diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,17694 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.99927797833935,
+  "eval_steps": 500,
+  "global_step": 12462,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0012033694344163659,
+      "grad_norm": 5.875,
+      "learning_rate": 1.1482631771428573e-05,
+      "loss": 1.9783,
+      "step": 5
+    },
+    {
+      "epoch": 0.0024067388688327317,
+      "grad_norm": 3.296875,
+      "learning_rate": 2.5835921485714284e-05,
+      "loss": 1.9013,
+      "step": 10
+    },
+    {
+      "epoch": 0.0036101083032490976,
+      "grad_norm": 3.03125,
+      "learning_rate": 4.0189211200000005e-05,
+      "loss": 1.8287,
+      "step": 15
+    },
+    {
+      "epoch": 0.0048134777376654635,
+      "grad_norm": 2.75,
+      "learning_rate": 5.454250091428571e-05,
+      "loss": 1.7177,
+      "step": 20
+    },
+    {
+      "epoch": 0.006016847172081829,
+      "grad_norm": 2.734375,
+      "learning_rate": 6.889579062857144e-05,
+      "loss": 1.5836,
+      "step": 25
+    },
+    {
+      "epoch": 0.007220216606498195,
+      "grad_norm": 3.0,
+      "learning_rate": 8.324908034285715e-05,
+      "loss": 1.5078,
+      "step": 30
+    },
+    {
+      "epoch": 0.00842358604091456,
+      "grad_norm": 3.125,
+      "learning_rate": 9.760237005714287e-05,
+      "loss": 1.4619,
+      "step": 35
+    },
+    {
+      "epoch": 0.009626955475330927,
+      "grad_norm": 3.15625,
+      "learning_rate": 0.0001004730087456795,
+      "loss": 1.447,
+      "step": 40
+    },
+    {
+      "epoch": 0.010830324909747292,
+      "grad_norm": 2.9375,
+      "learning_rate": 0.00010047293052503616,
+      "loss": 1.3868,
+      "step": 45
+    },
+    {
+      "epoch": 0.012033694344163659,
+      "grad_norm": 2.671875,
+      "learning_rate": 0.0001004727921347998,
+      "loss": 1.3767,
+      "step": 50
+    },
+    {
+      "epoch": 0.013237063778580024,
+      "grad_norm": 2.578125,
+      "learning_rate": 0.00010047259357519138,
+      "loss": 1.3253,
+      "step": 55
+    },
+    {
+      "epoch": 0.01444043321299639,
+      "grad_norm": 2.734375,
+      "learning_rate": 0.00010047233484652802,
+      "loss": 1.3186,
+      "step": 60
+    },
+    {
+      "epoch": 0.015643802647412757,
+      "grad_norm": 2.75,
+      "learning_rate": 0.00010047201594922292,
+      "loss": 1.2852,
+      "step": 65
+    },
+    {
+      "epoch": 0.01684717208182912,
+      "grad_norm": 2.734375,
+      "learning_rate": 0.00010047163688378533,
+      "loss": 1.2451,
+      "step": 70
+    },
+    {
+      "epoch": 0.018050541516245487,
+      "grad_norm": 2.328125,
+      "learning_rate": 0.0001004711976508206,
+      "loss": 1.2399,
+      "step": 75
+    },
+    {
+      "epoch": 0.019253910950661854,
+      "grad_norm": 2.390625,
+      "learning_rate": 0.00010047069825103019,
+      "loss": 1.2288,
+      "step": 80
+    },
+    {
+      "epoch": 0.02045728038507822,
+      "grad_norm": 2.53125,
+      "learning_rate": 0.00010047013868521161,
+      "loss": 1.2198,
+      "step": 85
+    },
+    {
+      "epoch": 0.021660649819494584,
+      "grad_norm": 2.5625,
+      "learning_rate": 0.00010046951895425849,
+      "loss": 1.2158,
+      "step": 90
+    },
