diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7031 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 200,
+  "global_step": 4852,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00020610057708161583,
+      "grad_norm": 0.7938007287198758,
+      "learning_rate": 4.1152263374485604e-07,
+      "loss": 1.2403,
+      "step": 1
+    },
+    {
+      "epoch": 0.001030502885408079,
+      "grad_norm": 0.9829139995457005,
+      "learning_rate": 2.05761316872428e-06,
+      "loss": 1.5592,
+      "step": 5
+    },
+    {
+      "epoch": 0.002061005770816158,
+      "grad_norm": 1.3104221596563024,
+      "learning_rate": 4.11522633744856e-06,
+      "loss": 1.9165,
+      "step": 10
+    },
+    {
+      "epoch": 0.0030915086562242375,
+      "grad_norm": 0.8263152262533461,
+      "learning_rate": 6.172839506172839e-06,
+      "loss": 1.6148,
+      "step": 15
+    },
+    {
+      "epoch": 0.004122011541632316,
+      "grad_norm": 1.2188669223330737,
+      "learning_rate": 8.23045267489712e-06,
+      "loss": 1.7383,
+      "step": 20
+    },
+    {
+      "epoch": 0.005152514427040396,
+      "grad_norm": 1.3669452124425536,
+      "learning_rate": 1.02880658436214e-05,
+      "loss": 1.8777,
+      "step": 25
+    },
+    {
+      "epoch": 0.006183017312448475,
+      "grad_norm": 0.6877557296913395,
+      "learning_rate": 1.2345679012345678e-05,
+      "loss": 1.3855,
+      "step": 30
+    },
+    {
+      "epoch": 0.007213520197856554,
+      "grad_norm": 0.6292369266751937,
+      "learning_rate": 1.440329218106996e-05,
+      "loss": 1.7622,
+      "step": 35
+    },
+    {
+      "epoch": 0.008244023083264633,
+      "grad_norm": 0.43117531258376596,
+      "learning_rate": 1.646090534979424e-05,
+      "loss": 1.4525,
+      "step": 40
+    },
+    {
+      "epoch": 0.009274525968672712,
+      "grad_norm": 0.3651450490233128,
+      "learning_rate": 1.8518518518518518e-05,
+      "loss": 1.546,
+      "step": 45
+    },
+    {
+      "epoch": 0.010305028854080791,
+      "grad_norm": 0.4844380918063832,
+      "learning_rate": 2.05761316872428e-05,
+      "loss": 1.6499,
+      "step": 50
+    },
+    {
+      "epoch": 0.01133553173948887,
+      "grad_norm": 0.33393658192356523,
+      "learning_rate": 2.2633744855967078e-05,
+      "loss": 1.1817,
+      "step": 55
+    },
+    {
+      "epoch": 0.01236603462489695,
+      "grad_norm": 0.34761501749088797,
+      "learning_rate": 2.4691358024691357e-05,
+      "loss": 1.5655,
+      "step": 60
+    },
+    {
+      "epoch": 0.01339653751030503,
+      "grad_norm": 0.3221134242168732,
+      "learning_rate": 2.6748971193415638e-05,
+      "loss": 1.2608,
+      "step": 65
+    },
+    {
+      "epoch": 0.014427040395713108,
+      "grad_norm": 0.31264722399213424,
+      "learning_rate": 2.880658436213992e-05,
+      "loss": 1.4104,
+      "step": 70
+    },
+    {
+      "epoch": 0.015457543281121188,
+      "grad_norm": 0.3955664926406516,
+      "learning_rate": 3.08641975308642e-05,
+      "loss": 1.4622,
+      "step": 75
+    },
+    {
+      "epoch": 0.016488046166529265,
+      "grad_norm": 0.3149436282257214,
+      "learning_rate": 3.292181069958848e-05,
+      "loss": 1.159,
+      "step": 80
+    },
+    {
+      "epoch": 0.017518549051937345,
+      "grad_norm": 0.2854140231011414,
+      "learning_rate": 3.497942386831276e-05,
+      "loss": 1.4829,
+      "step": 85
+    },
+    {
+      "epoch": 0.018549051937345424,
+      "grad_norm": 0.3326334208099821,
+      "learning_rate": 3.7037037037037037e-05,
+      "loss": 1.1516,
+      "step": 90
+    },
+    {
+      "epoch": 0.019579554822753503,
+      "grad_norm": 0.27192427116328566,
+      "learning_rate": 3.909465020576132e-05,
+      "loss": 1.3785,
+      "step": 95
+    },
+    {
+      "epoch": 0.020610057708161583,
+      "grad_norm": 0.3152208635246729,
+      "learning_rate": 4.11522633744856e-05,
+      "loss": 1.3886,
+      "step": 100
+    },
+    {
+      "epoch": 0.021640560593569662,
+      "grad_norm": 0.27663290557612785,
+      "learning_rate": 4.3209876543209875e-05,
+      "loss": 1.1185,
+      "step": 105
+    },
+    {
+      "epoch": 0.02267106347897774,
+      "grad_norm": 0.27185356829855356,
+      "learning_rate": 4.5267489711934157e-05,
+      "loss": 1.4773,
+      "step": 110
+    },
+    {
+      "epoch": 0.02370156636438582,
+      "grad_norm": 0.2935081378811297,
+      "learning_rate": 4.732510288065844e-05,
+      "loss": 1.261,
+      "step": 115
+    },
+    {
+      "epoch": 0.0247320692497939,
+      "grad_norm": 0.29236576081560334,
+      "learning_rate": 4.938271604938271e-05,
+      "loss": 1.3387,
+      "step": 120
+    },
+    {
+      "epoch": 0.02576257213520198,
+      "grad_norm": 0.38637732181078555,
+      "learning_rate": 5.1440329218106995e-05,
+      "loss": 1.4305,
+      "step": 125
+    },
+    {
+      "epoch": 0.02679307502061006,
+      "grad_norm": 0.2982040676995298,
+      "learning_rate": 5.3497942386831277e-05,
+      "loss": 1.0162,
+      "step": 130
+    },
+    {
+      "epoch": 0.027823577906018138,
+      "grad_norm": 0.2699066491657892,
+      "learning_rate": 5.555555555555556e-05,
+      "loss": 1.4066,
+      "step": 135
+    },
+    {
+      "epoch": 0.028854080791426217,
+      "grad_norm": 0.33203812944453787,
+      "learning_rate": 5.761316872427984e-05,
+      "loss": 1.1751,
+      "step": 140
+    },
+    {
+      "epoch": 0.029884583676834296,
+      "grad_norm": 0.3312647784314266,
+      "learning_rate": 5.9670781893004115e-05,
+      "loss": 1.2269,
+      "step": 145
+    },
+    {
+      "epoch": 0.030915086562242376,
+      "grad_norm": 0.541956195146165,
+      "learning_rate": 6.17283950617284e-05,
+      "loss": 1.354,
+      "step": 150
+    },
+    {
+      "epoch": 0.031945589447650455,
+      "grad_norm": 0.3172355233573687,
+      "learning_rate": 6.378600823045267e-05,
+      "loss": 1.0666,
+      "step": 155
+    },
+    {
+      "epoch": 0.03297609233305853,
+      "grad_norm": 0.32272337262102746,
+      "learning_rate": 6.584362139917696e-05,
+      "loss": 1.4928,
+      "step": 160
+    },
+    {
+      "epoch": 0.03400659521846661,
+      "grad_norm": 0.4252299718796606,
+      "learning_rate": 6.790123456790123e-05,
+      "loss": 1.1979,
+      "step": 165
+    },
+    {
+      "epoch": 0.03503709810387469,
+      "grad_norm": 0.2816942875302398,
+      "learning_rate": 6.995884773662552e-05,
+      "loss": 1.3223,
+      "step": 170
+    },
+    {
+      "epoch": 0.03606760098928277,
+      "grad_norm": 0.34311165874954624,
+      "learning_rate": 7.20164609053498e-05,
+      "loss": 1.4296,
+      "step": 175
+    },
+    {
+      "epoch": 0.03709810387469085,
+      "grad_norm": 0.297834971078222,
+      "learning_rate": 7.407407407407407e-05,
+      "loss": 1.087,
+      "step": 180
+    },
+    {
+      "epoch": 0.03812860676009893,
+      "grad_norm": 0.2602802208757066,
+      "learning_rate": 7.613168724279836e-05,
+      "loss": 1.4754,
+      "step": 185
+    },
+    {
+      "epoch": 0.039159109645507006,
+      "grad_norm": 0.40680201193562965,
+      "learning_rate": 7.818930041152264e-05,
+      "loss": 1.1889,
+      "step": 190
+    },
+    {
+      "epoch": 0.04018961253091509,
+      "grad_norm": 0.45086574163739,
+      "learning_rate": 8.024691358024692e-05,
+      "loss": 1.2851,
+      "step": 195
+    },
+    {
+      "epoch": 0.041220115416323165,
+      "grad_norm": 0.4308650820975275,
+      "learning_rate": 8.23045267489712e-05,
+      "loss": 1.3478,
+      "step": 200
+    },
+    {
+      "epoch": 0.041220115416323165,
+      "eval_loss": 1.2309638261795044,
+      "eval_runtime": 2469.2141,
+      "eval_samples_per_second": 3.24,
+      "eval_steps_per_second": 0.202,
+      "step": 200
+    },
+    {
+      "epoch": 0.04225061830173125,
+      "grad_norm": 0.37845098126001203,
+      "learning_rate": 8.436213991769549e-05,
+      "loss": 1.0707,
+      "step": 205
+    },
+    {
+      "epoch": 0.043281121187139324,
+      "grad_norm": 0.2851215824173796,
+      "learning_rate": 8.641975308641975e-05,
+      "loss": 1.4131,
+      "step": 210
+    },
+    {
+      "epoch": 0.044311624072547406,
+      "grad_norm": 0.3309919261494515,
+      "learning_rate": 8.847736625514404e-05,
+      "loss": 1.1951,
+      "step": 215
+    },
+    {
+      "epoch": 0.04534212695795548,
+      "grad_norm": 0.3065911094811643,
+      "learning_rate": 9.053497942386831e-05,
+      "loss": 1.2554,
+      "step": 220
+    },
+    {
+      "epoch": 0.04637262984336356,
+      "grad_norm": 0.32586015367310234,
+      "learning_rate": 9.25925925925926e-05,
+      "loss": 1.4263,
+      "step": 225
+    },
+    {
+      "epoch": 0.04740313272877164,
+      "grad_norm": 0.277075233563208,
+      "learning_rate": 9.465020576131688e-05,
+      "loss": 1.0216,
+      "step": 230
+    },
+    {
+      "epoch": 0.04843363561417972,
+      "grad_norm": 0.2769072933842902,
+      "learning_rate": 9.670781893004116e-05,
+      "loss": 1.3689,
+      "step": 235
+    },
+    {
+      "epoch": 0.0494641384995878,
+      "grad_norm": 0.27922329154610587,
+      "learning_rate": 9.876543209876543e-05,
+      "loss": 1.2181,
+      "step": 240
+    },
+    {
+      "epoch": 0.050494641384995875,
+      "grad_norm": 0.30982282181292814,
+      "learning_rate": 0.00010082304526748971,
+      "loss": 1.2183,
+      "step": 245
+    },
+    {
+      "epoch": 0.05152514427040396,
+      "grad_norm": 0.33257211954241994,
+      "learning_rate": 0.00010288065843621399,
+      "loss": 1.4305,
+      "step": 250
+    },
+    {
+      "epoch": 0.052555647155812034,
+      "grad_norm": 0.2838288504939334,
+      "learning_rate": 0.00010493827160493828,
+      "loss": 1.0051,
+      "step": 255
+    },
+    {
+      "epoch": 0.05358615004122012,
+      "grad_norm": 0.28623162042613864,
+      "learning_rate": 0.00010699588477366255,
+      "loss": 1.436,
+      "step": 260
+    },
+    {
+      "epoch": 0.05461665292662819,
+      "grad_norm": 0.2720219498268959,
+      "learning_rate": 0.00010905349794238684,
+      "loss": 1.1381,
+      "step": 265
+    },
+    {
+      "epoch": 0.055647155812036275,
+      "grad_norm": 0.30404224920143047,
+      "learning_rate": 0.00011111111111111112,
+      "loss": 1.1905,
+      "step": 270
+    },
+    {
+      "epoch": 0.05667765869744435,
+      "grad_norm": 0.3469618797327675,
+      "learning_rate": 0.0001131687242798354,
+      "loss": 1.3797,
+      "step": 275
+    },
+    {
+      "epoch": 0.057708161582852434,
+      "grad_norm": 0.30848130199190793,
+      "learning_rate": 0.00011522633744855968,
+      "loss": 1.0163,
+      "step": 280
+    },
+    {
+      "epoch": 0.05873866446826051,
+      "grad_norm": 0.2587435053896794,
+      "learning_rate": 0.00011728395061728397,
+      "loss": 1.401,
+      "step": 285
+    },
+    {
+      "epoch": 0.05976916735366859,
+      "grad_norm": 0.30665664697045764,
+      "learning_rate": 0.00011934156378600823,
+      "loss": 1.197,
+      "step": 290
+    },
+    {
+      "epoch": 0.06079967023907667,
+      "grad_norm": 0.28688303187237163,
+      "learning_rate": 0.0001213991769547325,
+      "loss": 1.2817,
+      "step": 295
+    },
+    {
+      "epoch": 0.06183017312448475,
+      "grad_norm": 0.32502880249151705,
+      "learning_rate": 0.0001234567901234568,
+      "loss": 1.388,
+      "step": 300
+    },
+    {
+      "epoch": 0.06286067600989283,
+      "grad_norm": 0.30699533228385806,
+      "learning_rate": 0.00012551440329218108,
+      "loss": 1.0403,
+      "step": 305
+    },
+    {
+      "epoch": 0.06389117889530091,
+      "grad_norm": 0.29902009304688737,
+      "learning_rate": 0.00012757201646090534,
+      "loss": 1.3604,
+      "step": 310
+    },
+    {
+      "epoch": 0.06492168178070899,
+      "grad_norm": 0.26224004465269324,
+      "learning_rate": 0.00012962962962962963,
+      "loss": 1.12,
+      "step": 315
+    },
+    {
+      "epoch": 0.06595218466611706,
+      "grad_norm": 0.2697856637362638,
+      "learning_rate": 0.00013168724279835392,
+      "loss": 1.295,
+      "step": 320
+    },
+    {
+      "epoch": 0.06698268755152514,
+      "grad_norm": 0.3118309921774858,
+      "learning_rate": 0.0001337448559670782,
+      "loss": 1.3498,
+      "step": 325
+    },
+    {
+      "epoch": 0.06801319043693323,
+      "grad_norm": 0.3774735054690596,
+      "learning_rate": 0.00013580246913580247,
+      "loss": 1.1412,
+      "step": 330
+    },
+    {
+      "epoch": 0.0690436933223413,
+      "grad_norm": 0.27963249445347155,
+      "learning_rate": 0.00013786008230452676,
+      "loss": 1.4175,
+      "step": 335
+    },
+    {
+      "epoch": 0.07007419620774938,
+      "grad_norm": 0.25343833022804746,
+      "learning_rate": 0.00013991769547325105,
+      "loss": 1.1228,
+      "step": 340
+    },
+    {
+      "epoch": 0.07110469909315746,
+      "grad_norm": 0.2676785839608282,
+      "learning_rate": 0.00014197530864197534,
+      "loss": 1.2235,
+      "step": 345
+    },
+    {
+      "epoch": 0.07213520197856554,
+      "grad_norm": 0.33130141183815665,
+      "learning_rate": 0.0001440329218106996,
+      "loss": 1.3623,
+      "step": 350
+    },
+    {
+      "epoch": 0.07316570486397361,
+      "grad_norm": 0.2676095204791478,
+      "learning_rate": 0.00014609053497942386,
+      "loss": 1.0355,
+      "step": 355
+    },
+    {
+      "epoch": 0.0741962077493817,
+      "grad_norm": 0.31354711887470305,
+      "learning_rate": 0.00014814814814814815,
+      "loss": 1.3388,
+      "step": 360
+    },
+    {
+      "epoch": 0.07522671063478978,
+      "grad_norm": 0.26048475689786116,
+      "learning_rate": 0.00015020576131687243,
+      "loss": 1.1111,
+      "step": 365
+    },
+    {
+      "epoch": 0.07625721352019786,
+      "grad_norm": 0.3339375055235073,
+      "learning_rate": 0.00015226337448559672,
+      "loss": 1.1734,
+      "step": 370
+    },
+    {
+      "epoch": 0.07728771640560593,
+      "grad_norm": 0.327943444067688,
+      "learning_rate": 0.00015432098765432098,
+      "loss": 1.3208,
+      "step": 375
+    },
+    {
+      "epoch": 0.07831821929101401,
+      "grad_norm": 0.2830129836407308,
+      "learning_rate": 0.00015637860082304527,
+      "loss": 1.0787,
+      "step": 380
+    },
+    {
+      "epoch": 0.0793487221764221,
+      "grad_norm": 0.2751640321887791,
+      "learning_rate": 0.00015843621399176956,
+      "loss": 1.3772,
+      "step": 385
+    },
+    {
+      "epoch": 0.08037922506183018,
+      "grad_norm": 0.27545532948015855,
+      "learning_rate": 0.00016049382716049385,
+      "loss": 1.1672,
+      "step": 390
+    },
+    {
+      "epoch": 0.08140972794723825,
+      "grad_norm": 0.300224560861019,
+      "learning_rate": 0.0001625514403292181,
+      "loss": 1.3224,
+      "step": 395
+    },
+    {
+      "epoch": 0.08244023083264633,
+      "grad_norm": 0.3227759504569782,
+      "learning_rate": 0.0001646090534979424,
+      "loss": 1.3495,
+      "step": 400
+    },
+    {
+      "epoch": 0.08244023083264633,
+      "eval_loss": 1.1826071739196777,
+      "eval_runtime": 2476.8142,
+      "eval_samples_per_second": 3.23,
+      "eval_steps_per_second": 0.202,
+      "step": 400
+    },
+    {
+      "epoch": 0.08347073371805441,
+      "grad_norm": 0.27853687771921737,
+      "learning_rate": 0.0001666666666666667,
+      "loss": 1.1022,
+      "step": 405
+    },
+    {
+      "epoch": 0.0845012366034625,
+      "grad_norm": 0.2685951286639512,
+      "learning_rate": 0.00016872427983539098,
+      "loss": 1.4061,
+      "step": 410
+    },
+    {
+      "epoch": 0.08553173948887056,
+      "grad_norm": 0.2608899320776935,
+      "learning_rate": 0.00017078189300411524,
+      "loss": 1.1174,
+      "step": 415
+    },
+    {
+      "epoch": 0.08656224237427865,
+      "grad_norm": 0.3397550472709015,
+      "learning_rate": 0.0001728395061728395,
+      "loss": 1.2198,
+      "step": 420
+    },
+    {
+      "epoch": 0.08759274525968673,
+      "grad_norm": 0.25033631170694937,
+      "learning_rate": 0.0001748971193415638,
+      "loss": 1.4029,
+      "step": 425
+    },
+    {
+      "epoch": 0.08862324814509481,
+      "grad_norm": 0.3049472388523693,
+      "learning_rate": 0.00017695473251028808,
+      "loss": 1.0265,
+      "step": 430
+    },
+    {
+      "epoch": 0.08965375103050288,
+      "grad_norm": 0.29598713943285077,
+      "learning_rate": 0.00017901234567901234,
+      "loss": 1.3719,
+      "step": 435
+    },
+    {
+      "epoch": 0.09068425391591096,
+      "grad_norm": 0.2656526738754943,
+      "learning_rate": 0.00018106995884773663,
+      "loss": 1.1238,
+      "step": 440
+    },
+    {
+      "epoch": 0.09171475680131905,
+      "grad_norm": 0.2546218728984369,
+      "learning_rate": 0.00018312757201646091,
+      "loss": 1.1853,
+      "step": 445
+    },
+    {
+      "epoch": 0.09274525968672712,
+      "grad_norm": 0.309485029099833,
+      "learning_rate": 0.0001851851851851852,
+      "loss": 1.3064,
+      "step": 450
+    },
+    {
+      "epoch": 0.0937757625721352,
+      "grad_norm": 0.28184488141448555,
+      "learning_rate": 0.00018724279835390946,
+      "loss": 1.0605,
+      "step": 455
+    },
+    {
+      "epoch": 0.09480626545754328,
+      "grad_norm": 0.2913075285039461,
+      "learning_rate": 0.00018930041152263375,
+      "loss": 1.4311,
+      "step": 460
+    },
+    {
+      "epoch": 0.09583676834295136,
+      "grad_norm": 0.7186836779179142,
+      "learning_rate": 0.00019135802469135804,
+      "loss": 1.1263,
+      "step": 465
+    },
+    {
+      "epoch": 0.09686727122835943,
+      "grad_norm": 0.29590695855715343,
+      "learning_rate": 0.00019341563786008233,
+      "loss": 1.2589,
+      "step": 470
+    },
+    {
+      "epoch": 0.09789777411376752,
+      "grad_norm": 0.31832697133586063,
+      "learning_rate": 0.0001954732510288066,
+      "loss": 1.395,
+      "step": 475
+    },
+    {
+      "epoch": 0.0989282769991756,
+      "grad_norm": 0.23627896192430478,
+      "learning_rate": 0.00019753086419753085,
+      "loss": 1.0703,
+      "step": 480
+    },
+    {
+      "epoch": 0.09995877988458368,
+      "grad_norm": 0.27354848483986705,
+      "learning_rate": 0.00019958847736625514,
+      "loss": 1.3205,
+      "step": 485
+    },
+    {
+      "epoch": 0.10098928276999175,
+      "grad_norm": 0.4895153568779187,
+      "learning_rate": 0.00019999958578867407,
+      "loss": 1.1488,
+      "step": 490
+    },
+    {
+      "epoch": 0.10201978565539983,
+      "grad_norm": 0.2811706808521417,
+      "learning_rate": 0.00019999790306104336,
+      "loss": 1.2442,
+      "step": 495
+    },
+    {
+      "epoch": 0.10305028854080792,
+      "grad_norm": 0.2572161259717821,
+      "learning_rate": 0.0001999949259506647,
+      "loss": 1.3355,
+      "step": 500
+    },
+    {
+      "epoch": 0.104080791426216,
+      "grad_norm": 0.41643641721819225,
+      "learning_rate": 0.00019999065449607402,
+      "loss": 1.0238,
+      "step": 505
+    },
+    {
+      "epoch": 0.10511129431162407,
+      "grad_norm": 0.2871689949893079,
+      "learning_rate": 0.00019998508875256158,
+      "loss": 1.3608,
+      "step": 510
+    },
+    {
+      "epoch": 0.10614179719703215,
+      "grad_norm": 0.3219574868373838,
+      "learning_rate": 0.0001999782287921708,
+      "loss": 1.0774,
+      "step": 515
+    },
+    {
+      "epoch": 0.10717230008244023,
+      "grad_norm": 0.23385146626580391,
+      "learning_rate": 0.00019997007470369773,
+      "loss": 1.184,
+      "step": 520
+    },
+    {
+      "epoch": 0.10820280296784832,
+      "grad_norm": 0.3585994112684759,
+      "learning_rate": 0.0001999606265926897,
+      "loss": 1.3426,
+      "step": 525
+    },
+    {
+      "epoch": 0.10923330585325638,
+      "grad_norm": 0.2894131584925935,
+      "learning_rate": 0.000199949884581444,
+      "loss": 1.099,
+      "step": 530
+    },
+    {
+      "epoch": 0.11026380873866447,
+      "grad_norm": 0.2824735567473504,
+      "learning_rate": 0.00019993784880900623,
+      "loss": 1.3997,
+      "step": 535
+    },
+    {
+      "epoch": 0.11129431162407255,
+      "grad_norm": 0.23322094258169807,
+      "learning_rate": 0.0001999245194311687,
+      "loss": 1.1019,
+      "step": 540
+    },
+    {
+      "epoch": 0.11232481450948063,
+      "grad_norm": 0.30614162605617107,
+      "learning_rate": 0.00019990989662046818,
+      "loss": 1.2576,
+      "step": 545
+    },
+    {
+      "epoch": 0.1133553173948887,
+      "grad_norm": 0.3389548425122287,
+      "learning_rate": 0.0001998939805661837,
+      "loss": 1.301,
+      "step": 550
+    },
+    {
+      "epoch": 0.11438582028029678,
+      "grad_norm": 0.25179179489194975,
+      "learning_rate": 0.00019987677147433432,
+      "loss": 0.9871,
+      "step": 555
+    },
+    {
+      "epoch": 0.11541632316570487,
+      "grad_norm": 0.24522921674032325,
+      "learning_rate": 0.0001998582695676762,
+      "loss": 1.4205,
+      "step": 560
+    },
+    {
+      "epoch": 0.11644682605111294,
+      "grad_norm": 0.2528881733347297,
+      "learning_rate": 0.00019983847508569987,
+      "loss": 1.1283,
+      "step": 565
+    },
+    {
+      "epoch": 0.11747732893652102,
+      "grad_norm": 0.28075131598249065,
+      "learning_rate": 0.00019981738828462703,
+      "loss": 1.2467,
+      "step": 570
+    },
+    {
+      "epoch": 0.1185078318219291,
+      "grad_norm": 0.3223992578484773,
+      "learning_rate": 0.00019979500943740735,
+      "loss": 1.3673,
+      "step": 575
+    },
+    {
+      "epoch": 0.11953833470733718,
+      "grad_norm": 0.2804260590851507,
+      "learning_rate": 0.00019977133883371478,
+      "loss": 1.1133,
+      "step": 580
+    },
+    {
+      "epoch": 0.12056883759274525,
+      "grad_norm": 0.25183674337041767,
+      "learning_rate": 0.00019974637677994404,
+      "loss": 1.3493,
+      "step": 585
+    },
+    {
+      "epoch": 0.12159934047815334,
+      "grad_norm": 0.23352143639236672,
+      "learning_rate": 0.00019972012359920638,
+      "loss": 1.1183,
+      "step": 590
+    },
+    {
+      "epoch": 0.12262984336356142,
+      "grad_norm": 0.2597060461259452,
+      "learning_rate": 0.0001996925796313256,
+      "loss": 1.2021,
+      "step": 595
+    },
+    {
+      "epoch": 0.1236603462489695,
+      "grad_norm": 0.31429525704141625,
+      "learning_rate": 0.00019966374523283347,
+      "loss": 1.3753,
+      "step": 600
+    },
+    {
+      "epoch": 0.1236603462489695,
+      "eval_loss": 1.1556562185287476,
+      "eval_runtime": 2465.6561,
+      "eval_samples_per_second": 3.245,
+      "eval_steps_per_second": 0.203,
+      "step": 600
+    },
