2.691934178931898,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 1.4931,
+      "step": 100
+    },
+    {
+      "epoch": 0.028059861036878676,
+      "grad_norm": 1.5821332066035256,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 1.4285,
+      "step": 105
+    },
+    {
+      "epoch": 0.02939604489577766,
+      "grad_norm": 1.7190020107955757,
+      "learning_rate": 5.8666666666666675e-06,
+      "loss": 1.4481,
+      "step": 110
+    },
+    {
+      "epoch": 0.030732228754676644,
+      "grad_norm": 1.6645689315016792,
+      "learning_rate": 6.133333333333334e-06,
+      "loss": 1.465,
+      "step": 115
+    },
+    {
+      "epoch": 0.032068412613575625,
+      "grad_norm": 1.6462290185572064,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 1.4665,
+      "step": 120
+    },
+    {
+      "epoch": 0.03340459647247461,
+      "grad_norm": 1.649228040380627,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.4209,
+      "step": 125
+    },
+    {
+      "epoch": 0.034740780331373594,
+      "grad_norm": 1.6335763283148064,
+      "learning_rate": 6.9333333333333344e-06,
+      "loss": 1.4151,
+      "step": 130
+    },
+    {
+      "epoch": 0.03607696419027258,
+      "grad_norm": 1.6194942414847415,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 1.4377,
+      "step": 135
+    },
+    {
+      "epoch": 0.03741314804917156,
+      "grad_norm": 1.7105888808318117,
+      "learning_rate": 7.4666666666666675e-06,
+      "loss": 1.3934,
+      "step": 140
+    },
+    {
+      "epoch": 0.03874933190807055,
+      "grad_norm": 1.4814484480334174,
+      "learning_rate": 7.733333333333334e-06,
+      "loss": 1.4015,
+      "step": 145
+    },
+    {
+      "epoch": 0.04008551576696953,
+      "grad_norm": 1.5777144338014444,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.4021,
+      "step": 150
+    },
+    {
+      "epoch": 0.041421699625868516,
+      "grad_norm": 1.5930794844441725,
+      "learning_rate": 8.266666666666667e-06,
+      "loss": 1.434,
+      "step": 155
+    },
+    {
+      "epoch": 0.04275788348476751,
+      "grad_norm": 1.6091950467097091,
+      "learning_rate": 8.533333333333335e-06,
+      "loss": 1.3956,
+      "step": 160
+    },
+    {
+      "epoch": 0.04409406734366649,
+      "grad_norm": 1.6975335096598412,
+      "learning_rate": 8.8e-06,
+      "loss": 1.3996,
+      "step": 165
+    },
+    {
+      "epoch": 0.045430251202565476,
+      "grad_norm": 1.5343847156625996,
+      "learning_rate": 9.066666666666667e-06,
+      "loss": 1.3799,
+      "step": 170
+    },
+    {
+      "epoch": 0.04676643506146446,
+      "grad_norm": 1.770466644954361,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 1.3545,
+      "step": 175
+    },
+    {
+      "epoch": 0.048102618920363445,
+      "grad_norm": 1.5660422280689605,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 1.4292,
+      "step": 180
+    },
+    {
+      "epoch": 0.04943880277926243,
+      "grad_norm": 1.536676360877846,
+      "learning_rate": 9.866666666666668e-06,
+      "loss": 1.4028,
+      "step": 185
+    },
+    {
+      "epoch": 0.050774986638161414,
+      "grad_norm": 1.4804803919916396,
+      "learning_rate": 1.0133333333333335e-05,
+      "loss": 1.3899,
+      "step": 190
+    },
+    {
+      "epoch": 0.0521111704970604,
+      "grad_norm": 1.6476917950123358,
+      "learning_rate": 1.04e-05,
+      "loss": 1.4214,
+      "step": 195
+    },
+    {
+      "epoch": 0.05344735435595938,
+      "grad_norm": 1.5933336407770893,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 1.3663,
+      "step": 200
+    },
+    {
+      "epoch": 0.05344735435595938,
+      "eval_loss": 1.3954896926879883,
+      "eval_runtime": 525.9836,
+      "eval_samples_per_second": 25.183,
+      "eval_steps_per_second": 3.148,
+      "step": 200
+    },
+    {
+      "epoch": 0.05478353821485837,
+      "grad_norm": 1.55465727793257,
+      "learning_rate": 1.0933333333333334e-05,
+      "loss": 1.3566,
+      "step": 205
+    },
+    {
+      "epoch": 0.05611972207375735,
+      "grad_norm": 1.6465837885812638,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 1.3572,
+      "step": 210
+    },
+    {
+      "epoch": 0.057455905932656336,
+      "grad_norm": 1.6314586937231732,
+      "learning_rate": 1.1466666666666668e-05,
+      "loss": 1.4066,
+      "step": 215
+    },
+    {
+      "epoch": 0.05879208979155532,
+      "grad_norm": 1.5940186416338837,
+      "learning_rate": 1.1733333333333335e-05,
+      "loss": 1.3727,
+      "step": 220
+    },
+    {
+      "epoch": 0.060128273650454304,
+      "grad_norm": 1.587832961506543,
+      "learning_rate": 1.2e-05,
+      "loss": 1.3755,
+      "step": 225
+    },
+    {
+      "epoch": 0.06146445750935329,
+      "grad_norm": 1.5731554504380694,
+      "learning_rate": 1.2266666666666667e-05,
+      "loss": 1.4007,
+      "step": 230
+    },
+    {
+      "epoch": 0.06280064136825227,
+      "grad_norm": 1.5722181818634398,
+      "learning_rate": 1.2533333333333336e-05,
+      "loss": 1.3783,
+      "step": 235
+    },
+    {
+      "epoch": 0.06413682522715125,
+      "grad_norm": 1.467532784965517,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 1.4386,
+      "step": 240
+    },
+    {
+      "epoch": 0.06547300908605024,
+      "grad_norm": 1.5125828644122057,
+      "learning_rate": 1.3066666666666668e-05,
+      "loss": 1.3186,
+      "step": 245
+    },
+    {
+      "epoch": 0.06680919294494922,
+      "grad_norm": 1.5183168300505157,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.3551,
+      "step": 250
+    },
+    {
+      "epoch": 0.0681453768038482,
+      "grad_norm": 1.4941218649910812,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 1.3664,
+      "step": 255
+    },
+    {
+      "epoch": 0.06948156066274719,
+      "grad_norm": 1.4869296436066686,
+      "learning_rate": 1.3866666666666669e-05,
+      "loss": 1.3576,
+      "step": 260
+    },
+    {
+      "epoch": 0.07081774452164617,
+      "grad_norm": 1.4293831791321558,
+      "learning_rate": 1.4133333333333334e-05,
+      "loss": 1.3556,
+      "step": 265
+    },
+    {
+      "epoch": 0.07215392838054516,
+      "grad_norm": 1.5191533813334088,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 1.3366,
+      "step": 270
+    },
+    {
+      "epoch": 0.07349011223944414,
+      "grad_norm": 1.4810082339187982,
+      "learning_rate": 1.4666666666666666e-05,
+      "loss": 1.3573,
+      "step": 275
+    },
+    {
+      "epoch": 0.07482629609834313,
+      "grad_norm": 1.5063612979974181,
+      "learning_rate": 1.4933333333333335e-05,
+      "loss": 1.3427,
+      "step": 280
+    },
+    {
+      "epoch": 0.07616247995724211,
+      "grad_norm": 1.618950454636469,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 1.3554,
+      "step": 285
+    },
+    {
+      "epoch": 0.0774986638161411,
+      "grad_norm": 1.5626468747954396,
+      "learning_rate": 1.546666666666667e-05,
+      "loss": 1.3771,
+      "step": 290
+    },
+    {
+      "epoch": 0.07883484767504008,
+      "grad_norm": 1.5292548773347097,
+      "learning_rate": 1.5733333333333334e-05,
+      "loss": 1.3918,
+      "step": 295
+    },
+    {
+      "epoch": 0.08017103153393906,
+      "grad_norm": 1.5358359721754233,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.3815,
+      "step": 300
+    },
+    {
+      "epoch": 0.08150721539283805,
+      "grad_norm": 1.3953477584618263,
+      "learning_rate": 1.6266666666666668e-05,
+      "loss": 1.3854,
+      "step": 305
+    },
+    {
+      "epoch": 0.08284339925173703,
+      "grad_norm": 1.4468372312833726,
+      "learning_rate": 1.6533333333333333e-05,
+      "loss": 1.3727,
+      "step": 310
+    },
+    {
+      "epoch": 0.08417958311063603,
+      "grad_norm": 1.488344751961309,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 1.3706,
+      "step": 315
+    },
+    {
+      "epoch": 0.08551576696953501,
+      "grad_norm": 1.4093480949221249,
+      "learning_rate": 1.706666666666667e-05,
+      "loss": 1.3491,
+      "step": 320
+    },
+    {
+      "epoch": 0.086851950828434,
+      "grad_norm": 1.5432094268322356,
+      "learning_rate": 1.7333333333333336e-05,
+      "loss": 1.3586,
+      "step": 325
+    },
+    {
+      "epoch": 0.08818813468733298,
+      "grad_norm": 1.4550377653252384,
+      "learning_rate": 1.76e-05,
+      "loss": 1.39,
+      "step": 330
+    },
+    {
+      "epoch": 0.08952431854623197,
+      "grad_norm": 1.4281640050412778,
+      "learning_rate": 1.7866666666666666e-05,
+      "loss": 1.3669,
+      "step": 335
+    },
+    {
+      "epoch": 0.09086050240513095,
+      "grad_norm": 1.4426556416465324,
+      "learning_rate": 1.8133333333333335e-05,
+      "loss": 1.3876,
+      "step": 340
+    },
+    {
+      "epoch": 0.09219668626402994,
+      "grad_norm": 1.406168800248235,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 1.3686,
+      "step": 345
+    },
+    {
+      "epoch": 0.09353287012292892,
+      "grad_norm": 1.4301962338673828,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 1.3689,
+      "step": 350
+    },
+    {
+      "epoch": 0.0948690539818279,
+      "grad_norm": 1.6044731560961005,
+      "learning_rate": 1.8933333333333334e-05,
+      "loss": 1.4118,
+      "step": 355
+    },
+    {
+      "epoch": 0.09620523784072689,
+      "grad_norm": 1.4879888050516272,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 1.3791,
+      "step": 360
+    },
+    {
+      "epoch": 0.09754142169962587,
+      "grad_norm": 1.4698368446133456,
+      "learning_rate": 1.9466666666666668e-05,
+      "loss": 1.348,
+      "step": 365
+    },
+    {
+      "epoch": 0.09887760555852486,
+      "grad_norm": 1.4186637789128058,
+      "learning_rate": 1.9733333333333336e-05,
+      "loss": 1.3325,
+      "step": 370
+    },
+    {
+      "epoch": 0.10021378941742384,
+      "grad_norm": 1.5476694754206417,
+      "learning_rate": 2e-05,
+      "loss": 1.3642,
+      "step": 375
+    },
+    {
+      "epoch": 0.10154997327632283,
+      "grad_norm": 1.4733865828906345,
+      "learning_rate": 1.9999891176487904e-05,
+      "loss": 1.3807,
+      "step": 380
+    },
+    {
+      "epoch": 0.10288615713522181,
+      "grad_norm": 1.4551876993221122,
+      "learning_rate": 1.9999564708320124e-05,
+      "loss": 1.3679,
+      "step": 385
+    },
+    {
+      "epoch": 0.1042223409941208,
+      "grad_norm": 1.5257129447187303,
+      "learning_rate": 1.999902060260214e-05,
+      "loss": 1.4111,
+      "step": 390
+    },
+    {
+      "epoch": 0.10555852485301978,
+      "grad_norm": 1.3923714980063469,
+      "learning_rate": 1.9998258871176252e-05,
+      "loss": 1.3729,
+      "step": 395
+    },
+    {
+      "epoch": 0.10689470871191876,
+      "grad_norm": 1.4311098965377238,
+      "learning_rate": 1.999727953062132e-05,
+      "loss": 1.3413,
+      "step": 400
+    },
+    {
+      "epoch": 0.10689470871191876,
+      "eval_loss": 1.3721706867218018,
+      "eval_runtime": 526.6891,
+      "eval_samples_per_second": 25.15,
+      "eval_steps_per_second": 3.144,
+      "step": 400
+    },
+    {
+      "epoch": 0.10823089257081775,
+      "grad_norm": 1.3829932271044623,
+      "learning_rate": 1.99960826022524e-05,
+      "loss": 1.3466,
+      "step": 405
+    },
+    {
+      "epoch": 0.10956707642971673,
+      "grad_norm": 1.3521931502381987,
+      "learning_rate": 1.9994668112120283e-05,
+      "loss": 1.3713,
+      "step": 410
+    },
+    {
+      "epoch": 0.11090326028861572,
+      "grad_norm": 1.4976307839881127,
+      "learning_rate": 1.9993036091010924e-05,
+      "loss": 1.3564,
+      "step": 415
+    },
+    {
+      "epoch": 0.1122394441475147,
+      "grad_norm": 1.3701705887791993,
+      "learning_rate": 1.999118657444477e-05,
+      "loss": 1.3666,
+      "step": 420
+    },
+    {
+      "epoch": 0.11357562800641369,
+      "grad_norm": 1.4279953676541446,
+      "learning_rate": 1.9989119602676007e-05,
+      "loss": 1.3498,
+      "step": 425
+    },
+    {
+      "epoch": 0.11491181186531267,
+      "grad_norm": 1.4077392862076545,
+      "learning_rate": 1.9986835220691662e-05,
+      "loss": 1.3655,
+      "step": 430
+    },
+    {
+      "epoch": 0.11624799572421166,
+      "grad_norm": 1.5303947503986481,
+      "learning_rate": 1.9984333478210622e-05,
+      "loss": 1.3758,
+      "step": 435
+    },
+    {
+      "epoch": 0.11758417958311064,
+      "grad_norm": 1.4002227761633332,
+      "learning_rate": 1.9981614429682576e-05,
+      "loss": 1.338,
+      "step": 440
+    },
+    {
+      "epoch": 0.11892036344200962,
+      "grad_norm": 1.3305325078036485,
+      "learning_rate": 1.9978678134286796e-05,
+      "loss": 1.3502,
+      "step": 445
+    },
+    {
+      "epoch": 0.12025654730090861,
+      "grad_norm": 1.4218348465021207,
+      "learning_rate": 1.9975524655930884e-05,
+      "loss": 1.3526,
+      "step": 450
+    },
+    {
+      "epoch": 0.1215927311598076,
+      "grad_norm": 1.3929817535643747,
+      "learning_rate": 1.997215406324936e-05,
+      "loss": 1.3575,
+      "step": 455
+    },
+    {
+      "epoch": 0.12292891501870658,
+      "grad_norm": 1.4164344515855345,
+      "learning_rate": 1.9968566429602166e-05,
+      "loss": 1.3548,
+      "step": 460
+    },
+    {
+      "epoch": 0.12426509887760556,
+      "grad_norm": 1.3940688971110364,
+      "learning_rate": 1.996476183307308e-05,
+      "loss": 1.3651,
+      "step": 465
+    },
+    {
+      "epoch": 0.12560128273650453,
+      "grad_norm": 1.4429246742185509,
+      "learning_rate": 1.996074035646802e-05,
+      "loss": 1.3566,
+      "step": 470
+    },
+    {
+      "epoch": 0.12693746659540353,
+      "grad_norm": 1.4139926227619217,
+      "learning_rate": 1.9956502087313217e-05,
+      "loss": 1.4029,
+      "step": 475
+    },
+    {
+      "epoch": 0.1282736504543025,
+      "grad_norm": 1.36708702924336,
+      "learning_rate": 1.9952047117853345e-05,
+      "loss": 1.3558,
+      "step": 480
+    },
+    {
+      "epoch": 0.1296098343132015,
+      "grad_norm": 1.4333918113560515,
+      "learning_rate": 1.994737554504949e-05,
+      "loss": 1.3735,
+      "step": 485
+    },
+    {
+      "epoch": 0.13094601817210047,
+      "grad_norm": 1.3176020208151544,
+      "learning_rate": 1.994248747057704e-05,
+      "loss": 1.341,
+      "step": 490
+    },
+    {
+      "epoch": 0.13228220203099947,
+      "grad_norm": 1.4308466916521525,
+      "learning_rate": 1.9937383000823485e-05,
+      "loss": 1.405,
+      "step": 495
+    },
+    {
+      "epoch": 0.13361838588989844,
+      "grad_norm": 1.4229715319544036,
+      "learning_rate": 1.9932062246886087e-05,
+      "loss": 1.3655,
+      "step": 500
+    },
+    {
+      "epoch": 0.13495456974879744,
+      "grad_norm": 1.3858117778473207,
+      "learning_rate": 1.992652532456947e-05,
+      "loss": 1.3533,
+      "step": 505
+    },
+    {
+      "epoch": 0.1362907536076964,
+      "grad_norm": 1.3134241462309153,
+      "learning_rate": 1.992077235438311e-05,
+      "loss": 1.3152,
+      "step": 510
+    },
+    {
+      "epoch": 0.1376269374665954,
+      "grad_norm": 1.365018139049316,
+      "learning_rate": 1.991480346153868e-05,
+      "loss": 1.3399,
+      "step": 515
+    },
+    {
+      "epoch": 0.13896312132549438,
+      "grad_norm": 1.426928066531938,
+      "learning_rate": 1.9908618775947364e-05,
+      "loss": 1.3933,
+      "step": 520
+    },
+    {
+      "epoch": 0.14029930518439337,
+      "grad_norm": 1.428258632489371,
+      "learning_rate": 1.9902218432216996e-05,
+      "loss": 1.4028,
+      "step": 525
+    },
+    {
+      "epoch": 0.14163548904329235,
+      "grad_norm": 1.4241733392483251,
+      "learning_rate": 1.989560256964916e-05,
+      "loss": 1.4106,
+      "step": 530
+    },
+    {
+      "epoch": 0.14297167290219134,
+      "grad_norm": 1.3754571749169886,
+      "learning_rate": 1.9888771332236137e-05,
+      "loss": 1.383,
+      "step": 535
+    },
+    {
+      "epoch": 0.14430785676109031,
+      "grad_norm": 1.3253761034149971,
+      "learning_rate": 1.9881724868657768e-05,
+      "loss": 1.3846,
+      "step": 540
+    },
+    {
+      "epoch": 0.1456440406199893,
+      "grad_norm": 1.3292314064664583,
+      "learning_rate": 1.9874463332278245e-05,
+      "loss": 1.3831,
+      "step": 545
+    },
+    {
+      "epoch": 0.14698022447888828,
+      "grad_norm": 1.5134929576884686,
+      "learning_rate": 1.9866986881142737e-05,
+      "loss": 1.3535,
+      "step": 550
+    },
+    {
+      "epoch": 0.14831640833778728,
+      "grad_norm": 1.4571661850393143,
+      "learning_rate": 1.9859295677973988e-05,
+      "loss": 1.3482,
+      "step": 555
+    },
+    {
+      "epoch": 0.14965259219668625,
+      "grad_norm": 1.3720756399964669,
+      "learning_rate": 1.9851389890168738e-05,
+      "loss": 1.3677,
+      "step": 560
+    },
+    {