+    {
+      "epoch": 0.02286401925391095,
+      "grad_norm": 2.515625,
+      "learning_rate": 0.00010046883905916052,
+      "loss": 1.2246,
+      "step": 95
+    },
+    {
+      "epoch": 0.024067388688327317,
+      "grad_norm": 2.484375,
+      "learning_rate": 0.00010046809900100347,
+      "loss": 1.1625,
+      "step": 100
+    },
+    {
+      "epoch": 0.02527075812274368,
+      "grad_norm": 2.515625,
+      "learning_rate": 0.00010046729878096918,
+      "loss": 1.1858,
+      "step": 105
+    },
+    {
+      "epoch": 0.026474127557160047,
+      "grad_norm": 2.625,
+      "learning_rate": 0.0001004664384003356,
+      "loss": 1.1756,
+      "step": 110
+    },
+    {
+      "epoch": 0.027677496991576414,
+      "grad_norm": 2.5,
+      "learning_rate": 0.00010046551786047672,
+      "loss": 1.1696,
+      "step": 115
+    },
+    {
+      "epoch": 0.02888086642599278,
+      "grad_norm": 2.390625,
+      "learning_rate": 0.00010046453716286263,
+      "loss": 1.1177,
+      "step": 120
+    },
+    {
+      "epoch": 0.030084235860409144,
+      "grad_norm": 2.375,
+      "learning_rate": 0.00010046349630905946,
+      "loss": 1.1426,
+      "step": 125
+    },
+    {
+      "epoch": 0.031287605294825514,
+      "grad_norm": 2.609375,
+      "learning_rate": 0.00010046239530072942,
+      "loss": 1.1572,
+      "step": 130
+    },
+    {
+      "epoch": 0.032490974729241874,
+      "grad_norm": 2.546875,
+      "learning_rate": 0.00010046123413963083,
+      "loss": 1.1154,
+      "step": 135
+    },
+    {
+      "epoch": 0.03369434416365824,
+      "grad_norm": 2.46875,
+      "learning_rate": 0.00010046001282761799,
+      "loss": 1.1424,
+      "step": 140
+    },
+    {
+      "epoch": 0.03489771359807461,
+      "grad_norm": 2.921875,
+      "learning_rate": 0.00010045873136664134,
+      "loss": 1.1045,
+      "step": 145
+    },
+    {
+      "epoch": 0.036101083032490974,
+      "grad_norm": 2.59375,
+      "learning_rate": 0.00010045738975874728,
+      "loss": 1.1311,
+      "step": 150
+    },
+    {
+      "epoch": 0.03730445246690734,
+      "grad_norm": 2.546875,
+      "learning_rate": 0.00010045598800607839,
+      "loss": 1.1298,
+      "step": 155
+    },
+    {
+      "epoch": 0.03850782190132371,
+      "grad_norm": 2.53125,
+      "learning_rate": 0.00010045452611087318,
+      "loss": 1.0965,
+      "step": 160
+    },
+    {
+      "epoch": 0.039711191335740074,
+      "grad_norm": 2.75,
+      "learning_rate": 0.00010045300407546628,
+      "loss": 1.0964,
+      "step": 165
+    },
+    {
+      "epoch": 0.04091456077015644,
+      "grad_norm": 2.984375,
+      "learning_rate": 0.00010045142190228834,
+      "loss": 1.0847,
+      "step": 170
+    },
+    {
+      "epoch": 0.0421179302045728,
+      "grad_norm": 2.421875,
+      "learning_rate": 0.00010044977959386601,
+      "loss": 1.073,
+      "step": 175
+    },
+    {
+      "epoch": 0.04332129963898917,
+      "grad_norm": 2.671875,
+      "learning_rate": 0.00010044807715282207,
+      "loss": 1.1091,
+      "step": 180
+    },
+    {
+      "epoch": 0.044524669073405534,
+      "grad_norm": 2.296875,
+      "learning_rate": 0.00010044631458187523,
+      "loss": 1.0794,
+      "step": 185
+    },
+    {