+    {
+      "epoch": 0.12469084913437757,
+      "grad_norm": 0.23970501230526484,
+      "learning_rate": 0.00019963362077696537,
+      "loss": 0.9978,
+      "step": 605
+    },
+    {
+      "epoch": 0.12572135201978565,
+      "grad_norm": 0.24126751266896648,
+      "learning_rate": 0.00019960220665365518,
+      "loss": 1.3576,
+      "step": 610
+    },
+    {
+      "epoch": 0.12675185490519372,
+      "grad_norm": 0.29806682020411596,
+      "learning_rate": 0.0001995695032695305,
+      "loss": 1.0721,
+      "step": 615
+    },
+    {
+      "epoch": 0.12778235779060182,
+      "grad_norm": 0.3121199170110613,
+      "learning_rate": 0.0001995355110479071,
+      "loss": 1.2159,
+      "step": 620
+    },
+    {
+      "epoch": 0.1288128606760099,
+      "grad_norm": 0.3534465377015386,
+      "learning_rate": 0.00019950023042878366,
+      "loss": 1.3058,
+      "step": 625
+    },
+    {
+      "epoch": 0.12984336356141798,
+      "grad_norm": 0.23203831442914624,
+      "learning_rate": 0.00019946366186883604,
+      "loss": 0.9527,
+      "step": 630
+    },
+    {
+      "epoch": 0.13087386644682605,
+      "grad_norm": 0.2720965062058241,
+      "learning_rate": 0.00019942580584141127,
+      "loss": 1.3558,
+      "step": 635
+    },
+    {
+      "epoch": 0.13190436933223412,
+      "grad_norm": 0.2629592480398592,
+      "learning_rate": 0.0001993866628365215,
+      "loss": 1.1318,
+      "step": 640
+    },
+    {
+      "epoch": 0.13293487221764222,
+      "grad_norm": 0.2920226997136243,
+      "learning_rate": 0.00019934623336083772,
+      "loss": 1.122,
+      "step": 645
+    },
+    {
+      "epoch": 0.1339653751030503,
+      "grad_norm": 0.27168430341439026,
+      "learning_rate": 0.00019930451793768298,
+      "loss": 1.3241,
+      "step": 650
+    },
+    {
+      "epoch": 0.13499587798845836,
+      "grad_norm": 0.2845184130718667,
+      "learning_rate": 0.00019926151710702588,
+      "loss": 0.9734,
+      "step": 655
+    },
+    {
+      "epoch": 0.13602638087386645,
+      "grad_norm": 0.2888242420563682,
+      "learning_rate": 0.00019921723142547347,
+      "loss": 1.3656,
+      "step": 660
+    },
+    {
+      "epoch": 0.13705688375927452,
+      "grad_norm": 0.27299177987397777,
+      "learning_rate": 0.00019917166146626392,
+      "loss": 1.0691,
+      "step": 665
+    },
+    {
+      "epoch": 0.1380873866446826,
+      "grad_norm": 0.35236952647741737,
+      "learning_rate": 0.0001991248078192593,
+      "loss": 1.2263,
+      "step": 670
+    },
+    {
+      "epoch": 0.1391178895300907,
+      "grad_norm": 0.2681417629469303,
+      "learning_rate": 0.00019907667109093794,
+      "loss": 1.354,
+      "step": 675
+    },
+    {
+      "epoch": 0.14014839241549876,
+      "grad_norm": 0.310226199539697,
+      "learning_rate": 0.00019902725190438627,
+      "loss": 1.0365,
+      "step": 680
+    },
+    {
+      "epoch": 0.14117889530090685,
+      "grad_norm": 0.3036115548984946,
+      "learning_rate": 0.00019897655089929126,
+      "loss": 1.3279,
+      "step": 685
+    },
+    {
+      "epoch": 0.14220939818631492,
+      "grad_norm": 0.24182676057547745,
+      "learning_rate": 0.00019892456873193165,
+      "loss": 1.0803,
+      "step": 690
+    },
+    {
+      "epoch": 0.143239901071723,
+      "grad_norm": 0.27410693799977515,
+      "learning_rate": 0.00019887130607516978,
+      "loss": 1.2236,
+      "step": 695
+    },
+    {
+      "epoch": 0.1442704039571311,
+      "grad_norm": 0.3158493448762225,
+      "learning_rate": 0.00019881676361844275,
+      "loss": 1.3796,
+      "step": 700
+    },
+    {
+      "epoch": 0.14530090684253916,
+      "grad_norm": 0.48562045991681607,
+      "learning_rate": 0.0001987609420677535,
+      "loss": 1.0014,
+      "step": 705
+    },
+    {
+      "epoch": 0.14633140972794723,
+      "grad_norm": 0.24730942495184344,
+      "learning_rate": 0.00019870384214566174,
+      "loss": 1.3463,
+      "step": 710
+    },
+    {
+      "epoch": 0.14736191261335532,
+      "grad_norm": 0.3153226513268582,
+      "learning_rate": 0.00019864546459127448,
+      "loss": 1.0977,
+      "step": 715
+    },
+    {
+      "epoch": 0.1483924154987634,
+      "grad_norm": 0.249510122134111,
+      "learning_rate": 0.0001985858101602366,
+      "loss": 1.2687,
+      "step": 720
+    },
+    {
+      "epoch": 0.1494229183841715,
+      "grad_norm": 0.30842159531538627,
+      "learning_rate": 0.0001985248796247209,
+      "loss": 1.2815,
+      "step": 725
+    },
+    {
+      "epoch": 0.15045342126957956,
+      "grad_norm": 0.26379223642130784,
+      "learning_rate": 0.00019846267377341827,
+      "loss": 0.9403,
+      "step": 730
+    },
+    {
+      "epoch": 0.15148392415498763,
+      "grad_norm": 0.23891641481355658,
+      "learning_rate": 0.00019839919341152742,
+      "loss": 1.3505,
+      "step": 735
+    },
+    {
+      "epoch": 0.15251442704039572,
+      "grad_norm": 0.3238473098861904,
+      "learning_rate": 0.00019833443936074442,
+      "loss": 1.0409,
+      "step": 740
+    },
+    {
+      "epoch": 0.1535449299258038,
+      "grad_norm": 0.2746647114109431,
+      "learning_rate": 0.00019826841245925212,
+      "loss": 1.0984,
+      "step": 745
+    },
+    {
+      "epoch": 0.15457543281121186,
+      "grad_norm": 0.37674302298987083,
+      "learning_rate": 0.00019820111356170923,
+      "loss": 1.3147,
+      "step": 750
+    },
+    {
+      "epoch": 0.15560593569661996,
+      "grad_norm": 0.25219340224453657,
+      "learning_rate": 0.00019813254353923937,
+      "loss": 1.0348,
+      "step": 755
+    },
+    {
+      "epoch": 0.15663643858202803,
+      "grad_norm": 0.26735510810071406,
+      "learning_rate": 0.00019806270327941971,
+      "loss": 1.3468,
+      "step": 760
+    },
+    {
+      "epoch": 0.15766694146743612,
+      "grad_norm": 0.25452976189367954,
+      "learning_rate": 0.00019799159368626945,
+      "loss": 1.0699,
+      "step": 765
+    },
+    {
+      "epoch": 0.1586974443528442,
+      "grad_norm": 0.2863967455600455,
+      "learning_rate": 0.00019791921568023822,
+      "loss": 1.1045,
+      "step": 770
+    },
+    {
+      "epoch": 0.15972794723825226,
+      "grad_norm": 0.32564915808035255,
+      "learning_rate": 0.00019784557019819404,
+      "loss": 1.301,
+      "step": 775
+    },
+    {
+      "epoch": 0.16075845012366036,
+      "grad_norm": 0.2598862519671672,
+      "learning_rate": 0.00019777065819341137,
+      "loss": 0.9559,
+      "step": 780
+    },
+    {
+      "epoch": 0.16178895300906843,
+      "grad_norm": 0.2694147600296894,
+      "learning_rate": 0.00019769448063555856,
+      "loss": 1.346,
+      "step": 785
+    },
+    {
+      "epoch": 0.1628194558944765,
+      "grad_norm": 0.2464767798391986,
+      "learning_rate": 0.00019761703851068553,
+      "loss": 1.102,
+      "step": 790
+    },
+    {
+      "epoch": 0.1638499587798846,
+      "grad_norm": 0.2753953016773011,
+      "learning_rate": 0.0001975383328212107,
+      "loss": 1.2188,
+      "step": 795
+    },
+    {
+      "epoch": 0.16488046166529266,
+      "grad_norm": 0.38049304343681456,
+      "learning_rate": 0.00019745836458590836,
+      "loss": 1.3454,
+      "step": 800
+    },
+    {
+      "epoch": 0.16488046166529266,
+      "eval_loss": 1.1297026872634888,
+      "eval_runtime": 2464.1297,
+      "eval_samples_per_second": 3.247,
+      "eval_steps_per_second": 0.203,
+      "step": 800
+    },
+    {
+      "epoch": 0.16591096455070073,
+      "grad_norm": 0.25205784909199425,
+      "learning_rate": 0.0001973771348398953,
+      "loss": 0.9905,
+      "step": 805
+    },
+    {
+      "epoch": 0.16694146743610883,
+      "grad_norm": 0.2719665322536544,
+      "learning_rate": 0.0001972946446346173,
+      "loss": 1.3772,
+      "step": 810
+    },
+    {
+      "epoch": 0.1679719703215169,
+      "grad_norm": 0.23804058554763297,
+      "learning_rate": 0.00019721089503783577,
+      "loss": 1.0834,
+      "step": 815
+    },
+    {
+      "epoch": 0.169002473206925,
+      "grad_norm": 0.28963422898407165,
+      "learning_rate": 0.00019712588713361378,
+      "loss": 1.2064,
+      "step": 820
+    },
+    {
+      "epoch": 0.17003297609233306,
+      "grad_norm": 0.3442324930908095,
+      "learning_rate": 0.00019703962202230203,
+      "loss": 1.2967,
+      "step": 825
+    },
+    {
+      "epoch": 0.17106347897774113,
+      "grad_norm": 0.2547770116672794,
+      "learning_rate": 0.00019695210082052472,
+      "loss": 0.949,
+      "step": 830
+    },
+    {
+      "epoch": 0.17209398186314923,
+      "grad_norm": 0.2832243503202021,
+      "learning_rate": 0.00019686332466116487,
+      "loss": 1.2852,
+      "step": 835
+    },
+    {
+      "epoch": 0.1731244847485573,
+      "grad_norm": 0.2332582520614932,
+      "learning_rate": 0.0001967732946933499,
+      "loss": 0.979,
+      "step": 840
+    },
+    {
+      "epoch": 0.17415498763396536,
+      "grad_norm": 0.29363618940675,
+      "learning_rate": 0.00019668201208243658,
+      "loss": 1.223,
+      "step": 845
+    },
+    {
+      "epoch": 0.17518549051937346,
+      "grad_norm": 0.28680318445015207,
+      "learning_rate": 0.0001965894780099961,
+      "loss": 1.3119,
+      "step": 850
+    },
+    {
+      "epoch": 0.17621599340478153,
+      "grad_norm": 0.2415531891983295,
+      "learning_rate": 0.00019649569367379867,
+      "loss": 0.9811,
+      "step": 855
+    },
+    {
+      "epoch": 0.17724649629018963,
+      "grad_norm": 0.2766629334909396,
+      "learning_rate": 0.00019640066028779794,
+      "loss": 1.2464,
+      "step": 860
+    },
+    {
+      "epoch": 0.1782769991755977,
+      "grad_norm": 0.24460494296561755,
+      "learning_rate": 0.00019630437908211548,
+      "loss": 1.0896,
+      "step": 865
+    },
+    {
+      "epoch": 0.17930750206100576,
+      "grad_norm": 0.28580998835949145,
+      "learning_rate": 0.00019620685130302478,
+      "loss": 1.1636,
+      "step": 870
+    },
+    {
+      "epoch": 0.18033800494641386,
+      "grad_norm": 0.29222255776500833,
+      "learning_rate": 0.00019610807821293503,
+      "loss": 1.3137,
+      "step": 875
+    },
+    {
+      "epoch": 0.18136850783182193,
+      "grad_norm": 0.2684207256549492,
+      "learning_rate": 0.00019600806109037485,
+      "loss": 0.9964,
+      "step": 880
+    },
+    {
+      "epoch": 0.18239901071723,
+      "grad_norm": 0.2559298498236225,
+      "learning_rate": 0.00019590680122997582,
+      "loss": 1.3378,
+      "step": 885
+    },
+    {
+      "epoch": 0.1834295136026381,
+      "grad_norm": 0.2349865929838881,
+      "learning_rate": 0.00019580429994245555,
+      "loss": 1.0436,
+      "step": 890
+    },
+    {
+      "epoch": 0.18446001648804616,
+      "grad_norm": 0.29736696922743316,
+      "learning_rate": 0.0001957005585546009,
+      "loss": 1.1778,
+      "step": 895
+    },
+    {
+      "epoch": 0.18549051937345423,
+      "grad_norm": 0.34641352119598734,
+      "learning_rate": 0.00019559557840925055,
+      "loss": 1.289,
+      "step": 900
+    },
+    {
+      "epoch": 0.18652102225886233,
+      "grad_norm": 0.2589416225643458,
+      "learning_rate": 0.00019548936086527798,
+      "loss": 0.9491,
+      "step": 905
+    },
+    {
+      "epoch": 0.1875515251442704,
+      "grad_norm": 0.2778855204059982,
+      "learning_rate": 0.00019538190729757356,
+      "loss": 1.3701,
+      "step": 910
+    },
+    {
+      "epoch": 0.1885820280296785,
+      "grad_norm": 0.2433946170800378,
+      "learning_rate": 0.00019527321909702688,
+      "loss": 1.0454,
+      "step": 915
+    },
+    {
+      "epoch": 0.18961253091508656,
+      "grad_norm": 0.28491470830869087,
+      "learning_rate": 0.00019516329767050878,
+      "loss": 1.2001,
+      "step": 920
+    },
+    {
+      "epoch": 0.19064303380049463,
+      "grad_norm": 0.30637698470530034,
+      "learning_rate": 0.00019505214444085308,
+      "loss": 1.3096,
+      "step": 925
+    },
+    {
+      "epoch": 0.19167353668590273,
+      "grad_norm": 0.24736218088176834,
+      "learning_rate": 0.00019493976084683813,
+      "loss": 0.9663,
+      "step": 930
+    },
+    {
+      "epoch": 0.1927040395713108,
+      "grad_norm": 0.24983643400010644,
+      "learning_rate": 0.00019482614834316836,
+      "loss": 1.3698,
+      "step": 935
+    },
+    {
+      "epoch": 0.19373454245671887,
+      "grad_norm": 0.2420833122771942,
+      "learning_rate": 0.00019471130840045518,
+      "loss": 1.0335,
+      "step": 940
+    },
+    {
+      "epoch": 0.19476504534212696,
+      "grad_norm": 0.2573293132603339,
+      "learning_rate": 0.00019459524250519826,
+      "loss": 1.1905,
+      "step": 945
+    },
+    {
+      "epoch": 0.19579554822753503,
+      "grad_norm": 0.3710740261808263,
+      "learning_rate": 0.00019447795215976594,
+      "loss": 1.3102,
+      "step": 950
+    },
+    {
+      "epoch": 0.19682605111294313,
+      "grad_norm": 0.2569919144815,
+      "learning_rate": 0.0001943594388823761,
+      "loss": 0.8868,
+      "step": 955
+    },
+    {
+      "epoch": 0.1978565539983512,
+      "grad_norm": 0.3020196940157676,
+      "learning_rate": 0.00019423970420707627,
+      "loss": 1.3362,
+      "step": 960
+    },
+    {
+      "epoch": 0.19888705688375927,
+      "grad_norm": 0.23638629002825132,
+      "learning_rate": 0.00019411874968372402,
+      "loss": 1.0545,
+      "step": 965
+    },
+    {
+      "epoch": 0.19991755976916736,
+      "grad_norm": 0.2690903988670187,
+      "learning_rate": 0.00019399657687796658,
+      "loss": 1.1581,
+      "step": 970
+    },
+    {
+      "epoch": 0.20094806265457543,
+      "grad_norm": 0.3219945588976219,
+      "learning_rate": 0.00019387318737122092,
+      "loss": 1.2816,
+      "step": 975
+    },
+    {
+      "epoch": 0.2019785655399835,
+      "grad_norm": 0.2510582302230213,
+      "learning_rate": 0.0001937485827606529,
+      "loss": 0.9363,
+      "step": 980
+    },
+    {
+      "epoch": 0.2030090684253916,
+      "grad_norm": 0.3075547394305651,
+      "learning_rate": 0.00019362276465915702,
+      "loss": 1.2897,
+      "step": 985
+    },
+    {
+      "epoch": 0.20403957131079967,
+      "grad_norm": 0.24639351565359302,
+      "learning_rate": 0.0001934957346953352,
+      "loss": 1.0314,
+      "step": 990
+    },
+    {
+      "epoch": 0.20507007419620774,
+      "grad_norm": 0.31059587986354126,
+      "learning_rate": 0.00019336749451347586,
+      "loss": 1.2041,
+      "step": 995
+    },
+    {
+      "epoch": 0.20610057708161583,
+      "grad_norm": 0.2842021215906897,
+      "learning_rate": 0.0001932380457735326,
+      "loss": 1.2731,
+      "step": 1000
+    },
+    {
+      "epoch": 0.20610057708161583,
+      "eval_loss": 1.1071134805679321,
+      "eval_runtime": 2476.6697,
+      "eval_samples_per_second": 3.23,
+      "eval_steps_per_second": 0.202,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2071310799670239,
+      "grad_norm": 0.25497821921270564,
+      "learning_rate": 0.00019310739015110267,
+      "loss": 0.9285,
+      "step": 1005
+    },
+    {
+      "epoch": 0.208161582852432,
+      "grad_norm": 0.2793650388701531,
+      "learning_rate": 0.00019297552933740547,
+      "loss": 1.3337,
+      "step": 1010
+    },
+    {
+      "epoch": 0.20919208573784007,
+      "grad_norm": 0.2516697285323032,
+      "learning_rate": 0.0001928424650392603,
+      "loss": 1.0655,
+      "step": 1015
+    },
+    {
+      "epoch": 0.21022258862324814,
+      "grad_norm": 0.30212273915620885,
+      "learning_rate": 0.00019270819897906468,
+      "loss": 1.2153,
+      "step": 1020
+    },
+    {
+      "epoch": 0.21125309150865623,
+      "grad_norm": 0.33419010602150057,
+      "learning_rate": 0.00019257273289477174,
+      "loss": 1.3252,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2122835943940643,
+      "grad_norm": 0.27421309631436125,
+      "learning_rate": 0.00019243606853986786,
+      "loss": 1.0082,
+      "step": 1030
+    },
+    {
+      "epoch": 0.21331409727947237,
+      "grad_norm": 0.34717245381281975,
+      "learning_rate": 0.00019229820768335,
+      "loss": 1.3901,
+      "step": 1035
+    },
+    {
+      "epoch": 0.21434460016488047,
+      "grad_norm": 0.23427202649137227,
+      "learning_rate": 0.00019215915210970267,
+      "loss": 0.9838,
+      "step": 1040
+    },
+    {
+      "epoch": 0.21537510305028854,
+      "grad_norm": 0.2923332885069089,
+      "learning_rate": 0.00019201890361887506,
+      "loss": 1.1746,
+      "step": 1045
+    },
+    {
+      "epoch": 0.21640560593569663,
+      "grad_norm": 0.34553332711604556,
+      "learning_rate": 0.0001918774640262574,
+      "loss": 1.2803,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2174361088211047,
+      "grad_norm": 0.282937796924011,
+      "learning_rate": 0.00019173483516265788,
+      "loss": 1.0297,
+      "step": 1055
+    },
+    {
+      "epoch": 0.21846661170651277,
+      "grad_norm": 0.3064500972204145,
+      "learning_rate": 0.00019159101887427854,
+      "loss": 1.3082,
+      "step": 1060
+    },
+    {
+      "epoch": 0.21949711459192087,
+      "grad_norm": 0.24140887151281765,
+      "learning_rate": 0.00019144601702269162,
+      "loss": 1.0166,
+      "step": 1065
+    },
+    {
+      "epoch": 0.22052761747732894,
+      "grad_norm": 0.2751668782987028,
+      "learning_rate": 0.00019129983148481552,
+      "loss": 1.149,
+      "step": 1070
+    },
+    {
+      "epoch": 0.221558120362737,
+      "grad_norm": 0.32570210618301154,
+      "learning_rate": 0.0001911524641528902,
+      "loss": 1.287,
+      "step": 1075
+    },
+    {
+      "epoch": 0.2225886232481451,
+      "grad_norm": 0.24698689309459554,
+      "learning_rate": 0.00019100391693445306,
+      "loss": 0.9291,
+      "step": 1080
+    },
+    {
+      "epoch": 0.22361912613355317,
+      "grad_norm": 0.26397978494526464,
+      "learning_rate": 0.00019085419175231394,
+      "loss": 1.2573,
+      "step": 1085
+    },
+    {
+      "epoch": 0.22464962901896127,
+      "grad_norm": 0.23809647134840975,
+      "learning_rate": 0.00019070329054453046,
+      "loss": 1.0226,
+      "step": 1090
+    },
+    {
+      "epoch": 0.22568013190436934,
+      "grad_norm": 0.31058086961600134,
+      "learning_rate": 0.00019055121526438272,
+      "loss": 1.1561,
+      "step": 1095
+    },
+    {
+      "epoch": 0.2267106347897774,
+      "grad_norm": 0.2778718312695914,
+      "learning_rate": 0.00019039796788034822,
+      "loss": 1.3034,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2277411376751855,
+      "grad_norm": 0.2417800787196308,
+      "learning_rate": 0.00019024355037607622,
+      "loss": 0.8994,
+      "step": 1105
+    },
+    {
+      "epoch": 0.22877164056059357,
+      "grad_norm": 0.28904453048763723,
+      "learning_rate": 0.00019008796475036222,
+      "loss": 1.2439,
+      "step": 1110
+    },
+    {
+      "epoch": 0.22980214344600164,
+      "grad_norm": 0.23624741886843423,
+      "learning_rate": 0.00018993121301712193,
+      "loss": 1.0735,
+      "step": 1115
+    },
+    {
+      "epoch": 0.23083264633140974,
+      "grad_norm": 0.27985038336320134,
+      "learning_rate": 0.00018977329720536529,
+      "loss": 1.1098,
+      "step": 1120
+    },
+    {
+      "epoch": 0.2318631492168178,
+      "grad_norm": 0.2891758603238283,
+      "learning_rate": 0.00018961421935917016,
+      "loss": 1.2739,
+      "step": 1125
+    },
+    {
+      "epoch": 0.23289365210222587,
+      "grad_norm": 0.2635168830732569,
+      "learning_rate": 0.00018945398153765597,
+      "loss": 1.0068,
+      "step": 1130
+    },
+    {
+      "epoch": 0.23392415498763397,
+      "grad_norm": 0.2773479010066933,
+      "learning_rate": 0.00018929258581495685,
+      "loss": 1.3011,
+      "step": 1135
+    },
+    {
+      "epoch": 0.23495465787304204,
+      "grad_norm": 0.2560481495589164,
+      "learning_rate": 0.00018913003428019506,
+      "loss": 1.0617,
+      "step": 1140
+    },
+    {
+      "epoch": 0.23598516075845014,
+      "grad_norm": 0.277847323563774,
+      "learning_rate": 0.00018896632903745374,
+      "loss": 1.0843,
+      "step": 1145
+    },
+    {
+      "epoch": 0.2370156636438582,
+      "grad_norm": 0.3618488575426716,
+      "learning_rate": 0.00018880147220574976,
+      "loss": 1.2909,
+      "step": 1150
+    },
+    {
+      "epoch": 0.23804616652926627,
+      "grad_norm": 0.2840442439659897,
+      "learning_rate": 0.00018863546591900622,
+      "loss": 0.9916,
+      "step": 1155
+    },
+    {
+      "epoch": 0.23907666941467437,
+      "grad_norm": 0.2904562870162801,
+      "learning_rate": 0.00018846831232602492,
+      "loss": 1.3014,
+      "step": 1160
+    },
+    {
+      "epoch": 0.24010717230008244,
+      "grad_norm": 0.24724715741223574,
+      "learning_rate": 0.00018830001359045845,
+      "loss": 0.9967,
+      "step": 1165
+    },
+    {
+      "epoch": 0.2411376751854905,
+      "grad_norm": 0.26593630753000524,
+      "learning_rate": 0.00018813057189078243,
+      "loss": 1.0626,
+      "step": 1170
+    },
+    {
+      "epoch": 0.2421681780708986,
+      "grad_norm": 0.3276447476001242,
+      "learning_rate": 0.00018795998942026685,
+      "loss": 1.2979,
+      "step": 1175
+    },
+    {
+      "epoch": 0.24319868095630667,
+      "grad_norm": 0.26299318912598185,
+      "learning_rate": 0.00018778826838694812,
+      "loss": 0.9544,
+      "step": 1180
+    },
+    {
+      "epoch": 0.24422918384171477,
+      "grad_norm": 0.281736858661113,
+      "learning_rate": 0.0001876154110136003,
+      "loss": 1.3208,
+      "step": 1185
+    },
+    {
+      "epoch": 0.24525968672712284,
+      "grad_norm": 0.2855862979125554,
+      "learning_rate": 0.0001874414195377063,
+      "loss": 1.073,
+      "step": 1190
+    },
+    {
+      "epoch": 0.2462901896125309,
+      "grad_norm": 0.29014069310260776,
+      "learning_rate": 0.000187266296211429,
+      "loss": 1.1681,
+      "step": 1195
+    },
+    {
+      "epoch": 0.247320692497939,
+      "grad_norm": 0.3159103313554715,
+      "learning_rate": 0.0001870900433015821,
+      "loss": 1.3863,
+      "step": 1200
+    },
+    {
+      "epoch": 0.247320692497939,
+      "eval_loss": 1.0878444910049438,
+      "eval_runtime": 2463.5391,
+      "eval_samples_per_second": 3.247,
+      "eval_steps_per_second": 0.203,
+      "step": 1200
+    },
+    {
+      "epoch": 0.24835119538334707,
+      "grad_norm": 0.2980434519288696,
+      "learning_rate": 0.00018691266308960066,
+      "loss": 1.0099,
+      "step": 1205
+    },
+    {
+      "epoch": 0.24938169826875514,
+      "grad_norm": 0.2786508426961495,
+      "learning_rate": 0.00018673415787151166,