+      "epoch": 0.15098877605558525,
+      "grad_norm": 1.3902503536690078,
+      "learning_rate": 1.9843269689794114e-05,
+      "loss": 1.3201,
+      "step": 565
+    },
+    {
+      "epoch": 0.15232495991448422,
+      "grad_norm": 1.4845499289340192,
+      "learning_rate": 1.983493525358385e-05,
+      "loss": 1.4121,
+      "step": 570
+    },
+    {
+      "epoch": 0.15366114377338322,
+      "grad_norm": 1.5074167691063687,
+      "learning_rate": 1.982638676293448e-05,
+      "loss": 1.3842,
+      "step": 575
+    },
+    {
+      "epoch": 0.1549973276322822,
+      "grad_norm": 1.386731826246787,
+      "learning_rate": 1.981762440390136e-05,
+      "loss": 1.3248,
+      "step": 580
+    },
+    {
+      "epoch": 0.1563335114911812,
+      "grad_norm": 1.398496405386941,
+      "learning_rate": 1.9808648367194614e-05,
+      "loss": 1.4116,
+      "step": 585
+    },
+    {
+      "epoch": 0.15766969535008016,
+      "grad_norm": 1.3353006312197904,
+      "learning_rate": 1.9799458848175023e-05,
+      "loss": 1.3557,
+      "step": 590
+    },
+    {
+      "epoch": 0.15900587920897916,
+      "grad_norm": 1.3229490481790953,
+      "learning_rate": 1.9790056046849726e-05,
+      "loss": 1.3425,
+      "step": 595
+    },
+    {
+      "epoch": 0.16034206306787813,
+      "grad_norm": 1.3413421989191214,
+      "learning_rate": 1.97804401678679e-05,
+      "loss": 1.365,
+      "step": 600
+    },
+    {
+      "epoch": 0.16034206306787813,
+      "eval_loss": 1.3631840944290161,
+      "eval_runtime": 523.8522,
+      "eval_samples_per_second": 25.286,
+      "eval_steps_per_second": 3.161,
+      "step": 600
+    },
+    {
+      "epoch": 0.16167824692677712,
+      "grad_norm": 1.3574397972892758,
+      "learning_rate": 1.9770611420516286e-05,
+      "loss": 1.3677,
+      "step": 605
+    },
+    {
+      "epoch": 0.1630144307856761,
+      "grad_norm": 1.3451601369327404,
+      "learning_rate": 1.9760570018714647e-05,
+      "loss": 1.3509,
+      "step": 610
+    },
+    {
+      "epoch": 0.1643506146445751,
+      "grad_norm": 1.384864628214446,
+      "learning_rate": 1.975031618101111e-05,
+      "loss": 1.3258,
+      "step": 615
+    },
+    {
+      "epoch": 0.16568679850347406,
+      "grad_norm": 1.3682344543460991,
+      "learning_rate": 1.9739850130577393e-05,
+      "loss": 1.3552,
+      "step": 620
+    },
+    {
+      "epoch": 0.16702298236237306,
+      "grad_norm": 1.4305243117138575,
+      "learning_rate": 1.9729172095203977e-05,
+      "loss": 1.3849,
+      "step": 625
+    },
+    {
+      "epoch": 0.16835916622127206,
+      "grad_norm": 1.3723669913018262,
+      "learning_rate": 1.9718282307295115e-05,
+      "loss": 1.3659,
+      "step": 630
+    },
+    {
+      "epoch": 0.16969535008017103,
+      "grad_norm": 1.3516044026149312,
+      "learning_rate": 1.970718100386381e-05,
+      "loss": 1.3736,
+      "step": 635
+    },
+    {
+      "epoch": 0.17103153393907003,
+      "grad_norm": 1.436359994097034,
+      "learning_rate": 1.969586842652662e-05,
+      "loss": 1.3818,
+      "step": 640
+    },
+    {
+      "epoch": 0.172367717797969,
+      "grad_norm": 1.4157888272139716,
+      "learning_rate": 1.9684344821498432e-05,
+      "loss": 1.3342,
+      "step": 645
+    },
+    {
+      "epoch": 0.173703901656868,
+      "grad_norm": 1.3227193318099222,
+      "learning_rate": 1.9672610439587073e-05,
+      "loss": 1.3862,
+      "step": 650
+    },
+    {
+      "epoch": 0.17504008551576697,
+      "grad_norm": 1.3534338266553816,
+      "learning_rate": 1.9660665536187875e-05,
+      "loss": 1.3617,
+      "step": 655
+    },
+    {
+      "epoch": 0.17637626937466597,
+      "grad_norm": 1.3075745113464525,
+      "learning_rate": 1.9648510371278106e-05,
+      "loss": 1.351,
+      "step": 660
+    },
+    {
+      "epoch": 0.17771245323356494,
+      "grad_norm": 1.3543581869904175,
+      "learning_rate": 1.9636145209411318e-05,
+      "loss": 1.3866,
+      "step": 665
+    },
+    {
+      "epoch": 0.17904863709246394,
+      "grad_norm": 1.3462890650035784,
+      "learning_rate": 1.9623570319711574e-05,
+      "loss": 1.3754,
+      "step": 670
+    },
+    {
+      "epoch": 0.1803848209513629,
+      "grad_norm": 1.2991571392936196,
+      "learning_rate": 1.9610785975867608e-05,
+      "loss": 1.323,
+      "step": 675
+    },
+    {
+      "epoch": 0.1817210048102619,
+      "grad_norm": 1.434385743767408,
+      "learning_rate": 1.9597792456126855e-05,
+      "loss": 1.383,
+      "step": 680
+    },
+    {
+      "epoch": 0.18305718866916088,
+      "grad_norm": 1.3595317685749773,
+      "learning_rate": 1.9584590043289416e-05,
+      "loss": 1.2787,
+      "step": 685
+    },
+    {
+      "epoch": 0.18439337252805987,
+      "grad_norm": 1.6484220578907431,
+      "learning_rate": 1.957117902470187e-05,
+      "loss": 1.3747,
+      "step": 690
+    },
+    {
+      "epoch": 0.18572955638695884,
+      "grad_norm": 1.2945139560173622,
+      "learning_rate": 1.9557559692251047e-05,
+      "loss": 1.3621,
+      "step": 695
+    },
+    {
+      "epoch": 0.18706574024585784,
+      "grad_norm": 1.343012210724551,
+      "learning_rate": 1.9543732342357664e-05,
+      "loss": 1.3205,
+      "step": 700
+    },
+    {
+      "epoch": 0.1884019241047568,
+      "grad_norm": 1.3223543378479201,
+      "learning_rate": 1.9529697275969876e-05,
+      "loss": 1.3101,
+      "step": 705
+    },
+    {
+      "epoch": 0.1897381079636558,
+      "grad_norm": 1.3069901527523717,
+      "learning_rate": 1.951545479855673e-05,
+      "loss": 1.3752,
+      "step": 710
+    },
+    {
+      "epoch": 0.19107429182255478,
+      "grad_norm": 1.3767101616598136,
+      "learning_rate": 1.9501005220101507e-05,
+      "loss": 1.3854,
+      "step": 715
+    },
+    {
+      "epoch": 0.19241047568145378,
+      "grad_norm": 1.3595423389248131,
+      "learning_rate": 1.948634885509498e-05,
+      "loss": 1.3708,
+      "step": 720
+    },
+    {
+      "epoch": 0.19374665954035275,
+      "grad_norm": 1.269545504198035,
+      "learning_rate": 1.947148602252858e-05,
+      "loss": 1.3443,
+      "step": 725
+    },
+    {
+      "epoch": 0.19508284339925175,
+      "grad_norm": 1.2745742850192638,
+      "learning_rate": 1.9456417045887423e-05,
+      "loss": 1.3339,
+      "step": 730
+    },
+    {
+      "epoch": 0.19641902725815072,
+      "grad_norm": 1.4182072631536495,
+      "learning_rate": 1.944114225314331e-05,
+      "loss": 1.3103,
+      "step": 735
+    },
+    {
+      "epoch": 0.19775521111704972,
+      "grad_norm": 1.3233791325230955,
+      "learning_rate": 1.9425661976747552e-05,
+      "loss": 1.3625,
+      "step": 740
+    },
+    {
+      "epoch": 0.1990913949759487,
+      "grad_norm": 1.4760222437081327,
+      "learning_rate": 1.9409976553623767e-05,
+      "loss": 1.3624,
+      "step": 745
+    },
+    {
+      "epoch": 0.20042757883484769,
+      "grad_norm": 1.4210858490294298,
+      "learning_rate": 1.9394086325160515e-05,
+      "loss": 1.3599,
+      "step": 750
+    },
+    {
+      "epoch": 0.20176376269374666,
+      "grad_norm": 1.3258331784580841,
+      "learning_rate": 1.9377991637203894e-05,
+      "loss": 1.3411,
+      "step": 755
+    },
+    {
+      "epoch": 0.20309994655264565,
+      "grad_norm": 1.5486410395081083,
+      "learning_rate": 1.9361692840049997e-05,
+      "loss": 1.4124,
+      "step": 760
+    },
+    {
+      "epoch": 0.20443613041154463,
+      "grad_norm": 1.3108858505753092,
+      "learning_rate": 1.9345190288437292e-05,
+      "loss": 1.3323,
+      "step": 765
+    },
+    {
+      "epoch": 0.20577231427044362,
+      "grad_norm": 1.3401186763256265,
+      "learning_rate": 1.9328484341538903e-05,
+      "loss": 1.3589,
+      "step": 770
+    },
+    {
+      "epoch": 0.2071084981293426,
+      "grad_norm": 1.3258497911117144,
+      "learning_rate": 1.93115753629548e-05,
+      "loss": 1.3263,
+      "step": 775
+    },
+    {
+      "epoch": 0.2084446819882416,
+      "grad_norm": 1.279705605242652,
+      "learning_rate": 1.929446372070386e-05,
+      "loss": 1.3577,
+      "step": 780
+    },
+    {
+      "epoch": 0.20978086584714056,
+      "grad_norm": 1.3688317662377307,
+      "learning_rate": 1.9277149787215893e-05,
+      "loss": 1.3325,
+      "step": 785
+    },
+    {
+      "epoch": 0.21111704970603956,
+      "grad_norm": 1.3308089715501914,
+      "learning_rate": 1.9259633939323504e-05,
+      "loss": 1.3594,
+      "step": 790
+    },
+    {
+      "epoch": 0.21245323356493853,
+      "grad_norm": 1.3306187188699843,
+      "learning_rate": 1.924191655825391e-05,
+      "loss": 1.3706,
+      "step": 795
+    },
+    {
+      "epoch": 0.21378941742383753,
+      "grad_norm": 1.3079863145055592,
+      "learning_rate": 1.922399802962064e-05,
+      "loss": 1.33,
+      "step": 800
+    },
+    {
+      "epoch": 0.21378941742383753,
+      "eval_loss": 1.3532133102416992,
+      "eval_runtime": 523.8533,
+      "eval_samples_per_second": 25.286,
+      "eval_steps_per_second": 3.161,
+      "step": 800
+    },
+    {
+      "epoch": 0.2151256012827365,
+      "grad_norm": 1.2760851031416467,
+      "learning_rate": 1.9205878743415137e-05,
+      "loss": 1.321,
+      "step": 805
+    },
+    {
+      "epoch": 0.2164617851416355,
+      "grad_norm": 1.278778697243769,
+      "learning_rate": 1.9187559093998275e-05,
+      "loss": 1.3632,
+      "step": 810
+    },
+    {
+      "epoch": 0.21779796900053447,
+      "grad_norm": 1.3321470791378587,
+      "learning_rate": 1.916903948009177e-05,
+      "loss": 1.3079,
+      "step": 815
+    },
+    {
+      "epoch": 0.21913415285943347,
+      "grad_norm": 1.3698830796691455,
+      "learning_rate": 1.915032030476951e-05,
+      "loss": 1.3835,
+      "step": 820
+    },
+    {
+      "epoch": 0.22047033671833244,
+      "grad_norm": 1.314469593484434,
+      "learning_rate": 1.913140197544877e-05,
+      "loss": 1.3423,
+      "step": 825
+    },
+    {
+      "epoch": 0.22180652057723144,
+      "grad_norm": 1.3585983787357763,
+      "learning_rate": 1.911228490388136e-05,
+      "loss": 1.3038,
+      "step": 830
+    },
+    {
+      "epoch": 0.2231427044361304,
+      "grad_norm": 1.309351148797077,
+      "learning_rate": 1.9092969506144653e-05,
+      "loss": 1.3711,
+      "step": 835
+    },
+    {
+      "epoch": 0.2244788882950294,
+      "grad_norm": 1.2522351293958003,
+      "learning_rate": 1.907345620263254e-05,
+      "loss": 1.3436,
+      "step": 840
+    },
+    {
+      "epoch": 0.22581507215392838,
+      "grad_norm": 1.2176016475207827,
+      "learning_rate": 1.9053745418046257e-05,
+      "loss": 1.3544,
+      "step": 845
+    },
+    {
+      "epoch": 0.22715125601282737,
+      "grad_norm": 1.3937619861187576,
+      "learning_rate": 1.903383758138517e-05,
+      "loss": 1.3563,
+      "step": 850
+    },
+    {
+      "epoch": 0.22848743987172634,
+      "grad_norm": 1.3085996735753536,
+      "learning_rate": 1.9013733125937412e-05,
+      "loss": 1.3149,
+      "step": 855
+    },
+    {
+      "epoch": 0.22982362373062534,
+      "grad_norm": 1.5210379330796175,
+      "learning_rate": 1.8993432489270484e-05,
+      "loss": 1.3202,
+      "step": 860
+    },
+    {
+      "epoch": 0.2311598075895243,
+      "grad_norm": 1.3435267655155434,
+      "learning_rate": 1.8972936113221696e-05,
+      "loss": 1.357,
+      "step": 865
+    },
+    {
+      "epoch": 0.2324959914484233,
+      "grad_norm": 1.32208190242327,
+      "learning_rate": 1.8952244443888573e-05,
+      "loss": 1.3838,
+      "step": 870
+    },
+    {
+      "epoch": 0.23383217530732228,
+      "grad_norm": 1.3123583179163099,
+      "learning_rate": 1.8931357931619143e-05,
+      "loss": 1.3576,
+      "step": 875
+    },
+    {
+      "epoch": 0.23516835916622128,
+      "grad_norm": 1.2321415331674137,
+      "learning_rate": 1.8910277031002125e-05,
+      "loss": 1.3413,
+      "step": 880
+    },
+    {
+      "epoch": 0.23650454302512025,
+      "grad_norm": 1.2717606000257338,
+      "learning_rate": 1.888900220085706e-05,
+      "loss": 1.3412,
+      "step": 885
+    },
+    {
+      "epoch": 0.23784072688401925,
+      "grad_norm": 1.3184808872543174,
+      "learning_rate": 1.886753390422428e-05,
+      "loss": 1.3616,
+      "step": 890
+    },
+    {
+      "epoch": 0.23917691074291822,
+      "grad_norm": 1.2849347258583887,
+      "learning_rate": 1.8845872608354877e-05,
+      "loss": 1.3021,
+      "step": 895
+    },
+    {
+      "epoch": 0.24051309460181722,
+      "grad_norm": 1.2973872521270147,
+      "learning_rate": 1.882401878470052e-05,
+      "loss": 1.3365,
+      "step": 900
+    },
+    {
+      "epoch": 0.2418492784607162,
+      "grad_norm": 1.3250631489181373,
+      "learning_rate": 1.8801972908903162e-05,
+      "loss": 1.3433,
+      "step": 905
+    },
+    {
+      "epoch": 0.2431854623196152,
+      "grad_norm": 1.263181234454277,
+      "learning_rate": 1.877973546078474e-05,
+      "loss": 1.3703,
+      "step": 910
+    },
+    {
+      "epoch": 0.24452164617851416,
+      "grad_norm": 1.379416457226466,
+      "learning_rate": 1.875730692433669e-05,
+      "loss": 1.353,
+      "step": 915
+    },
+    {
+      "epoch": 0.24585783003741316,
+      "grad_norm": 1.293087775842146,
+      "learning_rate": 1.873468778770944e-05,
+      "loss": 1.3255,
+      "step": 920
+    },
+    {
+      "epoch": 0.24719401389631213,
+      "grad_norm": 1.3395508283607358,
+      "learning_rate": 1.8711878543201757e-05,
+      "loss": 1.2999,
+      "step": 925
+    },
+    {
+      "epoch": 0.24853019775521112,
+      "grad_norm": 1.368607821812935,
+      "learning_rate": 1.8688879687250067e-05,
+      "loss": 1.3305,
+      "step": 930
+    },
+    {
+      "epoch": 0.2498663816141101,
+      "grad_norm": 1.323555417532407,
+      "learning_rate": 1.8665691720417624e-05,
+      "loss": 1.3662,
+      "step": 935
+    },
+    {
+      "epoch": 0.25120256547300907,
+      "grad_norm": 1.367652347855113,
+      "learning_rate": 1.8642315147383628e-05,
+      "loss": 1.3469,
+      "step": 940
+    },
+    {
+      "epoch": 0.25253874933190806,
+      "grad_norm": 1.2966904825916514,
+      "learning_rate": 1.8618750476932237e-05,
+      "loss": 1.3071,
+      "step": 945
+    },
+    {
+      "epoch": 0.25387493319080706,
+      "grad_norm": 1.4450003246232057,
+      "learning_rate": 1.8594998221941482e-05,
+      "loss": 1.3721,
+      "step": 950
+    },
+    {
+      "epoch": 0.25521111704970606,
+      "grad_norm": 1.284360919841786,
+      "learning_rate": 1.857105889937213e-05,
+      "loss": 1.3638,
+      "step": 955
+    },
+    {
+      "epoch": 0.256547300908605,
+      "grad_norm": 1.288974034075832,
+      "learning_rate": 1.8546933030256417e-05,
+      "loss": 1.3349,
+      "step": 960
+    },
+    {
+      "epoch": 0.257883484767504,
+      "grad_norm": 1.3405220877515678,
+      "learning_rate": 1.85226211396867e-05,
+      "loss": 1.3052,
+      "step": 965
+    },
+    {
+      "epoch": 0.259219668626403,
+      "grad_norm": 1.3868830711147844,
+      "learning_rate": 1.8498123756804038e-05,
+      "loss": 1.3609,
+      "step": 970
+    },
+    {
+      "epoch": 0.260555852485302,
+      "grad_norm": 1.3022776914197354,
+      "learning_rate": 1.8473441414786692e-05,
+      "loss": 1.3385,
+      "step": 975
+    },
+    {
+      "epoch": 0.26189203634420094,
+      "grad_norm": 1.4058613068210792,
+      "learning_rate": 1.8448574650838477e-05,
+      "loss": 1.3234,
+      "step": 980
+    },
+    {
+      "epoch": 0.26322822020309994,
+      "grad_norm": 1.3008663043779487,
+      "learning_rate": 1.842352400617712e-05,
+      "loss": 1.3201,
+      "step": 985
+    },
+    {
+      "epoch": 0.26456440406199894,
+      "grad_norm": 1.274233066139097,
+      "learning_rate": 1.8398290026022444e-05,
+      "loss": 1.3372,
+      "step": 990
+    },
+    {
+      "epoch": 0.26590058792089794,
+      "grad_norm": 1.2895370237805952,
+      "learning_rate": 1.8372873259584517e-05,
+      "loss": 1.3082,
+      "step": 995
+    },
+    {
+      "epoch": 0.2672367717797969,
+      "grad_norm": 1.2250435736281928,
+      "learning_rate": 1.83472742600517e-05,
+      "loss": 1.3219,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2672367717797969,
+      "eval_loss": 1.3463348150253296,
+      "eval_runtime": 523.6844,
+      "eval_samples_per_second": 25.294,
+      "eval_steps_per_second": 3.162,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2685729556386959,
+      "grad_norm": 1.335279277755242,
+      "learning_rate": 1.83214935845786e-05,
+      "loss": 1.3195,
+      "step": 1005
+    },
+    {
+      "epoch": 0.2699091394975949,
+      "grad_norm": 1.3383554327189036,
+      "learning_rate": 1.8295531794273948e-05,
+      "loss": 1.3471,
+      "step": 1010
+    },
+    {
+      "epoch": 0.2712453233564939,