+      "epoch": 0.0457280385078219,
+      "grad_norm": 2.453125,
+      "learning_rate": 0.00010044449188384028,
+      "loss": 1.1212,
+      "step": 190
+    },
+    {
+      "epoch": 0.04693140794223827,
+      "grad_norm": 2.421875,
+      "learning_rate": 0.00010044260906162802,
+      "loss": 1.0835,
+      "step": 195
+    },
+    {
+      "epoch": 0.048134777376654635,
+      "grad_norm": 2.5,
+      "learning_rate": 0.00010044066611824526,
+      "loss": 1.0714,
+      "step": 200
+    },
+    {
+      "epoch": 0.049338146811071,
+      "grad_norm": 2.328125,
+      "learning_rate": 0.00010043866305679484,
+      "loss": 1.0716,
+      "step": 205
+    },
+    {
+      "epoch": 0.05054151624548736,
+      "grad_norm": 2.46875,
+      "learning_rate": 0.00010043659988047559,
+      "loss": 1.0586,
+      "step": 210
+    },
+    {
+      "epoch": 0.05174488567990373,
+      "grad_norm": 2.265625,
+      "learning_rate": 0.00010043447659258233,
+      "loss": 1.0528,
+      "step": 215
+    },
+    {
+      "epoch": 0.052948255114320095,
+      "grad_norm": 3.03125,
+      "learning_rate": 0.0001004322931965059,
+      "loss": 1.0601,
+      "step": 220
+    },
+    {
+      "epoch": 0.05415162454873646,
+      "grad_norm": 2.59375,
+      "learning_rate": 0.00010043004969573312,
+      "loss": 1.0495,
+      "step": 225
+    },
+    {
+      "epoch": 0.05535499398315283,
+      "grad_norm": 2.3125,
+      "learning_rate": 0.00010042774609384681,
+      "loss": 1.058,
+      "step": 230
+    },
+    {
+      "epoch": 0.056558363417569195,
+      "grad_norm": 2.578125,
+      "learning_rate": 0.00010042538239452578,
+      "loss": 1.0285,
+      "step": 235
+    },
+    {
+      "epoch": 0.05776173285198556,
+      "grad_norm": 2.515625,
+      "learning_rate": 0.00010042295860154474,
+      "loss": 1.0332,
+      "step": 240
+    },
+    {
+      "epoch": 0.05896510228640193,
+      "grad_norm": 2.484375,
+      "learning_rate": 0.00010042047471877447,
+      "loss": 1.0177,
+      "step": 245
+    },
+    {
+      "epoch": 0.06016847172081829,
+      "grad_norm": 2.796875,
+      "learning_rate": 0.00010041793075018161,
+      "loss": 1.0364,
+      "step": 250
+    },
+    {
+      "epoch": 0.061371841155234655,
+      "grad_norm": 2.3125,
+      "learning_rate": 0.00010041532669982886,
+      "loss": 1.0268,
+      "step": 255
+    },
+    {
+      "epoch": 0.06257521058965103,
+      "grad_norm": 2.515625,
+      "learning_rate": 0.00010041266257187478,
+      "loss": 0.9869,
+      "step": 260
+    },
+    {
+      "epoch": 0.06377858002406739,
+      "grad_norm": 2.703125,
+      "learning_rate": 0.00010040993837057395,
+      "loss": 1.0372,
+      "step": 265
+    },
+    {
+      "epoch": 0.06498194945848375,
+      "grad_norm": 2.40625,
+      "learning_rate": 0.00010040715410027679,
+      "loss": 0.987,
+      "step": 270
+    },
+    {
+      "epoch": 0.06618531889290012,
+      "grad_norm": 2.265625,
+      "learning_rate": 0.00010040430976542976,
+      "loss": 1.0099,
+      "step": 275
+    },
+    {
+      "epoch": 0.06738868832731648,
+      "grad_norm": 2.734375,
+      "learning_rate": 0.00010040140537057514,
+      "loss": 1.0188,
+      "step": 280
+    },
+    {
+      "epoch": 0.06859205776173286,