+      "loss": 1.3381,
+      "step": 1210
+    },
+    {
+      "epoch": 0.25041220115416324,
+      "grad_norm": 0.2676203832313212,
+      "learning_rate": 0.00018655452995790435,
+      "loss": 1.0199,
+      "step": 1215
+    },
+    {
+      "epoch": 0.2514427040395713,
+      "grad_norm": 0.28577827654986465,
+      "learning_rate": 0.00018637378167390018,
+      "loss": 1.2001,
+      "step": 1220
+    },
+    {
+      "epoch": 0.2524732069249794,
+      "grad_norm": 0.3398905155731309,
+      "learning_rate": 0.0001861919153591228,
+      "loss": 1.2629,
+      "step": 1225
+    },
+    {
+      "epoch": 0.25350370981038745,
+      "grad_norm": 0.26851195221770446,
+      "learning_rate": 0.00018600893336766786,
+      "loss": 0.9409,
+      "step": 1230
+    },
+    {
+      "epoch": 0.25453421269579557,
+      "grad_norm": 0.26580580512721763,
+      "learning_rate": 0.00018582483806807228,
+      "loss": 1.3071,
+      "step": 1235
+    },
+    {
+      "epoch": 0.25556471558120364,
+      "grad_norm": 0.2571726332870005,
+      "learning_rate": 0.0001856396318432838,
+      "loss": 1.0674,
+      "step": 1240
+    },
+    {
+      "epoch": 0.2565952184666117,
+      "grad_norm": 0.28205467317170896,
+      "learning_rate": 0.0001854533170906302,
+      "loss": 1.1248,
+      "step": 1245
+    },
+    {
+      "epoch": 0.2576257213520198,
+      "grad_norm": 0.3312983996287865,
+      "learning_rate": 0.00018526589622178802,
+      "loss": 1.2796,
+      "step": 1250
+    },
+    {
+      "epoch": 0.25865622423742785,
+      "grad_norm": 0.2757071364626265,
+      "learning_rate": 0.00018507737166275154,
+      "loss": 0.9658,
+      "step": 1255
+    },
+    {
+      "epoch": 0.25968672712283597,
+      "grad_norm": 0.3459316658489597,
+      "learning_rate": 0.00018488774585380125,
+      "loss": 1.3143,
+      "step": 1260
+    },
+    {
+      "epoch": 0.26071723000824404,
+      "grad_norm": 0.2521379464850173,
+      "learning_rate": 0.00018469702124947245,
+      "loss": 0.9855,
+      "step": 1265
+    },
+    {
+      "epoch": 0.2617477328936521,
+      "grad_norm": 0.28337200244179356,
+      "learning_rate": 0.00018450520031852325,
+      "loss": 1.2029,
+      "step": 1270
+    },
+    {
+      "epoch": 0.2627782357790602,
+      "grad_norm": 0.3562233737662029,
+      "learning_rate": 0.0001843122855439027,
+      "loss": 1.2867,
+      "step": 1275
+    },
+    {
+      "epoch": 0.26380873866446825,
+      "grad_norm": 0.2883121950988055,
+      "learning_rate": 0.00018411827942271884,
+      "loss": 0.978,
+      "step": 1280
+    },
+    {
+      "epoch": 0.2648392415498763,
+      "grad_norm": 0.3157197550829857,
+      "learning_rate": 0.000183923184466206,
+      "loss": 1.2767,
+      "step": 1285
+    },
+    {
+      "epoch": 0.26586974443528444,
+      "grad_norm": 0.24013448810718416,
+      "learning_rate": 0.0001837270031996926,
+      "loss": 1.0749,
+      "step": 1290
+    },
+    {
+      "epoch": 0.2669002473206925,
+      "grad_norm": 0.32239639052379476,
+      "learning_rate": 0.00018352973816256838,
+      "loss": 1.1699,
+      "step": 1295
+    },
+    {
+      "epoch": 0.2679307502061006,
+      "grad_norm": 0.29188168243683993,
+      "learning_rate": 0.0001833313919082515,
+      "loss": 1.3297,
+      "step": 1300
+    },
+    {
+      "epoch": 0.26896125309150865,
+      "grad_norm": 0.2710883490462726,
+      "learning_rate": 0.0001831319670041555,
+      "loss": 0.9246,
+      "step": 1305
+    },
+    {
+      "epoch": 0.2699917559769167,
+      "grad_norm": 0.2949025846368701,
+      "learning_rate": 0.00018293146603165603,
+      "loss": 1.339,
+      "step": 1310
+    },
+    {
+      "epoch": 0.27102225886232484,
+      "grad_norm": 0.24960673732959246,
+      "learning_rate": 0.00018272989158605752,
+      "loss": 1.0767,
+      "step": 1315
+    },
+    {
+      "epoch": 0.2720527617477329,
+      "grad_norm": 0.31392425706558585,
+      "learning_rate": 0.00018252724627655954,
+      "loss": 1.1564,
+      "step": 1320
+    },
+    {
+      "epoch": 0.273083264633141,
+      "grad_norm": 0.3755895664237009,
+      "learning_rate": 0.00018232353272622302,
+      "loss": 1.2677,
+      "step": 1325
+    },
+    {
+      "epoch": 0.27411376751854905,
+      "grad_norm": 0.2711084044553726,
+      "learning_rate": 0.00018211875357193632,
+      "loss": 0.9061,
+      "step": 1330
+    },
+    {
+      "epoch": 0.2751442704039571,
+      "grad_norm": 0.33175445691564986,
+      "learning_rate": 0.00018191291146438105,
+      "loss": 1.273,
+      "step": 1335
+    },
+    {
+      "epoch": 0.2761747732893652,
+      "grad_norm": 0.2537090735923007,
+      "learning_rate": 0.0001817060090679978,
+      "loss": 1.0512,
+      "step": 1340
+    },
+    {
+      "epoch": 0.2772052761747733,
+      "grad_norm": 0.36552128539895057,
+      "learning_rate": 0.00018149804906095163,
+      "loss": 1.2273,
+      "step": 1345
+    },
+    {
+      "epoch": 0.2782357790601814,
+      "grad_norm": 0.3485807735804365,
+      "learning_rate": 0.00018128903413509756,
+      "loss": 1.2532,
+      "step": 1350
+    },
+    {
+      "epoch": 0.27926628194558945,
+      "grad_norm": 0.2737156223424568,
+      "learning_rate": 0.0001810789669959453,
+      "loss": 1.0208,
+      "step": 1355
+    },
+    {
+      "epoch": 0.2802967848309975,
+      "grad_norm": 0.3091305751030036,
+      "learning_rate": 0.0001808678503626248,
+      "loss": 1.3024,
+      "step": 1360
+    },
+    {
+      "epoch": 0.2813272877164056,
+      "grad_norm": 0.2448701581737246,
+      "learning_rate": 0.00018065568696785058,
+      "loss": 1.0065,
+      "step": 1365
+    },
+    {
+      "epoch": 0.2823577906018137,
+      "grad_norm": 0.33520109656163827,
+      "learning_rate": 0.00018044247955788662,
+      "loss": 1.1409,
+      "step": 1370
+    },
+    {
+      "epoch": 0.2833882934872218,
+      "grad_norm": 0.36094563501069604,
+      "learning_rate": 0.00018022823089251073,
+      "loss": 1.2687,
+      "step": 1375
+    },
+    {
+      "epoch": 0.28441879637262985,
+      "grad_norm": 0.29785608686411563,
+      "learning_rate": 0.00018001294374497882,
+      "loss": 0.9357,
+      "step": 1380
+    },
+    {
+      "epoch": 0.2854492992580379,
+      "grad_norm": 0.3875323214985989,
+      "learning_rate": 0.00017979662090198906,
+      "loss": 1.2922,
+      "step": 1385
+    },
+    {
+      "epoch": 0.286479802143446,
+      "grad_norm": 0.3058769686259136,
+      "learning_rate": 0.00017957926516364565,
+      "loss": 1.0047,
+      "step": 1390
+    },
+    {
+      "epoch": 0.2875103050288541,
+      "grad_norm": 0.3030215125843954,
+      "learning_rate": 0.00017936087934342283,
+      "loss": 1.0829,
+      "step": 1395
+    },
+    {
+      "epoch": 0.2885408079142622,
+      "grad_norm": 0.32726489328282893,
+      "learning_rate": 0.00017914146626812823,
+      "loss": 1.2567,
+      "step": 1400
+    },
+    {
+      "epoch": 0.2885408079142622,
+      "eval_loss": 1.077731728553772,
+      "eval_runtime": 2463.5195,
+      "eval_samples_per_second": 3.247,
+      "eval_steps_per_second": 0.203,
+      "step": 1400
+    },
+    {
+      "epoch": 0.28957131079967025,
+      "grad_norm": 0.2992981015754632,
+      "learning_rate": 0.0001789210287778664,
+      "loss": 0.9792,
+      "step": 1405
+    },
+    {
+      "epoch": 0.2906018136850783,
+      "grad_norm": 0.3418337934197468,
+      "learning_rate": 0.00017869956972600202,
+      "loss": 1.3454,
+      "step": 1410
+    },
+    {
+      "epoch": 0.2916323165704864,
+      "grad_norm": 0.2526294763625045,
+      "learning_rate": 0.00017847709197912296,
+      "loss": 0.9715,
+      "step": 1415
+    },
+    {
+      "epoch": 0.29266281945589445,
+      "grad_norm": 0.3445050915146703,
+      "learning_rate": 0.0001782535984170032,
+      "loss": 1.113,
+      "step": 1420
+    },
+    {
+      "epoch": 0.2936933223413026,
+      "grad_norm": 0.32389102430563055,
+      "learning_rate": 0.00017802909193256547,
+      "loss": 1.2316,
+      "step": 1425
+    },
+    {
+      "epoch": 0.29472382522671065,
+      "grad_norm": 0.2613877323741928,
+      "learning_rate": 0.00017780357543184397,
+      "loss": 0.9269,
+      "step": 1430
+    },
+    {
+      "epoch": 0.2957543281121187,
+      "grad_norm": 0.2752040315178045,
+      "learning_rate": 0.00017757705183394653,
+      "loss": 1.2481,
+      "step": 1435
+    },
+    {
+      "epoch": 0.2967848309975268,
+      "grad_norm": 0.2535871144663709,
+      "learning_rate": 0.00017734952407101706,
+      "loss": 1.0301,
+      "step": 1440
+    },
+    {
+      "epoch": 0.29781533388293485,
+      "grad_norm": 0.28927983501411936,
+      "learning_rate": 0.0001771209950881974,
+      "loss": 1.1218,
+      "step": 1445
+    },
+    {
+      "epoch": 0.298845836768343,
+      "grad_norm": 0.37005769485403384,
+      "learning_rate": 0.00017689146784358927,
+      "loss": 1.2699,
+      "step": 1450
+    },
+    {
+      "epoch": 0.29987633965375105,
+      "grad_norm": 0.257690573166382,
+      "learning_rate": 0.0001766609453082161,
+      "loss": 0.9097,
+      "step": 1455
+    },
+    {
+      "epoch": 0.3009068425391591,
+      "grad_norm": 0.2708622421638633,
+      "learning_rate": 0.00017642943046598436,
+      "loss": 1.2685,
+      "step": 1460
+    },
+    {
+      "epoch": 0.3019373454245672,
+      "grad_norm": 0.2553661493662262,
+      "learning_rate": 0.00017619692631364506,
+      "loss": 1.062,
+      "step": 1465
+    },
+    {
+      "epoch": 0.30296784830997525,
+      "grad_norm": 0.3268803054784382,
+      "learning_rate": 0.00017596343586075497,
+      "loss": 1.1081,
+      "step": 1470
+    },
+    {
+      "epoch": 0.3039983511953833,
+      "grad_norm": 0.36412223215382217,
+      "learning_rate": 0.00017572896212963754,
+      "loss": 1.3221,
+      "step": 1475
+    },
+    {
+      "epoch": 0.30502885408079145,
+      "grad_norm": 0.2624897168675397,
+      "learning_rate": 0.000175493508155344,
+      "loss": 0.854,
+      "step": 1480
+    },
+    {
+      "epoch": 0.3060593569661995,
+      "grad_norm": 0.2910755462081272,
+      "learning_rate": 0.00017525707698561385,
+      "loss": 1.2961,
+      "step": 1485
+    },
+    {
+      "epoch": 0.3070898598516076,
+      "grad_norm": 0.26020570413888594,
+      "learning_rate": 0.00017501967168083557,
+      "loss": 1.0475,
+      "step": 1490
+    },
+    {
+      "epoch": 0.30812036273701565,
+      "grad_norm": 0.31355952905674644,
+      "learning_rate": 0.00017478129531400688,
+      "loss": 1.1383,
+      "step": 1495
+    },
+    {
+      "epoch": 0.3091508656224237,
+      "grad_norm": 0.3394529288865745,
+      "learning_rate": 0.00017454195097069505,
+      "loss": 1.3237,
+      "step": 1500
+    },
+    {
+      "epoch": 0.31018136850783184,
+      "grad_norm": 0.2623708218727371,
+      "learning_rate": 0.00017430164174899696,
+      "loss": 0.8949,
+      "step": 1505
+    },
+    {
+      "epoch": 0.3112118713932399,
+      "grad_norm": 0.3129719424005796,
+      "learning_rate": 0.0001740603707594989,
+      "loss": 1.2513,
+      "step": 1510
+    },
+    {
+      "epoch": 0.312242374278648,
+      "grad_norm": 0.2457430728630137,
+      "learning_rate": 0.00017381814112523648,
+      "loss": 1.0458,
+      "step": 1515
+    },
+    {
+      "epoch": 0.31327287716405605,
+      "grad_norm": 0.3235157301586437,
+      "learning_rate": 0.000173574955981654,
+      "loss": 1.1418,
+      "step": 1520
+    },
+    {
+      "epoch": 0.3143033800494641,
+      "grad_norm": 0.38068266969796094,
+      "learning_rate": 0.00017333081847656397,
+      "loss": 1.2703,
+      "step": 1525
+    },
+    {
+      "epoch": 0.31533388293487224,
+      "grad_norm": 0.2845050833721142,
+      "learning_rate": 0.00017308573177010652,
+      "loss": 0.8513,
+      "step": 1530
+    },
+    {
+      "epoch": 0.3163643858202803,
+      "grad_norm": 0.3011956997232069,
+      "learning_rate": 0.00017283969903470815,
+      "loss": 1.2671,
+      "step": 1535
+    },
+    {
+      "epoch": 0.3173948887056884,
+      "grad_norm": 0.2548678800376021,
+      "learning_rate": 0.0001725927234550409,
+      "loss": 0.9834,
+      "step": 1540
+    },
+    {
+      "epoch": 0.31842539159109645,
+      "grad_norm": 0.33070817321371104,
+      "learning_rate": 0.00017234480822798113,
+      "loss": 1.204,
+      "step": 1545
+    },
+    {
+      "epoch": 0.3194558944765045,
+      "grad_norm": 0.3739990601669961,
+      "learning_rate": 0.00017209595656256807,
+      "loss": 1.2874,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3204863973619126,
+      "grad_norm": 0.28630568916559335,
+      "learning_rate": 0.0001718461716799623,
+      "loss": 0.9404,
+      "step": 1555
+    },
+    {
+      "epoch": 0.3215169002473207,
+      "grad_norm": 0.2984332733087229,
+      "learning_rate": 0.000171595456813404,
+      "loss": 1.3065,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3225474031327288,
+      "grad_norm": 0.25297494310840185,
+      "learning_rate": 0.00017134381520817127,
+      "loss": 0.9533,
+      "step": 1565
+    },
+    {
+      "epoch": 0.32357790601813685,
+      "grad_norm": 0.3035622251707662,
+      "learning_rate": 0.00017109125012153783,
+      "loss": 1.1281,
+      "step": 1570
+    },
+    {
+      "epoch": 0.3246084089035449,
+      "grad_norm": 0.3401968875424911,
+      "learning_rate": 0.00017083776482273126,
+      "loss": 1.2439,
+      "step": 1575
+    },
+    {
+      "epoch": 0.325638911788953,
+      "grad_norm": 0.2742633328984466,
+      "learning_rate": 0.00017058336259289026,
+      "loss": 0.9067,
+      "step": 1580
+    },
+    {
+      "epoch": 0.3266694146743611,
+      "grad_norm": 0.33201726031181766,
+      "learning_rate": 0.0001703280467250225,
+      "loss": 1.2643,
+      "step": 1585
+    },
+    {
+      "epoch": 0.3276999175597692,
+      "grad_norm": 0.2740668506157443,
+      "learning_rate": 0.0001700718205239618,
+      "loss": 0.9757,
+      "step": 1590
+    },
+    {
+      "epoch": 0.32873042044517725,
+      "grad_norm": 0.28447062660977135,
+      "learning_rate": 0.0001698146873063255,
+      "loss": 1.0629,
+      "step": 1595
+    },
+    {
+      "epoch": 0.3297609233305853,
+      "grad_norm": 0.3855059276442332,
+      "learning_rate": 0.00016955665040047134,
+      "loss": 1.257,
+      "step": 1600
+    },
+    {
+      "epoch": 0.3297609233305853,
+      "eval_loss": 1.0629926919937134,
+      "eval_runtime": 2467.5248,
+      "eval_samples_per_second": 3.242,
+      "eval_steps_per_second": 0.203,
+      "step": 1600
+    },
+    {
+      "epoch": 0.3307914262159934,
+      "grad_norm": 0.30766662381135884,
+      "learning_rate": 0.00016929771314645454,
+      "loss": 0.9716,
+      "step": 1605
+    },
+    {
+      "epoch": 0.33182192910140146,
+      "grad_norm": 0.3070232969296392,
+      "learning_rate": 0.00016903787889598458,
+      "loss": 1.3027,
+      "step": 1610
+    },
+    {
+      "epoch": 0.3328524319868096,
+      "grad_norm": 0.29718017467704144,
+      "learning_rate": 0.00016877715101238172,
+      "loss": 1.053,
+      "step": 1615
+    },
+    {
+      "epoch": 0.33388293487221765,
+      "grad_norm": 0.29124927902233755,
+      "learning_rate": 0.00016851553287053342,
+      "loss": 1.1027,
+      "step": 1620
+    },
+    {
+      "epoch": 0.3349134377576257,
+      "grad_norm": 0.37490768596522844,
+      "learning_rate": 0.00016825302785685077,
+      "loss": 1.2816,
+      "step": 1625
+    },
+    {
+      "epoch": 0.3359439406430338,
+      "grad_norm": 0.2909222245308242,
+      "learning_rate": 0.00016798963936922467,
+      "loss": 0.9477,
+      "step": 1630
+    },
+    {
+      "epoch": 0.33697444352844186,
+      "grad_norm": 0.369759262778725,
+      "learning_rate": 0.00016772537081698175,
+      "loss": 1.2659,
+      "step": 1635
+    },
+    {
+      "epoch": 0.33800494641385,
+      "grad_norm": 0.25527189055893823,
+      "learning_rate": 0.00016746022562084026,
+      "loss": 1.0374,
+      "step": 1640
+    },
+    {
+      "epoch": 0.33903544929925805,
+      "grad_norm": 0.332617684071012,
+      "learning_rate": 0.0001671942072128659,
+      "loss": 1.1066,
+      "step": 1645
+    },
+    {
+      "epoch": 0.3400659521846661,
+      "grad_norm": 0.3184886656626706,
+      "learning_rate": 0.00016692731903642725,
+      "loss": 1.2649,
+      "step": 1650
+    },
+    {
+      "epoch": 0.3410964550700742,
+      "grad_norm": 0.2799293838329678,
+      "learning_rate": 0.0001666595645461512,
+      "loss": 0.9379,
+      "step": 1655
+    },
+    {
+      "epoch": 0.34212695795548226,
+      "grad_norm": 0.3517762307499943,
+      "learning_rate": 0.0001663909472078784,
+      "loss": 1.252,
+      "step": 1660
+    },
+    {
+      "epoch": 0.3431574608408903,
+      "grad_norm": 0.25384775837778967,
+      "learning_rate": 0.0001661214704986182,
+      "loss": 0.9847,
+      "step": 1665
+    },
+    {
+      "epoch": 0.34418796372629845,
+      "grad_norm": 0.29743299437921455,
+      "learning_rate": 0.00016585113790650388,
+      "loss": 1.1201,
+      "step": 1670
+    },
+    {
+      "epoch": 0.3452184666117065,
+      "grad_norm": 0.35589431095248486,
+      "learning_rate": 0.00016557995293074715,
+      "loss": 1.2206,
+      "step": 1675
+    },
+    {
+      "epoch": 0.3462489694971146,
+      "grad_norm": 0.2911495340019745,
+      "learning_rate": 0.00016530791908159323,
+      "loss": 0.9072,
+      "step": 1680
+    },
+    {
+      "epoch": 0.34727947238252266,
+      "grad_norm": 0.28789636914166633,
+      "learning_rate": 0.0001650350398802751,
+      "loss": 1.2946,
+      "step": 1685
+    },
+    {
+      "epoch": 0.3483099752679307,
+      "grad_norm": 0.26335155835600166,
+      "learning_rate": 0.0001647613188589682,
+      "loss": 1.0242,
+      "step": 1690
+    },
+    {
+      "epoch": 0.34934047815333885,
+      "grad_norm": 0.3325623900144522,
+      "learning_rate": 0.00016448675956074444,
+      "loss": 1.0931,
+      "step": 1695
+    },
+    {
+      "epoch": 0.3503709810387469,
+      "grad_norm": 0.3773837775653675,
+      "learning_rate": 0.0001642113655395266,
+      "loss": 1.2456,
+      "step": 1700
+    },
+    {
+      "epoch": 0.351401483924155,
+      "grad_norm": 0.285753347787117,
+      "learning_rate": 0.00016393514036004204,
+      "loss": 0.9662,
+      "step": 1705
+    },
+    {
+      "epoch": 0.35243198680956306,
+      "grad_norm": 0.3046613746172406,
+      "learning_rate": 0.0001636580875977769,
+      "loss": 1.3162,
+      "step": 1710
+    },
+    {
+      "epoch": 0.3534624896949711,
+      "grad_norm": 0.25099480129499735,
+      "learning_rate": 0.0001633802108389295,
+      "loss": 1.0061,
+      "step": 1715
+    },
+    {
+      "epoch": 0.35449299258037925,
+      "grad_norm": 0.34150662422413947,
+      "learning_rate": 0.00016310151368036408,
+      "loss": 1.1411,
+      "step": 1720
+    },
+    {
+      "epoch": 0.3555234954657873,
+      "grad_norm": 0.33392898782461256,
+      "learning_rate": 0.00016282199972956425,
+      "loss": 1.2154,
+      "step": 1725
+    },
+    {
+      "epoch": 0.3565539983511954,
+      "grad_norm": 0.3016273908980527,
+      "learning_rate": 0.00016254167260458622,
+      "loss": 0.9269,
+      "step": 1730
+    },
+    {
+      "epoch": 0.35758450123660346,
+      "grad_norm": 0.3297238572031139,
+      "learning_rate": 0.000162260535934012,
+      "loss": 1.2298,
+      "step": 1735
+    },
+    {
+      "epoch": 0.3586150041220115,
+      "grad_norm": 0.26311212102325543,
+      "learning_rate": 0.00016197859335690247,
+      "loss": 1.0416,
+      "step": 1740
+    },
+    {
+      "epoch": 0.3596455070074196,
+      "grad_norm": 0.35401650365663206,
+      "learning_rate": 0.0001616958485227503,
+      "loss": 1.0915,
+      "step": 1745
+    },
+    {
+      "epoch": 0.3606760098928277,
+      "grad_norm": 0.3946861144577044,
+      "learning_rate": 0.0001614123050914325,
+      "loss": 1.2156,
+      "step": 1750
+    },
+    {
+      "epoch": 0.3617065127782358,
+      "grad_norm": 0.31649222443853475,
+      "learning_rate": 0.00016112796673316336,
+      "loss": 0.9126,
+      "step": 1755
+    },
+    {
+      "epoch": 0.36273701566364386,
+      "grad_norm": 0.3591291538700441,
+      "learning_rate": 0.00016084283712844666,
+      "loss": 1.2438,
+      "step": 1760
+    },
+    {
+      "epoch": 0.3637675185490519,
+      "grad_norm": 0.2567906575863548,
+      "learning_rate": 0.00016055691996802823,
+      "loss": 0.9908,
+      "step": 1765
+    },
+    {
+      "epoch": 0.36479802143446,
+      "grad_norm": 0.34039693260231785,
+      "learning_rate": 0.00016027021895284808,
+      "loss": 1.0576,
+      "step": 1770
+    },
+    {
+      "epoch": 0.3658285243198681,
+      "grad_norm": 0.45929799156275763,
+      "learning_rate": 0.0001599827377939925,
+      "loss": 1.3099,
+      "step": 1775
+    },
+    {
+      "epoch": 0.3668590272052762,
+      "grad_norm": 0.2736394916456097,
+      "learning_rate": 0.00015969448021264606,
+      "loss": 0.8622,
+      "step": 1780
+    },
+    {
+      "epoch": 0.36788953009068426,
+      "grad_norm": 0.2884997182361836,
+      "learning_rate": 0.00015940544994004334,
+      "loss": 1.2272,
+      "step": 1785
+    },
+    {
+      "epoch": 0.3689200329760923,
+      "grad_norm": 0.2392312698637367,
+      "learning_rate": 0.00015911565071742088,
+      "loss": 0.9626,
+      "step": 1790
+    },
+    {
+      "epoch": 0.3699505358615004,
+      "grad_norm": 0.339213144942226,
+      "learning_rate": 0.00015882508629596836,
+      "loss": 1.1233,
+      "step": 1795
+    },
+    {
+      "epoch": 0.37098103874690846,
+      "grad_norm": 0.40881279626532063,
+      "learning_rate": 0.00015853376043678053,
+      "loss": 1.2129,
+      "step": 1800
+    },
+    {
+      "epoch": 0.37098103874690846,
+      "eval_loss": 1.051790475845337,
+      "eval_runtime": 2468.3063,
+      "eval_samples_per_second": 3.241,
+      "eval_steps_per_second": 0.203,
+      "step": 1800
+    },
+    {
+      "epoch": 0.3720115416323166,
+      "grad_norm": 0.27225360643392454,
+      "learning_rate": 0.00015824167691080802,
+      "loss": 0.8599,
+      "step": 1805
+    },
+    {
+      "epoch": 0.37304204451772466,
+      "grad_norm": 0.3655053849913508,
+      "learning_rate": 0.00015794883949880894,
+      "loss": 1.2174,
+      "step": 1810
+    },
+    {
+      "epoch": 0.3740725474031327,
+      "grad_norm": 0.2810810355550597,
+      "learning_rate": 0.00015765525199129966,
+      "loss": 1.0166,
+      "step": 1815
+    },
+    {
+      "epoch": 0.3751030502885408,