+      "grad_norm": 1.272300682321573,
+      "learning_rate": 1.826938945418837e-05,
+      "loss": 1.3156,
+      "step": 1015
+    },
+    {
+      "epoch": 0.2725815072153928,
+      "grad_norm": 1.4984409833622696,
+      "learning_rate": 1.8243067133302143e-05,
+      "loss": 1.3528,
+      "step": 1020
+    },
+    {
+      "epoch": 0.2739176910742918,
+      "grad_norm": 1.2475449536511878,
+      "learning_rate": 1.8216565404512732e-05,
+      "loss": 1.341,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2752538749331908,
+      "grad_norm": 1.251734878893183,
+      "learning_rate": 1.818988484462238e-05,
+      "loss": 1.3106,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2765900587920898,
+      "grad_norm": 1.322345839521324,
+      "learning_rate": 1.8163026034325532e-05,
+      "loss": 1.294,
+      "step": 1035
+    },
+    {
+      "epoch": 0.27792624265098875,
+      "grad_norm": 1.2838160402093648,
+      "learning_rate": 1.8135989558196207e-05,
+      "loss": 1.3484,
+      "step": 1040
+    },
+    {
+      "epoch": 0.27926242650988775,
+      "grad_norm": 1.2943916155535866,
+      "learning_rate": 1.8108776004675255e-05,
+      "loss": 1.3052,
+      "step": 1045
+    },
+    {
+      "epoch": 0.28059861036878675,
+      "grad_norm": 1.2762154467420836,
+      "learning_rate": 1.808138596605758e-05,
+      "loss": 1.3284,
+      "step": 1050
+    },
+    {
+      "epoch": 0.28193479422768575,
+      "grad_norm": 1.255384319187328,
+      "learning_rate": 1.8053820038479214e-05,
+      "loss": 1.3686,
+      "step": 1055
+    },
+    {
+      "epoch": 0.2832709780865847,
+      "grad_norm": 1.2673273813526882,
+      "learning_rate": 1.802607882190437e-05,
+      "loss": 1.3382,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2846071619454837,
+      "grad_norm": 1.2940248666834164,
+      "learning_rate": 1.799816292011237e-05,
+      "loss": 1.3498,
+      "step": 1065
+    },
+    {
+      "epoch": 0.2859433458043827,
+      "grad_norm": 1.2046683889500032,
+      "learning_rate": 1.7970072940684514e-05,
+      "loss": 1.3329,
+      "step": 1070
+    },
+    {
+      "epoch": 0.2872795296632817,
+      "grad_norm": 1.366921247819638,
+      "learning_rate": 1.7941809494990838e-05,
+      "loss": 1.3152,
+      "step": 1075
+    },
+    {
+      "epoch": 0.28861571352218063,
+      "grad_norm": 1.2494608185941236,
+      "learning_rate": 1.7913373198176832e-05,
+      "loss": 1.3242,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2899518973810796,
+      "grad_norm": 1.2279808317447078,
+      "learning_rate": 1.7884764669150035e-05,
+      "loss": 1.3359,
+      "step": 1085
+    },
+    {
+      "epoch": 0.2912880812399786,
+      "grad_norm": 1.2317749726693008,
+      "learning_rate": 1.7855984530566564e-05,
+      "loss": 1.3186,
+      "step": 1090
+    },
+    {
+      "epoch": 0.2926242650988776,
+      "grad_norm": 1.3570393319531886,
+      "learning_rate": 1.7827033408817573e-05,
+      "loss": 1.3449,
+      "step": 1095
+    },
+    {
+      "epoch": 0.29396044895777657,
+      "grad_norm": 1.247405852772324,
+      "learning_rate": 1.779791193401561e-05,
+      "loss": 1.3416,
+      "step": 1100
+    },
+    {
+      "epoch": 0.29529663281667556,
+      "grad_norm": 1.2893644804363795,
+      "learning_rate": 1.776862073998091e-05,
+      "loss": 1.3674,
+      "step": 1105
+    },
+    {
+      "epoch": 0.29663281667557456,
+      "grad_norm": 1.2741258542191471,
+      "learning_rate": 1.7739160464227593e-05,
+      "loss": 1.3291,
+      "step": 1110
+    },
+    {
+      "epoch": 0.29796900053447356,
+      "grad_norm": 1.4051921967322472,
+      "learning_rate": 1.7709531747949796e-05,
+      "loss": 1.3592,
+      "step": 1115
+    },
+    {
+      "epoch": 0.2993051843933725,
+      "grad_norm": 1.2562313322258833,
+      "learning_rate": 1.7679735236007715e-05,
+      "loss": 1.3259,
+      "step": 1120
+    },
+    {
+      "epoch": 0.3006413682522715,
+      "grad_norm": 1.316830880638165,
+      "learning_rate": 1.7649771576913553e-05,
+      "loss": 1.3448,
+      "step": 1125
+    },
+    {
+      "epoch": 0.3019775521111705,
+      "grad_norm": 1.3518115028546631,
+      "learning_rate": 1.7619641422817446e-05,
+      "loss": 1.3291,
+      "step": 1130
+    },
+    {
+      "epoch": 0.3033137359700695,
+      "grad_norm": 1.3279266806644043,
+      "learning_rate": 1.758934542949323e-05,
+      "loss": 1.3589,
+      "step": 1135
+    },
+    {
+      "epoch": 0.30464991982896844,
+      "grad_norm": 1.2655330414005816,
+      "learning_rate": 1.755888425632418e-05,
+      "loss": 1.3267,
+      "step": 1140
+    },
+    {
+      "epoch": 0.30598610368786744,
+      "grad_norm": 1.2109654835513972,
+      "learning_rate": 1.7528258566288666e-05,
+      "loss": 1.3264,
+      "step": 1145
+    },
+    {
+      "epoch": 0.30732228754676644,
+      "grad_norm": 1.269808597072721,
+      "learning_rate": 1.7497469025945722e-05,
+      "loss": 1.2766,
+      "step": 1150
+    },
+    {
+      "epoch": 0.30865847140566544,
+      "grad_norm": 1.2456464717450948,
+      "learning_rate": 1.7466516305420524e-05,
+      "loss": 1.3352,
+      "step": 1155
+    },
+    {
+      "epoch": 0.3099946552645644,
+      "grad_norm": 1.2359484363944293,
+      "learning_rate": 1.743540107838983e-05,
+      "loss": 1.3295,
+      "step": 1160
+    },
+    {
+      "epoch": 0.3113308391234634,
+      "grad_norm": 1.2244387854880765,
+      "learning_rate": 1.74041240220673e-05,
+      "loss": 1.2754,
+      "step": 1165
+    },
+    {
+      "epoch": 0.3126670229823624,
+      "grad_norm": 1.34164790224033,
+      "learning_rate": 1.7372685817188747e-05,
+      "loss": 1.3066,
+      "step": 1170
+    },
+    {
+      "epoch": 0.3140032068412614,
+      "grad_norm": 1.274512676357063,
+      "learning_rate": 1.734108714799735e-05,
+      "loss": 1.337,
+      "step": 1175
+    },
+    {
+      "epoch": 0.3153393907001603,
+      "grad_norm": 1.1877061220307676,
+      "learning_rate": 1.7309328702228742e-05,
+      "loss": 1.304,
+      "step": 1180
+    },
+    {
+      "epoch": 0.3166755745590593,
+      "grad_norm": 1.307991502348843,
+      "learning_rate": 1.7277411171096042e-05,
+      "loss": 1.3234,
+      "step": 1185
+    },
+    {
+      "epoch": 0.3180117584179583,
+      "grad_norm": 1.2459886266896583,
+      "learning_rate": 1.7245335249274818e-05,
+      "loss": 1.344,
+      "step": 1190
+    },
+    {
+      "epoch": 0.3193479422768573,
+      "grad_norm": 1.3088574044809078,
+      "learning_rate": 1.7213101634887968e-05,
+      "loss": 1.3213,
+      "step": 1195
+    },
+    {
+      "epoch": 0.32068412613575625,
+      "grad_norm": 1.401051770726762,
+      "learning_rate": 1.718071102949051e-05,
+      "loss": 1.3355,
+      "step": 1200
+    },
+    {
+      "epoch": 0.32068412613575625,
+      "eval_loss": 1.3390916585922241,
+      "eval_runtime": 523.6061,
+      "eval_samples_per_second": 25.298,
+      "eval_steps_per_second": 3.163,
+      "step": 1200
+    },
+    {
+      "epoch": 0.32202030999465525,
+      "grad_norm": 1.2972562697211982,
+      "learning_rate": 1.7148164138054333e-05,
+      "loss": 1.3181,
+      "step": 1205
+    },
+    {
+      "epoch": 0.32335649385355425,
+      "grad_norm": 1.2825002290000433,
+      "learning_rate": 1.7115461668952848e-05,
+      "loss": 1.3422,
+      "step": 1210
+    },
+    {
+      "epoch": 0.32469267771245325,
+      "grad_norm": 1.2161188921788924,
+      "learning_rate": 1.7082604333945557e-05,
+      "loss": 1.3505,
+      "step": 1215
+    },
+    {
+      "epoch": 0.3260288615713522,
+      "grad_norm": 1.2075785701291617,
+      "learning_rate": 1.7049592848162583e-05,
+      "loss": 1.331,
+      "step": 1220
+    },
+    {
+      "epoch": 0.3273650454302512,
+      "grad_norm": 1.3030995176122255,
+      "learning_rate": 1.701642793008909e-05,
+      "loss": 1.3266,
+      "step": 1225
+    },
+    {
+      "epoch": 0.3287012292891502,
+      "grad_norm": 1.2590894105748518,
+      "learning_rate": 1.6983110301549652e-05,
+      "loss": 1.3199,
+      "step": 1230
+    },
+    {
+      "epoch": 0.3300374131480492,
+      "grad_norm": 1.2284782881775649,
+      "learning_rate": 1.6949640687692535e-05,
+      "loss": 1.3243,
+      "step": 1235
+    },
+    {
+      "epoch": 0.33137359700694813,
+      "grad_norm": 1.2623721858490347,
+      "learning_rate": 1.691601981697393e-05,
+      "loss": 1.3299,
+      "step": 1240
+    },
+    {
+      "epoch": 0.3327097808658471,
+      "grad_norm": 1.226343404396135,
+      "learning_rate": 1.688224842114208e-05,
+      "loss": 1.3031,
+      "step": 1245
+    },
+    {
+      "epoch": 0.3340459647247461,
+      "grad_norm": 1.2453664354622114,
+      "learning_rate": 1.6848327235221368e-05,
+      "loss": 1.3047,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3353821485836451,
+      "grad_norm": 1.2917212982794357,
+      "learning_rate": 1.681425699749631e-05,
+      "loss": 1.3367,
+      "step": 1255
+    },
+    {
+      "epoch": 0.3367183324425441,
+      "grad_norm": 1.1431548737930348,
+      "learning_rate": 1.6780038449495492e-05,
+      "loss": 1.2894,
+      "step": 1260
+    },
+    {
+      "epoch": 0.33805451630144306,
+      "grad_norm": 1.305490289585266,
+      "learning_rate": 1.674567233597542e-05,
+      "loss": 1.3303,
+      "step": 1265
+    },
+    {
+      "epoch": 0.33939070016034206,
+      "grad_norm": 1.3295423667445367,
+      "learning_rate": 1.6711159404904346e-05,
+      "loss": 1.3795,
+      "step": 1270
+    },
+    {
+      "epoch": 0.34072688401924106,
+      "grad_norm": 1.2557371261898338,
+      "learning_rate": 1.667650040744593e-05,
+      "loss": 1.3923,
+      "step": 1275
+    },
+    {
+      "epoch": 0.34206306787814006,
+      "grad_norm": 1.2709385762973946,
+      "learning_rate": 1.6641696097942937e-05,
+      "loss": 1.3633,
+      "step": 1280
+    },
+    {
+      "epoch": 0.343399251737039,
+      "grad_norm": 1.2297417601415581,
+      "learning_rate": 1.6606747233900816e-05,
+      "loss": 1.3475,
+      "step": 1285
+    },
+    {
+      "epoch": 0.344735435595938,
+      "grad_norm": 1.270772993810448,
+      "learning_rate": 1.6571654575971186e-05,
+      "loss": 1.3215,
+      "step": 1290
+    },
+    {
+      "epoch": 0.346071619454837,
+      "grad_norm": 1.243136160915021,
+      "learning_rate": 1.6536418887935307e-05,
+      "loss": 1.2909,
+      "step": 1295
+    },
+    {
+      "epoch": 0.347407803313736,
+      "grad_norm": 1.2453939190767478,
+      "learning_rate": 1.6501040936687444e-05,
+      "loss": 1.299,
+      "step": 1300
+    },
+    {
+      "epoch": 0.34874398717263494,
+      "grad_norm": 1.3190132616463197,
+      "learning_rate": 1.6465521492218175e-05,
+      "loss": 1.3242,
+      "step": 1305
+    },
+    {
+      "epoch": 0.35008017103153394,
+      "grad_norm": 1.2836129069270563,
+      "learning_rate": 1.6429861327597643e-05,
+      "loss": 1.3257,
+      "step": 1310
+    },
+    {
+      "epoch": 0.35141635489043294,
+      "grad_norm": 1.3046224543331126,
+      "learning_rate": 1.6394061218958714e-05,
+      "loss": 1.3735,
+      "step": 1315
+    },
+    {
+      "epoch": 0.35275253874933193,
+      "grad_norm": 1.195969330334029,
+      "learning_rate": 1.63581219454801e-05,
+      "loss": 1.3547,
+      "step": 1320
+    },
+    {
+      "epoch": 0.3540887226082309,
+      "grad_norm": 1.2293037135379614,
+      "learning_rate": 1.63220442893694e-05,
+      "loss": 1.3258,
+      "step": 1325
+    },
+    {
+      "epoch": 0.3554249064671299,
+      "grad_norm": 1.221432152844297,
+      "learning_rate": 1.6285829035846057e-05,
+      "loss": 1.3368,
+      "step": 1330
+    },
+    {
+      "epoch": 0.3567610903260289,
+      "grad_norm": 1.2802671063330058,
+      "learning_rate": 1.624947697312429e-05,
+      "loss": 1.3497,
+      "step": 1335
+    },
+    {
+      "epoch": 0.35809727418492787,
+      "grad_norm": 1.408250486602362,
+      "learning_rate": 1.621298889239592e-05,
+      "loss": 1.3856,
+      "step": 1340
+    },
+    {
+      "epoch": 0.3594334580438268,
+      "grad_norm": 1.2601768753005949,
+      "learning_rate": 1.617636558781318e-05,
+      "loss": 1.3151,
+      "step": 1345
+    },
+    {
+      "epoch": 0.3607696419027258,
+      "grad_norm": 1.2756859712312687,
+      "learning_rate": 1.6139607856471377e-05,
+      "loss": 1.2932,
+      "step": 1350
+    },
+    {
+      "epoch": 0.3621058257616248,
+      "grad_norm": 1.2546446591863418,
+      "learning_rate": 1.610271649839161e-05,
+      "loss": 1.3112,
+      "step": 1355
+    },
+    {
+      "epoch": 0.3634420096205238,
+      "grad_norm": 1.2883557482491337,
+      "learning_rate": 1.6065692316503306e-05,
+      "loss": 1.3155,
+      "step": 1360
+    },
+    {
+      "epoch": 0.36477819347942275,
+      "grad_norm": 1.2350958249428101,
+      "learning_rate": 1.6028536116626763e-05,
+      "loss": 1.2951,
+      "step": 1365
+    },
+    {
+      "epoch": 0.36611437733832175,
+      "grad_norm": 1.209684203322871,
+      "learning_rate": 1.5991248707455614e-05,
+      "loss": 1.3159,
+      "step": 1370
+    },
+    {
+      "epoch": 0.36745056119722075,
+      "grad_norm": 1.3434534905824467,
+      "learning_rate": 1.595383090053923e-05,
+      "loss": 1.3079,
+      "step": 1375
+    },
+    {
+      "epoch": 0.36878674505611975,
+      "grad_norm": 1.2129683097633361,
+      "learning_rate": 1.5916283510265037e-05,
+      "loss": 1.2949,
+      "step": 1380
+    },
+    {
+      "epoch": 0.3701229289150187,
+      "grad_norm": 1.3085166024177262,
+      "learning_rate": 1.5878607353840814e-05,
+      "loss": 1.3557,
+      "step": 1385
+    },
+    {
+      "epoch": 0.3714591127739177,
+      "grad_norm": 1.2946759032750124,
+      "learning_rate": 1.5840803251276892e-05,
+      "loss": 1.3005,
+      "step": 1390
+    },
+    {
+      "epoch": 0.3727952966328167,
+      "grad_norm": 1.283448494880763,
+      "learning_rate": 1.5802872025368316e-05,
+      "loss": 1.2877,
+      "step": 1395
+    },
+    {
+      "epoch": 0.3741314804917157,
+      "grad_norm": 1.2309029767719335,
+      "learning_rate": 1.576481450167693e-05,
+      "loss": 1.334,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3741314804917157,
+      "eval_loss": 1.330536127090454,
+      "eval_runtime": 523.7724,
+      "eval_samples_per_second": 25.29,
+      "eval_steps_per_second": 3.162,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3754676643506146,
+      "grad_norm": 1.2496837334635877,
+      "learning_rate": 1.5726631508513412e-05,
+      "loss": 1.3089,
+      "step": 1405
+    },
+    {
+      "epoch": 0.3768038482095136,
+      "grad_norm": 1.374132042278566,
+      "learning_rate": 1.568832387691924e-05,
+      "loss": 1.3465,
+      "step": 1410
+    },
+    {
+      "epoch": 0.3781400320684126,
+      "grad_norm": 1.2412100387519944,
+      "learning_rate": 1.5649892440648625e-05,
+      "loss": 1.338,
+      "step": 1415
+    },
+    {
+      "epoch": 0.3794762159273116,
+      "grad_norm": 1.2515878984452657,
+      "learning_rate": 1.5611338036150338e-05,
+      "loss": 1.302,
+      "step": 1420
+    },
+    {
+      "epoch": 0.38081239978621056,
+      "grad_norm": 1.3524756121075554,
+      "learning_rate": 1.5572661502549514e-05,
+      "loss": 1.3297,
+      "step": 1425
+    },
+    {
+      "epoch": 0.38214858364510956,
+      "grad_norm": 1.3613331896664973,
+      "learning_rate": 1.5533863681629404e-05,
+      "loss": 1.3375,
+      "step": 1430
+    },
+    {
+      "epoch": 0.38348476750400856,
+      "grad_norm": 1.3234108123677346,
+      "learning_rate": 1.5494945417813034e-05,
+      "loss": 1.28,
+      "step": 1435
+    },
+    {
+      "epoch": 0.38482095136290756,
+      "grad_norm": 1.2464116209347194,
+      "learning_rate": 1.545590755814483e-05,
+      "loss": 1.339,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3861571352218065,
+      "grad_norm": 1.2082007561781767,
+      "learning_rate": 1.5416750952272198e-05,
+      "loss": 1.2924,
+      "step": 1445
+    },
+    {
+      "epoch": 0.3874933190807055,
+      "grad_norm": 1.2460568233344762,
+      "learning_rate": 1.537747645242701e-05,
+      "loss": 1.2913,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3888295029396045,
+      "grad_norm": 1.2643768624756646,
+      "learning_rate": 1.5338084913407067e-05,
+      "loss": 1.3385,
+      "step": 1455
+    },
+    {
+      "epoch": 0.3901656867985035,
+      "grad_norm": 1.2393822310314082,
+      "learning_rate": 1.5298577192557487e-05,
+      "loss": 1.2918,
+      "step": 1460
+    },
+    {
+      "epoch": 0.39150187065740244,
+      "grad_norm": 1.2212527378031937,
+      "learning_rate": 1.525895414975207e-05,
+      "loss": 1.3496,
+      "step": 1465
+    },
+    {
+      "epoch": 0.39283805451630144,
+      "grad_norm": 1.2539579056942822,