+      "grad_norm": 2.484375,
+      "learning_rate": 0.00010039844092035118,
+      "loss": 1.0453,
+      "step": 285
+    },
+    {
+      "epoch": 0.06979542719614922,
+      "grad_norm": 2.296875,
+      "learning_rate": 0.00010039541641949205,
+      "loss": 1.0081,
+      "step": 290
+    },
+    {
+      "epoch": 0.07099879663056559,
+      "grad_norm": 2.46875,
+      "learning_rate": 0.00010039233187282776,
+      "loss": 0.994,
+      "step": 295
+    },
+    {
+      "epoch": 0.07220216606498195,
+      "grad_norm": 2.375,
+      "learning_rate": 0.00010038918728528426,
+      "loss": 1.0029,
+      "step": 300
+    },
+    {
+      "epoch": 0.07340553549939831,
+      "grad_norm": 2.453125,
+      "learning_rate": 0.00010038598266188339,
+      "loss": 1.0095,
+      "step": 305
+    },
+    {
+      "epoch": 0.07460890493381468,
+      "grad_norm": 2.25,
+      "learning_rate": 0.00010038271800774282,
+      "loss": 0.9699,
+      "step": 310
+    },
+    {
+      "epoch": 0.07581227436823104,
+      "grad_norm": 2.40625,
+      "learning_rate": 0.0001003793933280761,
+      "loss": 0.9616,
+      "step": 315
+    },
+    {
+      "epoch": 0.07701564380264742,
+      "grad_norm": 2.34375,
+      "learning_rate": 0.0001003760086281927,
+      "loss": 1.0072,
+      "step": 320
+    },
+    {
+      "epoch": 0.07821901323706378,
+      "grad_norm": 2.203125,
+      "learning_rate": 0.00010037256391349785,
+      "loss": 0.9806,
+      "step": 325
+    },
+    {
+      "epoch": 0.07942238267148015,
+      "grad_norm": 2.484375,
+      "learning_rate": 0.00010036905918949269,
+      "loss": 0.9474,
+      "step": 330
+    },
+    {
+      "epoch": 0.08062575210589651,
+      "grad_norm": 2.515625,
+      "learning_rate": 0.00010036549446177414,
+      "loss": 1.0093,
+      "step": 335
+    },
+    {
+      "epoch": 0.08182912154031288,
+      "grad_norm": 2.5625,
+      "learning_rate": 0.000100361869736035,
+      "loss": 0.952,
+      "step": 340
+    },
+    {
+      "epoch": 0.08303249097472924,
+      "grad_norm": 2.171875,
+      "learning_rate": 0.00010035818501806385,
+      "loss": 0.9729,
+      "step": 345
+    },
+    {
+      "epoch": 0.0842358604091456,
+      "grad_norm": 2.375,
+      "learning_rate": 0.00010035444031374507,
+      "loss": 0.947,
+      "step": 350
+    },
+    {
+      "epoch": 0.08543922984356198,
+      "grad_norm": 2.125,
+      "learning_rate": 0.00010035063562905885,
+      "loss": 0.9918,
+      "step": 355
+    },
+    {
+      "epoch": 0.08664259927797834,
+      "grad_norm": 2.09375,
+      "learning_rate": 0.00010034677097008121,
+      "loss": 0.9803,
+      "step": 360
+    },
+    {
+      "epoch": 0.08784596871239471,
+      "grad_norm": 2.09375,
+      "learning_rate": 0.00010034284634298385,
+      "loss": 0.9303,
+      "step": 365
+    },
+    {
+      "epoch": 0.08904933814681107,
+      "grad_norm": 2.40625,
+      "learning_rate": 0.0001003388617540343,
+      "loss": 0.9267,
+      "step": 370
+    },
+    {
+      "epoch": 0.09025270758122744,
+      "grad_norm": 2.15625,
+      "learning_rate": 0.00010033481720959588,
+      "loss": 0.9313,
+      "step": 375
+    },
+    {
+      "epoch": 0.0914560770156438,