+      "grad_norm": 0.3332282628999955,
+      "learning_rate": 0.000157360918188506,
+      "loss": 1.1182,
+      "step": 1820
+    },
+    {
+      "epoch": 0.37613355317394886,
+      "grad_norm": 0.3175764657865911,
+      "learning_rate": 0.0001570658419003137,
+      "loss": 1.2687,
+      "step": 1825
+    },
+    {
+      "epoch": 0.377164056059357,
+      "grad_norm": 0.3120363543994925,
+      "learning_rate": 0.00015677002694621948,
+      "loss": 0.9812,
+      "step": 1830
+    },
+    {
+      "epoch": 0.37819455894476506,
+      "grad_norm": 0.327602667442098,
+      "learning_rate": 0.00015647347715528137,
+      "loss": 1.1919,
+      "step": 1835
+    },
+    {
+      "epoch": 0.3792250618301731,
+      "grad_norm": 0.2668526402245522,
+      "learning_rate": 0.00015617619636606924,
+      "loss": 0.9995,
+      "step": 1840
+    },
+    {
+      "epoch": 0.3802555647155812,
+      "grad_norm": 0.31419704503426554,
+      "learning_rate": 0.00015587818842661494,
+      "loss": 1.1765,
+      "step": 1845
+    },
+    {
+      "epoch": 0.38128606760098926,
+      "grad_norm": 0.4282942251054684,
+      "learning_rate": 0.00015557945719436278,
+      "loss": 1.2522,
+      "step": 1850
+    },
+    {
+      "epoch": 0.3823165704863974,
+      "grad_norm": 0.308069434736153,
+      "learning_rate": 0.00015528000653611935,
+      "loss": 0.8995,
+      "step": 1855
+    },
+    {
+      "epoch": 0.38334707337180546,
+      "grad_norm": 0.272826512737121,
+      "learning_rate": 0.0001549798403280036,
+      "loss": 1.2827,
+      "step": 1860
+    },
+    {
+      "epoch": 0.3843775762572135,
+      "grad_norm": 0.2498018349436216,
+      "learning_rate": 0.0001546789624553966,
+      "loss": 1.0308,
+      "step": 1865
+    },
+    {
+      "epoch": 0.3854080791426216,
+      "grad_norm": 0.32988195624784494,
+      "learning_rate": 0.00015437737681289128,
+      "loss": 1.1726,
+      "step": 1870
+    },
+    {
+      "epoch": 0.38643858202802966,
+      "grad_norm": 0.38213734379034847,
+      "learning_rate": 0.00015407508730424206,
+      "loss": 1.2046,
+      "step": 1875
+    },
+    {
+      "epoch": 0.38746908491343773,
+      "grad_norm": 0.2839035313263252,
+      "learning_rate": 0.00015377209784231424,
+      "loss": 0.9713,
+      "step": 1880
+    },
+    {
+      "epoch": 0.38849958779884586,
+      "grad_norm": 0.26153622554750505,
+      "learning_rate": 0.00015346841234903337,
+      "loss": 1.2771,
+      "step": 1885
+    },
+    {
+      "epoch": 0.3895300906842539,
+      "grad_norm": 0.25504870927422546,
+      "learning_rate": 0.0001531640347553345,
+      "loss": 1.0141,
+      "step": 1890
+    },
+    {
+      "epoch": 0.390560593569662,
+      "grad_norm": 0.33076548230805997,
+      "learning_rate": 0.00015285896900111133,
+      "loss": 1.1153,
+      "step": 1895
+    },
+    {
+      "epoch": 0.39159109645507006,
+      "grad_norm": 0.3682940168037353,
+      "learning_rate": 0.00015255321903516516,
+      "loss": 1.2387,
+      "step": 1900
+    },
+    {
+      "epoch": 0.39262159934047813,
+      "grad_norm": 0.3030559581752516,
+      "learning_rate": 0.00015224678881515377,
+      "loss": 0.8932,
+      "step": 1905
+    },
+    {
+      "epoch": 0.39365210222588626,
+      "grad_norm": 0.3016417378121846,
+      "learning_rate": 0.00015193968230754024,
+      "loss": 1.2566,
+      "step": 1910
+    },
+    {
+      "epoch": 0.3946826051112943,
+      "grad_norm": 0.2562267990440701,
+      "learning_rate": 0.00015163190348754162,
+      "loss": 1.0584,
+      "step": 1915
+    },
+    {
+      "epoch": 0.3957131079967024,
+      "grad_norm": 0.3152397653362513,
+      "learning_rate": 0.00015132345633907734,
+      "loss": 1.1119,
+      "step": 1920
+    },
+    {
+      "epoch": 0.39674361088211046,
+      "grad_norm": 0.3550077716096075,
+      "learning_rate": 0.0001510143448547178,
+      "loss": 1.2196,
+      "step": 1925
+    },
+    {
+      "epoch": 0.39777411376751853,
+      "grad_norm": 0.2713673784146358,
+      "learning_rate": 0.00015070457303563268,
+      "loss": 0.8809,
+      "step": 1930
+    },
+    {
+      "epoch": 0.3988046166529266,
+      "grad_norm": 0.32482913652095136,
+      "learning_rate": 0.000150394144891539,
+      "loss": 1.2114,
+      "step": 1935
+    },
+    {
+      "epoch": 0.3998351195383347,
+      "grad_norm": 0.2681392320249827,
+      "learning_rate": 0.00015008306444064942,
+      "loss": 1.0092,
+      "step": 1940
+    },
+    {
+      "epoch": 0.4008656224237428,
+      "grad_norm": 0.30813511160423507,
+      "learning_rate": 0.00014977133570961997,
+      "loss": 1.0587,
+      "step": 1945
+    },
+    {
+      "epoch": 0.40189612530915086,
+      "grad_norm": 0.3045685029840853,
+      "learning_rate": 0.00014945896273349827,
+      "loss": 1.3074,
+      "step": 1950
+    },
+    {
+      "epoch": 0.40292662819455893,
+      "grad_norm": 0.3178744984824384,
+      "learning_rate": 0.00014914594955567099,
+      "loss": 0.9477,
+      "step": 1955
+    },
+    {
+      "epoch": 0.403957131079967,
+      "grad_norm": 0.36931867482484526,
+      "learning_rate": 0.00014883230022781163,
+      "loss": 1.2359,
+      "step": 1960
+    },
+    {
+      "epoch": 0.4049876339653751,
+      "grad_norm": 0.27801383031198734,
+      "learning_rate": 0.00014851801880982814,
+      "loss": 1.0366,
+      "step": 1965
+    },
+    {
+      "epoch": 0.4060181368507832,
+      "grad_norm": 0.31213479911432085,
+      "learning_rate": 0.00014820310936981026,
+      "loss": 1.1624,
+      "step": 1970
+    },
+    {
+      "epoch": 0.40704863973619126,
+      "grad_norm": 0.33227538300256093,
+      "learning_rate": 0.000147887575983977,
+      "loss": 1.2724,
+      "step": 1975
+    },
+    {
+      "epoch": 0.40807914262159933,
+      "grad_norm": 0.298687515913839,
+      "learning_rate": 0.00014757142273662358,
+      "loss": 0.8436,
+      "step": 1980
+    },
+    {
+      "epoch": 0.4091096455070074,
+      "grad_norm": 0.36351887792750426,
+      "learning_rate": 0.00014725465372006905,
+      "loss": 1.2507,
+      "step": 1985
+    },
+    {
+      "epoch": 0.41014014839241547,
+      "grad_norm": 0.2702448610300903,
+      "learning_rate": 0.0001469372730346028,
+      "loss": 1.0191,
+      "step": 1990
+    },
+    {
+      "epoch": 0.4111706512778236,
+      "grad_norm": 0.3224079902083336,
+      "learning_rate": 0.00014661928478843186,
+      "loss": 1.1598,
+      "step": 1995
+    },
+    {
+      "epoch": 0.41220115416323166,
+      "grad_norm": 0.41798586070796323,
+      "learning_rate": 0.00014630069309762753,
+      "loss": 1.1939,
+      "step": 2000
+    },
+    {
+      "epoch": 0.41220115416323166,
+      "eval_loss": 1.0405410528182983,
+      "eval_runtime": 2474.052,
+      "eval_samples_per_second": 3.234,
+      "eval_steps_per_second": 0.202,
+      "step": 2000
+    },
+    {
+      "epoch": 0.41323165704863973,
+      "grad_norm": 0.28365317528526945,
+      "learning_rate": 0.00014598150208607212,
+      "loss": 0.8639,
+      "step": 2005
+    },
+    {
+      "epoch": 0.4142621599340478,
+      "grad_norm": 0.3214047753618349,
+      "learning_rate": 0.00014566171588540572,
+      "loss": 1.2183,
+      "step": 2010
+    },
+    {
+      "epoch": 0.41529266281945587,
+      "grad_norm": 0.2644278825870432,
+      "learning_rate": 0.00014534133863497252,
+      "loss": 0.9564,
+      "step": 2015
+    },
+    {
+      "epoch": 0.416323165704864,
+      "grad_norm": 0.32226077969266037,
+      "learning_rate": 0.00014502037448176734,
+      "loss": 1.0942,
+      "step": 2020
+    },
+    {
+      "epoch": 0.41735366859027206,
+      "grad_norm": 0.4101159870787422,
+      "learning_rate": 0.00014469882758038193,
+      "loss": 1.244,
+      "step": 2025
+    },
+    {
+      "epoch": 0.41838417147568013,
+      "grad_norm": 0.28153576445753414,
+      "learning_rate": 0.00014437670209295112,
+      "loss": 0.8947,
+      "step": 2030
+    },
+    {
+      "epoch": 0.4194146743610882,
+      "grad_norm": 0.3785396934306716,
+      "learning_rate": 0.0001440540021890992,
+      "loss": 1.2432,
+      "step": 2035
+    },
+    {
+      "epoch": 0.42044517724649627,
+      "grad_norm": 0.2542390029018319,
+      "learning_rate": 0.00014373073204588556,
+      "loss": 1.0083,
+      "step": 2040
+    },
+    {
+      "epoch": 0.4214756801319044,
+      "grad_norm": 0.3123729353287892,
+      "learning_rate": 0.0001434068958477509,
+      "loss": 1.1195,
+      "step": 2045
+    },
+    {
+      "epoch": 0.42250618301731246,
+      "grad_norm": 0.39530507310232316,
+      "learning_rate": 0.00014308249778646306,
+      "loss": 1.2036,
+      "step": 2050
+    },
+    {
+      "epoch": 0.42353668590272053,
+      "grad_norm": 0.2818105109815067,
+      "learning_rate": 0.0001427575420610626,
+      "loss": 0.8847,
+      "step": 2055
+    },
+    {
+      "epoch": 0.4245671887881286,
+      "grad_norm": 0.3219044657747011,
+      "learning_rate": 0.00014243203287780856,
+      "loss": 1.2623,
+      "step": 2060
+    },
+    {
+      "epoch": 0.42559769167353667,
+      "grad_norm": 0.2611111361591686,
+      "learning_rate": 0.00014210597445012398,
+      "loss": 1.0083,
+      "step": 2065
+    },
+    {
+      "epoch": 0.42662819455894474,
+      "grad_norm": 0.33485314347370027,
+      "learning_rate": 0.0001417793709985415,
+      "loss": 1.1823,
+      "step": 2070
+    },
+    {
+      "epoch": 0.42765869744435286,
+      "grad_norm": 0.34925710089096873,
+      "learning_rate": 0.0001414522267506484,
+      "loss": 1.2728,
+      "step": 2075
+    },
+    {
+      "epoch": 0.42868920032976093,
+      "grad_norm": 0.25546314251507235,
+      "learning_rate": 0.0001411245459410322,
+      "loss": 0.8703,
+      "step": 2080
+    },
+    {
+      "epoch": 0.429719703215169,
+      "grad_norm": 0.38324554510188286,
+      "learning_rate": 0.00014079633281122573,
+      "loss": 1.2551,
+      "step": 2085
+    },
+    {
+      "epoch": 0.43075020610057707,
+      "grad_norm": 0.27810663295608457,
+      "learning_rate": 0.00014046759160965224,
+      "loss": 1.0267,
+      "step": 2090
+    },
+    {
+      "epoch": 0.43178070898598514,
+      "grad_norm": 0.35335784100847833,
+      "learning_rate": 0.0001401383265915703,
+      "loss": 1.1345,
+      "step": 2095
+    },
+    {
+      "epoch": 0.43281121187139326,
+      "grad_norm": 0.4118966005270665,
+      "learning_rate": 0.00013980854201901886,
+      "loss": 1.1936,
+      "step": 2100
+    },
+    {
+      "epoch": 0.43384171475680133,
+      "grad_norm": 0.3062362602687425,
+      "learning_rate": 0.00013947824216076207,
+      "loss": 0.9152,
+      "step": 2105
+    },
+    {
+      "epoch": 0.4348722176422094,
+      "grad_norm": 0.29620498290186464,
+      "learning_rate": 0.00013914743129223405,
+      "loss": 1.2309,
+      "step": 2110
+    },
+    {
+      "epoch": 0.43590272052761747,
+      "grad_norm": 0.2710224032495773,
+      "learning_rate": 0.00013881611369548325,
+      "loss": 1.0408,
+      "step": 2115
+    },
+    {
+      "epoch": 0.43693322341302554,
+      "grad_norm": 0.3056464151796937,
+      "learning_rate": 0.00013848429365911753,
+      "loss": 1.1743,
+      "step": 2120
+    },
+    {
+      "epoch": 0.4379637262984336,
+      "grad_norm": 0.3843468598801491,
+      "learning_rate": 0.00013815197547824824,
+      "loss": 1.2395,
+      "step": 2125
+    },
+    {
+      "epoch": 0.43899422918384173,
+      "grad_norm": 0.2798396925326106,
+      "learning_rate": 0.00013781916345443474,
+      "loss": 0.8612,
+      "step": 2130
+    },
+    {
+      "epoch": 0.4400247320692498,
+      "grad_norm": 0.3279250283208323,
+      "learning_rate": 0.00013748586189562878,
+      "loss": 1.2354,
+      "step": 2135
+    },
+    {
+      "epoch": 0.44105523495465787,
+      "grad_norm": 0.2735131812328209,
+      "learning_rate": 0.00013715207511611876,
+      "loss": 1.0026,
+      "step": 2140
+    },
+    {
+      "epoch": 0.44208573784006594,
+      "grad_norm": 0.37219144396512693,
+      "learning_rate": 0.0001368178074364737,
+      "loss": 1.0704,
+      "step": 2145
+    },
+    {
+      "epoch": 0.443116240725474,
+      "grad_norm": 0.46408445168045037,
+      "learning_rate": 0.00013648306318348762,
+      "loss": 1.2511,
+      "step": 2150
+    },
+    {
+      "epoch": 0.44414674361088213,
+      "grad_norm": 0.2720164174270696,
+      "learning_rate": 0.0001361478466901231,
+      "loss": 0.8989,
+      "step": 2155
+    },
+    {
+      "epoch": 0.4451772464962902,
+      "grad_norm": 0.3747167130586965,
+      "learning_rate": 0.0001358121622954557,
+      "loss": 1.3666,
+      "step": 2160
+    },
+    {
+      "epoch": 0.44620774938169827,
+      "grad_norm": 0.2962954738156199,
+      "learning_rate": 0.00013547601434461733,
+      "loss": 0.9376,
+      "step": 2165
+    },
+    {
+      "epoch": 0.44723825226710634,
+      "grad_norm": 0.3347575447270046,
+      "learning_rate": 0.0001351394071887404,
+      "loss": 1.0754,
+      "step": 2170
+    },
+    {
+      "epoch": 0.4482687551525144,
+      "grad_norm": 0.37805918523823334,
+      "learning_rate": 0.0001348023451849012,
+      "loss": 1.2106,
+      "step": 2175
+    },
+    {
+      "epoch": 0.44929925803792253,
+      "grad_norm": 0.302277885917108,
+      "learning_rate": 0.00013446483269606362,
+      "loss": 0.8481,
+      "step": 2180
+    },
+    {
+      "epoch": 0.4503297609233306,
+      "grad_norm": 0.35890494559800096,
+      "learning_rate": 0.00013412687409102277,
+      "loss": 1.2101,
+      "step": 2185
+    },
+    {
+      "epoch": 0.45136026380873867,
+      "grad_norm": 0.27832175193098935,
+      "learning_rate": 0.00013378847374434814,
+      "loss": 0.9563,
+      "step": 2190
+    },
+    {
+      "epoch": 0.45239076669414674,
+      "grad_norm": 0.35752890078545335,
+      "learning_rate": 0.0001334496360363274,
+      "loss": 1.1166,
+      "step": 2195
+    },
+    {
+      "epoch": 0.4534212695795548,
+      "grad_norm": 0.3208813709791692,
+      "learning_rate": 0.00013311036535290925,
+      "loss": 1.2658,
+      "step": 2200
+    },
+    {
+      "epoch": 0.4534212695795548,
+      "eval_loss": 1.0313003063201904,
+      "eval_runtime": 2462.8678,
+      "eval_samples_per_second": 3.248,
+      "eval_steps_per_second": 0.203,
+      "step": 2200
+    },
+    {
+      "epoch": 0.4544517724649629,
+      "grad_norm": 0.3164662130947771,
+      "learning_rate": 0.00013277066608564713,
+      "loss": 0.9046,
+      "step": 2205
+    },
+    {
+      "epoch": 0.455482275350371,
+      "grad_norm": 0.32760397119015733,
+      "learning_rate": 0.0001324305426316418,
+      "loss": 1.2596,
+      "step": 2210
+    },
+    {
+      "epoch": 0.45651277823577907,
+      "grad_norm": 0.26879676205264913,
+      "learning_rate": 0.000132089999393485,
+      "loss": 0.9842,
+      "step": 2215
+    },
+    {
+      "epoch": 0.45754328112118714,
+      "grad_norm": 0.35211465947054194,
+      "learning_rate": 0.000131749040779202,
+      "loss": 1.1149,
+      "step": 2220
+    },
+    {
+      "epoch": 0.4585737840065952,
+      "grad_norm": 0.3640438187361373,
+      "learning_rate": 0.0001314076712021949,
+      "loss": 1.2521,
+      "step": 2225
+    },
+    {
+      "epoch": 0.4596042868920033,
+      "grad_norm": 0.28859583596302096,
+      "learning_rate": 0.0001310658950811852,
+      "loss": 0.8689,
+      "step": 2230
+    },
+    {
+      "epoch": 0.4606347897774114,
+      "grad_norm": 0.3245571156879321,
+      "learning_rate": 0.00013072371684015688,
+      "loss": 1.2013,
+      "step": 2235
+    },
+    {
+      "epoch": 0.46166529266281947,
+      "grad_norm": 0.2630264452723511,
+      "learning_rate": 0.00013038114090829892,
+      "loss": 1.0054,
+      "step": 2240
+    },
+    {
+      "epoch": 0.46269579554822754,
+      "grad_norm": 0.3629689887711568,
+      "learning_rate": 0.00013003817171994807,
+      "loss": 1.1201,
+      "step": 2245
+    },
+    {
+      "epoch": 0.4637262984336356,
+      "grad_norm": 0.38051210660623236,
+      "learning_rate": 0.00012969481371453135,
+      "loss": 1.2291,
+      "step": 2250
+    },
+    {
+      "epoch": 0.4647568013190437,
+      "grad_norm": 0.3038049492726604,
+      "learning_rate": 0.00012935107133650885,
+      "loss": 0.8923,
+      "step": 2255
+    },
+    {
+      "epoch": 0.46578730420445175,
+      "grad_norm": 0.34319934420858916,
+      "learning_rate": 0.00012900694903531586,
+      "loss": 1.2325,
+      "step": 2260
+    },
+    {
+      "epoch": 0.46681780708985987,
+      "grad_norm": 0.2729387501421985,
+      "learning_rate": 0.0001286624512653055,
+      "loss": 0.9893,
+      "step": 2265
+    },
+    {
+      "epoch": 0.46784830997526794,
+      "grad_norm": 0.37018830848703915,
+      "learning_rate": 0.00012831758248569097,
+      "loss": 1.1021,
+      "step": 2270
+    },
+    {
+      "epoch": 0.468878812860676,
+      "grad_norm": 0.36278206444854155,
+      "learning_rate": 0.00012797234716048784,
+      "loss": 1.2005,
+      "step": 2275
+    },
+    {
+      "epoch": 0.4699093157460841,
+      "grad_norm": 0.2976065955292209,
+      "learning_rate": 0.00012762674975845637,
+      "loss": 0.8405,
+      "step": 2280
+    },
+    {
+      "epoch": 0.47093981863149215,
+      "grad_norm": 0.3358390029204746,
+      "learning_rate": 0.00012728079475304345,
+      "loss": 1.2722,
+      "step": 2285
+    },
+    {
+      "epoch": 0.47197032151690027,
+      "grad_norm": 0.24689134910878455,
+      "learning_rate": 0.0001269344866223249,
+      "loss": 0.9856,
+      "step": 2290
+    },
+    {
+      "epoch": 0.47300082440230834,
+      "grad_norm": 0.3183219956731036,
+      "learning_rate": 0.00012658782984894743,
+      "loss": 1.0672,
+      "step": 2295
+    },
+    {
+      "epoch": 0.4740313272877164,
+      "grad_norm": 0.3220468985387059,
+      "learning_rate": 0.00012624082892007064,
+      "loss": 1.2301,
+      "step": 2300
+    },
+    {
+      "epoch": 0.4750618301731245,
+      "grad_norm": 0.3012629746942751,
+      "learning_rate": 0.00012589348832730882,
+      "loss": 0.9345,
+      "step": 2305
+    },
+    {
+      "epoch": 0.47609233305853255,
+      "grad_norm": 0.34208223291291867,
+      "learning_rate": 0.00012554581256667296,
+      "loss": 1.1783,
+      "step": 2310
+    },
+    {
+      "epoch": 0.47712283594394067,
+      "grad_norm": 0.2633458752929658,
+      "learning_rate": 0.00012519780613851254,
+      "loss": 1.0127,
+      "step": 2315
+    },
+    {
+      "epoch": 0.47815333882934874,
+      "grad_norm": 0.3245639160611015,
+      "learning_rate": 0.00012484947354745714,
+      "loss": 1.074,
+      "step": 2320
+    },
+    {
+      "epoch": 0.4791838417147568,
+      "grad_norm": 0.4155838481655266,
+      "learning_rate": 0.0001245008193023583,
+      "loss": 1.2446,
+      "step": 2325
+    },
+    {
+      "epoch": 0.4802143446001649,
+      "grad_norm": 0.29474111212781395,
+      "learning_rate": 0.00012415184791623101,
+      "loss": 0.8756,
+      "step": 2330
+    },
+    {
+      "epoch": 0.48124484748557295,
+      "grad_norm": 0.38251199632272437,
+      "learning_rate": 0.00012380256390619548,
+      "loss": 1.2367,
+      "step": 2335
+    },
+    {
+      "epoch": 0.482275350370981,
+      "grad_norm": 0.2876263709986664,
+      "learning_rate": 0.00012345297179341844,
+      "loss": 0.942,
+      "step": 2340
+    },
+    {
+      "epoch": 0.48330585325638914,
+      "grad_norm": 0.3470650266011227,
+      "learning_rate": 0.00012310307610305477,
+      "loss": 1.1312,
+      "step": 2345
+    },
+    {
+      "epoch": 0.4843363561417972,
+      "grad_norm": 0.3437397458970076,
+      "learning_rate": 0.00012275288136418889,
+      "loss": 1.2013,
+      "step": 2350
+    },
+    {
+      "epoch": 0.4853668590272053,
+      "grad_norm": 0.2891400010122533,
+      "learning_rate": 0.0001224023921097762,
+      "loss": 0.875,
+      "step": 2355
+    },
+    {
+      "epoch": 0.48639736191261335,
+      "grad_norm": 0.3343377419438735,
+      "learning_rate": 0.0001220516128765842,
+      "loss": 1.2263,
+      "step": 2360
+    },
+    {
+      "epoch": 0.4874278647980214,
+      "grad_norm": 0.2815731947053776,
+      "learning_rate": 0.00012170054820513401,
+      "loss": 1.0279,
+      "step": 2365
+    },
+    {
+      "epoch": 0.48845836768342954,
+      "grad_norm": 0.34918143666071294,
+      "learning_rate": 0.00012134920263964147,
+      "loss": 1.1073,
+      "step": 2370
+    },
+    {
+      "epoch": 0.4894888705688376,
+      "grad_norm": 0.3378508040938572,
+      "learning_rate": 0.00012099758072795832,
+      "loss": 1.2131,
+      "step": 2375
+    },
+    {
+      "epoch": 0.4905193734542457,
+      "grad_norm": 0.3404097695213611,
+      "learning_rate": 0.00012064568702151335,
+      "loss": 0.9207,
+      "step": 2380
+    },
+    {
+      "epoch": 0.49154987633965375,
+      "grad_norm": 0.31871974527595615,
+      "learning_rate": 0.0001202935260752535,
+      "loss": 1.2815,
+      "step": 2385
+    },
+    {
+      "epoch": 0.4925803792250618,
+      "grad_norm": 0.27586523947951513,
+      "learning_rate": 0.00011994110244758496,
+      "loss": 0.9798,
+      "step": 2390
+    },
+    {
+      "epoch": 0.4936108821104699,
+      "grad_norm": 0.29258038219303606,
+      "learning_rate": 0.00011958842070031395,
+      "loss": 1.0612,
+      "step": 2395
+    },
+    {
+      "epoch": 0.494641384995878,
+      "grad_norm": 0.37118694749476105,
+      "learning_rate": 0.0001192354853985879,
+      "loss": 1.1718,
+      "step": 2400
+    },
+    {
+      "epoch": 0.494641384995878,
+      "eval_loss": 1.0186035633087158,
+      "eval_runtime": 2474.4655,
+      "eval_samples_per_second": 3.233,
+      "eval_steps_per_second": 0.202,
+      "step": 2400
+    },
+    {
+      "epoch": 0.4956718878812861,
+      "grad_norm": 0.3041203248300929,
+      "learning_rate": 0.00011888230111083627,
+      "loss": 0.8718,
+      "step": 2405
+    },
+    {
+      "epoch": 0.49670239076669415,
+      "grad_norm": 0.34626490332091997,
+      "learning_rate": 0.00011852887240871145,
+      "loss": 1.2384,
+      "step": 2410
+    },
+    {
+      "epoch": 0.4977328936521022,
+      "grad_norm": 0.2627628128026553,
+      "learning_rate": 0.00011817520386702947,
+      "loss": 0.9474,
+      "step": 2415
+    },
+    {
+      "epoch": 0.4987633965375103,
+      "grad_norm": 0.3243643599593401,
+      "learning_rate": 0.00011782130006371092,
+      "loss": 1.0514,
+      "step": 2420
+    },
+    {
+      "epoch": 0.4997938994229184,
+      "grad_norm": 0.38727382866372845,
+      "learning_rate": 0.00011746716557972167,