+      "learning_rate": 1.5219216647374546e-05,
+      "loss": 1.3285,
+      "step": 1470
+    },
+    {
+      "epoch": 0.39417423837520044,
+      "grad_norm": 1.194899743310905,
+      "learning_rate": 1.5179365550299823e-05,
+      "loss": 1.2987,
+      "step": 1475
+    },
+    {
+      "epoch": 0.39551042223409943,
+      "grad_norm": 1.2361033901459717,
+      "learning_rate": 1.513940172587518e-05,
+      "loss": 1.3018,
+      "step": 1480
+    },
+    {
+      "epoch": 0.3968466060929984,
+      "grad_norm": 1.2040496820635649,
+      "learning_rate": 1.5099326043901361e-05,
+      "loss": 1.3144,
+      "step": 1485
+    },
+    {
+      "epoch": 0.3981827899518974,
+      "grad_norm": 1.2436110116124757,
+      "learning_rate": 1.5059139376613652e-05,
+      "loss": 1.3147,
+      "step": 1490
+    },
+    {
+      "epoch": 0.3995189738107964,
+      "grad_norm": 1.229222024704698,
+      "learning_rate": 1.5018842598662913e-05,
+      "loss": 1.3084,
+      "step": 1495
+    },
+    {
+      "epoch": 0.40085515766969537,
+      "grad_norm": 1.2280023763221926,
+      "learning_rate": 1.4978436587096526e-05,
+      "loss": 1.2993,
+      "step": 1500
+    },
+    {
+      "epoch": 0.4021913415285943,
+      "grad_norm": 1.2379509698653712,
+      "learning_rate": 1.4937922221339303e-05,
+      "loss": 1.3448,
+      "step": 1505
+    },
+    {
+      "epoch": 0.4035275253874933,
+      "grad_norm": 1.1819300792108958,
+      "learning_rate": 1.4897300383174362e-05,
+      "loss": 1.3093,
+      "step": 1510
+    },
+    {
+      "epoch": 0.4048637092463923,
+      "grad_norm": 1.2646351619539313,
+      "learning_rate": 1.4856571956723924e-05,
+      "loss": 1.3292,
+      "step": 1515
+    },
+    {
+      "epoch": 0.4061998931052913,
+      "grad_norm": 1.20105031681724,
+      "learning_rate": 1.4815737828430068e-05,
+      "loss": 1.3275,
+      "step": 1520
+    },
+    {
+      "epoch": 0.40753607696419025,
+      "grad_norm": 1.2322477247241055,
+      "learning_rate": 1.4774798887035446e-05,
+      "loss": 1.3273,
+      "step": 1525
+    },
+    {
+      "epoch": 0.40887226082308925,
+      "grad_norm": 1.2203562838881301,
+      "learning_rate": 1.4733756023563932e-05,
+      "loss": 1.3046,
+      "step": 1530
+    },
+    {
+      "epoch": 0.41020844468198825,
+      "grad_norm": 1.3701923681759842,
+      "learning_rate": 1.4692610131301242e-05,
+      "loss": 1.3468,
+      "step": 1535
+    },
+    {
+      "epoch": 0.41154462854088725,
+      "grad_norm": 1.2104295209543379,
+      "learning_rate": 1.4651362105775471e-05,
+      "loss": 1.3292,
+      "step": 1540
+    },
+    {
+      "epoch": 0.4128808123997862,
+      "grad_norm": 1.178262832900831,
+      "learning_rate": 1.4610012844737622e-05,
+      "loss": 1.3095,
+      "step": 1545
+    },
+    {
+      "epoch": 0.4142169962586852,
+      "grad_norm": 1.2034684530693827,
+      "learning_rate": 1.4568563248142058e-05,
+      "loss": 1.277,
+      "step": 1550
+    },
+    {
+      "epoch": 0.4155531801175842,
+      "grad_norm": 1.2346570794875227,
+      "learning_rate": 1.4527014218126913e-05,
+      "loss": 1.3323,
+      "step": 1555
+    },
+    {
+      "epoch": 0.4168893639764832,
+      "grad_norm": 1.2924089782717,
+      "learning_rate": 1.4485366658994463e-05,
+      "loss": 1.3117,
+      "step": 1560
+    },
+    {
+      "epoch": 0.4182255478353821,
+      "grad_norm": 1.222906017451485,
+      "learning_rate": 1.4443621477191434e-05,
+      "loss": 1.3185,
+      "step": 1565
+    },
+    {
+      "epoch": 0.4195617316942811,
+      "grad_norm": 1.2534027609567953,
+      "learning_rate": 1.440177958128929e-05,
+      "loss": 1.2978,
+      "step": 1570
+    },
+    {
+      "epoch": 0.4208979155531801,
+      "grad_norm": 1.2140810113206562,
+      "learning_rate": 1.4359841881964445e-05,
+      "loss": 1.3162,
+      "step": 1575
+    },
+    {
+      "epoch": 0.4222340994120791,
+      "grad_norm": 1.1684011954978797,
+      "learning_rate": 1.4317809291978442e-05,
+      "loss": 1.3333,
+      "step": 1580
+    },
+    {
+      "epoch": 0.42357028327097807,
+      "grad_norm": 1.214319379461352,
+      "learning_rate": 1.4275682726158092e-05,
+      "loss": 1.2867,
+      "step": 1585
+    },
+    {
+      "epoch": 0.42490646712987706,
+      "grad_norm": 1.182224144852539,
+      "learning_rate": 1.4233463101375569e-05,
+      "loss": 1.3039,
+      "step": 1590
+    },
+    {
+      "epoch": 0.42624265098877606,
+      "grad_norm": 1.2314041097519686,
+      "learning_rate": 1.4191151336528441e-05,
+      "loss": 1.3488,
+      "step": 1595
+    },
+    {
+      "epoch": 0.42757883484767506,
+      "grad_norm": 1.2641228680933205,
+      "learning_rate": 1.4148748352519677e-05,
+      "loss": 1.3183,
+      "step": 1600
+    },
+    {
+      "epoch": 0.42757883484767506,
+      "eval_loss": 1.3232654333114624,
+      "eval_runtime": 523.7168,
+      "eval_samples_per_second": 25.292,
+      "eval_steps_per_second": 3.162,
+      "step": 1600
+    },
+    {
+      "epoch": 0.428915018706574,
+      "grad_norm": 1.2410818763020894,
+      "learning_rate": 1.4106255072237605e-05,
+      "loss": 1.2977,
+      "step": 1605
+    },
+    {
+      "epoch": 0.430251202565473,
+      "grad_norm": 1.2216210579429683,
+      "learning_rate": 1.406367242053583e-05,
+      "loss": 1.3388,
+      "step": 1610
+    },
+    {
+      "epoch": 0.431587386424372,
+      "grad_norm": 1.2380524456449333,
+      "learning_rate": 1.402100132421309e-05,
+      "loss": 1.3048,
+      "step": 1615
+    },
+    {
+      "epoch": 0.432923570283271,
+      "grad_norm": 1.222444425938849,
+      "learning_rate": 1.39782427119931e-05,
+      "loss": 1.3129,
+      "step": 1620
+    },
+    {
+      "epoch": 0.43425975414216994,
+      "grad_norm": 1.2084516798428055,
+      "learning_rate": 1.3935397514504332e-05,
+      "loss": 1.3312,
+      "step": 1625
+    },
+    {
+      "epoch": 0.43559593800106894,
+      "grad_norm": 1.3138679343999906,
+      "learning_rate": 1.3892466664259756e-05,
+      "loss": 1.3358,
+      "step": 1630
+    },
+    {
+      "epoch": 0.43693212185996794,
+      "grad_norm": 1.325108022318488,
+      "learning_rate": 1.3849451095636555e-05,
+      "loss": 1.3197,
+      "step": 1635
+    },
+    {
+      "epoch": 0.43826830571886694,
+      "grad_norm": 1.239099901058549,
+      "learning_rate": 1.3806351744855781e-05,
+      "loss": 1.3276,
+      "step": 1640
+    },
+    {
+      "epoch": 0.4396044895777659,
+      "grad_norm": 1.2701728016347087,
+      "learning_rate": 1.3763169549961976e-05,
+      "loss": 1.3087,
+      "step": 1645
+    },
+    {
+      "epoch": 0.4409406734366649,
+      "grad_norm": 1.3093633737758703,
+      "learning_rate": 1.371990545080276e-05,
+      "loss": 1.3334,
+      "step": 1650
+    },
+    {
+      "epoch": 0.4422768572955639,
+      "grad_norm": 1.2721317308394575,
+      "learning_rate": 1.3676560389008378e-05,
+      "loss": 1.3568,
+      "step": 1655
+    },
+    {
+      "epoch": 0.4436130411544629,
+      "grad_norm": 1.2184083694463035,
+      "learning_rate": 1.3633135307971204e-05,
+      "loss": 1.3279,
+      "step": 1660
+    },
+    {
+      "epoch": 0.4449492250133618,
+      "grad_norm": 1.3001651314508893,
+      "learning_rate": 1.3589631152825197e-05,
+      "loss": 1.3176,
+      "step": 1665
+    },
+    {
+      "epoch": 0.4462854088722608,
+      "grad_norm": 1.2226816442485073,
+      "learning_rate": 1.3546048870425356e-05,
+      "loss": 1.2705,
+      "step": 1670
+    },
+    {
+      "epoch": 0.4476215927311598,
+      "grad_norm": 1.4176804415374429,
+      "learning_rate": 1.3502389409327087e-05,
+      "loss": 1.2679,
+      "step": 1675
+    },
+    {
+      "epoch": 0.4489577765900588,
+      "grad_norm": 1.2388040805053258,
+      "learning_rate": 1.3458653719765564e-05,
+      "loss": 1.3046,
+      "step": 1680
+    },
+    {
+      "epoch": 0.45029396044895775,
+      "grad_norm": 1.2129122448318554,
+      "learning_rate": 1.341484275363506e-05,
+      "loss": 1.3149,
+      "step": 1685
+    },
+    {
+      "epoch": 0.45163014430785675,
+      "grad_norm": 1.244763644565545,
+      "learning_rate": 1.3370957464468213e-05,
+      "loss": 1.3436,
+      "step": 1690
+    },
+    {
+      "epoch": 0.45296632816675575,
+      "grad_norm": 1.2759739357160127,
+      "learning_rate": 1.332699880741528e-05,
+      "loss": 1.3569,
+      "step": 1695
+    },
+    {
+      "epoch": 0.45430251202565475,
+      "grad_norm": 1.2281614718249263,
+      "learning_rate": 1.3282967739223357e-05,
+      "loss": 1.3075,
+      "step": 1700
+    },
+    {
+      "epoch": 0.4556386958845537,
+      "grad_norm": 1.2189899125191879,
+      "learning_rate": 1.3238865218215535e-05,
+      "loss": 1.3017,
+      "step": 1705
+    },
+    {
+      "epoch": 0.4569748797434527,
+      "grad_norm": 1.231210053539351,
+      "learning_rate": 1.3194692204270063e-05,
+      "loss": 1.3366,
+      "step": 1710
+    },
+    {
+      "epoch": 0.4583110636023517,
+      "grad_norm": 1.2018193616808162,
+      "learning_rate": 1.3150449658799442e-05,
+      "loss": 1.3535,
+      "step": 1715
+    },
+    {
+      "epoch": 0.4596472474612507,
+      "grad_norm": 1.2128311963623164,
+      "learning_rate": 1.3106138544729511e-05,
+      "loss": 1.3495,
+      "step": 1720
+    },
+    {
+      "epoch": 0.46098343132014963,
+      "grad_norm": 1.2304855721971648,
+      "learning_rate": 1.3061759826478477e-05,
+      "loss": 1.308,
+      "step": 1725
+    },
+    {
+      "epoch": 0.4623196151790486,
+      "grad_norm": 1.183612891624708,
+      "learning_rate": 1.3017314469935942e-05,
+      "loss": 1.2771,
+      "step": 1730
+    },
+    {
+      "epoch": 0.4636557990379476,
+      "grad_norm": 1.2035215718306922,
+      "learning_rate": 1.2972803442441863e-05,
+      "loss": 1.3374,
+      "step": 1735
+    },
+    {
+      "epoch": 0.4649919828968466,
+      "grad_norm": 1.2078408808291714,
+      "learning_rate": 1.2928227712765504e-05,
+      "loss": 1.3155,
+      "step": 1740
+    },
+    {
+      "epoch": 0.46632816675574557,
+      "grad_norm": 1.2685296349837316,
+      "learning_rate": 1.2883588251084362e-05,
+      "loss": 1.3224,
+      "step": 1745
+    },
+    {
+      "epoch": 0.46766435061464456,
+      "grad_norm": 1.2298326258192636,
+      "learning_rate": 1.2838886028963038e-05,
+      "loss": 1.3332,
+      "step": 1750
+    },
+    {
+      "epoch": 0.46900053447354356,
+      "grad_norm": 1.1803857708163348,
+      "learning_rate": 1.2794122019332087e-05,
+      "loss": 1.2889,
+      "step": 1755
+    },
+    {
+      "epoch": 0.47033671833244256,
+      "grad_norm": 1.274081083989054,
+      "learning_rate": 1.2749297196466861e-05,
+      "loss": 1.289,
+      "step": 1760
+    },
+    {
+      "epoch": 0.4716729021913415,
+      "grad_norm": 1.3238957747747422,
+      "learning_rate": 1.270441253596629e-05,
+      "loss": 1.3189,
+      "step": 1765
+    },
+    {
+      "epoch": 0.4730090860502405,
+      "grad_norm": 1.2097611285289955,
+      "learning_rate": 1.265946901473166e-05,
+      "loss": 1.3071,
+      "step": 1770
+    },
+    {
+      "epoch": 0.4743452699091395,
+      "grad_norm": 1.1625021555183794,
+      "learning_rate": 1.2614467610945323e-05,
+      "loss": 1.2987,
+      "step": 1775
+    },
+    {
+      "epoch": 0.4756814537680385,
+      "grad_norm": 1.1887076661153424,
+      "learning_rate": 1.256940930404945e-05,
+      "loss": 1.2783,
+      "step": 1780
+    },
+    {
+      "epoch": 0.47701763762693744,
+      "grad_norm": 1.2645632943006766,
+      "learning_rate": 1.2524295074724683e-05,
+      "loss": 1.3448,
+      "step": 1785
+    },
+    {
+      "epoch": 0.47835382148583644,
+      "grad_norm": 1.2361406321440211,
+      "learning_rate": 1.2479125904868795e-05,
+      "loss": 1.2726,
+      "step": 1790
+    },
+    {
+      "epoch": 0.47969000534473544,
+      "grad_norm": 1.274165207428913,
+      "learning_rate": 1.2433902777575326e-05,
+      "loss": 1.2907,
+      "step": 1795
+    },
+    {
+      "epoch": 0.48102618920363444,
+      "grad_norm": 1.2418881744061736,
+      "learning_rate": 1.2388626677112185e-05,
+      "loss": 1.334,
+      "step": 1800
+    },
+    {
+      "epoch": 0.48102618920363444,
+      "eval_loss": 1.316095232963562,
+      "eval_runtime": 523.7124,
+      "eval_samples_per_second": 25.293,
+      "eval_steps_per_second": 3.162,
+      "step": 1800
+    },
+    {
+      "epoch": 0.4823623730625334,
+      "grad_norm": 1.4406988030285752,
+      "learning_rate": 1.2343298588900226e-05,
+      "loss": 1.3175,
+      "step": 1805
+    },
+    {
+      "epoch": 0.4836985569214324,
+      "grad_norm": 1.1708167122953599,
+      "learning_rate": 1.2297919499491797e-05,
+      "loss": 1.3048,
+      "step": 1810
+    },
+    {
+      "epoch": 0.4850347407803314,
+      "grad_norm": 1.2437310171766487,
+      "learning_rate": 1.2252490396549282e-05,
+      "loss": 1.3287,
+      "step": 1815
+    },
+    {
+      "epoch": 0.4863709246392304,
+      "grad_norm": 1.5815332852740502,
+      "learning_rate": 1.220701226882358e-05,
+      "loss": 1.3344,
+      "step": 1820
+    },
+    {
+      "epoch": 0.4877071084981293,
+      "grad_norm": 1.2057093168260966,
+      "learning_rate": 1.2161486106132612e-05,
+      "loss": 1.3623,
+      "step": 1825
+    },
+    {
+      "epoch": 0.4890432923570283,
+      "grad_norm": 1.1494411879995097,
+      "learning_rate": 1.2115912899339757e-05,
+      "loss": 1.283,
+      "step": 1830
+    },
+    {
+      "epoch": 0.4903794762159273,
+      "grad_norm": 1.1721905423708292,
+      "learning_rate": 1.2070293640332306e-05,
+      "loss": 1.3124,
+      "step": 1835
+    },
+    {
+      "epoch": 0.4917156600748263,
+      "grad_norm": 1.165222069970401,
+      "learning_rate": 1.202462932199985e-05,
+      "loss": 1.3094,
+      "step": 1840
+    },
+    {
+      "epoch": 0.49305184393372525,
+      "grad_norm": 1.250148064441355,
+      "learning_rate": 1.1978920938212691e-05,
+      "loss": 1.3363,
+      "step": 1845
+    },
+    {
+      "epoch": 0.49438802779262425,
+      "grad_norm": 1.1969490785809054,
+      "learning_rate": 1.1933169483800203e-05,
+      "loss": 1.3057,
+      "step": 1850
+    },
+    {
+      "epoch": 0.49572421165152325,
+      "grad_norm": 1.2469012012408995,
+      "learning_rate": 1.1887375954529167e-05,
+      "loss": 1.3053,
+      "step": 1855
+    },
+    {
+      "epoch": 0.49706039551042225,
+      "grad_norm": 1.1816859874185814,
+      "learning_rate": 1.1841541347082134e-05,
+      "loss": 1.2982,
+      "step": 1860
+    },
+    {
+      "epoch": 0.4983965793693212,
+      "grad_norm": 1.258761404972035,
+      "learning_rate": 1.1795666659035682e-05,
+      "loss": 1.2963,
+      "step": 1865
+    },
+    {
+      "epoch": 0.4997327632282202,
+      "grad_norm": 1.2450588185319733,
+      "learning_rate": 1.1749752888838754e-05,
+      "loss": 1.3272,
+      "step": 1870
+    },
+    {
+      "epoch": 0.5010689470871192,
+      "grad_norm": 1.2052446498628289,
+      "learning_rate": 1.1703801035790897e-05,
+      "loss": 1.2946,
+      "step": 1875
+    },
+    {
+      "epoch": 0.5024051309460181,
+      "grad_norm": 1.2142678023982578,
+      "learning_rate": 1.1657812100020507e-05,
+      "loss": 1.302,
+      "step": 1880
+    },
+    {
+      "epoch": 0.5037413148049171,
+      "grad_norm": 1.2721562041773744,
+      "learning_rate": 1.1611787082463095e-05,
+      "loss": 1.3236,
+      "step": 1885
+    },
+    {
+      "epoch": 0.5050774986638161,
+      "grad_norm": 1.3019577911767723,
+      "learning_rate": 1.1565726984839471e-05,
+      "loss": 1.3,
+      "step": 1890
+    },
+    {
+      "epoch": 0.5064136825227151,
+      "grad_norm": 1.2326550134057832,
+      "learning_rate": 1.1519632809633954e-05,
+      "loss": 1.3449,
+      "step": 1895
+    },
+    {
+      "epoch": 0.5077498663816141,
+      "grad_norm": 1.2108121941048178,
+      "learning_rate": 1.1473505560072543e-05,
+      "loss": 1.3003,
+      "step": 1900
+    },
+    {
+      "epoch": 0.5090860502405131,
+      "grad_norm": 1.2351796577016483,
+      "learning_rate": 1.1427346240101105e-05,
+      "loss": 1.294,
+      "step": 1905
+    },
+    {
+      "epoch": 0.5104222340994121,
+      "grad_norm": 1.210376427103751,
+      "learning_rate": 1.1381155854363503e-05,
+      "loss": 1.269,
+      "step": 1910
+    },
+    {
+      "epoch": 0.5117584179583111,
+      "grad_norm": 1.1902823799298627,
+      "learning_rate": 1.1334935408179736e-05,
+      "loss": 1.2871,
+      "step": 1915
+    },
+    {
+      "epoch": 0.51309460181721,
+      "grad_norm": 1.2389975794814998,
+      "learning_rate": 1.1288685907524057e-05,
+      "loss": 1.3092,
+      "step": 1920
+    },
+    {
+      "epoch": 0.514430785676109,
+      "grad_norm": 1.228008698129948,