+      "grad_norm": 2.3125,
+      "learning_rate": 0.00010033071271612757,
+      "loss": 0.9154,
+      "step": 380
+    },
+    {
+      "epoch": 0.09265944645006016,
+      "grad_norm": 2.4375,
+      "learning_rate": 0.00010032654828018417,
+      "loss": 0.9725,
+      "step": 385
+    },
+    {
+      "epoch": 0.09386281588447654,
+      "grad_norm": 2.3125,
+      "learning_rate": 0.00010032232390841613,
+      "loss": 0.9499,
+      "step": 390
+    },
+    {
+      "epoch": 0.0950661853188929,
+      "grad_norm": 2.359375,
+      "learning_rate": 0.00010031803960756968,
+      "loss": 0.926,
+      "step": 395
+    },
+    {
+      "epoch": 0.09626955475330927,
+      "grad_norm": 2.359375,
+      "learning_rate": 0.00010031369538448673,
+      "loss": 0.9864,
+      "step": 400
+    },
+    {
+      "epoch": 0.09747292418772563,
+      "grad_norm": 2.15625,
+      "learning_rate": 0.00010030929124610487,
+      "loss": 0.9206,
+      "step": 405
+    },
+    {
+      "epoch": 0.098676293622142,
+      "grad_norm": 2.28125,
+      "learning_rate": 0.0001003048271994574,
+      "loss": 0.9297,
+      "step": 410
+    },
+    {
+      "epoch": 0.09987966305655836,
+      "grad_norm": 2.421875,
+      "learning_rate": 0.0001003003032516733,
+      "loss": 0.942,
+      "step": 415
+    },
+    {
+      "epoch": 0.10108303249097472,
+      "grad_norm": 2.140625,
+      "learning_rate": 0.00010029571940997716,
+      "loss": 0.937,
+      "step": 420
+    },
+    {
+      "epoch": 0.1022864019253911,
+      "grad_norm": 2.25,
+      "learning_rate": 0.00010029107568168925,
+      "loss": 0.9598,
+      "step": 425
+    },
+    {
+      "epoch": 0.10348977135980746,
+      "grad_norm": 2.4375,
+      "learning_rate": 0.0001002863720742255,
+      "loss": 0.9444,
+      "step": 430
+    },
+    {
+      "epoch": 0.10469314079422383,
+      "grad_norm": 2.109375,
+      "learning_rate": 0.00010028160859509742,
+      "loss": 0.9544,
+      "step": 435
+    },
+    {
+      "epoch": 0.10589651022864019,
+      "grad_norm": 2.40625,
+      "learning_rate": 0.00010027678525191216,
+      "loss": 0.9032,
+      "step": 440
+    },
+    {
+      "epoch": 0.10709987966305656,
+      "grad_norm": 2.171875,
+      "learning_rate": 0.00010027190205237246,
+      "loss": 0.9468,
+      "step": 445
+    },
+    {
+      "epoch": 0.10830324909747292,
+      "grad_norm": 2.203125,
+      "learning_rate": 0.00010026695900427668,
+      "loss": 0.9244,
+      "step": 450
+    },
+    {
+      "epoch": 0.1095066185318893,
+      "grad_norm": 2.390625,
+      "learning_rate": 0.00010026195611551872,
+      "loss": 0.9286,
+      "step": 455
+    },
+    {
+      "epoch": 0.11070998796630566,
+      "grad_norm": 2.390625,
+      "learning_rate": 0.00010025689339408803,
+      "loss": 0.9255,
+      "step": 460
+    },
+    {
+      "epoch": 0.11191335740072202,
+      "grad_norm": 2.484375,
+      "learning_rate": 0.00010025177084806967,
+      "loss": 0.9652,
+      "step": 465
+    },
+    {
+      "epoch": 0.11311672683513839,
+      "grad_norm": 2.125,
+      "learning_rate": 0.00010024658848564418,
+      "loss": 0.921,
+      "step": 470
+    },
+    {
+      "epoch": 0.11432009626955475,
+      "grad_norm": 2.546875,