+      "loss": 1.2191,
+      "step": 2425
+    },
+    {
+      "epoch": 0.5008244023083265,
+      "grad_norm": 0.29252070214667514,
+      "learning_rate": 0.00011711280499901347,
+      "loss": 0.8687,
+      "step": 2430
+    },
+    {
+      "epoch": 0.5018549051937345,
+      "grad_norm": 0.32103295732093995,
+      "learning_rate": 0.00011675822290846474,
+      "loss": 1.2169,
+      "step": 2435
+    },
+    {
+      "epoch": 0.5028854080791426,
+      "grad_norm": 0.2619181848445574,
+      "learning_rate": 0.00011640342389782114,
+      "loss": 0.9916,
+      "step": 2440
+    },
+    {
+      "epoch": 0.5039159109645507,
+      "grad_norm": 0.36227995308769084,
+      "learning_rate": 0.00011604841255963616,
+      "loss": 1.0797,
+      "step": 2445
+    },
+    {
+      "epoch": 0.5049464138499588,
+      "grad_norm": 0.4472357927883422,
+      "learning_rate": 0.00011569319348921168,
+      "loss": 1.2004,
+      "step": 2450
+    },
+    {
+      "epoch": 0.5059769167353668,
+      "grad_norm": 0.2990090832488795,
+      "learning_rate": 0.00011533777128453844,
+      "loss": 0.905,
+      "step": 2455
+    },
+    {
+      "epoch": 0.5070074196207749,
+      "grad_norm": 0.3421802694540464,
+      "learning_rate": 0.00011498215054623664,
+      "loss": 1.2502,
+      "step": 2460
+    },
+    {
+      "epoch": 0.5080379225061831,
+      "grad_norm": 0.2619877856697061,
+      "learning_rate": 0.00011462633587749629,
+      "loss": 0.9878,
+      "step": 2465
+    },
+    {
+      "epoch": 0.5090684253915911,
+      "grad_norm": 0.3573733070627249,
+      "learning_rate": 0.00011427033188401768,
+      "loss": 1.0799,
+      "step": 2470
+    },
+    {
+      "epoch": 0.5100989282769992,
+      "grad_norm": 0.36110594973295185,
+      "learning_rate": 0.00011391414317395167,
+      "loss": 1.1831,
+      "step": 2475
+    },
+    {
+      "epoch": 0.5111294311624073,
+      "grad_norm": 0.29726968151818156,
+      "learning_rate": 0.00011355777435784016,
+      "loss": 0.9268,
+      "step": 2480
+    },
+    {
+      "epoch": 0.5121599340478153,
+      "grad_norm": 0.38992649702696064,
+      "learning_rate": 0.0001132012300485564,
+      "loss": 1.2403,
+      "step": 2485
+    },
+    {
+      "epoch": 0.5131904369332234,
+      "grad_norm": 0.31217559249274757,
+      "learning_rate": 0.00011284451486124514,
+      "loss": 0.9683,
+      "step": 2490
+    },
+    {
+      "epoch": 0.5142209398186315,
+      "grad_norm": 0.3309516167120589,
+      "learning_rate": 0.00011248763341326307,
+      "loss": 1.018,
+      "step": 2495
+    },
+    {
+      "epoch": 0.5152514427040396,
+      "grad_norm": 0.3669152523683029,
+      "learning_rate": 0.00011213059032411897,
+      "loss": 1.2018,
+      "step": 2500
+    },
+    {
+      "epoch": 0.5162819455894476,
+      "grad_norm": 0.26794057215849676,
+      "learning_rate": 0.00011177339021541387,
+      "loss": 0.8112,
+      "step": 2505
+    },
+    {
+      "epoch": 0.5173124484748557,
+      "grad_norm": 0.35964442987172796,
+      "learning_rate": 0.00011141603771078133,
+      "loss": 1.1878,
+      "step": 2510
+    },
+    {
+      "epoch": 0.5183429513602638,
+      "grad_norm": 0.26756565994743753,
+      "learning_rate": 0.00011105853743582751,
+      "loss": 1.0244,
+      "step": 2515
+    },
+    {
+      "epoch": 0.5193734542456719,
+      "grad_norm": 0.37262291618379195,
+      "learning_rate": 0.00011070089401807129,
+      "loss": 1.0956,
+      "step": 2520
+    },
+    {
+      "epoch": 0.52040395713108,
+      "grad_norm": 0.3862693430821137,
+      "learning_rate": 0.0001103431120868845,
+      "loss": 1.1777,
+      "step": 2525
+    },
+    {
+      "epoch": 0.5214344600164881,
+      "grad_norm": 0.34456489334354296,
+      "learning_rate": 0.00010998519627343182,
+      "loss": 0.9219,
+      "step": 2530
+    },
+    {
+      "epoch": 0.5224649629018961,
+      "grad_norm": 0.3363644750152242,
+      "learning_rate": 0.00010962715121061095,
+      "loss": 1.2219,
+      "step": 2535
+    },
+    {
+      "epoch": 0.5234954657873042,
+      "grad_norm": 0.2773020803302085,
+      "learning_rate": 0.00010926898153299259,
+      "loss": 0.9829,
+      "step": 2540
+    },
+    {
+      "epoch": 0.5245259686727123,
+      "grad_norm": 0.3387440320396881,
+      "learning_rate": 0.00010891069187676051,
+      "loss": 1.1028,
+      "step": 2545
+    },
+    {
+      "epoch": 0.5255564715581204,
+      "grad_norm": 0.4105637833575777,
+      "learning_rate": 0.00010855228687965138,
+      "loss": 1.1622,
+      "step": 2550
+    },
+    {
+      "epoch": 0.5265869744435284,
+      "grad_norm": 0.3203362105742117,
+      "learning_rate": 0.000108193771180895,
+      "loss": 0.8631,
+      "step": 2555
+    },
+    {
+      "epoch": 0.5276174773289365,
+      "grad_norm": 0.3176938770218858,
+      "learning_rate": 0.00010783514942115398,
+      "loss": 1.1615,
+      "step": 2560
+    },
+    {
+      "epoch": 0.5286479802143446,
+      "grad_norm": 0.27188600955614695,
+      "learning_rate": 0.0001074764262424639,
+      "loss": 1.0099,
+      "step": 2565
+    },
+    {
+      "epoch": 0.5296784830997526,
+      "grad_norm": 0.3225534680387049,
+      "learning_rate": 0.00010711760628817304,
+      "loss": 1.1351,
+      "step": 2570
+    },
+    {
+      "epoch": 0.5307089859851608,
+      "grad_norm": 0.39558907715108127,
+      "learning_rate": 0.0001067586942028824,
+      "loss": 1.1706,
+      "step": 2575
+    },
+    {
+      "epoch": 0.5317394888705689,
+      "grad_norm": 0.28753930251222504,
+      "learning_rate": 0.00010639969463238553,
+      "loss": 0.8745,
+      "step": 2580
+    },
+    {
+      "epoch": 0.532769991755977,
+      "grad_norm": 0.30734397473616565,
+      "learning_rate": 0.00010604061222360828,
+      "loss": 1.1448,
+      "step": 2585
+    },
+    {
+      "epoch": 0.533800494641385,
+      "grad_norm": 0.26804645458820414,
+      "learning_rate": 0.00010568145162454896,
+      "loss": 0.9725,
+      "step": 2590
+    },
+    {
+      "epoch": 0.5348309975267931,
+      "grad_norm": 0.3263053082797193,
+      "learning_rate": 0.00010532221748421787,
+      "loss": 1.0632,
+      "step": 2595
+    },
+    {
+      "epoch": 0.5358615004122012,
+      "grad_norm": 0.3650347794639528,
+      "learning_rate": 0.00010496291445257725,
+      "loss": 1.1795,
+      "step": 2600
+    },
+    {
+      "epoch": 0.5358615004122012,
+      "eval_loss": 1.0102049112319946,
+      "eval_runtime": 2464.7805,
+      "eval_samples_per_second": 3.246,
+      "eval_steps_per_second": 0.203,
+      "step": 2600
+    },
+    {
+      "epoch": 0.5368920032976092,
+      "grad_norm": 0.299206263893403,
+      "learning_rate": 0.00010460354718048109,
+      "loss": 0.8798,
+      "step": 2605
+    },
+    {
+      "epoch": 0.5379225061830173,
+      "grad_norm": 0.32212757306557593,
+      "learning_rate": 0.00010424412031961484,
+      "loss": 1.2768,
+      "step": 2610
+    },
+    {
+      "epoch": 0.5389530090684254,
+      "grad_norm": 0.2686223265118172,
+      "learning_rate": 0.0001038846385224354,
+      "loss": 0.9964,
+      "step": 2615
+    },
+    {
+      "epoch": 0.5399835119538334,
+      "grad_norm": 0.31440179887584824,
+      "learning_rate": 0.00010352510644211074,
+      "loss": 1.0418,
+      "step": 2620
+    },
+    {
+      "epoch": 0.5410140148392415,
+      "grad_norm": 0.3262137336119927,
+      "learning_rate": 0.0001031655287324596,
+      "loss": 1.2262,
+      "step": 2625
+    },
+    {
+      "epoch": 0.5420445177246497,
+      "grad_norm": 0.2977115868800609,
+      "learning_rate": 0.00010280591004789144,
+      "loss": 0.867,
+      "step": 2630
+    },
+    {
+      "epoch": 0.5430750206100577,
+      "grad_norm": 0.3481967708447838,
+      "learning_rate": 0.00010244625504334609,
+      "loss": 1.1666,
+      "step": 2635
+    },
+    {
+      "epoch": 0.5441055234954658,
+      "grad_norm": 0.27201591179231566,
+      "learning_rate": 0.00010208656837423351,
+      "loss": 0.9024,
+      "step": 2640
+    },
+    {
+      "epoch": 0.5451360263808739,
+      "grad_norm": 0.3829985152487614,
+      "learning_rate": 0.00010172685469637351,
+      "loss": 1.0713,
+      "step": 2645
+    },
+    {
+      "epoch": 0.546166529266282,
+      "grad_norm": 0.36090749295144076,
+      "learning_rate": 0.00010136711866593551,
+      "loss": 1.1479,
+      "step": 2650
+    },
+    {
+      "epoch": 0.54719703215169,
+      "grad_norm": 0.300126161900349,
+      "learning_rate": 0.00010100736493937828,
+      "loss": 0.8899,
+      "step": 2655
+    },
+    {
+      "epoch": 0.5482275350370981,
+      "grad_norm": 0.33344984792579047,
+      "learning_rate": 0.00010064759817338965,
+      "loss": 1.2477,
+      "step": 2660
+    },
+    {
+      "epoch": 0.5492580379225062,
+      "grad_norm": 0.2615632901768006,
+      "learning_rate": 0.00010028782302482617,
+      "loss": 0.9453,
+      "step": 2665
+    },
+    {
+      "epoch": 0.5502885408079142,
+      "grad_norm": 0.36602450939184455,
+      "learning_rate": 9.992804415065305e-05,
+      "loss": 1.0671,
+      "step": 2670
+    },
+    {
+      "epoch": 0.5513190436933223,
+      "grad_norm": 0.34898414138240497,
+      "learning_rate": 9.956826620788352e-05,
+      "loss": 1.1969,
+      "step": 2675
+    },
+    {
+      "epoch": 0.5523495465787304,
+      "grad_norm": 0.28926741748856266,
+      "learning_rate": 9.920849385351889e-05,
+      "loss": 0.8264,
+      "step": 2680
+    },
+    {
+      "epoch": 0.5533800494641385,
+      "grad_norm": 0.3470665476375764,
+      "learning_rate": 9.884873174448811e-05,
+      "loss": 1.2393,
+      "step": 2685
+    },
+    {
+      "epoch": 0.5544105523495466,
+      "grad_norm": 0.25248382277496384,
+      "learning_rate": 9.848898453758752e-05,
+      "loss": 0.9752,
+      "step": 2690
+    },
+    {
+      "epoch": 0.5554410552349547,
+      "grad_norm": 0.322941880174542,
+      "learning_rate": 9.812925688942054e-05,
+      "loss": 1.0431,
+      "step": 2695
+    },
+    {
+      "epoch": 0.5564715581203628,
+      "grad_norm": 0.5004506241518043,
+      "learning_rate": 9.776955345633739e-05,
+      "loss": 1.1417,
+      "step": 2700
+    },
+    {
+      "epoch": 0.5575020610057708,
+      "grad_norm": 0.2897297596123754,
+      "learning_rate": 9.740987889437492e-05,
+      "loss": 0.9032,
+      "step": 2705
+    },
+    {
+      "epoch": 0.5585325638911789,
+      "grad_norm": 0.3767677442786407,
+      "learning_rate": 9.705023785919623e-05,
+      "loss": 1.1532,
+      "step": 2710
+    },
+    {
+      "epoch": 0.559563066776587,
+      "grad_norm": 0.25503990950782773,
+      "learning_rate": 9.669063500603049e-05,
+      "loss": 0.9414,
+      "step": 2715
+    },
+    {
+      "epoch": 0.560593569661995,
+      "grad_norm": 0.34636076916747177,
+      "learning_rate": 9.633107498961251e-05,
+      "loss": 1.0689,
+      "step": 2720
+    },
+    {
+      "epoch": 0.5616240725474031,
+      "grad_norm": 0.39789228504761426,
+      "learning_rate": 9.597156246412277e-05,
+      "loss": 1.1464,
+      "step": 2725
+    },
+    {
+      "epoch": 0.5626545754328112,
+      "grad_norm": 0.289581783982976,
+      "learning_rate": 9.561210208312694e-05,
+      "loss": 0.818,
+      "step": 2730
+    },
+    {
+      "epoch": 0.5636850783182193,
+      "grad_norm": 0.31157714978145534,
+      "learning_rate": 9.525269849951578e-05,
+      "loss": 1.2706,
+      "step": 2735
+    },
+    {
+      "epoch": 0.5647155812036274,
+      "grad_norm": 0.28372271464928434,
+      "learning_rate": 9.489335636544476e-05,
+      "loss": 0.9805,
+      "step": 2740
+    },
+    {
+      "epoch": 0.5657460840890355,
+      "grad_norm": 0.33997031384483695,
+      "learning_rate": 9.453408033227406e-05,
+      "loss": 1.0937,
+      "step": 2745
+    },
+    {
+      "epoch": 0.5667765869744436,
+      "grad_norm": 0.33417374019033036,
+      "learning_rate": 9.417487505050816e-05,
+      "loss": 1.2517,
+      "step": 2750
+    },
+    {
+      "epoch": 0.5678070898598516,
+      "grad_norm": 0.3379621672677856,
+      "learning_rate": 9.38157451697358e-05,
+      "loss": 0.8737,
+      "step": 2755
+    },
+    {
+      "epoch": 0.5688375927452597,
+      "grad_norm": 0.3589988298317977,
+      "learning_rate": 9.345669533856961e-05,
+      "loss": 1.1273,
+      "step": 2760
+    },
+    {
+      "epoch": 0.5698680956306678,
+      "grad_norm": 0.26916205391576065,
+      "learning_rate": 9.309773020458616e-05,
+      "loss": 0.9603,
+      "step": 2765
+    },
+    {
+      "epoch": 0.5708985985160758,
+      "grad_norm": 0.3159986677198098,
+      "learning_rate": 9.273885441426562e-05,
+      "loss": 1.077,
+      "step": 2770
+    },
+    {
+      "epoch": 0.5719291014014839,
+      "grad_norm": 0.3412008285509142,
+      "learning_rate": 9.238007261293176e-05,
+      "loss": 1.1973,
+      "step": 2775
+    },
+    {
+      "epoch": 0.572959604286892,
+      "grad_norm": 0.28834466612116516,
+      "learning_rate": 9.202138944469168e-05,
+      "loss": 0.8306,
+      "step": 2780
+    },
+    {
+      "epoch": 0.5739901071723,
+      "grad_norm": 0.38911649440218604,
+      "learning_rate": 9.16628095523758e-05,
+      "loss": 1.141,
+      "step": 2785
+    },
+    {
+      "epoch": 0.5750206100577082,
+      "grad_norm": 0.29683818561664493,
+      "learning_rate": 9.130433757747772e-05,
+      "loss": 0.9729,
+      "step": 2790
+    },
+    {
+      "epoch": 0.5760511129431163,
+      "grad_norm": 0.3445235007246779,
+      "learning_rate": 9.094597816009409e-05,
+      "loss": 0.9819,
+      "step": 2795
+    },
+    {
+      "epoch": 0.5770816158285244,
+      "grad_norm": 0.3661330469833344,
+      "learning_rate": 9.058773593886469e-05,
+      "loss": 1.1984,
+      "step": 2800
+    },
+    {
+      "epoch": 0.5770816158285244,
+      "eval_loss": 1.0007954835891724,
+      "eval_runtime": 2470.6884,
+      "eval_samples_per_second": 3.238,
+      "eval_steps_per_second": 0.202,
+      "step": 2800
+    },
+    {
+      "epoch": 0.5781121187139324,
+      "grad_norm": 0.32153311729783396,
+      "learning_rate": 9.022961555091226e-05,
+      "loss": 0.8678,
+      "step": 2805
+    },
+    {
+      "epoch": 0.5791426215993405,
+      "grad_norm": 0.36605355908566417,
+      "learning_rate": 8.987162163178256e-05,
+      "loss": 1.2198,
+      "step": 2810
+    },
+    {
+      "epoch": 0.5801731244847486,
+      "grad_norm": 0.2691696483789003,
+      "learning_rate": 8.951375881538421e-05,
+      "loss": 0.9082,
+      "step": 2815
+    },
+    {
+      "epoch": 0.5812036273701566,
+      "grad_norm": 0.35640081108055943,
+      "learning_rate": 8.915603173392895e-05,
+      "loss": 1.0764,
+      "step": 2820
+    },
+    {
+      "epoch": 0.5822341302555647,
+      "grad_norm": 0.33493558837647336,
+      "learning_rate": 8.87984450178715e-05,
+      "loss": 1.1807,
+      "step": 2825
+    },
+    {
+      "epoch": 0.5832646331409728,
+      "grad_norm": 0.31748865735569,
+      "learning_rate": 8.84410032958497e-05,
+      "loss": 0.8788,
+      "step": 2830
+    },
+    {
+      "epoch": 0.5842951360263808,
+      "grad_norm": 0.36563160381436116,
+      "learning_rate": 8.808371119462452e-05,
+      "loss": 1.1127,
+      "step": 2835
+    },
+    {
+      "epoch": 0.5853256389117889,
+      "grad_norm": 0.26343660014410963,
+      "learning_rate": 8.772657333902027e-05,
+      "loss": 0.9977,
+      "step": 2840
+    },
+    {
+      "epoch": 0.5863561417971971,
+      "grad_norm": 0.3695249008293323,
+      "learning_rate": 8.736959435186466e-05,
+      "loss": 1.095,
+      "step": 2845
+    },
+    {
+      "epoch": 0.5873866446826052,
+      "grad_norm": 0.35773020515251225,
+      "learning_rate": 8.701277885392909e-05,
+      "loss": 1.2322,
+      "step": 2850
+    },
+    {
+      "epoch": 0.5884171475680132,
+      "grad_norm": 0.3269028767537368,
+      "learning_rate": 8.665613146386854e-05,
+      "loss": 0.8593,
+      "step": 2855
+    },
+    {
+      "epoch": 0.5894476504534213,
+      "grad_norm": 0.3544711574787662,
+      "learning_rate": 8.629965679816217e-05,
+      "loss": 1.1952,
+      "step": 2860
+    },
+    {
+      "epoch": 0.5904781533388294,
+      "grad_norm": 0.29639050674293943,
+      "learning_rate": 8.594335947105328e-05,
+      "loss": 0.9516,
+      "step": 2865
+    },
+    {
+      "epoch": 0.5915086562242374,
+      "grad_norm": 0.3426897718147058,
+      "learning_rate": 8.558724409448974e-05,
+      "loss": 1.0779,
+      "step": 2870
+    },
+    {
+      "epoch": 0.5925391591096455,
+      "grad_norm": 0.3740367235927869,
+      "learning_rate": 8.523131527806423e-05,
+      "loss": 1.1865,
+      "step": 2875
+    },
+    {
+      "epoch": 0.5935696619950536,
+      "grad_norm": 0.31189806294414407,
+      "learning_rate": 8.48755776289545e-05,
+      "loss": 0.8847,
+      "step": 2880
+    },
+    {
+      "epoch": 0.5946001648804616,
+      "grad_norm": 0.37495206238491713,
+      "learning_rate": 8.452003575186394e-05,
+      "loss": 1.1804,
+      "step": 2885
+    },
+    {
+      "epoch": 0.5956306677658697,
+      "grad_norm": 0.2797206274129371,
+      "learning_rate": 8.416469424896167e-05,
+      "loss": 0.9762,
+      "step": 2890
+    },
+    {
+      "epoch": 0.5966611706512778,
+      "grad_norm": 0.33348859645525075,
+      "learning_rate": 8.380955771982332e-05,
+      "loss": 1.0883,
+      "step": 2895
+    },
+    {
+      "epoch": 0.597691673536686,
+      "grad_norm": 0.44145796098661705,
+      "learning_rate": 8.345463076137125e-05,
+      "loss": 1.166,
+      "step": 2900
+    },
+    {
+      "epoch": 0.598722176422094,
+      "grad_norm": 0.286114945740939,
+      "learning_rate": 8.309991796781511e-05,
+      "loss": 0.8535,
+      "step": 2905
+    },
+    {
+      "epoch": 0.5997526793075021,
+      "grad_norm": 0.34343359665934886,
+      "learning_rate": 8.274542393059235e-05,
+      "loss": 1.1742,
+      "step": 2910
+    },
+    {
+      "epoch": 0.6007831821929102,
+      "grad_norm": 0.2624287823547147,
+      "learning_rate": 8.239115323830889e-05,
+      "loss": 0.955,
+      "step": 2915
+    },
+    {
+      "epoch": 0.6018136850783182,
+      "grad_norm": 0.35154822360336263,
+      "learning_rate": 8.203711047667958e-05,
+      "loss": 1.0426,
+      "step": 2920
+    },
+    {
+      "epoch": 0.6028441879637263,
+      "grad_norm": 0.3344934169585091,
+      "learning_rate": 8.1683300228469e-05,
+      "loss": 1.1785,
+      "step": 2925
+    },
+    {
+      "epoch": 0.6038746908491344,
+      "grad_norm": 0.3209999819274115,
+      "learning_rate": 8.132972707343192e-05,
+      "loss": 0.8842,
+      "step": 2930
+    },
+    {
+      "epoch": 0.6049051937345424,
+      "grad_norm": 0.39454283654575434,
+      "learning_rate": 8.097639558825427e-05,
+      "loss": 1.1693,
+      "step": 2935
+    },
+    {
+      "epoch": 0.6059356966199505,
+      "grad_norm": 0.2878286096920221,
+      "learning_rate": 8.062331034649376e-05,
+      "loss": 0.8965,
+      "step": 2940
+    },
+    {
+      "epoch": 0.6069661995053586,
+      "grad_norm": 0.33688522439048935,
+      "learning_rate": 8.027047591852069e-05,
+      "loss": 1.1381,
+      "step": 2945
+    },
+    {
+      "epoch": 0.6079967023907666,
+      "grad_norm": 0.38189870355144323,
+      "learning_rate": 7.991789687145873e-05,
+      "loss": 1.1257,
+      "step": 2950
+    },
+    {
+      "epoch": 0.6090272052761748,
+      "grad_norm": 0.3051536804305413,
+      "learning_rate": 7.956557776912596e-05,
+      "loss": 0.9044,
+      "step": 2955
+    },
+    {
+      "epoch": 0.6100577081615829,
+      "grad_norm": 0.3451327149435641,
+      "learning_rate": 7.921352317197574e-05,
+      "loss": 1.2264,
+      "step": 2960
+    },
+    {
+      "epoch": 0.611088211046991,
+      "grad_norm": 0.2827004152527063,
+      "learning_rate": 7.886173763703757e-05,
+      "loss": 0.9794,
+      "step": 2965
+    },
+    {
+      "epoch": 0.612118713932399,
+      "grad_norm": 0.40169478150821714,
+      "learning_rate": 7.851022571785819e-05,
+      "loss": 1.1179,
+      "step": 2970
+    },
+    {
+      "epoch": 0.6131492168178071,
+      "grad_norm": 0.4438701028140135,
+      "learning_rate": 7.815899196444267e-05,
+      "loss": 1.1662,
+      "step": 2975
+    },
+    {
+      "epoch": 0.6141797197032152,
+      "grad_norm": 0.3254217005122457,
+      "learning_rate": 7.780804092319547e-05,
+      "loss": 0.8434,
+      "step": 2980
+    },
+    {
+      "epoch": 0.6152102225886232,
+      "grad_norm": 0.3284145654312824,
+      "learning_rate": 7.745737713686152e-05,
+      "loss": 1.2109,
+      "step": 2985
+    },
+    {
+      "epoch": 0.6162407254740313,
+      "grad_norm": 0.28117376745819467,
+      "learning_rate": 7.710700514446762e-05,
+      "loss": 0.955,
+      "step": 2990
+    },
+    {
+      "epoch": 0.6172712283594394,
+      "grad_norm": 0.35764651325723473,
+      "learning_rate": 7.675692948126345e-05,
+      "loss": 1.032,
+      "step": 2995
+    },
+    {
+      "epoch": 0.6183017312448474,
+      "grad_norm": 0.4045276967582808,
+      "learning_rate": 7.640715467866307e-05,
+      "loss": 1.157,
+      "step": 3000
+    },
+    {
+      "epoch": 0.6183017312448474,
+      "eval_loss": 0.9929932951927185,
+      "eval_runtime": 2471.8115,
+      "eval_samples_per_second": 3.236,
+      "eval_steps_per_second": 0.202,
+      "step": 3000
+    },
+    {
+      "epoch": 0.6193322341302555,
+      "grad_norm": 0.3148405737434604,
+      "learning_rate": 7.605768526418605e-05,
+      "loss": 0.8743,
+      "step": 3005
+    },
+    {
+      "epoch": 0.6203627370156637,
+      "grad_norm": 0.35877325985884195,
+      "learning_rate": 7.57085257613991e-05,
+      "loss": 1.2573,
+      "step": 3010
+    },
+    {
+      "epoch": 0.6213932399010718,
+      "grad_norm": 0.2728312132512685,
+      "learning_rate": 7.535968068985737e-05,
+      "loss": 0.9663,
+      "step": 3015
+    },
+    {
+      "epoch": 0.6224237427864798,
+      "grad_norm": 0.3236792895371986,
+      "learning_rate": 7.501115456504595e-05,
+      "loss": 1.1109,
+      "step": 3020
+    },
+    {
+      "epoch": 0.6234542456718879,
+      "grad_norm": 0.40547856146336464,
+      "learning_rate": 7.466295189832148e-05,
+      "loss": 1.1882,
+      "step": 3025
+    },
+    {
+      "epoch": 0.624484748557296,
+      "grad_norm": 0.26523329910607873,
+      "learning_rate": 7.431507719685371e-05,
+      "loss": 0.8277,
+      "step": 3030
+    },
+    {
+      "epoch": 0.625515251442704,
+      "grad_norm": 0.3252197516332192,
+      "learning_rate": 7.396753496356718e-05,