+      "learning_rate": 1.1242408359003091e-05,
+      "loss": 1.287,
+      "step": 1925
+    },
+    {
+      "epoch": 0.515766969535008,
+      "grad_norm": 1.1816697296808294,
+      "learning_rate": 1.1196103769833908e-05,
+      "loss": 1.3128,
+      "step": 1930
+    },
+    {
+      "epoch": 0.517103153393907,
+      "grad_norm": 1.1835653007689506,
+      "learning_rate": 1.1149773147822112e-05,
+      "loss": 1.2925,
+      "step": 1935
+    },
+    {
+      "epoch": 0.518439337252806,
+      "grad_norm": 1.2380413372980732,
+      "learning_rate": 1.1103417501339903e-05,
+      "loss": 1.3405,
+      "step": 1940
+    },
+    {
+      "epoch": 0.519775521111705,
+      "grad_norm": 1.2784541489247507,
+      "learning_rate": 1.1057037839304135e-05,
+      "loss": 1.2585,
+      "step": 1945
+    },
+    {
+      "epoch": 0.521111704970604,
+      "grad_norm": 1.2402864740991135,
+      "learning_rate": 1.1010635171154342e-05,
+      "loss": 1.2867,
+      "step": 1950
+    },
+    {
+      "epoch": 0.522447888829503,
+      "grad_norm": 1.2346852083660054,
+      "learning_rate": 1.0964210506830802e-05,
+      "loss": 1.2781,
+      "step": 1955
+    },
+    {
+      "epoch": 0.5237840726884019,
+      "grad_norm": 1.2195618023918173,
+      "learning_rate": 1.0917764856752509e-05,
+      "loss": 1.289,
+      "step": 1960
+    },
+    {
+      "epoch": 0.5251202565473009,
+      "grad_norm": 1.1817447386492328,
+      "learning_rate": 1.0871299231795214e-05,
+      "loss": 1.2883,
+      "step": 1965
+    },
+    {
+      "epoch": 0.5264564404061999,
+      "grad_norm": 1.1727533453851813,
+      "learning_rate": 1.082481464326942e-05,
+      "loss": 1.32,
+      "step": 1970
+    },
+    {
+      "epoch": 0.5277926242650989,
+      "grad_norm": 1.181864666573128,
+      "learning_rate": 1.077831210289837e-05,
+      "loss": 1.3068,
+      "step": 1975
+    },
+    {
+      "epoch": 0.5291288081239979,
+      "grad_norm": 1.2249148804887326,
+      "learning_rate": 1.0731792622796007e-05,
+      "loss": 1.3044,
+      "step": 1980
+    },
+    {
+      "epoch": 0.5304649919828969,
+      "grad_norm": 1.2263040268107137,
+      "learning_rate": 1.0685257215444975e-05,
+      "loss": 1.3002,
+      "step": 1985
+    },
+    {
+      "epoch": 0.5318011758417959,
+      "grad_norm": 1.2237762133974603,
+      "learning_rate": 1.0638706893674571e-05,
+      "loss": 1.3138,
+      "step": 1990
+    },
+    {
+      "epoch": 0.5331373597006949,
+      "grad_norm": 1.324928279536264,
+      "learning_rate": 1.059214267063869e-05,
+      "loss": 1.3364,
+      "step": 1995
+    },
+    {
+      "epoch": 0.5344735435595938,
+      "grad_norm": 1.2493747053579303,
+      "learning_rate": 1.0545565559793796e-05,
+      "loss": 1.3013,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5344735435595938,
+      "eval_loss": 1.3086645603179932,
+      "eval_runtime": 523.908,
+      "eval_samples_per_second": 25.283,
+      "eval_steps_per_second": 3.161,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5358097274184928,
+      "grad_norm": 1.2142470124401028,
+      "learning_rate": 1.0498976574876838e-05,
+      "loss": 1.2563,
+      "step": 2005
+    },
+    {
+      "epoch": 0.5371459112773918,
+      "grad_norm": 1.1948400261490024,
+      "learning_rate": 1.0452376729883216e-05,
+      "loss": 1.3066,
+      "step": 2010
+    },
+    {
+      "epoch": 0.5384820951362908,
+      "grad_norm": 1.206399971819588,
+      "learning_rate": 1.0405767039044688e-05,
+      "loss": 1.2928,
+      "step": 2015
+    },
+    {
+      "epoch": 0.5398182789951897,
+      "grad_norm": 1.2064545160074232,
+      "learning_rate": 1.0359148516807302e-05,
+      "loss": 1.3377,
+      "step": 2020
+    },
+    {
+      "epoch": 0.5411544628540887,
+      "grad_norm": 1.2352197196893993,
+      "learning_rate": 1.0312522177809326e-05,
+      "loss": 1.32,
+      "step": 2025
+    },
+    {
+      "epoch": 0.5424906467129877,
+      "grad_norm": 1.1837145257724762,
+      "learning_rate": 1.0265889036859146e-05,
+      "loss": 1.2636,
+      "step": 2030
+    },
+    {
+      "epoch": 0.5438268305718867,
+      "grad_norm": 1.2342337040554485,
+      "learning_rate": 1.0219250108913206e-05,
+      "loss": 1.2928,
+      "step": 2035
+    },
+    {
+      "epoch": 0.5451630144307856,
+      "grad_norm": 1.2371903390877288,
+      "learning_rate": 1.0172606409053887e-05,
+      "loss": 1.3451,
+      "step": 2040
+    },
+    {
+      "epoch": 0.5464991982896846,
+      "grad_norm": 1.2027906126810077,
+      "learning_rate": 1.0125958952467439e-05,
+      "loss": 1.3274,
+      "step": 2045
+    },
+    {
+      "epoch": 0.5478353821485836,
+      "grad_norm": 1.176177465912769,
+      "learning_rate": 1.0079308754421877e-05,
+      "loss": 1.2996,
+      "step": 2050
+    },
+    {
+      "epoch": 0.5491715660074826,
+      "grad_norm": 1.2416193442138757,
+      "learning_rate": 1.003265683024487e-05,
+      "loss": 1.2798,
+      "step": 2055
+    },
+    {
+      "epoch": 0.5505077498663816,
+      "grad_norm": 1.2583071246909772,
+      "learning_rate": 9.986004195301679e-06,
+      "loss": 1.2727,
+      "step": 2060
+    },
+    {
+      "epoch": 0.5518439337252806,
+      "grad_norm": 1.1935438605799258,
+      "learning_rate": 9.939351864973006e-06,
+      "loss": 1.256,
+      "step": 2065
+    },
+    {
+      "epoch": 0.5531801175841796,
+      "grad_norm": 1.1633338976111967,
+      "learning_rate": 9.892700854632945e-06,
+      "loss": 1.2946,
+      "step": 2070
+    },
+    {
+      "epoch": 0.5545163014430786,
+      "grad_norm": 1.2341873796862335,
+      "learning_rate": 9.846052179626854e-06,
+      "loss": 1.2695,
+      "step": 2075
+    },
+    {
+      "epoch": 0.5558524853019775,
+      "grad_norm": 1.2110901028037866,
+      "learning_rate": 9.799406855249261e-06,
+      "loss": 1.2939,
+      "step": 2080
+    },
+    {
+      "epoch": 0.5571886691608765,
+      "grad_norm": 1.2285980383554018,
+      "learning_rate": 9.75276589672177e-06,
+      "loss": 1.2646,
+      "step": 2085
+    },
+    {
+      "epoch": 0.5585248530197755,
+      "grad_norm": 1.2442431465515202,
+      "learning_rate": 9.706130319170968e-06,
+      "loss": 1.284,
+      "step": 2090
+    },
+    {
+      "epoch": 0.5598610368786745,
+      "grad_norm": 1.2387684585320866,
+      "learning_rate": 9.659501137606317e-06,
+      "loss": 1.245,
+      "step": 2095
+    },
+    {
+      "epoch": 0.5611972207375735,
+      "grad_norm": 1.1969981629519848,
+      "learning_rate": 9.61287936689808e-06,
+      "loss": 1.3233,
+      "step": 2100
+    },
+    {
+      "epoch": 0.5625334045964725,
+      "grad_norm": 1.2647984251708813,
+      "learning_rate": 9.566266021755224e-06,
+      "loss": 1.3229,
+      "step": 2105
+    },
+    {
+      "epoch": 0.5638695884553715,
+      "grad_norm": 1.211725302821127,
+      "learning_rate": 9.519662116703333e-06,
+      "loss": 1.2725,
+      "step": 2110
+    },
+    {
+      "epoch": 0.5652057723142705,
+      "grad_norm": 1.1879063545622028,
+      "learning_rate": 9.473068666062535e-06,
+      "loss": 1.2781,
+      "step": 2115
+    },
+    {
+      "epoch": 0.5665419561731694,
+      "grad_norm": 1.3276654575701659,
+      "learning_rate": 9.426486683925412e-06,
+      "loss": 1.3065,
+      "step": 2120
+    },
+    {
+      "epoch": 0.5678781400320684,
+      "grad_norm": 1.237229888093536,
+      "learning_rate": 9.379917184134949e-06,
+      "loss": 1.2736,
+      "step": 2125
+    },
+    {
+      "epoch": 0.5692143238909674,
+      "grad_norm": 1.1968148529216311,
+      "learning_rate": 9.33336118026245e-06,
+      "loss": 1.3061,
+      "step": 2130
+    },
+    {
+      "epoch": 0.5705505077498664,
+      "grad_norm": 1.178529508943571,
+      "learning_rate": 9.286819685585482e-06,
+      "loss": 1.285,
+      "step": 2135
+    },
+    {
+      "epoch": 0.5718866916087654,
+      "grad_norm": 1.1976436960246757,
+      "learning_rate": 9.240293713065826e-06,
+      "loss": 1.3261,
+      "step": 2140
+    },
+    {
+      "epoch": 0.5732228754676644,
+      "grad_norm": 1.293767414098443,
+      "learning_rate": 9.193784275327439e-06,
+      "loss": 1.3081,
+      "step": 2145
+    },
+    {
+      "epoch": 0.5745590593265634,
+      "grad_norm": 1.1715456586189346,
+      "learning_rate": 9.14729238463438e-06,
+      "loss": 1.2754,
+      "step": 2150
+    },
+    {
+      "epoch": 0.5758952431854624,
+      "grad_norm": 1.2266737106027843,
+      "learning_rate": 9.100819052868818e-06,
+      "loss": 1.3039,
+      "step": 2155
+    },
+    {
+      "epoch": 0.5772314270443613,
+      "grad_norm": 1.2249867841645912,
+      "learning_rate": 9.054365291508998e-06,
+      "loss": 1.2862,
+      "step": 2160
+    },
+    {
+      "epoch": 0.5785676109032603,
+      "grad_norm": 1.2137804099395924,
+      "learning_rate": 9.007932111607202e-06,
+      "loss": 1.3038,
+      "step": 2165
+    },
+    {
+      "epoch": 0.5799037947621593,
+      "grad_norm": 1.1366732855918937,
+      "learning_rate": 8.961520523767777e-06,
+      "loss": 1.2751,
+      "step": 2170
+    },
+    {
+      "epoch": 0.5812399786210583,
+      "grad_norm": 1.2241697536325484,
+      "learning_rate": 8.915131538125124e-06,
+      "loss": 1.3154,
+      "step": 2175
+    },
+    {
+      "epoch": 0.5825761624799572,
+      "grad_norm": 1.1917760108841098,
+      "learning_rate": 8.868766164321704e-06,
+      "loss": 1.3126,
+      "step": 2180
+    },
+    {
+      "epoch": 0.5839123463388562,
+      "grad_norm": 1.1679731332413519,
+      "learning_rate": 8.822425411486087e-06,
+      "loss": 1.2923,
+      "step": 2185
+    },
+    {
+      "epoch": 0.5852485301977552,
+      "grad_norm": 1.160012615589374,
+      "learning_rate": 8.776110288210964e-06,
+      "loss": 1.295,
+      "step": 2190
+    },
+    {
+      "epoch": 0.5865847140566542,
+      "grad_norm": 1.2221113014345002,
+      "learning_rate": 8.729821802531213e-06,
+      "loss": 1.2981,
+      "step": 2195
+    },
+    {
+      "epoch": 0.5879208979155531,
+      "grad_norm": 1.1659553272788548,
+      "learning_rate": 8.683560961901952e-06,
+      "loss": 1.3156,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5879208979155531,
+      "eval_loss": 1.301594853401184,
+      "eval_runtime": 526.3671,
+      "eval_samples_per_second": 25.165,
+      "eval_steps_per_second": 3.146,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5892570817744521,
+      "grad_norm": 1.171579385550598,
+      "learning_rate": 8.637328773176605e-06,
+      "loss": 1.3301,
+      "step": 2205
+    },
+    {
+      "epoch": 0.5905932656333511,
+      "grad_norm": 1.1943778685731268,
+      "learning_rate": 8.591126242585007e-06,
+      "loss": 1.3187,
+      "step": 2210
+    },
+    {
+      "epoch": 0.5919294494922501,
+      "grad_norm": 1.1944105035087198,
+      "learning_rate": 8.544954375711487e-06,
+      "loss": 1.2836,
+      "step": 2215
+    },
+    {
+      "epoch": 0.5932656333511491,
+      "grad_norm": 1.2318242169174625,
+      "learning_rate": 8.498814177472987e-06,
+      "loss": 1.3131,
+      "step": 2220
+    },
+    {
+      "epoch": 0.5946018172100481,
+      "grad_norm": 1.2254563935614524,
+      "learning_rate": 8.452706652097187e-06,
+      "loss": 1.2875,
+      "step": 2225
+    },
+    {
+      "epoch": 0.5959380010689471,
+      "grad_norm": 1.1704675679433372,
+      "learning_rate": 8.406632803100665e-06,
+      "loss": 1.2853,
+      "step": 2230
+    },
+    {
+      "epoch": 0.5972741849278461,
+      "grad_norm": 1.1396029283635678,
+      "learning_rate": 8.360593633267024e-06,
+      "loss": 1.29,
+      "step": 2235
+    },
+    {
+      "epoch": 0.598610368786745,
+      "grad_norm": 1.1872398900997034,
+      "learning_rate": 8.314590144625102e-06,
+      "loss": 1.3049,
+      "step": 2240
+    },
+    {
+      "epoch": 0.599946552645644,
+      "grad_norm": 1.182289479761897,
+      "learning_rate": 8.268623338427139e-06,
+      "loss": 1.2826,
+      "step": 2245
+    },
+    {
+      "epoch": 0.601282736504543,
+      "grad_norm": 1.202378486908649,
+      "learning_rate": 8.222694215126985e-06,
+      "loss": 1.2919,
+      "step": 2250
+    },
+    {
+      "epoch": 0.602618920363442,
+      "grad_norm": 1.1561336328785552,
+      "learning_rate": 8.176803774358356e-06,
+      "loss": 1.2615,
+      "step": 2255
+    },
+    {
+      "epoch": 0.603955104222341,
+      "grad_norm": 1.2171931903507043,
+      "learning_rate": 8.130953014913025e-06,
+      "loss": 1.2887,
+      "step": 2260
+    },
+    {
+      "epoch": 0.60529128808124,
+      "grad_norm": 1.1818497689115386,
+      "learning_rate": 8.085142934719131e-06,
+      "loss": 1.2627,
+      "step": 2265
+    },
+    {
+      "epoch": 0.606627471940139,
+      "grad_norm": 1.2158971480650371,
+      "learning_rate": 8.039374530819444e-06,
+      "loss": 1.2829,
+      "step": 2270
+    },
+    {
+      "epoch": 0.607963655799038,
+      "grad_norm": 1.275466968098271,
+      "learning_rate": 7.993648799349646e-06,
+      "loss": 1.3204,
+      "step": 2275
+    },
+    {
+      "epoch": 0.6092998396579369,
+      "grad_norm": 1.2170453783162276,
+      "learning_rate": 7.947966735516682e-06,
+      "loss": 1.3045,
+      "step": 2280
+    },
+    {
+      "epoch": 0.6106360235168359,
+      "grad_norm": 1.204876466973939,
+      "learning_rate": 7.902329333577072e-06,
+      "loss": 1.285,
+      "step": 2285
+    },
+    {
+      "epoch": 0.6119722073757349,
+      "grad_norm": 1.2259462458581631,
+      "learning_rate": 7.856737586815292e-06,
+      "loss": 1.2937,
+      "step": 2290
+    },
+    {
+      "epoch": 0.6133083912346339,
+      "grad_norm": 1.2108988682821065,
+      "learning_rate": 7.811192487522141e-06,
+      "loss": 1.2896,
+      "step": 2295
+    },
+    {
+      "epoch": 0.6146445750935329,
+      "grad_norm": 1.1922326929306413,
+      "learning_rate": 7.765695026973155e-06,
+      "loss": 1.2888,
+      "step": 2300
+    },
+    {
+      "epoch": 0.6159807589524319,
+      "grad_norm": 1.2082232502962738,
+      "learning_rate": 7.72024619540702e-06,
+      "loss": 1.3027,
+      "step": 2305
+    },
+    {
+      "epoch": 0.6173169428113309,
+      "grad_norm": 1.1935181964760695,
+      "learning_rate": 7.674846982004033e-06,
+      "loss": 1.2726,
+      "step": 2310
+    },
+    {
+      "epoch": 0.6186531266702299,
+      "grad_norm": 1.1988170924470665,
+      "learning_rate": 7.62949837486456e-06,
+      "loss": 1.3105,
+      "step": 2315
+    },
+    {
+      "epoch": 0.6199893105291288,
+      "grad_norm": 1.1788468720871008,
+      "learning_rate": 7.584201360987544e-06,
+      "loss": 1.2962,
+      "step": 2320
+    },
+    {
+      "epoch": 0.6213254943880278,
+      "grad_norm": 1.2164638957688612,
+      "learning_rate": 7.538956926249013e-06,
+      "loss": 1.2918,
+      "step": 2325
+    },
+    {
+      "epoch": 0.6226616782469268,
+      "grad_norm": 1.2540807555588527,
+      "learning_rate": 7.493766055380622e-06,
+      "loss": 1.3032,
+      "step": 2330
+    },
+    {
+      "epoch": 0.6239978621058258,
+      "grad_norm": 1.2021644980450277,
+      "learning_rate": 7.44862973194823e-06,
+      "loss": 1.2784,
+      "step": 2335
+    },
+    {
+      "epoch": 0.6253340459647247,
+      "grad_norm": 1.1374900868431708,
+      "learning_rate": 7.403548938330487e-06,
+      "loss": 1.2615,
+      "step": 2340
+    },
+    {
+      "epoch": 0.6266702298236237,
+      "grad_norm": 1.223926027057831,
+      "learning_rate": 7.358524655697445e-06,
+      "loss": 1.3324,
+      "step": 2345
+    },
+    {
+      "epoch": 0.6280064136825227,
+      "grad_norm": 1.1663009699556086,
+      "learning_rate": 7.31355786398922e-06,
+      "loss": 1.3073,
+      "step": 2350
+    },
+    {
+      "epoch": 0.6293425975414217,
+      "grad_norm": 1.1605825631923092,
+      "learning_rate": 7.268649541894658e-06,
+      "loss": 1.3167,
+      "step": 2355
+    },
+    {
+      "epoch": 0.6306787814003206,
+      "grad_norm": 1.181802501051011,
+      "learning_rate": 7.223800666830013e-06,
+      "loss": 1.272,
+      "step": 2360
+    },
+    {
+      "epoch": 0.6320149652592196,
+      "grad_norm": 1.2124656825123712,
+      "learning_rate": 7.1790122149177135e-06,
+      "loss": 1.2783,
+      "step": 2365
+    },
+    {
+      "epoch": 0.6333511491181186,
+      "grad_norm": 1.159478422052137,
+      "learning_rate": 7.134285160965091e-06,
+      "loss": 1.3114,
+      "step": 2370
+    },
+    {
+      "epoch": 0.6346873329770176,
+      "grad_norm": 1.2191673865530408,
+      "learning_rate": 7.089620478443152e-06,
+      "loss": 1.2873,
+      "step": 2375
+    },
+    {
+      "epoch": 0.6360235168359166,
+      "grad_norm": 1.2212422633172386,
+      "learning_rate": 7.045019139465434e-06,
+      "loss": 1.2976,
+      "step": 2380