+      "learning_rate": 0.00010024134631508768,
+      "loss": 0.9406,
+      "step": 475
+    },
+    {
+      "epoch": 0.11552346570397112,
+      "grad_norm": 2.21875,
+      "learning_rate": 0.00010023604434477176,
+      "loss": 0.9309,
+      "step": 480
+    },
+    {
+      "epoch": 0.11672683513838748,
+      "grad_norm": 2.4375,
+      "learning_rate": 0.00010023068258316352,
+      "loss": 0.925,
+      "step": 485
+    },
+    {
+      "epoch": 0.11793020457280386,
+      "grad_norm": 2.21875,
+      "learning_rate": 0.00010022526103882553,
+      "loss": 0.8907,
+      "step": 490
+    },
+    {
+      "epoch": 0.11913357400722022,
+      "grad_norm": 2.21875,
+      "learning_rate": 0.00010021977972041586,
+      "loss": 0.9072,
+      "step": 495
+    },
+    {
+      "epoch": 0.12033694344163658,
+      "grad_norm": 2.359375,
+      "learning_rate": 0.00010021423863668802,
+      "loss": 0.8906,
+      "step": 500
+    },
+    {
+      "epoch": 0.12033694344163658,
+      "eval_loss": 0.8048076033592224,
+      "eval_runtime": 2.7309,
+      "eval_samples_per_second": 73.235,
+      "eval_steps_per_second": 73.235,
+      "step": 500
+    },
+    {
+      "epoch": 0.12154031287605295,
+      "grad_norm": 2.203125,
+      "learning_rate": 0.000100208637796491,
+      "loss": 0.8702,
+      "step": 505
+    },
+    {
+      "epoch": 0.12274368231046931,
+      "grad_norm": 2.1875,
+      "learning_rate": 0.00010020297720876913,
+      "loss": 0.9234,
+      "step": 510
+    },
+    {
+      "epoch": 0.12394705174488568,
+      "grad_norm": 2.171875,
+      "learning_rate": 0.00010019725688256226,
+      "loss": 0.9286,
+      "step": 515
+    },
+    {
+      "epoch": 0.12515042117930206,
+      "grad_norm": 2.578125,
+      "learning_rate": 0.00010019147682700556,
+      "loss": 0.9069,
+      "step": 520
+    },
+    {
+      "epoch": 0.1263537906137184,
+      "grad_norm": 2.375,
+      "learning_rate": 0.00010018563705132966,
+      "loss": 0.8735,
+      "step": 525
+    },
+    {
+      "epoch": 0.12755716004813478,
+      "grad_norm": 2.125,
+      "learning_rate": 0.00010017973756486048,
+      "loss": 0.9177,
+      "step": 530
+    },
+    {
+      "epoch": 0.12876052948255115,
+      "grad_norm": 2.34375,
+      "learning_rate": 0.00010017377837701939,
+      "loss": 0.8844,
+      "step": 535
+    },
+    {
+      "epoch": 0.1299638989169675,
+      "grad_norm": 2.0625,
+      "learning_rate": 0.00010016775949732298,
+      "loss": 0.9013,
+      "step": 540
+    },
+    {
+      "epoch": 0.13116726835138387,
+      "grad_norm": 1.9609375,
+      "learning_rate": 0.0001001616809353833,
+      "loss": 0.8743,
+      "step": 545
+    },
+    {
+      "epoch": 0.13237063778580024,
+      "grad_norm": 2.546875,
+      "learning_rate": 0.00010015554270090759,
+      "loss": 0.9059,
+      "step": 550
+    },
+    {
+      "epoch": 0.13357400722021662,
+      "grad_norm": 2.109375,
+      "learning_rate": 0.00010014934480369846,
+      "loss": 0.8756,
+      "step": 555
+    },
+    {
+      "epoch": 0.13477737665463296,
+      "grad_norm": 2.203125,
+      "learning_rate": 0.00010014308725365383,
+      "loss": 0.8941,
+      "step": 560
+    },
+    {
+      "epoch": 0.13598074608904934,