+      "loss": 1.2303,
+      "step": 3035
+    },
+    {
+      "epoch": 0.6265457543281121,
+      "grad_norm": 0.2664136991956755,
+      "learning_rate": 7.362032969708297e-05,
+      "loss": 0.9909,
+      "step": 3040
+    },
+    {
+      "epoch": 0.6275762572135202,
+      "grad_norm": 0.3438706008672591,
+      "learning_rate": 7.327346589166035e-05,
+      "loss": 1.057,
+      "step": 3045
+    },
+    {
+      "epoch": 0.6286067600989282,
+      "grad_norm": 0.40393507740091983,
+      "learning_rate": 7.292694803713871e-05,
+      "loss": 1.1674,
+      "step": 3050
+    },
+    {
+      "epoch": 0.6296372629843363,
+      "grad_norm": 0.2966521374455327,
+      "learning_rate": 7.258078061887947e-05,
+      "loss": 0.8634,
+      "step": 3055
+    },
+    {
+      "epoch": 0.6306677658697445,
+      "grad_norm": 0.3313252298668368,
+      "learning_rate": 7.223496811770796e-05,
+      "loss": 1.1705,
+      "step": 3060
+    },
+    {
+      "epoch": 0.6316982687551526,
+      "grad_norm": 0.282517205910409,
+      "learning_rate": 7.188951500985533e-05,
+      "loss": 0.9568,
+      "step": 3065
+    },
+    {
+      "epoch": 0.6327287716405606,
+      "grad_norm": 0.3423739475910178,
+      "learning_rate": 7.154442576690083e-05,
+      "loss": 1.1215,
+      "step": 3070
+    },
+    {
+      "epoch": 0.6337592745259687,
+      "grad_norm": 0.3951313592694174,
+      "learning_rate": 7.119970485571375e-05,
+      "loss": 1.2196,
+      "step": 3075
+    },
+    {
+      "epoch": 0.6347897774113768,
+      "grad_norm": 0.27968067577198935,
+      "learning_rate": 7.08553567383956e-05,
+      "loss": 0.8517,
+      "step": 3080
+    },
+    {
+      "epoch": 0.6358202802967848,
+      "grad_norm": 0.3967375078625761,
+      "learning_rate": 7.051138587222255e-05,
+      "loss": 1.1345,
+      "step": 3085
+    },
+    {
+      "epoch": 0.6368507831821929,
+      "grad_norm": 0.275638071848918,
+      "learning_rate": 7.016779670958746e-05,
+      "loss": 0.9428,
+      "step": 3090
+    },
+    {
+      "epoch": 0.637881286067601,
+      "grad_norm": 0.3876563491000536,
+      "learning_rate": 6.982459369794247e-05,
+      "loss": 1.0172,
+      "step": 3095
+    },
+    {
+      "epoch": 0.638911788953009,
+      "grad_norm": 0.39932395010881844,
+      "learning_rate": 6.948178127974127e-05,
+      "loss": 1.136,
+      "step": 3100
+    },
+    {
+      "epoch": 0.6399422918384171,
+      "grad_norm": 0.344145718815036,
+      "learning_rate": 6.913936389238174e-05,
+      "loss": 0.9074,
+      "step": 3105
+    },
+    {
+      "epoch": 0.6409727947238252,
+      "grad_norm": 0.35702868127871346,
+      "learning_rate": 6.879734596814839e-05,
+      "loss": 1.2245,
+      "step": 3110
+    },
+    {
+      "epoch": 0.6420032976092334,
+      "grad_norm": 0.278189478468898,
+      "learning_rate": 6.845573193415508e-05,
+      "loss": 0.9795,
+      "step": 3115
+    },
+    {
+      "epoch": 0.6430338004946414,
+      "grad_norm": 0.3820554155556976,
+      "learning_rate": 6.811452621228766e-05,
+      "loss": 1.0815,
+      "step": 3120
+    },
+    {
+      "epoch": 0.6440643033800495,
+      "grad_norm": 0.42159702416774886,
+      "learning_rate": 6.777373321914671e-05,
+      "loss": 1.2058,
+      "step": 3125
+    },
+    {
+      "epoch": 0.6450948062654576,
+      "grad_norm": 0.30363813905867076,
+      "learning_rate": 6.743335736599045e-05,
+      "loss": 0.853,
+      "step": 3130
+    },
+    {
+      "epoch": 0.6461253091508656,
+      "grad_norm": 0.3626110839853577,
+      "learning_rate": 6.709340305867762e-05,
+      "loss": 1.1719,
+      "step": 3135
+    },
+    {
+      "epoch": 0.6471558120362737,
+      "grad_norm": 0.2741980722413254,
+      "learning_rate": 6.675387469761033e-05,
+      "loss": 0.9857,
+      "step": 3140
+    },
+    {
+      "epoch": 0.6481863149216818,
+      "grad_norm": 0.3458024372951965,
+      "learning_rate": 6.641477667767738e-05,
+      "loss": 1.0737,
+      "step": 3145
+    },
+    {
+      "epoch": 0.6492168178070898,
+      "grad_norm": 0.3659993842021884,
+      "learning_rate": 6.607611338819697e-05,
+      "loss": 1.2332,
+      "step": 3150
+    },
+    {
+      "epoch": 0.6502473206924979,
+      "grad_norm": 0.3134224056811165,
+      "learning_rate": 6.573788921286028e-05,
+      "loss": 0.833,
+      "step": 3155
+    },
+    {
+      "epoch": 0.651277823577906,
+      "grad_norm": 0.31557728275482577,
+      "learning_rate": 6.540010852967447e-05,
+      "loss": 1.169,
+      "step": 3160
+    },
+    {
+      "epoch": 0.652308326463314,
+      "grad_norm": 0.2745096877567603,
+      "learning_rate": 6.506277571090613e-05,
+      "loss": 0.9294,
+      "step": 3165
+    },
+    {
+      "epoch": 0.6533388293487222,
+      "grad_norm": 0.37277814205936943,
+      "learning_rate": 6.47258951230246e-05,
+      "loss": 1.081,
+      "step": 3170
+    },
+    {
+      "epoch": 0.6543693322341303,
+      "grad_norm": 0.38045711766200485,
+      "learning_rate": 6.438947112664555e-05,
+      "loss": 1.1493,
+      "step": 3175
+    },
+    {
+      "epoch": 0.6553998351195384,
+      "grad_norm": 0.2706340588373179,
+      "learning_rate": 6.405350807647444e-05,
+      "loss": 0.8498,
+      "step": 3180
+    },
+    {
+      "epoch": 0.6564303380049464,
+      "grad_norm": 0.3336018890725225,
+      "learning_rate": 6.371801032125026e-05,
+      "loss": 1.1778,
+      "step": 3185
+    },
+    {
+      "epoch": 0.6574608408903545,
+      "grad_norm": 0.2913475821459678,
+      "learning_rate": 6.338298220368912e-05,
+      "loss": 0.9571,
+      "step": 3190
+    },
+    {
+      "epoch": 0.6584913437757626,
+      "grad_norm": 0.36985494968453825,
+      "learning_rate": 6.304842806042812e-05,
+      "loss": 1.0778,
+      "step": 3195
+    },
+    {
+      "epoch": 0.6595218466611706,
+      "grad_norm": 0.40407830600531813,
+      "learning_rate": 6.271435222196916e-05,
+      "loss": 1.1542,
+      "step": 3200
+    },
+    {
+      "epoch": 0.6595218466611706,
+      "eval_loss": 0.9862346649169922,
+      "eval_runtime": 2465.9653,
+      "eval_samples_per_second": 3.244,
+      "eval_steps_per_second": 0.203,
+      "step": 3200
+    },
+    {
+      "epoch": 0.6605523495465787,
+      "grad_norm": 0.2524320710636017,
+      "learning_rate": 6.238075901262293e-05,
+      "loss": 0.8157,
+      "step": 3205
+    },
+    {
+      "epoch": 0.6615828524319868,
+      "grad_norm": 0.3278601577235125,
+      "learning_rate": 6.204765275045298e-05,
+      "loss": 1.1461,
+      "step": 3210
+    },
+    {
+      "epoch": 0.6626133553173948,
+      "grad_norm": 0.28938045624305053,
+      "learning_rate": 6.171503774721966e-05,
+      "loss": 0.9831,
+      "step": 3215
+    },
+    {
+      "epoch": 0.6636438582028029,
+      "grad_norm": 0.3531127055799286,
+      "learning_rate": 6.13829183083245e-05,
+      "loss": 1.0438,
+      "step": 3220
+    },
+    {
+      "epoch": 0.6646743610882111,
+      "grad_norm": 0.3337288007445976,
+      "learning_rate": 6.105129873275435e-05,
+      "loss": 1.2684,
+      "step": 3225
+    },
+    {
+      "epoch": 0.6657048639736192,
+      "grad_norm": 0.30432663000679183,
+      "learning_rate": 6.072018331302577e-05,
+      "loss": 0.8405,
+      "step": 3230
+    },
+    {
+      "epoch": 0.6667353668590272,
+      "grad_norm": 0.3481398036760903,
+      "learning_rate": 6.038957633512957e-05,
+      "loss": 1.1838,
+      "step": 3235
+    },
+    {
+      "epoch": 0.6677658697444353,
+      "grad_norm": 0.27336340413608334,
+      "learning_rate": 6.005948207847516e-05,
+      "loss": 0.9207,
+      "step": 3240
+    },
+    {
+      "epoch": 0.6687963726298434,
+      "grad_norm": 0.3440287779317674,
+      "learning_rate": 5.97299048158352e-05,
+      "loss": 1.0703,
+      "step": 3245
+    },
+    {
+      "epoch": 0.6698268755152514,
+      "grad_norm": 0.39581616124415014,
+      "learning_rate": 5.940084881329042e-05,
+      "loss": 1.2125,
+      "step": 3250
+    },
+    {
+      "epoch": 0.6708573784006595,
+      "grad_norm": 0.2935018842179493,
+      "learning_rate": 5.907231833017424e-05,
+      "loss": 0.8331,
+      "step": 3255
+    },
+    {
+      "epoch": 0.6718878812860676,
+      "grad_norm": 0.3836289298282413,
+      "learning_rate": 5.8744317619017755e-05,
+      "loss": 1.2524,
+      "step": 3260
+    },
+    {
+      "epoch": 0.6729183841714756,
+      "grad_norm": 0.2710041280727128,
+      "learning_rate": 5.841685092549456e-05,
+      "loss": 0.9405,
+      "step": 3265
+    },
+    {
+      "epoch": 0.6739488870568837,
+      "grad_norm": 0.32969149065243836,
+      "learning_rate": 5.8089922488365975e-05,
+      "loss": 1.0393,
+      "step": 3270
+    },
+    {
+      "epoch": 0.6749793899422918,
+      "grad_norm": 0.336443888291771,
+      "learning_rate": 5.776353653942602e-05,
+      "loss": 1.2373,
+      "step": 3275
+    },
+    {
+      "epoch": 0.6760098928277,
+      "grad_norm": 0.30665104617457667,
+      "learning_rate": 5.743769730344666e-05,
+      "loss": 0.8681,
+      "step": 3280
+    },
+    {
+      "epoch": 0.677040395713108,
+      "grad_norm": 0.3625802330226177,
+      "learning_rate": 5.7112408998123256e-05,
+      "loss": 1.2316,
+      "step": 3285
+    },
+    {
+      "epoch": 0.6780708985985161,
+      "grad_norm": 0.2953674696158516,
+      "learning_rate": 5.678767583401974e-05,
+      "loss": 0.9204,
+      "step": 3290
+    },
+    {
+      "epoch": 0.6791014014839242,
+      "grad_norm": 0.38225007623929924,
+      "learning_rate": 5.646350201451438e-05,
+      "loss": 1.066,
+      "step": 3295
+    },
+    {
+      "epoch": 0.6801319043693322,
+      "grad_norm": 0.4406303775855105,
+      "learning_rate": 5.613989173574512e-05,
+      "loss": 1.1569,
+      "step": 3300
+    },
+    {
+      "epoch": 0.6811624072547403,
+      "grad_norm": 0.2882802372851351,
+      "learning_rate": 5.5816849186555386e-05,
+      "loss": 0.86,
+      "step": 3305
+    },
+    {
+      "epoch": 0.6821929101401484,
+      "grad_norm": 0.3780252315323207,
+      "learning_rate": 5.549437854843995e-05,
+      "loss": 1.1646,
+      "step": 3310
+    },
+    {
+      "epoch": 0.6832234130255564,
+      "grad_norm": 0.2834828395630223,
+      "learning_rate": 5.517248399549063e-05,
+      "loss": 0.973,
+      "step": 3315
+    },
+    {
+      "epoch": 0.6842539159109645,
+      "grad_norm": 0.3721150712296605,
+      "learning_rate": 5.48511696943423e-05,
+      "loss": 1.0507,
+      "step": 3320
+    },
+    {
+      "epoch": 0.6852844187963726,
+      "grad_norm": 0.3748866522580312,
+      "learning_rate": 5.4530439804119096e-05,
+      "loss": 1.1834,
+      "step": 3325
+    },
+    {
+      "epoch": 0.6863149216817807,
+      "grad_norm": 0.2887534766618149,
+      "learning_rate": 5.4210298476380484e-05,
+      "loss": 0.888,
+      "step": 3330
+    },
+    {
+      "epoch": 0.6873454245671888,
+      "grad_norm": 0.3409152912595764,
+      "learning_rate": 5.38907498550674e-05,
+      "loss": 1.1606,
+      "step": 3335
+    },
+    {
+      "epoch": 0.6883759274525969,
+      "grad_norm": 0.29326388422606586,
+      "learning_rate": 5.357179807644887e-05,
+      "loss": 0.8845,
+      "step": 3340
+    },
+    {
+      "epoch": 0.689406430338005,
+      "grad_norm": 0.3589561770883435,
+      "learning_rate": 5.3253447269068245e-05,
+      "loss": 1.0743,
+      "step": 3345
+    },
+    {
+      "epoch": 0.690436933223413,
+      "grad_norm": 0.38383916804262924,
+      "learning_rate": 5.293570155368981e-05,
+      "loss": 1.183,
+      "step": 3350
+    },
+    {
+      "epoch": 0.6914674361088211,
+      "grad_norm": 0.3068092183719897,
+      "learning_rate": 5.261856504324563e-05,
+      "loss": 0.881,
+      "step": 3355
+    },
+    {
+      "epoch": 0.6924979389942292,
+      "grad_norm": 0.36385771675879847,
+      "learning_rate": 5.230204184278195e-05,
+      "loss": 1.2467,
+      "step": 3360
+    },
+    {
+      "epoch": 0.6935284418796372,
+      "grad_norm": 0.2907466758494882,
+      "learning_rate": 5.198613604940649e-05,
+      "loss": 0.9442,
+      "step": 3365
+    },
+    {
+      "epoch": 0.6945589447650453,
+      "grad_norm": 0.3566322978977828,
+      "learning_rate": 5.1670851752235025e-05,
+      "loss": 1.0632,
+      "step": 3370
+    },
+    {
+      "epoch": 0.6955894476504534,
+      "grad_norm": 0.36835384306712526,
+      "learning_rate": 5.135619303233867e-05,
+      "loss": 1.1819,
+      "step": 3375
+    },
+    {
+      "epoch": 0.6966199505358615,
+      "grad_norm": 0.30530163857595344,
+      "learning_rate": 5.104216396269109e-05,
+      "loss": 0.8697,
+      "step": 3380
+    },
+    {
+      "epoch": 0.6976504534212696,
+      "grad_norm": 0.3728398219157709,
+      "learning_rate": 5.072876860811553e-05,
+      "loss": 1.2065,
+      "step": 3385
+    },
+    {
+      "epoch": 0.6986809563066777,
+      "grad_norm": 0.2719186126381508,
+      "learning_rate": 5.0416011025232546e-05,
+      "loss": 0.9332,
+      "step": 3390
+    },
+    {
+      "epoch": 0.6997114591920858,
+      "grad_norm": 0.39560521280304406,
+      "learning_rate": 5.010389526240719e-05,
+      "loss": 1.0035,
+      "step": 3395
+    },
+    {
+      "epoch": 0.7007419620774938,
+      "grad_norm": 0.45096201363438326,
+      "learning_rate": 4.979242535969672e-05,
+      "loss": 1.1648,
+      "step": 3400
+    },
+    {
+      "epoch": 0.7007419620774938,
+      "eval_loss": 0.9801868796348572,
+      "eval_runtime": 2471.0518,
+      "eval_samples_per_second": 3.237,
+      "eval_steps_per_second": 0.202,
+      "step": 3400
+    },
+    {
+      "epoch": 0.7017724649629019,
+      "grad_norm": 0.3253388221391844,
+      "learning_rate": 4.9481605348798435e-05,
+      "loss": 0.8283,
+      "step": 3405
+    },
+    {
+      "epoch": 0.70280296784831,
+      "grad_norm": 0.3382283949543721,
+      "learning_rate": 4.917143925299728e-05,
+      "loss": 1.1642,
+      "step": 3410
+    },
+    {
+      "epoch": 0.703833470733718,
+      "grad_norm": 0.2650304501518648,
+      "learning_rate": 4.886193108711378e-05,
+      "loss": 0.9334,
+      "step": 3415
+    },
+    {
+      "epoch": 0.7048639736191261,
+      "grad_norm": 0.3163351222719128,
+      "learning_rate": 4.8553084857452426e-05,
+      "loss": 1.0316,
+      "step": 3420
+    },
+    {
+      "epoch": 0.7058944765045342,
+      "grad_norm": 0.4184991045727221,
+      "learning_rate": 4.824490456174926e-05,
+      "loss": 1.1313,
+      "step": 3425
+    },
+    {
+      "epoch": 0.7069249793899423,
+      "grad_norm": 0.31621686065106913,
+      "learning_rate": 4.7937394189120485e-05,
+      "loss": 0.8658,
+      "step": 3430
+    },
+    {
+      "epoch": 0.7079554822753503,
+      "grad_norm": 0.3252525212780396,
+      "learning_rate": 4.763055772001086e-05,
+      "loss": 1.1749,
+      "step": 3435
+    },
+    {
+      "epoch": 0.7089859851607585,
+      "grad_norm": 0.28077212260765644,
+      "learning_rate": 4.732439912614195e-05,
+      "loss": 0.9195,
+      "step": 3440
+    },
+    {
+      "epoch": 0.7100164880461666,
+      "grad_norm": 0.3404715984560569,
+      "learning_rate": 4.7018922370460835e-05,
+      "loss": 1.1004,
+      "step": 3445
+    },
+    {
+      "epoch": 0.7110469909315746,
+      "grad_norm": 0.3572114714960076,
+      "learning_rate": 4.671413140708893e-05,
+      "loss": 1.1555,
+      "step": 3450
+    },
+    {
+      "epoch": 0.7120774938169827,
+      "grad_norm": 0.31012926926152484,
+      "learning_rate": 4.6410030181270546e-05,
+      "loss": 0.8328,
+      "step": 3455
+    },
+    {
+      "epoch": 0.7131079967023908,
+      "grad_norm": 0.3814244438557788,
+      "learning_rate": 4.610662262932209e-05,
+      "loss": 1.2221,
+      "step": 3460
+    },
+    {
+      "epoch": 0.7141384995877988,
+      "grad_norm": 0.2587030449835817,
+      "learning_rate": 4.5803912678580906e-05,
+      "loss": 0.923,
+      "step": 3465
+    },
+    {
+      "epoch": 0.7151690024732069,
+      "grad_norm": 0.36964920266690043,
+      "learning_rate": 4.5501904247354474e-05,
+      "loss": 1.0765,
+      "step": 3470
+    },
+    {
+      "epoch": 0.716199505358615,
+      "grad_norm": 0.4007947343253882,
+      "learning_rate": 4.520060124486989e-05,
+      "loss": 1.0613,
+      "step": 3475
+    },
+    {
+      "epoch": 0.717230008244023,
+      "grad_norm": 0.31355619965035286,
+      "learning_rate": 4.4900007571222946e-05,
+      "loss": 0.8406,
+      "step": 3480
+    },
+    {
+      "epoch": 0.7182605111294311,
+      "grad_norm": 0.3709230553408382,
+      "learning_rate": 4.460012711732795e-05,
+      "loss": 1.202,
+      "step": 3485
+    },
+    {
+      "epoch": 0.7192910140148392,
+      "grad_norm": 0.2663391335521852,
+      "learning_rate": 4.430096376486713e-05,
+      "loss": 0.957,
+      "step": 3490
+    },
+    {
+      "epoch": 0.7203215169002474,
+      "grad_norm": 0.36149156644544367,
+      "learning_rate": 4.4002521386240466e-05,
+      "loss": 1.0885,
+      "step": 3495
+    },
+    {
+      "epoch": 0.7213520197856554,
+      "grad_norm": 0.4116390911425264,
+      "learning_rate": 4.3704803844515705e-05,
+      "loss": 1.1649,
+      "step": 3500
+    },
+    {
+      "epoch": 0.7223825226710635,
+      "grad_norm": 0.3337684750408035,
+      "learning_rate": 4.3407814993378095e-05,
+      "loss": 0.8993,
+      "step": 3505
+    },
+    {
+      "epoch": 0.7234130255564716,
+      "grad_norm": 0.3496736243866778,
+      "learning_rate": 4.311155867708071e-05,
+      "loss": 1.1648,
+      "step": 3510
+    },
+    {
+      "epoch": 0.7244435284418796,
+      "grad_norm": 0.28479774144881315,
+      "learning_rate": 4.2816038730394656e-05,
+      "loss": 0.9484,
+      "step": 3515
+    },
+    {
+      "epoch": 0.7254740313272877,
+      "grad_norm": 0.36918418554896276,
+      "learning_rate": 4.252125897855932e-05,
+      "loss": 1.0583,
+      "step": 3520
+    },
+    {
+      "epoch": 0.7265045342126958,
+      "grad_norm": 0.33704337919665334,
+      "learning_rate": 4.222722323723294e-05,
+      "loss": 1.1907,
+      "step": 3525
+    },
+    {
+      "epoch": 0.7275350370981039,
+      "grad_norm": 0.3148307470632366,
+      "learning_rate": 4.1933935312443286e-05,
+      "loss": 0.8783,
+      "step": 3530
+    },
+    {
+      "epoch": 0.7285655399835119,
+      "grad_norm": 0.3517810748272734,
+      "learning_rate": 4.164139900053824e-05,
+      "loss": 1.1717,
+      "step": 3535
+    },
+    {
+      "epoch": 0.72959604286892,
+      "grad_norm": 0.2855842847206538,
+      "learning_rate": 4.134961808813672e-05,
+      "loss": 0.9112,
+      "step": 3540
+    },
+    {
+      "epoch": 0.7306265457543281,
+      "grad_norm": 0.34108259740221386,
+      "learning_rate": 4.1058596352079805e-05,
+      "loss": 1.0142,
+      "step": 3545
+    },
+    {
+      "epoch": 0.7316570486397362,
+      "grad_norm": 0.45538302782398454,
+      "learning_rate": 4.076833755938153e-05,
+      "loss": 1.0974,
+      "step": 3550
+    },
+    {
+      "epoch": 0.7326875515251443,
+      "grad_norm": 0.31487990147891376,
+      "learning_rate": 4.0478845467180506e-05,
+      "loss": 0.8298,
+      "step": 3555
+    },
+    {
+      "epoch": 0.7337180544105524,
+      "grad_norm": 0.36935273464286145,
+      "learning_rate": 4.0190123822690965e-05,
+      "loss": 1.1436,
+      "step": 3560
+    },
+    {
+      "epoch": 0.7347485572959604,
+      "grad_norm": 0.27930316978222564,
+      "learning_rate": 3.990217636315441e-05,
+      "loss": 0.9989,
+      "step": 3565
+    },
+    {
+      "epoch": 0.7357790601813685,
+      "grad_norm": 0.3714521960753657,
+      "learning_rate": 3.961500681579129e-05,
+      "loss": 1.1123,
+      "step": 3570
+    },
+    {
+      "epoch": 0.7368095630667766,
+      "grad_norm": 0.43314908676280006,
+      "learning_rate": 3.9328618897752566e-05,
+      "loss": 1.1162,
+      "step": 3575
+    },
+    {
+      "epoch": 0.7378400659521847,
+      "grad_norm": 0.29947090161111584,
+      "learning_rate": 3.904301631607186e-05,
+      "loss": 0.8351,
+      "step": 3580
+    },
+    {
+      "epoch": 0.7388705688375927,
+      "grad_norm": 0.4143197770113969,
+      "learning_rate": 3.875820276761717e-05,
+      "loss": 1.2288,
+      "step": 3585
+    },
+    {
+      "epoch": 0.7399010717230008,
+      "grad_norm": 0.2821180575460458,
+      "learning_rate": 3.847418193904324e-05,
+      "loss": 0.9842,
+      "step": 3590
+    },
+    {
+      "epoch": 0.7409315746084089,
+      "grad_norm": 0.35222923714451126,
+      "learning_rate": 3.8190957506743806e-05,
+      "loss": 1.0377,
+      "step": 3595
+    },
+    {
+      "epoch": 0.7419620774938169,
+      "grad_norm": 0.36695400370029363,
+      "learning_rate": 3.790853313680389e-05,
+      "loss": 1.1403,
+      "step": 3600
+    },
+    {
+      "epoch": 0.7419620774938169,
+      "eval_loss": 0.9750496745109558,
+      "eval_runtime": 2464.5637,
+      "eval_samples_per_second": 3.246,
+      "eval_steps_per_second": 0.203,
+      "step": 3600
+    },
+    {
+      "epoch": 0.7429925803792251,
+      "grad_norm": 0.33168677598566876,
+      "learning_rate": 3.7626912484952495e-05,
+      "loss": 0.9223,
+      "step": 3605
+    },
+    {
+      "epoch": 0.7440230832646332,
+      "grad_norm": 0.3685099675858659,
+      "learning_rate": 3.734609919651523e-05,
+      "loss": 1.22,
+      "step": 3610
+    },
+    {
+      "epoch": 0.7450535861500412,
+      "grad_norm": 0.30519501707613905,
+      "learning_rate": 3.706609690636703e-05,
+      "loss": 0.9541,
+      "step": 3615
+    },
+    {
+      "epoch": 0.7460840890354493,
+      "grad_norm": 0.3227606827521623,
+      "learning_rate": 3.6786909238885215e-05,
+      "loss": 1.002,
+      "step": 3620
+    },
+    {
+      "epoch": 0.7471145919208574,
+      "grad_norm": 0.3907472086101187,
+      "learning_rate": 3.650853980790262e-05,
+      "loss": 1.1461,
+      "step": 3625
+    },
+    {
+      "epoch": 0.7481450948062655,
+      "grad_norm": 0.2919486775395443,
+      "learning_rate": 3.6230992216660664e-05,
+      "loss": 0.8063,
+      "step": 3630
+    },
+    {
+      "epoch": 0.7491755976916735,
+      "grad_norm": 0.32875886981694724,
+      "learning_rate": 3.595427005776281e-05,
+      "loss": 1.1514,
+      "step": 3635
+    },
+    {
+      "epoch": 0.7502061005770816,
+      "grad_norm": 0.29158449116331836,
+      "learning_rate": 3.5678376913128075e-05,
+      "loss": 0.9759,
+      "step": 3640
+    },
+    {
+      "epoch": 0.7512366034624897,
+      "grad_norm": 0.37631936130463955,
+      "learning_rate": 3.540331635394458e-05,
+      "loss": 0.9905,