+    },
+    {
+      "epoch": 0.6373597006948156,
+      "grad_norm": 1.2849058652407424,
+      "learning_rate": 7.000482114766798e-06,
+      "loss": 1.3026,
+      "step": 2385
+    },
+    {
+      "epoch": 0.6386958845537146,
+      "grad_norm": 1.1693633671456403,
+      "learning_rate": 6.956010373682334e-06,
+      "loss": 1.2827,
+      "step": 2390
+    },
+    {
+      "epoch": 0.6400320684126136,
+      "grad_norm": 1.160020552936478,
+      "learning_rate": 6.911604884126259e-06,
+      "loss": 1.2515,
+      "step": 2395
+    },
+    {
+      "epoch": 0.6413682522715125,
+      "grad_norm": 1.1828314546763101,
+      "learning_rate": 6.867266612570833e-06,
+      "loss": 1.3092,
+      "step": 2400
+    },
+    {
+      "epoch": 0.6413682522715125,
+      "eval_loss": 1.2952780723571777,
+      "eval_runtime": 525.8904,
+      "eval_samples_per_second": 25.188,
+      "eval_steps_per_second": 3.149,
+      "step": 2400
+    },
+    {
+      "epoch": 0.6427044361304115,
+      "grad_norm": 1.137455706013817,
+      "learning_rate": 6.822996524025343e-06,
+      "loss": 1.2891,
+      "step": 2405
+    },
+    {
+      "epoch": 0.6440406199893105,
+      "grad_norm": 1.3754307767427276,
+      "learning_rate": 6.778795582015096e-06,
+      "loss": 1.2754,
+      "step": 2410
+    },
+    {
+      "epoch": 0.6453768038482095,
+      "grad_norm": 1.216856677079751,
+      "learning_rate": 6.734664748560437e-06,
+      "loss": 1.3014,
+      "step": 2415
+    },
+    {
+      "epoch": 0.6467129877071085,
+      "grad_norm": 1.2592962476705394,
+      "learning_rate": 6.690604984155826e-06,
+      "loss": 1.3221,
+      "step": 2420
+    },
+    {
+      "epoch": 0.6480491715660075,
+      "grad_norm": 1.1929612361301374,
+      "learning_rate": 6.646617247748926e-06,
+      "loss": 1.2768,
+      "step": 2425
+    },
+    {
+      "epoch": 0.6493853554249065,
+      "grad_norm": 1.1709139264257042,
+      "learning_rate": 6.602702496719724e-06,
+      "loss": 1.2549,
+      "step": 2430
+    },
+    {
+      "epoch": 0.6507215392838055,
+      "grad_norm": 1.2183672529081806,
+      "learning_rate": 6.558861686859712e-06,
+      "loss": 1.2764,
+      "step": 2435
+    },
+    {
+      "epoch": 0.6520577231427044,
+      "grad_norm": 1.1975495934510272,
+      "learning_rate": 6.515095772351072e-06,
+      "loss": 1.2929,
+      "step": 2440
+    },
+    {
+      "epoch": 0.6533939070016034,
+      "grad_norm": 1.1613807938703284,
+      "learning_rate": 6.471405705745906e-06,
+      "loss": 1.2684,
+      "step": 2445
+    },
+    {
+      "epoch": 0.6547300908605024,
+      "grad_norm": 1.2736187330390891,
+      "learning_rate": 6.427792437945516e-06,
+      "loss": 1.262,
+      "step": 2450
+    },
+    {
+      "epoch": 0.6560662747194014,
+      "grad_norm": 1.174927022230355,
+      "learning_rate": 6.384256918179692e-06,
+      "loss": 1.2958,
+      "step": 2455
+    },
+    {
+      "epoch": 0.6574024585783004,
+      "grad_norm": 1.1737222195085288,
+      "learning_rate": 6.340800093986071e-06,
+      "loss": 1.2987,
+      "step": 2460
+    },
+    {
+      "epoch": 0.6587386424371994,
+      "grad_norm": 1.1988962500744893,
+      "learning_rate": 6.297422911189499e-06,
+      "loss": 1.2469,
+      "step": 2465
+    },
+    {
+      "epoch": 0.6600748262960984,
+      "grad_norm": 1.138343574152133,
+      "learning_rate": 6.254126313881448e-06,
+      "loss": 1.2564,
+      "step": 2470
+    },
+    {
+      "epoch": 0.6614110101549974,
+      "grad_norm": 1.161565466877101,
+      "learning_rate": 6.210911244399477e-06,
+      "loss": 1.305,
+      "step": 2475
+    },
+    {
+      "epoch": 0.6627471940138963,
+      "grad_norm": 1.1944909640364771,
+      "learning_rate": 6.167778643306717e-06,
+      "loss": 1.2749,
+      "step": 2480
+    },
+    {
+      "epoch": 0.6640833778727953,
+      "grad_norm": 1.2250102533436973,
+      "learning_rate": 6.1247294493713845e-06,
+      "loss": 1.2892,
+      "step": 2485
+    },
+    {
+      "epoch": 0.6654195617316943,
+      "grad_norm": 1.211997261711134,
+      "learning_rate": 6.0817645995463845e-06,
+      "loss": 1.3064,
+      "step": 2490
+    },
+    {
+      "epoch": 0.6667557455905933,
+      "grad_norm": 1.2199059761183304,
+      "learning_rate": 6.038885028948889e-06,
+      "loss": 1.2591,
+      "step": 2495
+    },
+    {
+      "epoch": 0.6680919294494923,
+      "grad_norm": 1.204780340659424,
+      "learning_rate": 5.996091670839983e-06,
+      "loss": 1.2981,
+      "step": 2500
+    },
+    {
+      "epoch": 0.6694281133083912,
+      "grad_norm": 1.1622142280662515,
+      "learning_rate": 5.953385456604377e-06,
+      "loss": 1.2837,
+      "step": 2505
+    },
+    {
+      "epoch": 0.6707642971672902,
+      "grad_norm": 1.2595949655572636,
+      "learning_rate": 5.910767315730119e-06,
+      "loss": 1.3098,
+      "step": 2510
+    },
+    {
+      "epoch": 0.6721004810261892,
+      "grad_norm": 1.1991193943391245,
+      "learning_rate": 5.868238175788355e-06,
+      "loss": 1.2844,
+      "step": 2515
+    },
+    {
+      "epoch": 0.6734366648850882,
+      "grad_norm": 1.1286538895272766,
+      "learning_rate": 5.825798962413164e-06,
+      "loss": 1.2963,
+      "step": 2520
+    },
+    {
+      "epoch": 0.6747728487439871,
+      "grad_norm": 1.2065687187980167,
+      "learning_rate": 5.783450599281395e-06,
+      "loss": 1.3002,
+      "step": 2525
+    },
+    {
+      "epoch": 0.6761090326028861,
+      "grad_norm": 1.167072268529669,
+      "learning_rate": 5.7411940080925685e-06,
+      "loss": 1.2479,
+      "step": 2530
+    },
+    {
+      "epoch": 0.6774452164617851,
+      "grad_norm": 1.1698208994032941,
+      "learning_rate": 5.6990301085488235e-06,
+      "loss": 1.2999,
+      "step": 2535
+    },
+    {
+      "epoch": 0.6787814003206841,
+      "grad_norm": 1.1710233370221095,
+      "learning_rate": 5.656959818334873e-06,
+      "loss": 1.2595,
+      "step": 2540
+    },
+    {
+      "epoch": 0.6801175841795831,
+      "grad_norm": 1.386755510116527,
+      "learning_rate": 5.614984053098076e-06,
+      "loss": 1.2879,
+      "step": 2545
+    },
+    {
+      "epoch": 0.6814537680384821,
+      "grad_norm": 1.132940281229991,
+      "learning_rate": 5.5731037264284735e-06,
+      "loss": 1.2559,
+      "step": 2550
+    },
+    {
+      "epoch": 0.6827899518973811,
+      "grad_norm": 1.1926751219485074,
+      "learning_rate": 5.531319749838903e-06,
+      "loss": 1.2906,
+      "step": 2555
+    },
+    {
+      "epoch": 0.6841261357562801,
+      "grad_norm": 1.2244655010133048,
+      "learning_rate": 5.489633032745185e-06,
+      "loss": 1.2589,
+      "step": 2560
+    },
+    {
+      "epoch": 0.685462319615179,
+      "grad_norm": 1.177667766929752,
+      "learning_rate": 5.448044482446317e-06,
+      "loss": 1.2896,
+      "step": 2565
+    },
+    {
+      "epoch": 0.686798503474078,
+      "grad_norm": 1.1792403166242948,
+      "learning_rate": 5.406555004104712e-06,
+      "loss": 1.2724,
+      "step": 2570
+    },
+    {
+      "epoch": 0.688134687332977,
+      "grad_norm": 1.1752737928249541,
+      "learning_rate": 5.365165500726518e-06,
+      "loss": 1.3095,
+      "step": 2575
+    },
+    {
+      "epoch": 0.689470871191876,
+      "grad_norm": 1.2006301934171553,
+      "learning_rate": 5.323876873141973e-06,
+      "loss": 1.2682,
+      "step": 2580
+    },
+    {
+      "epoch": 0.690807055050775,
+      "grad_norm": 1.232283420067014,
+      "learning_rate": 5.282690019985756e-06,
+      "loss": 1.2973,
+      "step": 2585
+    },
+    {
+      "epoch": 0.692143238909674,
+      "grad_norm": 1.8591318420869134,
+      "learning_rate": 5.241605837677481e-06,
+      "loss": 1.2748,
+      "step": 2590
+    },
+    {
+      "epoch": 0.693479422768573,
+      "grad_norm": 1.1409257082194615,
+      "learning_rate": 5.200625220402139e-06,
+      "loss": 1.2786,
+      "step": 2595
+    },
+    {
+      "epoch": 0.694815606627472,
+      "grad_norm": 1.1874265547469547,
+      "learning_rate": 5.159749060090675e-06,
+      "loss": 1.2518,
+      "step": 2600
+    },
+    {
+      "epoch": 0.694815606627472,
+      "eval_loss": 1.2894538640975952,
+      "eval_runtime": 524.2545,
+      "eval_samples_per_second": 25.266,
+      "eval_steps_per_second": 3.159,
+      "step": 2600
+    },
+    {
+      "epoch": 0.6961517904863709,
+      "grad_norm": 1.1436341244774122,
+      "learning_rate": 5.118978246400555e-06,
+      "loss": 1.2868,
+      "step": 2605
+    },
+    {
+      "epoch": 0.6974879743452699,
+      "grad_norm": 1.1773055221354787,
+      "learning_rate": 5.078313666696404e-06,
+      "loss": 1.2865,
+      "step": 2610
+    },
+    {
+      "epoch": 0.6988241582041689,
+      "grad_norm": 1.205692219736226,
+      "learning_rate": 5.0377562060307e-06,
+      "loss": 1.271,
+      "step": 2615
+    },
+    {
+      "epoch": 0.7001603420630679,
+      "grad_norm": 1.1681619335726294,
+      "learning_rate": 4.997306747124508e-06,
+      "loss": 1.281,
+      "step": 2620
+    },
+    {
+      "epoch": 0.7014965259219669,
+      "grad_norm": 1.1239087952042888,
+      "learning_rate": 4.9569661703482585e-06,
+      "loss": 1.2933,
+      "step": 2625
+    },
+    {
+      "epoch": 0.7028327097808659,
+      "grad_norm": 1.1618547501473921,
+      "learning_rate": 4.916735353702603e-06,
+      "loss": 1.2556,
+      "step": 2630
+    },
+    {
+      "epoch": 0.7041688936397649,
+      "grad_norm": 1.1446438877087823,
+      "learning_rate": 4.876615172799294e-06,
+      "loss": 1.2964,
+      "step": 2635
+    },
+    {
+      "epoch": 0.7055050774986639,
+      "grad_norm": 1.2196306868529612,
+      "learning_rate": 4.836606500842129e-06,
+      "loss": 1.2934,
+      "step": 2640
+    },
+    {
+      "epoch": 0.7068412613575628,
+      "grad_norm": 1.1966884099651143,
+      "learning_rate": 4.7967102086079485e-06,
+      "loss": 1.2432,
+      "step": 2645
+    },
+    {
+      "epoch": 0.7081774452164618,
+      "grad_norm": 1.1483128968540317,
+      "learning_rate": 4.756927164427685e-06,
+      "loss": 1.2795,
+      "step": 2650
+    },
+    {
+      "epoch": 0.7095136290753608,
+      "grad_norm": 1.1539714350859465,
+      "learning_rate": 4.717258234167448e-06,
+      "loss": 1.2494,
+      "step": 2655
+    },
+    {
+      "epoch": 0.7108498129342598,
+      "grad_norm": 1.1421161621669704,
+      "learning_rate": 4.677704281209707e-06,
+      "loss": 1.2721,
+      "step": 2660
+    },
+    {
+      "epoch": 0.7121859967931587,
+      "grad_norm": 1.1666778493912067,
+      "learning_rate": 4.6382661664344665e-06,
+      "loss": 1.2805,
+      "step": 2665
+    },
+    {
+      "epoch": 0.7135221806520577,
+      "grad_norm": 1.2152799023877705,
+      "learning_rate": 4.59894474820057e-06,
+      "loss": 1.2781,
+      "step": 2670
+    },
+    {
+      "epoch": 0.7148583645109567,
+      "grad_norm": 1.1802067902023374,
+      "learning_rate": 4.559740882326984e-06,
+      "loss": 1.2889,
+      "step": 2675
+    },
+    {
+      "epoch": 0.7161945483698557,
+      "grad_norm": 1.1615396840856154,
+      "learning_rate": 4.520655422074176e-06,
+      "loss": 1.2295,
+      "step": 2680
+    },
+    {
+      "epoch": 0.7175307322287546,
+      "grad_norm": 1.1290092637035325,
+      "learning_rate": 4.481689218125561e-06,
+      "loss": 1.2812,
+      "step": 2685
+    },
+    {
+      "epoch": 0.7188669160876536,
+      "grad_norm": 1.1477861204230764,
+      "learning_rate": 4.442843118568976e-06,
+      "loss": 1.2565,
+      "step": 2690
+    },
+    {
+      "epoch": 0.7202030999465526,
+      "grad_norm": 1.2347710658904285,
+      "learning_rate": 4.4041179688782095e-06,
+      "loss": 1.2959,
+      "step": 2695
+    },
+    {
+      "epoch": 0.7215392838054516,
+      "grad_norm": 1.2180531197328832,
+      "learning_rate": 4.365514611894623e-06,
+      "loss": 1.2976,
+      "step": 2700
+    },
+    {
+      "epoch": 0.7228754676643506,
+      "grad_norm": 1.1757278803891242,
+      "learning_rate": 4.327033887808802e-06,
+      "loss": 1.2761,
+      "step": 2705
+    },
+    {
+      "epoch": 0.7242116515232496,
+      "grad_norm": 1.1645726922846997,
+      "learning_rate": 4.288676634142247e-06,
+      "loss": 1.2595,
+      "step": 2710
+    },
+    {
+      "epoch": 0.7255478353821486,
+      "grad_norm": 1.1411088256784303,
+      "learning_rate": 4.25044368572917e-06,
+      "loss": 1.2714,
+      "step": 2715
+    },
+    {
+      "epoch": 0.7268840192410476,
+      "grad_norm": 1.1671327639274043,
+      "learning_rate": 4.2123358746983225e-06,
+      "loss": 1.2876,
+      "step": 2720
+    },
+    {
+      "epoch": 0.7282202030999465,
+      "grad_norm": 1.1581886797505927,
+      "learning_rate": 4.1743540304548615e-06,
+      "loss": 1.2859,
+      "step": 2725
+    },
+    {
+      "epoch": 0.7295563869588455,
+      "grad_norm": 1.1510537093782165,
+      "learning_rate": 4.13649897966233e-06,
+      "loss": 1.2437,
+      "step": 2730
+    },
+    {
+      "epoch": 0.7308925708177445,
+      "grad_norm": 1.1962597941965873,
+      "learning_rate": 4.098771546224643e-06,
+      "loss": 1.2697,
+      "step": 2735
+    },
+    {
+      "epoch": 0.7322287546766435,
+      "grad_norm": 1.176978028412522,
+      "learning_rate": 4.061172551268162e-06,
+      "loss": 1.3069,
+      "step": 2740
+    },
+    {
+      "epoch": 0.7335649385355425,
+      "grad_norm": 1.2344307343981975,
+      "learning_rate": 4.023702813123828e-06,
+      "loss": 1.3034,
+      "step": 2745
+    },
+    {
+      "epoch": 0.7349011223944415,
+      "grad_norm": 1.1490296827570874,
+      "learning_rate": 3.986363147309332e-06,
+      "loss": 1.2934,
+      "step": 2750
+    },
+    {
+      "epoch": 0.7362373062533405,
+      "grad_norm": 1.128655104477458,
+      "learning_rate": 3.949154366511395e-06,
+      "loss": 1.3227,
+      "step": 2755
+    },
+    {
+      "epoch": 0.7375734901122395,
+      "grad_norm": 1.1377152493000582,
+      "learning_rate": 3.9120772805680575e-06,
+      "loss": 1.2837,
+      "step": 2760
+    },
+    {
+      "epoch": 0.7389096739711384,
+      "grad_norm": 1.1144595543031406,
+      "learning_rate": 3.8751326964510615e-06,
+      "loss": 1.2451,
+      "step": 2765
+    },
+    {
+      "epoch": 0.7402458578300374,
+      "grad_norm": 1.1528921550358049,
+      "learning_rate": 3.838321418248288e-06,
+      "loss": 1.2615,
+      "step": 2770
+    },
+    {
+      "epoch": 0.7415820416889364,
+      "grad_norm": 1.2690943955678684,
+      "learning_rate": 3.8016442471462524e-06,
+      "loss": 1.2829,
+      "step": 2775
+    },
+    {
+      "epoch": 0.7429182255478354,
+      "grad_norm": 1.2308517139392547,
+      "learning_rate": 3.7651019814126656e-06,
+      "loss": 1.2617,
+      "step": 2780
+    },
+    {
+      "epoch": 0.7442544094067344,
+      "grad_norm": 1.2017097006317634,
+      "learning_rate": 3.7286954163790734e-06,
+      "loss": 1.2364,
+      "step": 2785
+    },
+    {
+      "epoch": 0.7455905932656334,
+      "grad_norm": 1.1828941772780268,
+      "learning_rate": 3.6924253444235224e-06,
+      "loss": 1.3087,
+      "step": 2790
+    },
+    {
+      "epoch": 0.7469267771245324,
+      "grad_norm": 1.1861330848081788,
+      "learning_rate": 3.6562925549533355e-06,
+      "loss": 1.2789,
+      "step": 2795
+    },
+    {
+      "epoch": 0.7482629609834314,
+      "grad_norm": 1.1526594671507095,
+      "learning_rate": 3.6202978343879337e-06,
+      "loss": 1.2617,
+      "step": 2800
+    },
+    {
+      "epoch": 0.7482629609834314,
+      "eval_loss": 1.2845991849899292,
+      "eval_runtime": 523.8103,
+      "eval_samples_per_second": 25.288,
+      "eval_steps_per_second": 3.161,
+      "step": 2800
+    },
+    {
+      "epoch": 0.7495991448423303,
+      "grad_norm": 1.1298081744323016,
+      "learning_rate": 3.5844419661416886e-06,
+      "loss": 1.2902,
+      "step": 2805
+    },
+    {
+      "epoch": 0.7509353287012293,
+      "grad_norm": 1.1561421830885714,
+      "learning_rate": 3.5487257306069054e-06,
+      "loss": 1.29,
+      "step": 2810
+    },
+    {
+      "epoch": 0.7522715125601283,
+      "grad_norm": 1.1536531647083885,
+      "learning_rate": 3.5131499051368254e-06,
+      "loss": 1.2614,
+      "step": 2815
+    },
+    {
+      "epoch": 0.7536076964190273,
+      "grad_norm": 1.1856993866727077,
+      "learning_rate": 3.4777152640286984e-06,
+      "loss": 1.2861,
+      "step": 2820
+    },
+    {
+      "epoch": 0.7549438802779262,
+      "grad_norm": 1.2042645879668878,
+      "learning_rate": 3.4424225785069444e-06,
+      "loss": 1.2658,
+      "step": 2825
+    },
+    {
+      "epoch": 0.7562800641368252,
+      "grad_norm": 1.1759218780648395,
+      "learning_rate": 3.4072726167063617e-06,
+      "loss": 1.2613,
+      "step": 2830
+    },
+    {