+      "grad_norm": 2.375,
+      "learning_rate": 0.00010013677006076677,
+      "loss": 0.8459,
+      "step": 565
+    },
+    {
+      "epoch": 0.1371841155234657,
+      "grad_norm": 2.28125,
+      "learning_rate": 0.00010013039323512569,
+      "loss": 0.8917,
+      "step": 570
+    },
+    {
+      "epoch": 0.13838748495788206,
+      "grad_norm": 2.421875,
+      "learning_rate": 0.00010012395678691421,
+      "loss": 0.8788,
+      "step": 575
+    },
+    {
+      "epoch": 0.13959085439229843,
+      "grad_norm": 2.28125,
+      "learning_rate": 0.00010011746072641115,
+      "loss": 0.878,
+      "step": 580
+    },
+    {
+      "epoch": 0.1407942238267148,
+      "grad_norm": 2.265625,
+      "learning_rate": 0.00010011090506399054,
+      "loss": 0.8567,
+      "step": 585
+    },
+    {
+      "epoch": 0.14199759326113118,
+      "grad_norm": 2.28125,
+      "learning_rate": 0.00010010428981012161,
+      "loss": 0.9212,
+      "step": 590
+    },
+    {
+      "epoch": 0.14320096269554752,
+      "grad_norm": 2.09375,
+      "learning_rate": 0.00010009761497536873,
+      "loss": 0.8739,
+      "step": 595
+    },
+    {
+      "epoch": 0.1444043321299639,
+      "grad_norm": 2.25,
+      "learning_rate": 0.00010009088057039142,
+      "loss": 0.8645,
+      "step": 600
+    },
+    {
+      "epoch": 0.14560770156438027,
+      "grad_norm": 2.109375,
+      "learning_rate": 0.00010008408660594433,
+      "loss": 0.8443,
+      "step": 605
+    },
+    {
+      "epoch": 0.14681107099879662,
+      "grad_norm": 2.28125,
+      "learning_rate": 0.00010007723309287726,
+      "loss": 0.8587,
+      "step": 610
+    },
+    {
+      "epoch": 0.148014440433213,
+      "grad_norm": 2.234375,
+      "learning_rate": 0.00010007032004213507,
+      "loss": 0.9038,
+      "step": 615
+    },
+    {
+      "epoch": 0.14921780986762936,
+      "grad_norm": 2.90625,
+      "learning_rate": 0.0001000633474647577,
+      "loss": 0.8359,
+      "step": 620
+    },
+    {
+      "epoch": 0.15042117930204574,
+      "grad_norm": 2.3125,
+      "learning_rate": 0.0001000563153718802,
+      "loss": 0.877,
+      "step": 625
+    },
+    {
+      "epoch": 0.15162454873646208,
+      "grad_norm": 2.1875,
+      "learning_rate": 0.00010004922377473258,
+      "loss": 0.8872,
+      "step": 630
+    },
+    {
+      "epoch": 0.15282791817087846,
+      "grad_norm": 2.140625,
+      "learning_rate": 0.00010004207268463997,
+      "loss": 0.85,
+      "step": 635
+    },
+    {
+      "epoch": 0.15403128760529483,
+      "grad_norm": 2.203125,
+      "learning_rate": 0.00010003486211302243,
+      "loss": 0.8759,
+      "step": 640
+    },
+    {
+      "epoch": 0.1552346570397112,
+      "grad_norm": 2.09375,
+      "learning_rate": 0.00010002759207139508,
+      "loss": 0.8689,
+      "step": 645
+    },
+    {
+      "epoch": 0.15643802647412755,
+      "grad_norm": 2.359375,
+      "learning_rate": 0.00010002026257136792,
+      "loss": 0.9227,
+      "step": 650
+    },
+    {
+      "epoch": 0.15764139590854392,
+      "grad_norm": 2.109375,
+      "learning_rate": 0.00010001287362464602,
+      "loss": 0.8438,
+      "step": 655
+    },
+    {
+      "epoch": 0.1588447653429603,
+      "grad_norm": 2.015625,