+      "step": 3645
+    },
+    {
+      "epoch": 0.7522671063478977,
+      "grad_norm": 0.3439508621054155,
+      "learning_rate": 3.512909194062347e-05,
+      "loss": 1.1949,
+      "step": 3650
+    },
+    {
+      "epoch": 0.7532976092333058,
+      "grad_norm": 0.3146397920651041,
+      "learning_rate": 3.485570722275264e-05,
+      "loss": 0.8706,
+      "step": 3655
+    },
+    {
+      "epoch": 0.754328112118714,
+      "grad_norm": 0.38258052872478915,
+      "learning_rate": 3.458316573905087e-05,
+      "loss": 1.2172,
+      "step": 3660
+    },
+    {
+      "epoch": 0.755358615004122,
+      "grad_norm": 0.2793437890823054,
+      "learning_rate": 3.4311471017322175e-05,
+      "loss": 0.9391,
+      "step": 3665
+    },
+    {
+      "epoch": 0.7563891178895301,
+      "grad_norm": 0.3643128847802364,
+      "learning_rate": 3.4040626574409815e-05,
+      "loss": 0.9938,
+      "step": 3670
+    },
+    {
+      "epoch": 0.7574196207749382,
+      "grad_norm": 0.3709320339736749,
+      "learning_rate": 3.377063591615113e-05,
+      "loss": 1.1904,
+      "step": 3675
+    },
+    {
+      "epoch": 0.7584501236603463,
+      "grad_norm": 0.33437847713940755,
+      "learning_rate": 3.350150253733186e-05,
+      "loss": 0.9233,
+      "step": 3680
+    },
+    {
+      "epoch": 0.7594806265457543,
+      "grad_norm": 0.3607224445907033,
+      "learning_rate": 3.3233229921641064e-05,
+      "loss": 1.2125,
+      "step": 3685
+    },
+    {
+      "epoch": 0.7605111294311624,
+      "grad_norm": 0.27198443746283474,
+      "learning_rate": 3.296582154162604e-05,
+      "loss": 0.9196,
+      "step": 3690
+    },
+    {
+      "epoch": 0.7615416323165705,
+      "grad_norm": 0.3668276085455235,
+      "learning_rate": 3.2699280858647327e-05,
+      "loss": 1.0335,
+      "step": 3695
+    },
+    {
+      "epoch": 0.7625721352019785,
+      "grad_norm": 0.431636600953154,
+      "learning_rate": 3.2433611322833845e-05,
+      "loss": 1.1378,
+      "step": 3700
+    },
+    {
+      "epoch": 0.7636026380873866,
+      "grad_norm": 0.3284008231342369,
+      "learning_rate": 3.216881637303839e-05,
+      "loss": 0.8308,
+      "step": 3705
+    },
+    {
+      "epoch": 0.7646331409727948,
+      "grad_norm": 0.3443882838132438,
+      "learning_rate": 3.190489943679297e-05,
+      "loss": 1.1797,
+      "step": 3710
+    },
+    {
+      "epoch": 0.7656636438582028,
+      "grad_norm": 0.30472428853778355,
+      "learning_rate": 3.164186393026445e-05,
+      "loss": 0.8953,
+      "step": 3715
+    },
+    {
+      "epoch": 0.7666941467436109,
+      "grad_norm": 0.346855761423586,
+      "learning_rate": 3.137971325821054e-05,
+      "loss": 1.1678,
+      "step": 3720
+    },
+    {
+      "epoch": 0.767724649629019,
+      "grad_norm": 0.3586318210457773,
+      "learning_rate": 3.111845081393542e-05,
+      "loss": 1.1801,
+      "step": 3725
+    },
+    {
+      "epoch": 0.768755152514427,
+      "grad_norm": 0.3012111522622655,
+      "learning_rate": 3.0858079979245965e-05,
+      "loss": 0.8259,
+      "step": 3730
+    },
+    {
+      "epoch": 0.7697856553998351,
+      "grad_norm": 0.3675771233203778,
+      "learning_rate": 3.059860412440811e-05,
+      "loss": 1.1379,
+      "step": 3735
+    },
+    {
+      "epoch": 0.7708161582852432,
+      "grad_norm": 0.2997972891151249,
+      "learning_rate": 3.0340026608102902e-05,
+      "loss": 0.9012,
+      "step": 3740
+    },
+    {
+      "epoch": 0.7718466611706513,
+      "grad_norm": 0.3350924890911528,
+      "learning_rate": 3.008235077738334e-05,
+      "loss": 1.0476,
+      "step": 3745
+    },
+    {
+      "epoch": 0.7728771640560593,
+      "grad_norm": 0.42578144522976874,
+      "learning_rate": 2.9825579967630846e-05,
+      "loss": 1.2014,
+      "step": 3750
+    },
+    {
+      "epoch": 0.7739076669414674,
+      "grad_norm": 0.31905668249953223,
+      "learning_rate": 2.956971750251215e-05,
+      "loss": 0.9089,
+      "step": 3755
+    },
+    {
+      "epoch": 0.7749381698268755,
+      "grad_norm": 0.3320270260748608,
+      "learning_rate": 2.9314766693936356e-05,
+      "loss": 1.1712,
+      "step": 3760
+    },
+    {
+      "epoch": 0.7759686727122836,
+      "grad_norm": 0.26880506131338233,
+      "learning_rate": 2.906073084201191e-05,
+      "loss": 0.9532,
+      "step": 3765
+    },
+    {
+      "epoch": 0.7769991755976917,
+      "grad_norm": 0.360515318139803,
+      "learning_rate": 2.8807613235004037e-05,
+      "loss": 1.0235,
+      "step": 3770
+    },
+    {
+      "epoch": 0.7780296784830998,
+      "grad_norm": 0.40012910311865413,
+      "learning_rate": 2.855541714929206e-05,
+      "loss": 1.1292,
+      "step": 3775
+    },
+    {
+      "epoch": 0.7790601813685079,
+      "grad_norm": 0.3056042121876613,
+      "learning_rate": 2.8304145849327036e-05,
+      "loss": 0.7814,
+      "step": 3780
+    },
+    {
+      "epoch": 0.7800906842539159,
+      "grad_norm": 0.3475474680120366,
+      "learning_rate": 2.8053802587589538e-05,
+      "loss": 1.1668,
+      "step": 3785
+    },
+    {
+      "epoch": 0.781121187139324,
+      "grad_norm": 0.2843659913958508,
+      "learning_rate": 2.7804390604547557e-05,
+      "loss": 0.9237,
+      "step": 3790
+    },
+    {
+      "epoch": 0.7821516900247321,
+      "grad_norm": 0.36497922605390365,
+      "learning_rate": 2.7555913128614398e-05,
+      "loss": 1.0276,
+      "step": 3795
+    },
+    {
+      "epoch": 0.7831821929101401,
+      "grad_norm": 0.39023528771913,
+      "learning_rate": 2.7308373376107142e-05,
+      "loss": 1.1268,
+      "step": 3800
+    },
+    {
+      "epoch": 0.7831821929101401,
+      "eval_loss": 0.9705116748809814,
+      "eval_runtime": 2662.1546,
+      "eval_samples_per_second": 3.005,
+      "eval_steps_per_second": 0.188,
+      "step": 3800
+    },
+    {
+      "epoch": 0.7842126957955482,
+      "grad_norm": 0.3292637316481058,
+      "learning_rate": 2.706177455120482e-05,
+      "loss": 0.8586,
+      "step": 3805
+    },
+    {
+      "epoch": 0.7852431986809563,
+      "grad_norm": 0.3463786306953725,
+      "learning_rate": 2.681611984590696e-05,
+      "loss": 1.1979,
+      "step": 3810
+    },
+    {
+      "epoch": 0.7862737015663643,
+      "grad_norm": 0.3033390300653325,
+      "learning_rate": 2.6571412439992437e-05,
+      "loss": 0.9448,
+      "step": 3815
+    },
+    {
+      "epoch": 0.7873042044517725,
+      "grad_norm": 0.3744585100233523,
+      "learning_rate": 2.6327655500978076e-05,
+      "loss": 1.0825,
+      "step": 3820
+    },
+    {
+      "epoch": 0.7883347073371806,
+      "grad_norm": 0.3888535219196645,
+      "learning_rate": 2.608485218407779e-05,
+      "loss": 1.1602,
+      "step": 3825
+    },
+    {
+      "epoch": 0.7893652102225887,
+      "grad_norm": 0.3281836015354183,
+      "learning_rate": 2.5843005632161787e-05,
+      "loss": 0.8071,
+      "step": 3830
+    },
+    {
+      "epoch": 0.7903957131079967,
+      "grad_norm": 0.44764397719301374,
+      "learning_rate": 2.5602118975715683e-05,
+      "loss": 1.223,
+      "step": 3835
+    },
+    {
+      "epoch": 0.7914262159934048,
+      "grad_norm": 0.2605243101554053,
+      "learning_rate": 2.5362195332800253e-05,
+      "loss": 0.9269,
+      "step": 3840
+    },
+    {
+      "epoch": 0.7924567188788129,
+      "grad_norm": 0.34490759623743283,
+      "learning_rate": 2.5123237809010836e-05,
+      "loss": 0.9951,
+      "step": 3845
+    },
+    {
+      "epoch": 0.7934872217642209,
+      "grad_norm": 0.45219071339330563,
+      "learning_rate": 2.4885249497437223e-05,
+      "loss": 1.1591,
+      "step": 3850
+    },
+    {
+      "epoch": 0.794517724649629,
+      "grad_norm": 0.3701824230659728,
+      "learning_rate": 2.4648233478623705e-05,
+      "loss": 0.9237,
+      "step": 3855
+    },
+    {
+      "epoch": 0.7955482275350371,
+      "grad_norm": 0.3540574000256143,
+      "learning_rate": 2.4412192820529034e-05,
+      "loss": 1.1936,
+      "step": 3860
+    },
+    {
+      "epoch": 0.7965787304204451,
+      "grad_norm": 0.2979316113245556,
+      "learning_rate": 2.4177130578486885e-05,
+      "loss": 0.8818,
+      "step": 3865
+    },
+    {
+      "epoch": 0.7976092333058532,
+      "grad_norm": 0.3972288442858004,
+      "learning_rate": 2.3943049795166126e-05,
+      "loss": 1.0352,
+      "step": 3870
+    },
+    {
+      "epoch": 0.7986397361912614,
+      "grad_norm": 0.43218350490127366,
+      "learning_rate": 2.370995350053157e-05,
+      "loss": 1.141,
+      "step": 3875
+    },
+    {
+      "epoch": 0.7996702390766695,
+      "grad_norm": 0.33536259992407524,
+      "learning_rate": 2.3477844711804708e-05,
+      "loss": 0.8845,
+      "step": 3880
+    },
+    {
+      "epoch": 0.8007007419620775,
+      "grad_norm": 0.3715539599255406,
+      "learning_rate": 2.3246726433424716e-05,
+      "loss": 1.1807,
+      "step": 3885
+    },
+    {
+      "epoch": 0.8017312448474856,
+      "grad_norm": 0.27142620043338195,
+      "learning_rate": 2.301660165700936e-05,
+      "loss": 0.9382,
+      "step": 3890
+    },
+    {
+      "epoch": 0.8027617477328937,
+      "grad_norm": 0.3200117930180246,
+      "learning_rate": 2.2787473361316592e-05,
+      "loss": 1.1648,
+      "step": 3895
+    },
+    {
+      "epoch": 0.8037922506183017,
+      "grad_norm": 0.3832079088616207,
+      "learning_rate": 2.2559344512205705e-05,
+      "loss": 1.1815,
+      "step": 3900
+    },
+    {
+      "epoch": 0.8048227535037098,
+      "grad_norm": 0.28512724726672417,
+      "learning_rate": 2.233221806259903e-05,
+      "loss": 0.7237,
+      "step": 3905
+    },
+    {
+      "epoch": 0.8058532563891179,
+      "grad_norm": 0.347779304749182,
+      "learning_rate": 2.2106096952443888e-05,
+      "loss": 1.1468,
+      "step": 3910
+    },
+    {
+      "epoch": 0.8068837592745259,
+      "grad_norm": 0.27344000922062256,
+      "learning_rate": 2.188098410867424e-05,
+      "loss": 0.967,
+      "step": 3915
+    },
+    {
+      "epoch": 0.807914262159934,
+      "grad_norm": 0.35776849097052316,
+      "learning_rate": 2.165688244517299e-05,
+      "loss": 1.0351,
+      "step": 3920
+    },
+    {
+      "epoch": 0.8089447650453421,
+      "grad_norm": 0.6146898482449163,
+      "learning_rate": 2.143379486273428e-05,
+      "loss": 1.1856,
+      "step": 3925
+    },
+    {
+      "epoch": 0.8099752679307503,
+      "grad_norm": 0.37385831712675915,
+      "learning_rate": 2.1211724249025787e-05,
+      "loss": 0.9441,
+      "step": 3930
+    },
+    {
+      "epoch": 0.8110057708161583,
+      "grad_norm": 0.33568983400788555,
+      "learning_rate": 2.099067347855157e-05,
+      "loss": 1.2175,
+      "step": 3935
+    },
+    {
+      "epoch": 0.8120362737015664,
+      "grad_norm": 0.2905285106790405,
+      "learning_rate": 2.077064541261462e-05,
+      "loss": 0.9725,
+      "step": 3940
+    },
+    {
+      "epoch": 0.8130667765869745,
+      "grad_norm": 0.3292103830025177,
+      "learning_rate": 2.0551642899279975e-05,
+      "loss": 1.0363,
+      "step": 3945
+    },
+    {
+      "epoch": 0.8140972794723825,
+      "grad_norm": 0.4380964265869986,
+      "learning_rate": 2.0333668773337866e-05,
+      "loss": 1.1014,
+      "step": 3950
+    },
+    {
+      "epoch": 0.8151277823577906,
+      "grad_norm": 0.3297438808755647,
+      "learning_rate": 2.0116725856266926e-05,
+      "loss": 0.8749,
+      "step": 3955
+    },
+    {
+      "epoch": 0.8161582852431987,
+      "grad_norm": 0.3404142147430593,
+      "learning_rate": 1.9900816956197698e-05,
+      "loss": 1.2154,
+      "step": 3960
+    },
+    {
+      "epoch": 0.8171887881286067,
+      "grad_norm": 0.28990266176476354,
+      "learning_rate": 1.9685944867876373e-05,
+      "loss": 0.9402,
+      "step": 3965
+    },
+    {
+      "epoch": 0.8182192910140148,
+      "grad_norm": 0.3750595800822031,
+      "learning_rate": 1.9472112372628536e-05,
+      "loss": 1.0682,
+      "step": 3970
+    },
+    {
+      "epoch": 0.8192497938994229,
+      "grad_norm": 0.4415444668426194,
+      "learning_rate": 1.9259322238323095e-05,
+      "loss": 1.1201,
+      "step": 3975
+    },
+    {
+      "epoch": 0.8202802967848309,
+      "grad_norm": 0.31061384334227865,
+      "learning_rate": 1.9047577219336665e-05,
+      "loss": 0.8734,
+      "step": 3980
+    },
+    {
+      "epoch": 0.8213107996702391,
+      "grad_norm": 0.3577711991948567,
+      "learning_rate": 1.8836880056517658e-05,
+      "loss": 1.1896,
+      "step": 3985
+    },
+    {
+      "epoch": 0.8223413025556472,
+      "grad_norm": 0.2862925377080917,
+      "learning_rate": 1.862723347715103e-05,
+      "loss": 0.9206,
+      "step": 3990
+    },
+    {
+      "epoch": 0.8233718054410553,
+      "grad_norm": 0.36692123404130295,
+      "learning_rate": 1.841864019492282e-05,
+      "loss": 1.0504,
+      "step": 3995
+    },
+    {
+      "epoch": 0.8244023083264633,
+      "grad_norm": 0.3600663907207253,
+      "learning_rate": 1.821110290988509e-05,
+      "loss": 1.2122,
+      "step": 4000
+    },
+    {
+      "epoch": 0.8244023083264633,
+      "eval_loss": 0.9671783447265625,
+      "eval_runtime": 2672.6929,
+      "eval_samples_per_second": 2.993,
+      "eval_steps_per_second": 0.187,
+      "step": 4000
+    },
+    {
+      "epoch": 0.8254328112118714,
+      "grad_norm": 0.3081888529721894,
+      "learning_rate": 1.8004624308421026e-05,
+      "loss": 0.7821,
+      "step": 4005
+    },
+    {
+      "epoch": 0.8264633140972795,
+      "grad_norm": 0.36648973755342457,
+      "learning_rate": 1.7799207063210044e-05,
+      "loss": 1.2144,
+      "step": 4010
+    },
+    {
+      "epoch": 0.8274938169826875,
+      "grad_norm": 0.2898792198270504,
+      "learning_rate": 1.759485383319326e-05,
+      "loss": 0.9427,
+      "step": 4015
+    },
+    {
+      "epoch": 0.8285243198680956,
+      "grad_norm": 0.38174814126983564,
+      "learning_rate": 1.7391567263539144e-05,
+      "loss": 1.0154,
+      "step": 4020
+    },
+    {
+      "epoch": 0.8295548227535037,
+      "grad_norm": 0.4742619716693875,
+      "learning_rate": 1.7189349985609115e-05,
+      "loss": 1.0813,
+      "step": 4025
+    },
+    {
+      "epoch": 0.8305853256389117,
+      "grad_norm": 0.3664614977301218,
+      "learning_rate": 1.6988204616923666e-05,
+      "loss": 0.8368,
+      "step": 4030
+    },
+    {
+      "epoch": 0.8316158285243199,
+      "grad_norm": 0.43325455353643755,
+      "learning_rate": 1.6788133761128312e-05,
+      "loss": 1.1778,
+      "step": 4035
+    },
+    {
+      "epoch": 0.832646331409728,
+      "grad_norm": 0.29420614649546306,
+      "learning_rate": 1.658914000795999e-05,
+      "loss": 0.922,
+      "step": 4040
+    },
+    {
+      "epoch": 0.8336768342951361,
+      "grad_norm": 0.3437287490008131,
+      "learning_rate": 1.639122593321357e-05,
+      "loss": 1.0901,
+      "step": 4045
+    },
+    {
+      "epoch": 0.8347073371805441,
+      "grad_norm": 0.41394098029113957,
+      "learning_rate": 1.6194394098708377e-05,
+      "loss": 1.1766,
+      "step": 4050
+    },
+    {
+      "epoch": 0.8357378400659522,
+      "grad_norm": 0.3399407346712763,
+      "learning_rate": 1.59986470522551e-05,
+      "loss": 0.8491,
+      "step": 4055
+    },
+    {
+      "epoch": 0.8367683429513603,
+      "grad_norm": 0.3655997093950039,
+      "learning_rate": 1.580398732762297e-05,
+      "loss": 1.1981,
+      "step": 4060
+    },
+    {
+      "epoch": 0.8377988458367683,
+      "grad_norm": 0.30216864477326527,
+      "learning_rate": 1.5610417444506664e-05,
+      "loss": 0.9703,
+      "step": 4065
+    },
+    {
+      "epoch": 0.8388293487221764,
+      "grad_norm": 0.3993955860213197,
+      "learning_rate": 1.541793990849387e-05,
+      "loss": 1.0866,
+      "step": 4070
+    },
+    {
+      "epoch": 0.8398598516075845,
+      "grad_norm": 0.40765142147078065,
+      "learning_rate": 1.522655721103291e-05,
+      "loss": 1.1578,
+      "step": 4075
+    },
+    {
+      "epoch": 0.8408903544929925,
+      "grad_norm": 0.3132479277395683,
+      "learning_rate": 1.5036271829400294e-05,
+      "loss": 0.8317,
+      "step": 4080
+    },
+    {
+      "epoch": 0.8419208573784006,
+      "grad_norm": 0.353057034757265,
+      "learning_rate": 1.4847086226668872e-05,
+      "loss": 1.1387,
+      "step": 4085
+    },
+    {
+      "epoch": 0.8429513602638088,
+      "grad_norm": 0.28714788090256477,
+      "learning_rate": 1.4659002851675774e-05,
+      "loss": 0.9363,
+      "step": 4090
+    },
+    {
+      "epoch": 0.8439818631492169,
+      "grad_norm": 0.3558290938292747,
+      "learning_rate": 1.447202413899078e-05,
+      "loss": 0.9974,
+      "step": 4095
+    },
+    {
+      "epoch": 0.8450123660346249,
+      "grad_norm": 0.36577045780008416,
+      "learning_rate": 1.4286152508884898e-05,
+      "loss": 1.1466,
+      "step": 4100
+    },
+    {
+      "epoch": 0.846042868920033,
+      "grad_norm": 0.3010779731099561,
+      "learning_rate": 1.4101390367298861e-05,
+      "loss": 0.7895,
+      "step": 4105
+    },
+    {
+      "epoch": 0.8470733718054411,
+      "grad_norm": 0.3541525630578856,
+      "learning_rate": 1.3917740105812094e-05,
+      "loss": 1.1429,
+      "step": 4110
+    },
+    {
+      "epoch": 0.8481038746908491,
+      "grad_norm": 0.28819494987192507,
+      "learning_rate": 1.3735204101611776e-05,
+      "loss": 0.9335,
+      "step": 4115
+    },
+    {
+      "epoch": 0.8491343775762572,
+      "grad_norm": 0.34823589325876386,
+      "learning_rate": 1.355378471746196e-05,
+      "loss": 1.0577,
+      "step": 4120
+    },
+    {
+      "epoch": 0.8501648804616653,
+      "grad_norm": 0.34950447449734157,
+      "learning_rate": 1.3373484301673145e-05,
+      "loss": 1.1231,
+      "step": 4125
+    },
+    {
+      "epoch": 0.8511953833470733,
+      "grad_norm": 0.29975822569477556,
+      "learning_rate": 1.3194305188071732e-05,
+      "loss": 0.8424,
+      "step": 4130
+    },
+    {
+      "epoch": 0.8522258862324814,
+      "grad_norm": 0.390420435835868,
+      "learning_rate": 1.301624969596985e-05,
+      "loss": 1.1329,
+      "step": 4135
+    },
+    {
+      "epoch": 0.8532563891178895,
+      "grad_norm": 0.2911183426984063,
+      "learning_rate": 1.2839320130135468e-05,
+      "loss": 0.9841,
+      "step": 4140
+    },
+    {
+      "epoch": 0.8542868920032977,
+      "grad_norm": 0.3500369252129129,
+      "learning_rate": 1.266351878076234e-05,
+      "loss": 1.0513,
+      "step": 4145
+    },
+    {
+      "epoch": 0.8553173948887057,
+      "grad_norm": 0.4017098058926308,
+      "learning_rate": 1.2488847923440483e-05,
+      "loss": 1.1624,
+      "step": 4150
+    },
+    {
+      "epoch": 0.8563478977741138,
+      "grad_norm": 0.3441542578051488,
+      "learning_rate": 1.2315309819126852e-05,
+      "loss": 0.8611,
+      "step": 4155
+    },
+    {
+      "epoch": 0.8573784006595219,
+      "grad_norm": 0.36811687971995544,
+      "learning_rate": 1.2142906714115787e-05,
+      "loss": 1.2298,
+      "step": 4160
+    },
+    {
+      "epoch": 0.8584089035449299,
+      "grad_norm": 0.2934040538277352,
+      "learning_rate": 1.197164084001009e-05,
+      "loss": 0.9825,
+      "step": 4165
+    },
+    {
+      "epoch": 0.859439406430338,
+      "grad_norm": 0.3571442419781514,
+      "learning_rate": 1.1801514413692239e-05,
+      "loss": 1.0296,
+      "step": 4170
+    },
+    {
+      "epoch": 0.8604699093157461,
+      "grad_norm": 0.4388041060818487,
+      "learning_rate": 1.1632529637295475e-05,
+      "loss": 1.1665,
+      "step": 4175
+    },
+    {
+      "epoch": 0.8615004122011541,
+      "grad_norm": 0.3211936804672997,
+      "learning_rate": 1.1464688698175497e-05,
+      "loss": 0.8568,
+      "step": 4180
+    },
+    {
+      "epoch": 0.8625309150865622,
+      "grad_norm": 0.38975409701266794,
+      "learning_rate": 1.1297993768881998e-05,
+      "loss": 1.1854,
+      "step": 4185
+    },
+    {
+      "epoch": 0.8635614179719703,
+      "grad_norm": 0.27934297465783847,
+      "learning_rate": 1.113244700713063e-05,
+      "loss": 0.8697,
+      "step": 4190
+    },
+    {
+      "epoch": 0.8645919208573783,
+      "grad_norm": 0.39072549944659307,
+      "learning_rate": 1.0968050555775067e-05,
+      "loss": 1.0164,
+      "step": 4195
+    },
+    {
+      "epoch": 0.8656224237427865,
+      "grad_norm": 0.4998393287633088,
+      "learning_rate": 1.0804806542779223e-05,
+      "loss": 1.0571,
+      "step": 4200
+    },
+    {
+      "epoch": 0.8656224237427865,
+      "eval_loss": 0.9649366140365601,
+      "eval_runtime": 2477.8624,
+      "eval_samples_per_second": 3.229,
+      "eval_steps_per_second": 0.202,
+      "step": 4200
+    },
+    {
+      "epoch": 0.8666529266281946,
+      "grad_norm": 0.30161833311373715,
+      "learning_rate": 1.0642717081189735e-05,
+      "loss": 0.8134,
+      "step": 4205
+    },
+    {
+      "epoch": 0.8676834295136027,
+      "grad_norm": 0.3714099835276263,
+      "learning_rate": 1.0481784269108664e-05,
+      "loss": 1.2136,
+      "step": 4210
+    },
+    {
+      "epoch": 0.8687139323990107,
+      "grad_norm": 0.29858155300200195,
+      "learning_rate": 1.032201018966621e-05,
+      "loss": 0.9334,
+      "step": 4215
+    },
+    {
+      "epoch": 0.8697444352844188,
+      "grad_norm": 0.36365152527356986,
+      "learning_rate": 1.0163396910993883e-05,
+      "loss": 1.0167,
+      "step": 4220
+    },
+    {
+      "epoch": 0.8707749381698269,
+      "grad_norm": 0.3926164945656461,
+      "learning_rate": 1.0005946486197648e-05,
+      "loss": 1.1597,
+      "step": 4225
+    },
+    {
+      "epoch": 0.8718054410552349,
+      "grad_norm": 0.31631364428210557,
+      "learning_rate": 9.849660953331363e-06,
+      "loss": 0.8626,
+      "step": 4230
+    },
+    {
+      "epoch": 0.872835943940643,
+      "grad_norm": 0.397012339843304,
+      "learning_rate": 9.694542335370437e-06,
+      "loss": 1.1559,
+      "step": 4235
+    },
+    {
+      "epoch": 0.8738664468260511,
+      "grad_norm": 0.27478860653538173,
+      "learning_rate": 9.540592640185597e-06,
+      "loss": 0.9427,
+      "step": 4240
+    },
+    {
+      "epoch": 0.8748969497114591,
+      "grad_norm": 0.34552863820949387,
+      "learning_rate": 9.387813860516915e-06,
+      "loss": 1.0141,
+      "step": 4245
+    },
+    {
+      "epoch": 0.8759274525968672,
+      "grad_norm": 0.4001075016221677,
+      "learning_rate": 9.236207973948063e-06,
+      "loss": 1.1693,
+      "step": 4250
+    },
+    {
+      "epoch": 0.8769579554822754,
+      "grad_norm": 0.3089362146851165,
+      "learning_rate": 9.085776942880608e-06,