+      "epoch": 0.7576162479957242,
+      "grad_norm": 1.1941131261396993,
+      "learning_rate": 3.3722661436554104e-06,
+      "loss": 1.2926,
+      "step": 2835
+    },
+    {
+      "epoch": 0.7589524318546232,
+      "grad_norm": 1.2349621736436946,
+      "learning_rate": 3.337403921259559e-06,
+      "loss": 1.2852,
+      "step": 2840
+    },
+    {
+      "epoch": 0.7602886157135221,
+      "grad_norm": 1.243962877866245,
+      "learning_rate": 3.3026867082847058e-06,
+      "loss": 1.3087,
+      "step": 2845
+    },
+    {
+      "epoch": 0.7616247995724211,
+      "grad_norm": 1.2755840307358004,
+      "learning_rate": 3.268115260340654e-06,
+      "loss": 1.2574,
+      "step": 2850
+    },
+    {
+      "epoch": 0.7629609834313201,
+      "grad_norm": 1.256694820928362,
+      "learning_rate": 3.233690329864684e-06,
+      "loss": 1.2477,
+      "step": 2855
+    },
+    {
+      "epoch": 0.7642971672902191,
+      "grad_norm": 1.153504157549798,
+      "learning_rate": 3.1994126661051628e-06,
+      "loss": 1.2669,
+      "step": 2860
+    },
+    {
+      "epoch": 0.7656333511491181,
+      "grad_norm": 1.2272283092280594,
+      "learning_rate": 3.1652830151052416e-06,
+      "loss": 1.2919,
+      "step": 2865
+    },
+    {
+      "epoch": 0.7669695350080171,
+      "grad_norm": 1.1512364936411144,
+      "learning_rate": 3.131302119686621e-06,
+      "loss": 1.2736,
+      "step": 2870
+    },
+    {
+      "epoch": 0.7683057188669161,
+      "grad_norm": 1.1564153021008827,
+      "learning_rate": 3.097470719433373e-06,
+      "loss": 1.2556,
+      "step": 2875
+    },
+    {
+      "epoch": 0.7696419027258151,
+      "grad_norm": 1.164737683509797,
+      "learning_rate": 3.0637895506758597e-06,
+      "loss": 1.2672,
+      "step": 2880
+    },
+    {
+      "epoch": 0.770978086584714,
+      "grad_norm": 1.1685661560913843,
+      "learning_rate": 3.0302593464746944e-06,
+      "loss": 1.2907,
+      "step": 2885
+    },
+    {
+      "epoch": 0.772314270443613,
+      "grad_norm": 1.1699267498621475,
+      "learning_rate": 2.9968808366047942e-06,
+      "loss": 1.2699,
+      "step": 2890
+    },
+    {
+      "epoch": 0.773650454302512,
+      "grad_norm": 1.211199795955048,
+      "learning_rate": 2.963654747539494e-06,
+      "loss": 1.2693,
+      "step": 2895
+    },
+    {
+      "epoch": 0.774986638161411,
+      "grad_norm": 1.1816669846421493,
+      "learning_rate": 2.9305818024347378e-06,
+      "loss": 1.2651,
+      "step": 2900
+    },
+    {
+      "epoch": 0.77632282202031,
+      "grad_norm": 1.1914669242505636,
+      "learning_rate": 2.897662721113328e-06,
+      "loss": 1.2737,
+      "step": 2905
+    },
+    {
+      "epoch": 0.777659005879209,
+      "grad_norm": 1.1738719241452067,
+      "learning_rate": 2.864898220049277e-06,
+      "loss": 1.2564,
+      "step": 2910
+    },
+    {
+      "epoch": 0.778995189738108,
+      "grad_norm": 1.1840077986894364,
+      "learning_rate": 2.832289012352203e-06,
+      "loss": 1.2522,
+      "step": 2915
+    },
+    {
+      "epoch": 0.780331373597007,
+      "grad_norm": 1.233343112626304,
+      "learning_rate": 2.7998358077517975e-06,
+      "loss": 1.2448,
+      "step": 2920
+    },
+    {
+      "epoch": 0.7816675574559059,
+      "grad_norm": 1.1877039129056828,
+      "learning_rate": 2.7675393125824144e-06,
+      "loss": 1.2563,
+      "step": 2925
+    },
+    {
+      "epoch": 0.7830037413148049,
+      "grad_norm": 1.1982422068810221,
+      "learning_rate": 2.735400229767652e-06,
+      "loss": 1.2403,
+      "step": 2930
+    },
+    {
+      "epoch": 0.7843399251737039,
+      "grad_norm": 1.1935298921878634,
+      "learning_rate": 2.7034192588050845e-06,
+      "loss": 1.2701,
+      "step": 2935
+    },
+    {
+      "epoch": 0.7856761090326029,
+      "grad_norm": 1.151254746126473,
+      "learning_rate": 2.671597095751033e-06,
+      "loss": 1.2509,
+      "step": 2940
+    },
+    {
+      "epoch": 0.7870122928915019,
+      "grad_norm": 1.190278414424794,
+      "learning_rate": 2.6399344332054e-06,
+      "loss": 1.2914,
+      "step": 2945
+    },
+    {
+      "epoch": 0.7883484767504009,
+      "grad_norm": 1.168908881155384,
+      "learning_rate": 2.6084319602966167e-06,
+      "loss": 1.2739,
+      "step": 2950
+    },
+    {
+      "epoch": 0.7896846606092999,
+      "grad_norm": 1.1495974699891471,
+      "learning_rate": 2.577090362666631e-06,
+      "loss": 1.2822,
+      "step": 2955
+    },
+    {
+      "epoch": 0.7910208444681989,
+      "grad_norm": 1.199562609813528,
+      "learning_rate": 2.54591032245599e-06,
+      "loss": 1.2833,
+      "step": 2960
+    },
+    {
+      "epoch": 0.7923570283270978,
+      "grad_norm": 1.172598488602994,
+      "learning_rate": 2.514892518288988e-06,
+      "loss": 1.2427,
+      "step": 2965
+    },
+    {
+      "epoch": 0.7936932121859968,
+      "grad_norm": 1.2023007809260173,
+      "learning_rate": 2.484037625258908e-06,
+      "loss": 1.3116,
+      "step": 2970
+    },
+    {
+      "epoch": 0.7950293960448958,
+      "grad_norm": 1.1808013233447734,
+      "learning_rate": 2.4533463149133073e-06,
+      "loss": 1.2367,
+      "step": 2975
+    },
+    {
+      "epoch": 0.7963655799037948,
+      "grad_norm": 1.2065243432367916,
+      "learning_rate": 2.422819255239427e-06,
+      "loss": 1.2995,
+      "step": 2980
+    },
+    {
+      "epoch": 0.7977017637626937,
+      "grad_norm": 1.1490648428460022,
+      "learning_rate": 2.392457110649634e-06,
+      "loss": 1.2676,
+      "step": 2985
+    },
+    {
+      "epoch": 0.7990379476215927,
+      "grad_norm": 1.136037346488201,
+      "learning_rate": 2.362260541966972e-06,
+      "loss": 1.2816,
+      "step": 2990
+    },
+    {
+      "epoch": 0.8003741314804917,
+      "grad_norm": 1.149577173830776,
+      "learning_rate": 2.3322302064107762e-06,
+      "loss": 1.2948,
+      "step": 2995
+    },
+    {
+      "epoch": 0.8017103153393907,
+      "grad_norm": 1.185488122515381,
+      "learning_rate": 2.302366757582355e-06,
+      "loss": 1.3041,
+      "step": 3000
+    },
+    {
+      "epoch": 0.8017103153393907,
+      "eval_loss": 1.2808870077133179,
+      "eval_runtime": 523.4307,
+      "eval_samples_per_second": 25.306,
+      "eval_steps_per_second": 3.164,
+      "step": 3000
+    },
+    {
+      "epoch": 0.8030464991982896,
+      "grad_norm": 1.1513855940527316,
+      "learning_rate": 2.272670845450791e-06,
+      "loss": 1.277,
+      "step": 3005
+    },
+    {
+      "epoch": 0.8043826830571886,
+      "grad_norm": 1.1602967857229551,
+      "learning_rate": 2.243143116338773e-06,
+      "loss": 1.2653,
+      "step": 3010
+    },
+    {
+      "epoch": 0.8057188669160876,
+      "grad_norm": 1.197568564758537,
+      "learning_rate": 2.2137842129085396e-06,
+      "loss": 1.3524,
+      "step": 3015
+    },
+    {
+      "epoch": 0.8070550507749866,
+      "grad_norm": 1.1421243027571122,
+      "learning_rate": 2.1845947741478857e-06,
+      "loss": 1.2651,
+      "step": 3020
+    },
+    {
+      "epoch": 0.8083912346338856,
+      "grad_norm": 1.167663307330092,
+      "learning_rate": 2.1555754353562643e-06,
+      "loss": 1.2337,
+      "step": 3025
+    },
+    {
+      "epoch": 0.8097274184927846,
+      "grad_norm": 1.1570704863531982,
+      "learning_rate": 2.1267268281309418e-06,
+      "loss": 1.2671,
+      "step": 3030
+    },
+    {
+      "epoch": 0.8110636023516836,
+      "grad_norm": 1.1646116509684428,
+      "learning_rate": 2.098049580353273e-06,
+      "loss": 1.2831,
+      "step": 3035
+    },
+    {
+      "epoch": 0.8123997862105826,
+      "grad_norm": 1.1929244270961756,
+      "learning_rate": 2.069544316175025e-06,
+      "loss": 1.2631,
+      "step": 3040
+    },
+    {
+      "epoch": 0.8137359700694815,
+      "grad_norm": 1.1495579309904203,
+      "learning_rate": 2.041211656004781e-06,
+      "loss": 1.2891,
+      "step": 3045
+    },
+    {
+      "epoch": 0.8150721539283805,
+      "grad_norm": 1.158971184848035,
+      "learning_rate": 2.0130522164944666e-06,
+      "loss": 1.2543,
+      "step": 3050
+    },
+    {
+      "epoch": 0.8164083377872795,
+      "grad_norm": 1.1993305810419228,
+      "learning_rate": 1.985066610525904e-06,
+      "loss": 1.3294,
+      "step": 3055
+    },
+    {
+      "epoch": 0.8177445216461785,
+      "grad_norm": 1.1523443888711082,
+      "learning_rate": 1.9572554471974723e-06,
+      "loss": 1.2771,
+      "step": 3060
+    },
+    {
+      "epoch": 0.8190807055050775,
+      "grad_norm": 1.1695216750777881,
+      "learning_rate": 1.9296193318108723e-06,
+      "loss": 1.2578,
+      "step": 3065
+    },
+    {
+      "epoch": 0.8204168893639765,
+      "grad_norm": 1.1732642497713914,
+      "learning_rate": 1.9021588658579249e-06,
+      "loss": 1.2513,
+      "step": 3070
+    },
+    {
+      "epoch": 0.8217530732228755,
+      "grad_norm": 1.1519213257119236,
+      "learning_rate": 1.8748746470075029e-06,
+      "loss": 1.2977,
+      "step": 3075
+    },
+    {
+      "epoch": 0.8230892570817745,
+      "grad_norm": 1.1516668595048642,
+      "learning_rate": 1.847767269092511e-06,
+      "loss": 1.2321,
+      "step": 3080
+    },
+    {
+      "epoch": 0.8244254409406734,
+      "grad_norm": 1.1909108579078609,
+      "learning_rate": 1.820837322096961e-06,
+      "loss": 1.2484,
+      "step": 3085
+    },
+    {
+      "epoch": 0.8257616247995724,
+      "grad_norm": 1.101081523016669,
+      "learning_rate": 1.7940853921431378e-06,
+      "loss": 1.2575,
+      "step": 3090
+    },
+    {
+      "epoch": 0.8270978086584714,
+      "grad_norm": 1.178074475938639,
+      "learning_rate": 1.7675120614788367e-06,
+      "loss": 1.2664,
+      "step": 3095
+    },
+    {
+      "epoch": 0.8284339925173704,
+      "grad_norm": 1.1731584765176883,
+      "learning_rate": 1.7411179084646879e-06,
+      "loss": 1.2804,
+      "step": 3100
+    },
+    {
+      "epoch": 0.8297701763762694,
+      "grad_norm": 1.1586731249421351,
+      "learning_rate": 1.7149035075615795e-06,
+      "loss": 1.2905,
+      "step": 3105
+    },
+    {
+      "epoch": 0.8311063602351684,
+      "grad_norm": 1.169289922029439,
+      "learning_rate": 1.6888694293181462e-06,
+      "loss": 1.2836,
+      "step": 3110
+    },
+    {
+      "epoch": 0.8324425440940674,
+      "grad_norm": 1.1719327462563847,
+      "learning_rate": 1.6630162403583538e-06,
+      "loss": 1.2513,
+      "step": 3115
+    },
+    {
+      "epoch": 0.8337787279529664,
+      "grad_norm": 1.1435061474054447,
+      "learning_rate": 1.637344503369167e-06,
+      "loss": 1.2635,
+      "step": 3120
+    },
+    {
+      "epoch": 0.8351149118118654,
+      "grad_norm": 1.1521864026183537,
+      "learning_rate": 1.6118547770883031e-06,
+      "loss": 1.3022,
+      "step": 3125
+    },
+    {
+      "epoch": 0.8364510956707643,
+      "grad_norm": 1.2218598184249856,
+      "learning_rate": 1.5865476162920658e-06,
+      "loss": 1.293,
+      "step": 3130
+    },
+    {
+      "epoch": 0.8377872795296633,
+      "grad_norm": 1.1461871951868958,
+      "learning_rate": 1.5614235717832838e-06,
+      "loss": 1.284,
+      "step": 3135
+    },
+    {
+      "epoch": 0.8391234633885623,
+      "grad_norm": 1.2002653441412816,
+      "learning_rate": 1.536483190379302e-06,
+      "loss": 1.2608,
+      "step": 3140
+    },
+    {
+      "epoch": 0.8404596472474613,
+      "grad_norm": 1.194888981517886,
+      "learning_rate": 1.5117270149001061e-06,
+      "loss": 1.2728,
+      "step": 3145
+    },
+    {
+      "epoch": 0.8417958311063602,
+      "grad_norm": 1.1669886636546563,
+      "learning_rate": 1.4871555841564889e-06,
+      "loss": 1.2936,
+      "step": 3150
+    },
+    {
+      "epoch": 0.8431320149652592,
+      "grad_norm": 1.178341085561249,
+      "learning_rate": 1.4627694329383245e-06,
+      "loss": 1.2737,
+      "step": 3155
+    },
+    {
+      "epoch": 0.8444681988241582,
+      "grad_norm": 1.173822019467738,
+      "learning_rate": 1.4385690920029394e-06,
+      "loss": 1.2389,
+      "step": 3160
+    },
+    {
+      "epoch": 0.8458043826830572,
+      "grad_norm": 1.172444929862991,
+      "learning_rate": 1.4145550880635551e-06,
+      "loss": 1.283,
+      "step": 3165
+    },
+    {
+      "epoch": 0.8471405665419561,
+      "grad_norm": 1.157959236772995,
+      "learning_rate": 1.3907279437778154e-06,
+      "loss": 1.3089,
+      "step": 3170
+    },
+    {
+      "epoch": 0.8484767504008551,
+      "grad_norm": 1.142956839312928,
+      "learning_rate": 1.3670881777364276e-06,
+      "loss": 1.2385,
+      "step": 3175
+    },
+    {
+      "epoch": 0.8498129342597541,
+      "grad_norm": 1.1394377032150336,
+      "learning_rate": 1.3436363044518685e-06,
+      "loss": 1.3002,
+      "step": 3180
+    },
+    {
+      "epoch": 0.8511491181186531,
+      "grad_norm": 1.1081112610690127,
+      "learning_rate": 1.3203728343471766e-06,
+      "loss": 1.2471,
+      "step": 3185
+    },
+    {
+      "epoch": 0.8524853019775521,
+      "grad_norm": 1.1233094984645868,
+      "learning_rate": 1.2972982737448582e-06,
+      "loss": 1.2511,
+      "step": 3190
+    },
+    {
+      "epoch": 0.8538214858364511,
+      "grad_norm": 1.168807245788649,
+      "learning_rate": 1.27441312485586e-06,
+      "loss": 1.2487,
+      "step": 3195
+    },
+    {
+      "epoch": 0.8551576696953501,
+      "grad_norm": 1.2213102334394346,
+      "learning_rate": 1.2517178857686318e-06,
+      "loss": 1.3102,
+      "step": 3200
+    },
+    {
+      "epoch": 0.8551576696953501,
+      "eval_loss": 1.278082013130188,
+      "eval_runtime": 523.3237,
+      "eval_samples_per_second": 25.311,
+      "eval_steps_per_second": 3.164,
+      "step": 3200
+    },
+    {
+      "epoch": 0.8564938535542491,
+      "grad_norm": 1.120564972312021,
+      "learning_rate": 1.2292130504383037e-06,
+      "loss": 1.2578,
+      "step": 3205
+    },
+    {
+      "epoch": 0.857830037413148,
+      "grad_norm": 1.1580496245252512,
+      "learning_rate": 1.2068991086759175e-06,
+      "loss": 1.2983,
+      "step": 3210
+    },
+    {
+      "epoch": 0.859166221272047,
+      "grad_norm": 1.1198656733420578,
+      "learning_rate": 1.1847765461377757e-06,
+      "loss": 1.2245,
+      "step": 3215
+    },
+    {
+      "epoch": 0.860502405130946,
+      "grad_norm": 1.2004671948391095,
+      "learning_rate": 1.1628458443148716e-06,
+      "loss": 1.2737,
+      "step": 3220
+    },
+    {
+      "epoch": 0.861838588989845,
+      "grad_norm": 1.114311359039494,
+      "learning_rate": 1.1411074805223997e-06,
+      "loss": 1.2583,
+      "step": 3225
+    },
+    {
+      "epoch": 0.863174772848744,
+      "grad_norm": 1.1608634318847668,
+      "learning_rate": 1.119561927889381e-06,
+      "loss": 1.2543,
+      "step": 3230
+    },
+    {
+      "epoch": 0.864510956707643,
+      "grad_norm": 1.178010917617475,
+      "learning_rate": 1.0982096553483568e-06,
+      "loss": 1.3171,
+      "step": 3235
+    },
+    {
+      "epoch": 0.865847140566542,
+      "grad_norm": 1.1868881875439512,
+      "learning_rate": 1.077051127625185e-06,
+      "loss": 1.2985,
+      "step": 3240
+    },
+    {
+      "epoch": 0.867183324425441,
+      "grad_norm": 1.1357672595216541,
+      "learning_rate": 1.0560868052289253e-06,
+      "loss": 1.2634,
+      "step": 3245
+    },
+    {
+      "epoch": 0.8685195082843399,
+      "grad_norm": 1.171433935969695,
+      "learning_rate": 1.0353171444418187e-06,
+      "loss": 1.2759,
+      "step": 3250
+    },
+    {
+      "epoch": 0.8698556921432389,
+      "grad_norm": 1.1224025445608299,
+      "learning_rate": 1.014742597309346e-06,
+      "loss": 1.267,
+      "step": 3255
+    },
+    {
+      "epoch": 0.8711918760021379,
+      "grad_norm": 1.1874141914261405,
+      "learning_rate": 9.943636116304068e-07,
+      "loss": 1.2509,
+      "step": 3260
+    },
+    {
+      "epoch": 0.8725280598610369,
+      "grad_norm": 1.1508324966317753,
+      "learning_rate": 9.741806309475588e-07,
+      "loss": 1.2647,
+      "step": 3265
+    },
+    {
+      "epoch": 0.8738642437199359,
+      "grad_norm": 1.1671717510783461,
+      "learning_rate": 9.541940945373718e-07,
+      "loss": 1.2357,
+      "step": 3270
+    },
+    {
+      "epoch": 0.8752004275788349,
+      "grad_norm": 1.1214376758027105,
+      "learning_rate": 9.344044374008632e-07,
+      "loss": 1.2598,
+      "step": 3275
+    },
+    {
+      "epoch": 0.8765366114377339,
+      "grad_norm": 1.1745237243989533,
+      "learning_rate": 9.148120902540281e-07,
+      "loss": 1.238,
+      "step": 3280
+    },
+    {
+      "epoch": 0.8778727952966329,
+      "grad_norm": 1.1566262573536532,
+      "learning_rate": 8.954174795184756e-07,
+      "loss": 1.2464,
+      "step": 3285
+    },
+    {
+      "epoch": 0.8792089791555318,
+      "grad_norm": 1.158731873705362,