+      "loss": 0.8552,
+      "step": 4255
+    },
+    {
+      "epoch": 0.8779884583676835,
+      "grad_norm": 0.3341234310512727,
+      "learning_rate": 8.936522714508678e-06,
+      "loss": 1.1562,
+      "step": 4260
+    },
+    {
+      "epoch": 0.8790189612530915,
+      "grad_norm": 0.2798328581056808,
+      "learning_rate": 8.788447220793806e-06,
+      "loss": 0.99,
+      "step": 4265
+    },
+    {
+      "epoch": 0.8800494641384996,
+      "grad_norm": 0.39837992045978476,
+      "learning_rate": 8.641552378439776e-06,
+      "loss": 1.1073,
+      "step": 4270
+    },
+    {
+      "epoch": 0.8810799670239077,
+      "grad_norm": 0.4106276064285754,
+      "learning_rate": 8.495840088868024e-06,
+      "loss": 1.1004,
+      "step": 4275
+    },
+    {
+      "epoch": 0.8821104699093157,
+      "grad_norm": 0.37910153433606,
+      "learning_rate": 8.351312238192787e-06,
+      "loss": 0.9667,
+      "step": 4280
+    },
+    {
+      "epoch": 0.8831409727947238,
+      "grad_norm": 0.3835389510278448,
+      "learning_rate": 8.207970697196866e-06,
+      "loss": 1.1637,
+      "step": 4285
+    },
+    {
+      "epoch": 0.8841714756801319,
+      "grad_norm": 0.2996106330164154,
+      "learning_rate": 8.065817321307367e-06,
+      "loss": 0.8998,
+      "step": 4290
+    },
+    {
+      "epoch": 0.88520197856554,
+      "grad_norm": 0.37985003315229504,
+      "learning_rate": 7.924853950571642e-06,
+      "loss": 0.9981,
+      "step": 4295
+    },
+    {
+      "epoch": 0.886232481450948,
+      "grad_norm": 0.3734475653137268,
+      "learning_rate": 7.78508240963347e-06,
+      "loss": 1.1901,
+      "step": 4300
+    },
+    {
+      "epoch": 0.8872629843363561,
+      "grad_norm": 0.32917519272254847,
+      "learning_rate": 7.646504507709563e-06,
+      "loss": 0.8026,
+      "step": 4305
+    },
+    {
+      "epoch": 0.8882934872217643,
+      "grad_norm": 0.3184623209600695,
+      "learning_rate": 7.50912203856593e-06,
+      "loss": 1.1705,
+      "step": 4310
+    },
+    {
+      "epoch": 0.8893239901071723,
+      "grad_norm": 0.27564137851589965,
+      "learning_rate": 7.372936780494877e-06,
+      "loss": 0.8968,
+      "step": 4315
+    },
+    {
+      "epoch": 0.8903544929925804,
+      "grad_norm": 0.37744723978180933,
+      "learning_rate": 7.237950496291856e-06,
+      "loss": 0.9825,
+      "step": 4320
+    },
+    {
+      "epoch": 0.8913849958779885,
+      "grad_norm": 0.3497470810206515,
+      "learning_rate": 7.104164933232649e-06,
+      "loss": 1.1498,
+      "step": 4325
+    },
+    {
+      "epoch": 0.8924154987633965,
+      "grad_norm": 0.314664899690108,
+      "learning_rate": 6.971581823050832e-06,
+      "loss": 0.8124,
+      "step": 4330
+    },
+    {
+      "epoch": 0.8934460016488046,
+      "grad_norm": 0.3437097611977056,
+      "learning_rate": 6.840202881915325e-06,
+      "loss": 1.1555,
+      "step": 4335
+    },
+    {
+      "epoch": 0.8944765045342127,
+      "grad_norm": 0.28333067118500105,
+      "learning_rate": 6.710029810408092e-06,
+      "loss": 0.899,
+      "step": 4340
+    },
+    {
+      "epoch": 0.8955070074196207,
+      "grad_norm": 0.39572557163512656,
+      "learning_rate": 6.581064293502293e-06,
+      "loss": 1.0465,
+      "step": 4345
+    },
+    {
+      "epoch": 0.8965375103050288,
+      "grad_norm": 0.37753981586149277,
+      "learning_rate": 6.453308000540304e-06,
+      "loss": 1.1223,
+      "step": 4350
+    },
+    {
+      "epoch": 0.8975680131904369,
+      "grad_norm": 0.33675472826608077,
+      "learning_rate": 6.326762585212209e-06,
+      "loss": 0.8248,
+      "step": 4355
+    },
+    {
+      "epoch": 0.8985985160758451,
+      "grad_norm": 0.3825197592387351,
+      "learning_rate": 6.20142968553441e-06,
+      "loss": 1.1401,
+      "step": 4360
+    },
+    {
+      "epoch": 0.8996290189612531,
+      "grad_norm": 0.2838180681849475,
+      "learning_rate": 6.077310923828328e-06,
+      "loss": 0.909,
+      "step": 4365
+    },
+    {
+      "epoch": 0.9006595218466612,
+      "grad_norm": 0.37686082536565757,
+      "learning_rate": 5.954407906699511e-06,
+      "loss": 1.0217,
+      "step": 4370
+    },
+    {
+      "epoch": 0.9016900247320693,
+      "grad_norm": 0.3380783448392427,
+      "learning_rate": 5.8327222250167735e-06,
+      "loss": 1.1523,
+      "step": 4375
+    },
+    {
+      "epoch": 0.9027205276174773,
+      "grad_norm": 0.2883596670864561,
+      "learning_rate": 5.71225545389158e-06,
+      "loss": 0.8144,
+      "step": 4380
+    },
+    {
+      "epoch": 0.9037510305028854,
+      "grad_norm": 0.3483906272756692,
+      "learning_rate": 5.59300915265778e-06,
+      "loss": 1.1855,
+      "step": 4385
+    },
+    {
+      "epoch": 0.9047815333882935,
+      "grad_norm": 0.2639750686093266,
+      "learning_rate": 5.4749848648512624e-06,
+      "loss": 0.9381,
+      "step": 4390
+    },
+    {
+      "epoch": 0.9058120362737015,
+      "grad_norm": 0.3972431149683815,
+      "learning_rate": 5.358184118190068e-06,
+      "loss": 1.0683,
+      "step": 4395
+    },
+    {
+      "epoch": 0.9068425391591096,
+      "grad_norm": 0.35801168305170344,
+      "learning_rate": 5.242608424554651e-06,
+      "loss": 1.0903,
+      "step": 4400
+    },
+    {
+      "epoch": 0.9068425391591096,
+      "eval_loss": 0.9635400176048279,
+      "eval_runtime": 2476.6467,
+      "eval_samples_per_second": 3.23,
+      "eval_steps_per_second": 0.202,
+      "step": 4400
+    },
+    {
+      "epoch": 0.9078730420445177,
+      "grad_norm": 0.33090712708760794,
+      "learning_rate": 5.128259279968195e-06,
+      "loss": 0.8644,
+      "step": 4405
+    },
+    {
+      "epoch": 0.9089035449299258,
+      "grad_norm": 0.35066225756340424,
+      "learning_rate": 5.01513816457736e-06,
+      "loss": 1.1482,
+      "step": 4410
+    },
+    {
+      "epoch": 0.9099340478153339,
+      "grad_norm": 0.2777858386287618,
+      "learning_rate": 4.903246542633033e-06,
+      "loss": 0.9529,
+      "step": 4415
+    },
+    {
+      "epoch": 0.910964550700742,
+      "grad_norm": 0.3807172458090473,
+      "learning_rate": 4.7925858624714215e-06,
+      "loss": 1.0499,
+      "step": 4420
+    },
+    {
+      "epoch": 0.9119950535861501,
+      "grad_norm": 0.36543715138130545,
+      "learning_rate": 4.683157556495343e-06,
+      "loss": 1.1883,
+      "step": 4425
+    },
+    {
+      "epoch": 0.9130255564715581,
+      "grad_norm": 0.3339553261899812,
+      "learning_rate": 4.574963041155622e-06,
+      "loss": 0.8319,
+      "step": 4430
+    },
+    {
+      "epoch": 0.9140560593569662,
+      "grad_norm": 0.3202714496531148,
+      "learning_rate": 4.468003716932734e-06,
+      "loss": 1.1678,
+      "step": 4435
+    },
+    {
+      "epoch": 0.9150865622423743,
+      "grad_norm": 0.27715040842554794,
+      "learning_rate": 4.362280968318777e-06,
+      "loss": 0.9364,
+      "step": 4440
+    },
+    {
+      "epoch": 0.9161170651277823,
+      "grad_norm": 0.3628969851723974,
+      "learning_rate": 4.257796163799455e-06,
+      "loss": 1.0704,
+      "step": 4445
+    },
+    {
+      "epoch": 0.9171475680131904,
+      "grad_norm": 0.4255339930477384,
+      "learning_rate": 4.154550655836409e-06,
+      "loss": 1.1622,
+      "step": 4450
+    },
+    {
+      "epoch": 0.9181780708985985,
+      "grad_norm": 0.3098446765156209,
+      "learning_rate": 4.052545780849715e-06,
+      "loss": 0.8363,
+      "step": 4455
+    },
+    {
+      "epoch": 0.9192085737840066,
+      "grad_norm": 0.38626454996818416,
+      "learning_rate": 3.9517828592005475e-06,
+      "loss": 1.1148,
+      "step": 4460
+    },
+    {
+      "epoch": 0.9202390766694146,
+      "grad_norm": 0.2786681923266397,
+      "learning_rate": 3.852263195174155e-06,
+      "loss": 0.8743,
+      "step": 4465
+    },
+    {
+      "epoch": 0.9212695795548228,
+      "grad_norm": 0.3627972033581097,
+      "learning_rate": 3.7539880769628998e-06,
+      "loss": 1.0305,
+      "step": 4470
+    },
+    {
+      "epoch": 0.9223000824402309,
+      "grad_norm": 0.3545740104488335,
+      "learning_rate": 3.6569587766496216e-06,
+      "loss": 1.1188,
+      "step": 4475
+    },
+    {
+      "epoch": 0.9233305853256389,
+      "grad_norm": 0.30405956916616017,
+      "learning_rate": 3.561176550191203e-06,
+      "loss": 0.7902,
+      "step": 4480
+    },
+    {
+      "epoch": 0.924361088211047,
+      "grad_norm": 0.40805109311840626,
+      "learning_rate": 3.46664263740224e-06,
+      "loss": 1.0949,
+      "step": 4485
+    },
+    {
+      "epoch": 0.9253915910964551,
+      "grad_norm": 0.2937050314911077,
+      "learning_rate": 3.3733582619390523e-06,
+      "loss": 0.9339,
+      "step": 4490
+    },
+    {
+      "epoch": 0.9264220939818631,
+      "grad_norm": 0.37795976499150646,
+      "learning_rate": 3.281324631283833e-06,
+      "loss": 1.0108,
+      "step": 4495
+    },
+    {
+      "epoch": 0.9274525968672712,
+      "grad_norm": 0.4168578135164906,
+      "learning_rate": 3.1905429367289795e-06,
+      "loss": 1.1267,
+      "step": 4500
+    },
+    {
+      "epoch": 0.9284830997526793,
+      "grad_norm": 0.30324094735934326,
+      "learning_rate": 3.101014353361753e-06,
+      "loss": 0.8109,
+      "step": 4505
+    },
+    {
+      "epoch": 0.9295136026380874,
+      "grad_norm": 0.3929210483289662,
+      "learning_rate": 3.012740040048978e-06,
+      "loss": 1.1848,
+      "step": 4510
+    },
+    {
+      "epoch": 0.9305441055234954,
+      "grad_norm": 0.2889829253469363,
+      "learning_rate": 2.9257211394220773e-06,
+      "loss": 0.8944,
+      "step": 4515
+    },
+    {
+      "epoch": 0.9315746084089035,
+      "grad_norm": 0.39922043302261206,
+      "learning_rate": 2.8399587778623505e-06,
+      "loss": 1.0599,
+      "step": 4520
+    },
+    {
+      "epoch": 0.9326051112943117,
+      "grad_norm": 0.41401354380939953,
+      "learning_rate": 2.755454065486263e-06,
+      "loss": 1.1731,
+      "step": 4525
+    },
+    {
+      "epoch": 0.9336356141797197,
+      "grad_norm": 0.3816124740915107,
+      "learning_rate": 2.672208096131157e-06,
+      "loss": 0.9285,
+      "step": 4530
+    },
+    {
+      "epoch": 0.9346661170651278,
+      "grad_norm": 0.365839138684449,
+      "learning_rate": 2.5902219473411204e-06,
+      "loss": 1.1613,
+      "step": 4535
+    },
+    {
+      "epoch": 0.9356966199505359,
+      "grad_norm": 0.30150334355809666,
+      "learning_rate": 2.509496680352963e-06,
+      "loss": 0.9157,
+      "step": 4540
+    },
+    {
+      "epoch": 0.936727122835944,
+      "grad_norm": 0.36964220055921937,
+      "learning_rate": 2.430033340082516e-06,
+      "loss": 1.0569,
+      "step": 4545
+    },
+    {
+      "epoch": 0.937757625721352,
+      "grad_norm": 0.41058988578726896,
+      "learning_rate": 2.3518329551111217e-06,
+      "loss": 1.1936,
+      "step": 4550
+    },
+    {
+      "epoch": 0.9387881286067601,
+      "grad_norm": 0.32143724004908797,
+      "learning_rate": 2.2748965376723e-06,
+      "loss": 0.8617,
+      "step": 4555
+    },
+    {
+      "epoch": 0.9398186314921682,
+      "grad_norm": 0.3695703528509068,
+      "learning_rate": 2.199225083638656e-06,
+      "loss": 1.2086,
+      "step": 4560
+    },
+    {
+      "epoch": 0.9408491343775762,
+      "grad_norm": 0.28544602650039125,
+      "learning_rate": 2.1248195725089624e-06,
+      "loss": 0.9673,
+      "step": 4565
+    },
+    {
+      "epoch": 0.9418796372629843,
+      "grad_norm": 0.33475042033286156,
+      "learning_rate": 2.0516809673955083e-06,
+      "loss": 1.0888,
+      "step": 4570
+    },
+    {
+      "epoch": 0.9429101401483924,
+      "grad_norm": 0.40597870855708795,
+      "learning_rate": 1.9798102150116573e-06,
+      "loss": 1.1073,
+      "step": 4575
+    },
+    {
+      "epoch": 0.9439406430338005,
+      "grad_norm": 0.35068928153409934,
+      "learning_rate": 1.909208245659522e-06,
+      "loss": 0.8919,
+      "step": 4580
+    },
+    {
+      "epoch": 0.9449711459192086,
+      "grad_norm": 0.3520873665570541,
+      "learning_rate": 1.8398759732179637e-06,
+      "loss": 1.1762,
+      "step": 4585
+    },
+    {
+      "epoch": 0.9460016488046167,
+      "grad_norm": 0.31075650924614495,
+      "learning_rate": 1.7718142951307914e-06,
+      "loss": 0.8997,
+      "step": 4590
+    },
+    {
+      "epoch": 0.9470321516900247,
+      "grad_norm": 0.3520700228079484,
+      "learning_rate": 1.705024092395091e-06,
+      "loss": 1.0955,
+      "step": 4595
+    },
+    {
+      "epoch": 0.9480626545754328,
+      "grad_norm": 0.3680157698495691,
+      "learning_rate": 1.6395062295498698e-06,
+      "loss": 1.178,
+      "step": 4600
+    },
+    {
+      "epoch": 0.9480626545754328,
+      "eval_loss": 0.9628809690475464,
+      "eval_runtime": 2669.1038,
+      "eval_samples_per_second": 2.997,
+      "eval_steps_per_second": 0.187,
+      "step": 4600
+    },
+    {
+      "epoch": 0.9490931574608409,
+      "grad_norm": 0.31126006459380184,
+      "learning_rate": 1.5752615546647975e-06,
+      "loss": 0.8694,
+      "step": 4605
+    },
+    {
+      "epoch": 0.950123660346249,
+      "grad_norm": 0.34126438848006335,
+      "learning_rate": 1.5122908993293273e-06,
+      "loss": 1.1496,
+      "step": 4610
+    },
+    {
+      "epoch": 0.951154163231657,
+      "grad_norm": 0.28191122396530277,
+      "learning_rate": 1.4505950786418255e-06,
+      "loss": 0.9136,
+      "step": 4615
+    },
+    {
+      "epoch": 0.9521846661170651,
+      "grad_norm": 0.3757522319369373,
+      "learning_rate": 1.3901748911991253e-06,
+      "loss": 1.0589,
+      "step": 4620
+    },
+    {
+      "epoch": 0.9532151690024732,
+      "grad_norm": 0.3812139506827204,
+      "learning_rate": 1.331031119086079e-06,
+      "loss": 1.1577,
+      "step": 4625
+    },
+    {
+      "epoch": 0.9542456718878813,
+      "grad_norm": 0.3285095663923165,
+      "learning_rate": 1.2731645278655445e-06,
+      "loss": 0.9256,
+      "step": 4630
+    },
+    {
+      "epoch": 0.9552761747732894,
+      "grad_norm": 0.3434798919525598,
+      "learning_rate": 1.2165758665683924e-06,
+      "loss": 1.1373,
+      "step": 4635
+    },
+    {
+      "epoch": 0.9563066776586975,
+      "grad_norm": 0.3017676114908958,
+      "learning_rate": 1.1612658676838473e-06,
+      "loss": 0.8882,
+      "step": 4640
+    },
+    {
+      "epoch": 0.9573371805441055,
+      "grad_norm": 0.3795284009217967,
+      "learning_rate": 1.107235247150018e-06,
+      "loss": 1.0639,
+      "step": 4645
+    },
+    {
+      "epoch": 0.9583676834295136,
+      "grad_norm": 0.4845153985581749,
+      "learning_rate": 1.0544847043445938e-06,
+      "loss": 1.0753,
+      "step": 4650
+    },
+    {
+      "epoch": 0.9593981863149217,
+      "grad_norm": 0.32363946255340464,
+      "learning_rate": 1.0030149220758288e-06,
+      "loss": 0.8307,
+      "step": 4655
+    },
+    {
+      "epoch": 0.9604286892003298,
+      "grad_norm": 0.4184299714975136,
+      "learning_rate": 9.528265665736502e-07,
+      "loss": 1.1371,
+      "step": 4660
+    },
+    {
+      "epoch": 0.9614591920857378,
+      "grad_norm": 0.28512105148343453,
+      "learning_rate": 9.039202874811192e-07,
+      "loss": 0.9634,
+      "step": 4665
+    },
+    {
+      "epoch": 0.9624896949711459,
+      "grad_norm": 0.360971597617964,
+      "learning_rate": 8.562967178459391e-07,
+      "loss": 1.0805,
+      "step": 4670
+    },
+    {
+      "epoch": 0.963520197856554,
+      "grad_norm": 0.3822471953295701,
+      "learning_rate": 8.099564741123166e-07,
+      "loss": 1.2147,
+      "step": 4675
+    },
+    {
+      "epoch": 0.964550700741962,
+      "grad_norm": 0.32385746632379775,
+      "learning_rate": 7.649001561129354e-07,
+      "loss": 0.8677,
+      "step": 4680
+    },
+    {
+      "epoch": 0.9655812036273702,
+      "grad_norm": 0.33325536515100207,
+      "learning_rate": 7.211283470612395e-07,
+      "loss": 1.1736,
+      "step": 4685
+    },
+    {
+      "epoch": 0.9666117065127783,
+      "grad_norm": 0.2789936033919106,
+      "learning_rate": 6.786416135438512e-07,
+      "loss": 0.9563,
+      "step": 4690
+    },
+    {
+      "epoch": 0.9676422093981863,
+      "grad_norm": 0.3347741972532808,
+      "learning_rate": 6.374405055132537e-07,
+      "loss": 0.9439,
+      "step": 4695
+    },
+    {
+      "epoch": 0.9686727122835944,
+      "grad_norm": 0.3903192561742831,
+      "learning_rate": 5.975255562806647e-07,
+      "loss": 1.1797,
+      "step": 4700
+    },
+    {
+      "epoch": 0.9697032151690025,
+      "grad_norm": 0.32771315967905756,
+      "learning_rate": 5.58897282509141e-07,
+      "loss": 0.8953,
+      "step": 4705
+    },
+    {
+      "epoch": 0.9707337180544106,
+      "grad_norm": 0.34884009544181055,
+      "learning_rate": 5.215561842068728e-07,
+      "loss": 1.1761,
+      "step": 4710
+    },
+    {
+      "epoch": 0.9717642209398186,
+      "grad_norm": 0.2775010942276059,
+      "learning_rate": 4.855027447207338e-07,
+      "loss": 0.9457,
+      "step": 4715
+    },
+    {
+      "epoch": 0.9727947238252267,
+      "grad_norm": 0.382431683738394,
+      "learning_rate": 4.507374307299972e-07,
+      "loss": 1.0359,
+      "step": 4720
+    },
+    {
+      "epoch": 0.9738252267106348,
+      "grad_norm": 0.4253814777599987,
+      "learning_rate": 4.172606922403399e-07,
+      "loss": 1.1746,
+      "step": 4725
+    },
+    {
+      "epoch": 0.9748557295960428,
+      "grad_norm": 0.3119359666789732,
+      "learning_rate": 3.8507296257798145e-07,
+      "loss": 0.8048,
+      "step": 4730
+    },
+    {
+      "epoch": 0.9758862324814509,
+      "grad_norm": 0.35641657318401637,
+      "learning_rate": 3.541746583840655e-07,
+      "loss": 1.1508,
+      "step": 4735
+    },
+    {
+      "epoch": 0.9769167353668591,
+      "grad_norm": 0.28277759207142494,
+      "learning_rate": 3.24566179609298e-07,
+      "loss": 0.9406,
+      "step": 4740
+    },
+    {
+      "epoch": 0.9779472382522671,
+      "grad_norm": 0.3649978639842664,
+      "learning_rate": 2.9624790950875113e-07,
+      "loss": 1.0074,
+      "step": 4745
+    },
+    {
+      "epoch": 0.9789777411376752,
+      "grad_norm": 0.43930720949024876,
+      "learning_rate": 2.692202146369338e-07,
+      "loss": 1.1133,
+      "step": 4750
+    },
+    {
+      "epoch": 0.9800082440230833,
+      "grad_norm": 0.3568172083915493,
+      "learning_rate": 2.434834448429957e-07,
+      "loss": 0.8548,
+      "step": 4755
+    },
+    {
+      "epoch": 0.9810387469084914,
+      "grad_norm": 0.3512721884204812,
+      "learning_rate": 2.1903793326621957e-07,
+      "loss": 1.1658,
+      "step": 4760
+    },
+    {
+      "epoch": 0.9820692497938994,
+      "grad_norm": 0.28260427199429944,
+      "learning_rate": 1.9588399633173605e-07,
+      "loss": 0.8872,
+      "step": 4765
+    },
+    {
+      "epoch": 0.9830997526793075,
+      "grad_norm": 0.3342814004364708,
+      "learning_rate": 1.740219337463822e-07,
+      "loss": 1.0368,
+      "step": 4770
+    },
+    {
+      "epoch": 0.9841302555647156,
+      "grad_norm": 0.39410468491852235,
+      "learning_rate": 1.534520284948715e-07,
+      "loss": 1.1301,
+      "step": 4775
+    },
+    {
+      "epoch": 0.9851607584501236,
+      "grad_norm": 0.32300550472176315,
+      "learning_rate": 1.3417454683608554e-07,
+      "loss": 0.8525,
+      "step": 4780
+    },
+    {
+      "epoch": 0.9861912613355317,
+      "grad_norm": 0.38783208707768235,
+      "learning_rate": 1.1618973829966572e-07,
+      "loss": 1.2139,
+      "step": 4785
+    },
+    {
+      "epoch": 0.9872217642209398,
+      "grad_norm": 0.27657069522605154,
+      "learning_rate": 9.949783568272697e-08,
+      "loss": 0.9094,
+      "step": 4790
+    },
+    {
+      "epoch": 0.988252267106348,
+      "grad_norm": 0.37764224067778107,
+      "learning_rate": 8.409905504693782e-08,
+      "loss": 1.0543,
+      "step": 4795
+    },
+    {
+      "epoch": 0.989282769991756,
+      "grad_norm": 0.5230932784753239,
+      "learning_rate": 6.999359571561171e-08,
+      "loss": 1.1661,
+      "step": 4800
+    },
+    {
+      "epoch": 0.989282769991756,
+      "eval_loss": 0.9627586603164673,
+      "eval_runtime": 2560.4353,
+      "eval_samples_per_second": 3.124,
+      "eval_steps_per_second": 0.195,
+      "step": 4800
+    },
+    {
+      "epoch": 0.9903132728771641,
+      "grad_norm": 0.2751657982671328,
+      "learning_rate": 5.718164027121997e-08,
+      "loss": 0.7531,
+      "step": 4805
+    },
+    {
+      "epoch": 0.9913437757625722,
+      "grad_norm": 0.3764599692340869,
+      "learning_rate": 4.566335455299387e-08,
+      "loss": 1.1338,
+      "step": 4810
+    },
+    {
+      "epoch": 0.9923742786479802,
+      "grad_norm": 0.2992197664538524,
+      "learning_rate": 3.5438887654737355e-08,
+      "loss": 0.8897,
+      "step": 4815
+    },
+    {
+      "epoch": 0.9934047815333883,
+      "grad_norm": 0.3535257610205098,
+      "learning_rate": 2.6508371922984166e-08,
+      "loss": 1.0454,
+      "step": 4820
+    },
+    {
+      "epoch": 0.9944352844187964,
+      "grad_norm": 0.37026464651118146,
+      "learning_rate": 1.887192295521034e-08,
+      "loss": 1.1346,
+      "step": 4825
+    },
+    {
+      "epoch": 0.9954657873042044,
+      "grad_norm": 0.3271412553655858,
+      "learning_rate": 1.252963959834652e-08,
+      "loss": 0.8182,
+      "step": 4830
+    },
+    {
+      "epoch": 0.9964962901896125,
+      "grad_norm": 0.4040096861009905,
+      "learning_rate": 7.481603947556703e-09,
+      "loss": 1.0549,
+      "step": 4835
+    },
+    {
+      "epoch": 0.9975267930750206,
+      "grad_norm": 0.28077083214803567,
+      "learning_rate": 3.727881345105821e-09,
+      "loss": 0.918,
+      "step": 4840
+    },
+    {
+      "epoch": 0.9985572959604286,
+      "grad_norm": 0.38213498100295196,
+      "learning_rate": 1.2685203795492762e-09,
+      "loss": 1.0451,
+      "step": 4845
+    },
+    {
+      "epoch": 0.9995877988458368,
+      "grad_norm": 0.4052030227111039,
+      "learning_rate": 1.0355288510011107e-10,
+      "loss": 1.1878,
+      "step": 4850
+    },
+    {
+      "epoch": 1.0,
+      "step": 4852,
+      "total_flos": 3.101569705521971e+16,
+      "train_loss": 1.1041986294158312,
+      "train_runtime": 123226.4419,
+      "train_samples_per_second": 0.63,
+      "train_steps_per_second": 0.039
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 4852,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 200,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.101569705521971e+16,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}