+      "learning_rate": 8.762210273121363e-07,
+      "loss": 1.2783,
+      "step": 3290
+    },
+    {
+      "epoch": 0.8805451630144308,
+      "grad_norm": 1.1187300362715888,
+      "learning_rate": 8.572231514400775e-07,
+      "loss": 1.2692,
+      "step": 3295
+    },
+    {
+      "epoch": 0.8818813468733298,
+      "grad_norm": 1.1820432928944034,
+      "learning_rate": 8.384242653854146e-07,
+      "loss": 1.3092,
+      "step": 3300
+    },
+    {
+      "epoch": 0.8832175307322288,
+      "grad_norm": 1.1315388199619425,
+      "learning_rate": 8.198247783003133e-07,
+      "loss": 1.2581,
+      "step": 3305
+    },
+    {
+      "epoch": 0.8845537145911277,
+      "grad_norm": 1.135539300633719,
+      "learning_rate": 8.014250949970704e-07,
+      "loss": 1.2726,
+      "step": 3310
+    },
+    {
+      "epoch": 0.8858898984500267,
+      "grad_norm": 1.147749224885632,
+      "learning_rate": 7.832256159393181e-07,
+      "loss": 1.2426,
+      "step": 3315
+    },
+    {
+      "epoch": 0.8872260823089257,
+      "grad_norm": 1.1681170491391923,
+      "learning_rate": 7.652267372333056e-07,
+      "loss": 1.2487,
+      "step": 3320
+    },
+    {
+      "epoch": 0.8885622661678247,
+      "grad_norm": 1.1503659735195155,
+      "learning_rate": 7.474288506192662e-07,
+      "loss": 1.2898,
+      "step": 3325
+    },
+    {
+      "epoch": 0.8898984500267236,
+      "grad_norm": 1.205235234079246,
+      "learning_rate": 7.298323434629095e-07,
+      "loss": 1.3094,
+      "step": 3330
+    },
+    {
+      "epoch": 0.8912346338856226,
+      "grad_norm": 1.1557951328425964,
+      "learning_rate": 7.124375987469767e-07,
+      "loss": 1.2326,
+      "step": 3335
+    },
+    {
+      "epoch": 0.8925708177445216,
+      "grad_norm": 1.1334618687477662,
+      "learning_rate": 6.952449950629103e-07,
+      "loss": 1.2679,
+      "step": 3340
+    },
+    {
+      "epoch": 0.8939070016034206,
+      "grad_norm": 1.14695924636925,
+      "learning_rate": 6.782549066026145e-07,
+      "loss": 1.3017,
+      "step": 3345
+    },
+    {
+      "epoch": 0.8952431854623196,
+      "grad_norm": 1.147346259057884,
+      "learning_rate": 6.614677031503059e-07,
+      "loss": 1.2705,
+      "step": 3350
+    },
+    {
+      "epoch": 0.8965793693212186,
+      "grad_norm": 1.1656452627988076,
+      "learning_rate": 6.448837500744742e-07,
+      "loss": 1.2475,
+      "step": 3355
+    },
+    {
+      "epoch": 0.8979155531801176,
+      "grad_norm": 1.1415605090418526,
+      "learning_rate": 6.285034083199216e-07,
+      "loss": 1.273,
+      "step": 3360
+    },
+    {
+      "epoch": 0.8992517370390166,
+      "grad_norm": 1.1324906184307248,
+      "learning_rate": 6.123270343999132e-07,
+      "loss": 1.2817,
+      "step": 3365
+    },
+    {
+      "epoch": 0.9005879208979155,
+      "grad_norm": 1.1353501329617395,
+      "learning_rate": 5.963549803884128e-07,
+      "loss": 1.2734,
+      "step": 3370
+    },
+    {
+      "epoch": 0.9019241047568145,
+      "grad_norm": 1.2284321012966415,
+      "learning_rate": 5.80587593912425e-07,
+      "loss": 1.276,
+      "step": 3375
+    },
+    {
+      "epoch": 0.9032602886157135,
+      "grad_norm": 1.1122913016411495,
+      "learning_rate": 5.650252181444215e-07,
+      "loss": 1.2524,
+      "step": 3380
+    },
+    {
+      "epoch": 0.9045964724746125,
+      "grad_norm": 1.1797725331714115,
+      "learning_rate": 5.496681917948809e-07,
+      "loss": 1.2336,
+      "step": 3385
+    },
+    {
+      "epoch": 0.9059326563335115,
+      "grad_norm": 1.159426377293814,
+      "learning_rate": 5.345168491049124e-07,
+      "loss": 1.242,
+      "step": 3390
+    },
+    {
+      "epoch": 0.9072688401924105,
+      "grad_norm": 1.1645037741741007,
+      "learning_rate": 5.195715198389784e-07,
+      "loss": 1.3,
+      "step": 3395
+    },
+    {
+      "epoch": 0.9086050240513095,
+      "grad_norm": 1.1480424103028632,
+      "learning_rate": 5.048325292777279e-07,
+      "loss": 1.2675,
+      "step": 3400
+    },
+    {
+      "epoch": 0.9086050240513095,
+      "eval_loss": 1.2765072584152222,
+      "eval_runtime": 523.9483,
+      "eval_samples_per_second": 25.281,
+      "eval_steps_per_second": 3.161,
+      "step": 3400
+    },
+    {
+      "epoch": 0.9099412079102085,
+      "grad_norm": 1.1694724018883438,
+      "learning_rate": 4.903001982109002e-07,
+      "loss": 1.2881,
+      "step": 3405
+    },
+    {
+      "epoch": 0.9112773917691074,
+      "grad_norm": 1.149736399066079,
+      "learning_rate": 4.759748429303579e-07,
+      "loss": 1.2993,
+      "step": 3410
+    },
+    {
+      "epoch": 0.9126135756280064,
+      "grad_norm": 1.1394564788621853,
+      "learning_rate": 4.618567752231962e-07,
+      "loss": 1.2486,
+      "step": 3415
+    },
+    {
+      "epoch": 0.9139497594869054,
+      "grad_norm": 1.1538669258463885,
+      "learning_rate": 4.479463023649555e-07,
+      "loss": 1.2618,
+      "step": 3420
+    },
+    {
+      "epoch": 0.9152859433458044,
+      "grad_norm": 1.1791184429266262,
+      "learning_rate": 4.342437271129396e-07,
+      "loss": 1.2583,
+      "step": 3425
+    },
+    {
+      "epoch": 0.9166221272047034,
+      "grad_norm": 1.1326513415617325,
+      "learning_rate": 4.207493476996205e-07,
+      "loss": 1.2408,
+      "step": 3430
+    },
+    {
+      "epoch": 0.9179583110636024,
+      "grad_norm": 1.1222236287232992,
+      "learning_rate": 4.074634578261516e-07,
+      "loss": 1.2379,
+      "step": 3435
+    },
+    {
+      "epoch": 0.9192944949225014,
+      "grad_norm": 1.1253440717297474,
+      "learning_rate": 3.9438634665597165e-07,
+      "loss": 1.2616,
+      "step": 3440
+    },
+    {
+      "epoch": 0.9206306787814004,
+      "grad_norm": 1.1510397092293756,
+      "learning_rate": 3.815182988085153e-07,
+      "loss": 1.2685,
+      "step": 3445
+    },
+    {
+      "epoch": 0.9219668626402993,
+      "grad_norm": 1.1879818159354434,
+      "learning_rate": 3.6885959435301156e-07,
+      "loss": 1.2365,
+      "step": 3450
+    },
+    {
+      "epoch": 0.9233030464991983,
+      "grad_norm": 1.1515610393737254,
+      "learning_rate": 3.564105088023984e-07,
+      "loss": 1.2501,
+      "step": 3455
+    },
+    {
+      "epoch": 0.9246392303580973,
+      "grad_norm": 1.1434666832922344,
+      "learning_rate": 3.441713131073177e-07,
+      "loss": 1.2604,
+      "step": 3460
+    },
+    {
+      "epoch": 0.9259754142169963,
+      "grad_norm": 1.2537303724295639,
+      "learning_rate": 3.3214227365022e-07,
+      "loss": 1.2954,
+      "step": 3465
+    },
+    {
+      "epoch": 0.9273115980758952,
+      "grad_norm": 1.1324300056445142,
+      "learning_rate": 3.2032365223957253e-07,
+      "loss": 1.2553,
+      "step": 3470
+    },
+    {
+      "epoch": 0.9286477819347942,
+      "grad_norm": 1.1623455952898252,
+      "learning_rate": 3.0871570610415124e-07,
+      "loss": 1.2578,
+      "step": 3475
+    },
+    {
+      "epoch": 0.9299839657936932,
+      "grad_norm": 1.1816451537266295,
+      "learning_rate": 2.97318687887449e-07,
+      "loss": 1.2512,
+      "step": 3480
+    },
+    {
+      "epoch": 0.9313201496525922,
+      "grad_norm": 1.1900717072648277,
+      "learning_rate": 2.861328456421775e-07,
+      "loss": 1.275,
+      "step": 3485
+    },
+    {
+      "epoch": 0.9326563335114911,
+      "grad_norm": 1.1275013968941647,
+      "learning_rate": 2.7515842282486274e-07,
+      "loss": 1.2653,
+      "step": 3490
+    },
+    {
+      "epoch": 0.9339925173703901,
+      "grad_norm": 1.1553817658383019,
+      "learning_rate": 2.6439565829055267e-07,
+      "loss": 1.2572,
+      "step": 3495
+    },
+    {
+      "epoch": 0.9353287012292891,
+      "grad_norm": 1.1368557006469668,
+      "learning_rate": 2.5384478628761586e-07,
+      "loss": 1.263,
+      "step": 3500
+    },
+    {
+      "epoch": 0.9366648850881881,
+      "grad_norm": 1.1505999909872684,
+      "learning_rate": 2.435060364526387e-07,
+      "loss": 1.2504,
+      "step": 3505
+    },
+    {
+      "epoch": 0.9380010689470871,
+      "grad_norm": 1.158821621163466,
+      "learning_rate": 2.3337963380543726e-07,
+      "loss": 1.2609,
+      "step": 3510
+    },
+    {
+      "epoch": 0.9393372528059861,
+      "grad_norm": 1.1699141096438583,
+      "learning_rate": 2.234657987441502e-07,
+      "loss": 1.2396,
+      "step": 3515
+    },
+    {
+      "epoch": 0.9406734366648851,
+      "grad_norm": 1.1460631269115045,
+      "learning_rate": 2.1376474704044693e-07,
+      "loss": 1.2503,
+      "step": 3520
+    },
+    {
+      "epoch": 0.9420096205237841,
+      "grad_norm": 1.1207186839548462,
+      "learning_rate": 2.0427668983483361e-07,
+      "loss": 1.246,
+      "step": 3525
+    },
+    {
+      "epoch": 0.943345804382683,
+      "grad_norm": 1.1268601862855379,
+      "learning_rate": 1.9500183363205029e-07,
+      "loss": 1.2367,
+      "step": 3530
+    },
+    {
+      "epoch": 0.944681988241582,
+      "grad_norm": 1.1458026011683493,
+      "learning_rate": 1.85940380296582e-07,
+      "loss": 1.2363,
+      "step": 3535
+    },
+    {
+      "epoch": 0.946018172100481,
+      "grad_norm": 1.126828597537144,
+      "learning_rate": 1.7709252704826485e-07,
+      "loss": 1.2771,
+      "step": 3540
+    },
+    {
+      "epoch": 0.94735435595938,
+      "grad_norm": 1.159174309108889,
+      "learning_rate": 1.6845846645799025e-07,
+      "loss": 1.2846,
+      "step": 3545
+    },
+    {
+      "epoch": 0.948690539818279,
+      "grad_norm": 1.144716451177437,
+      "learning_rate": 1.6003838644351843e-07,
+      "loss": 1.2711,
+      "step": 3550
+    },
+    {
+      "epoch": 0.950026723677178,
+      "grad_norm": 1.2115861792213716,
+      "learning_rate": 1.5183247026538505e-07,
+      "loss": 1.2658,
+      "step": 3555
+    },
+    {
+      "epoch": 0.951362907536077,
+      "grad_norm": 1.1183503620411726,
+      "learning_rate": 1.4384089652291544e-07,
+      "loss": 1.2687,
+      "step": 3560
+    },
+    {
+      "epoch": 0.952699091394976,
+      "grad_norm": 1.1482250830428042,
+      "learning_rate": 1.3606383915033217e-07,
+      "loss": 1.2471,
+      "step": 3565
+    },
+    {
+      "epoch": 0.9540352752538749,
+      "grad_norm": 1.1465209714509406,
+      "learning_rate": 1.2850146741297586e-07,
+      "loss": 1.2776,
+      "step": 3570
+    },
+    {
+      "epoch": 0.9553714591127739,
+      "grad_norm": 1.1496013307054171,
+      "learning_rate": 1.2115394590361595e-07,
+      "loss": 1.2509,
+      "step": 3575
+    },
+    {
+      "epoch": 0.9567076429716729,
+      "grad_norm": 1.1607454473870606,
+      "learning_rate": 1.1402143453887238e-07,
+      "loss": 1.2887,
+      "step": 3580
+    },
+    {
+      "epoch": 0.9580438268305719,
+      "grad_norm": 1.169980738830098,
+      "learning_rate": 1.0710408855573173e-07,
+      "loss": 1.2912,
+      "step": 3585
+    },
+    {
+      "epoch": 0.9593800106894709,
+      "grad_norm": 1.131820961749528,
+      "learning_rate": 1.0040205850817109e-07,
+      "loss": 1.2405,
+      "step": 3590
+    },
+    {
+      "epoch": 0.9607161945483699,
+      "grad_norm": 1.1643105760459518,
+      "learning_rate": 9.391549026387948e-08,
+      "loss": 1.2845,
+      "step": 3595
+    },
+    {
+      "epoch": 0.9620523784072689,
+      "grad_norm": 1.1545962184900476,
+      "learning_rate": 8.764452500108711e-08,
+      "loss": 1.2978,
+      "step": 3600
+    },
+    {
+      "epoch": 0.9620523784072689,
+      "eval_loss": 1.275867223739624,
+      "eval_runtime": 523.9419,
+      "eval_samples_per_second": 25.281,
+      "eval_steps_per_second": 3.161,
+      "step": 3600
+    },
+    {
+      "epoch": 0.9633885622661679,
+      "grad_norm": 1.1593993216116114,
+      "learning_rate": 8.158929920548342e-08,
+      "loss": 1.2391,
+      "step": 3605
+    },
+    {
+      "epoch": 0.9647247461250668,
+      "grad_norm": 1.1553562511912872,
+      "learning_rate": 7.574994466725827e-08,
+      "loss": 1.2495,
+      "step": 3610
+    },
+    {
+      "epoch": 0.9660609299839658,
+      "grad_norm": 1.177204123718693,
+      "learning_rate": 7.012658847822428e-08,
+      "loss": 1.2905,
+      "step": 3615
+    },
+    {
+      "epoch": 0.9673971138428648,
+      "grad_norm": 1.1296251792868242,
+      "learning_rate": 6.471935302905574e-08,
+      "loss": 1.2825,
+      "step": 3620
+    },
+    {
+      "epoch": 0.9687332977017638,
+      "grad_norm": 1.1818608486775184,
+      "learning_rate": 5.952835600662288e-08,
+      "loss": 1.2327,
+      "step": 3625
+    },
+    {
+      "epoch": 0.9700694815606627,
+      "grad_norm": 1.1499671622397054,
+      "learning_rate": 5.455371039143176e-08,
+      "loss": 1.2795,
+      "step": 3630
+    },
+    {
+      "epoch": 0.9714056654195617,
+      "grad_norm": 1.1349247593874663,
+      "learning_rate": 4.9795524455164e-08,
+      "loss": 1.2644,
+      "step": 3635
+    },
+    {
+      "epoch": 0.9727418492784607,
+      "grad_norm": 1.1311236324206604,
+      "learning_rate": 4.52539017583209e-08,
+      "loss": 1.2385,
+      "step": 3640
+    },
+    {
+      "epoch": 0.9740780331373597,
+      "grad_norm": 1.1732868443594786,
+      "learning_rate": 4.0928941147966306e-08,
+      "loss": 1.2908,
+      "step": 3645
+    },
+    {
+      "epoch": 0.9754142169962586,
+      "grad_norm": 1.1619085838301122,
+      "learning_rate": 3.682073675558395e-08,
+      "loss": 1.278,
+      "step": 3650
+    },
+    {
+      "epoch": 0.9767504008551576,
+      "grad_norm": 1.1539440287056817,
+      "learning_rate": 3.2929377995019054e-08,
+      "loss": 1.2386,
+      "step": 3655
+    },
+    {
+      "epoch": 0.9780865847140566,
+      "grad_norm": 1.1170137832632987,
+      "learning_rate": 2.9254949560535428e-08,
+      "loss": 1.2589,
+      "step": 3660
+    },
+    {
+      "epoch": 0.9794227685729556,
+      "grad_norm": 1.1423040789553527,
+      "learning_rate": 2.5797531424976983e-08,
+      "loss": 1.2896,
+      "step": 3665
+    },
+    {
+      "epoch": 0.9807589524318546,
+      "grad_norm": 1.1680612690612637,
+      "learning_rate": 2.2557198838019102e-08,
+      "loss": 1.2404,
+      "step": 3670
+    },
+    {
+      "epoch": 0.9820951362907536,
+      "grad_norm": 1.114027530261495,
+      "learning_rate": 1.9534022324536606e-08,
+      "loss": 1.2802,
+      "step": 3675
+    },
+    {
+      "epoch": 0.9834313201496526,
+      "grad_norm": 1.1636519127752734,
+      "learning_rate": 1.6728067683066117e-08,
+      "loss": 1.2701,
+      "step": 3680
+    },
+    {
+      "epoch": 0.9847675040085516,
+      "grad_norm": 1.1842605033839753,
+      "learning_rate": 1.4139395984377191e-08,
+      "loss": 1.2858,
+      "step": 3685
+    },
+    {
+      "epoch": 0.9861036878674505,
+      "grad_norm": 1.1431541821634774,
+      "learning_rate": 1.1768063570136712e-08,
+      "loss": 1.2662,
+      "step": 3690
+    },
+    {
+      "epoch": 0.9874398717263495,
+      "grad_norm": 1.1466055572304683,
+      "learning_rate": 9.614122051689878e-09,
+      "loss": 1.2773,
+      "step": 3695
+    },
+    {
+      "epoch": 0.9887760555852485,
+      "grad_norm": 1.180031565283721,
+      "learning_rate": 7.67761830893443e-09,
+      "loss": 1.2449,
+      "step": 3700
+    },
+    {
+      "epoch": 0.9901122394441475,
+      "grad_norm": 1.1492882566559037,
+      "learning_rate": 5.958594489295921e-09,
+      "loss": 1.2521,
+      "step": 3705
+    },
+    {
+      "epoch": 0.9914484233030465,
+      "grad_norm": 1.1745365142675883,
+      "learning_rate": 4.457088006816213e-09,
+      "loss": 1.2739,
+      "step": 3710
+    },
+    {
+      "epoch": 0.9927846071619455,
+      "grad_norm": 1.1360910701337543,
+      "learning_rate": 3.173131541338581e-09,
+      "loss": 1.2512,
+      "step": 3715
+    },
+    {
+      "epoch": 0.9941207910208445,
+      "grad_norm": 1.1874182237790705,
+      "learning_rate": 2.1067530377927305e-09,
+      "loss": 1.2587,
+      "step": 3720
+    },
+    {
+      "epoch": 0.9954569748797435,
+      "grad_norm": 1.1682623889051527,
+      "learning_rate": 1.2579757055897202e-09,
+      "loss": 1.2834,
+      "step": 3725
+    },
+    {
+      "epoch": 0.9967931587386424,
+      "grad_norm": 1.1413958118113652,
+      "learning_rate": 6.268180181157047e-10,
+      "loss": 1.2642,
+      "step": 3730
+    },
+    {
+      "epoch": 0.9981293425975414,
+      "grad_norm": 1.1586136947062726,
+      "learning_rate": 2.1329371232892138e-10,
+      "loss": 1.211,
+      "step": 3735
+    },
+    {
+      "epoch": 0.9994655264564404,
+      "grad_norm": 1.196251896759201,
+      "learning_rate": 1.7411788463261858e-11,
+      "loss": 1.2948,
+      "step": 3740
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 3742,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 99431605075968.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}