+      "learning_rate": 0.0001000054252430293,
+      "loss": 0.8453,
+      "step": 660
+    },
+    {
+      "epoch": 0.16004813477737664,
+      "grad_norm": 2.453125,
+      "learning_rate": 9.999791743841263e-05,
+      "loss": 0.8422,
+      "step": 665
+    },
+    {
+      "epoch": 0.16125150421179302,
+      "grad_norm": 2.1875,
+      "learning_rate": 9.999035022278575e-05,
+      "loss": 0.866,
+      "step": 670
+    },
+    {
+      "epoch": 0.1624548736462094,
+      "grad_norm": 2.1875,
+      "learning_rate": 9.99827236082333e-05,
+      "loss": 0.8817,
+      "step": 675
+    },
+    {
+      "epoch": 0.16365824308062576,
+      "grad_norm": 2.234375,
+      "learning_rate": 9.997503760693478e-05,
+      "loss": 0.9046,
+      "step": 680
+    },
+    {
+      "epoch": 0.1648616125150421,
+      "grad_norm": 2.40625,
+      "learning_rate": 9.996729223116452e-05,
+      "loss": 0.8713,
+      "step": 685
+    },
+    {
+      "epoch": 0.16606498194945848,
+      "grad_norm": 2.109375,
+      "learning_rate": 9.995948749329168e-05,
+      "loss": 0.8363,
+      "step": 690
+    },
+    {
+      "epoch": 0.16726835138387486,
+      "grad_norm": 2.3125,
+      "learning_rate": 9.99516234057802e-05,
+      "loss": 0.8764,
+      "step": 695
+    },
+    {
+      "epoch": 0.1684717208182912,
+      "grad_norm": 1.9921875,
+      "learning_rate": 9.994369998118882e-05,
+      "loss": 0.8316,
+      "step": 700
+    },
+    {
+      "epoch": 0.16967509025270758,
+      "grad_norm": 2.203125,
+      "learning_rate": 9.993571723217103e-05,
+      "loss": 0.8533,
+      "step": 705
+    },
+    {
+      "epoch": 0.17087845968712395,
+      "grad_norm": 2.140625,
+      "learning_rate": 9.992767517147505e-05,
+      "loss": 0.8919,
+      "step": 710
+    },
+    {
+      "epoch": 0.17208182912154033,
+      "grad_norm": 2.390625,
+      "learning_rate": 9.991957381194385e-05,
+      "loss": 0.8414,
+      "step": 715
+    },
+    {
+      "epoch": 0.17328519855595667,
+      "grad_norm": 2.140625,
+      "learning_rate": 9.991141316651505e-05,
+      "loss": 0.8667,
+      "step": 720
+    },
+    {
+      "epoch": 0.17448856799037304,
+      "grad_norm": 1.984375,
+      "learning_rate": 9.990319324822101e-05,
+      "loss": 0.8545,
+      "step": 725
+    },
+    {
+      "epoch": 0.17569193742478942,
+      "grad_norm": 2.4375,
+      "learning_rate": 9.989491407018868e-05,
+      "loss": 0.8622,
+      "step": 730
+    },
+    {
+      "epoch": 0.17689530685920576,
+      "grad_norm": 2.03125,
+      "learning_rate": 9.98865756456397e-05,
+      "loss": 0.8522,
+      "step": 735
+    },
+    {
+      "epoch": 0.17809867629362214,
+      "grad_norm": 2.453125,
+      "learning_rate": 9.98781779878903e-05,
+      "loss": 0.8441,
+      "step": 740
+    },
+    {
+      "epoch": 0.1793020457280385,
+      "grad_norm": 2.09375,
+      "learning_rate": 9.986972111035132e-05,
+      "loss": 0.8572,
+      "step": 745
+    },
+    {
+      "epoch": 0.18050541516245489,
+      "grad_norm": 2.296875,
+      "learning_rate": 9.986120502652816e-05,
+      "loss": 0.8486,
+      "step": 750
+    },
+    {
+      "epoch": 0.18170878459687123,
+      "grad_norm": 1.9296875,
+      "learning_rate": 9.985262975002073e-05,
+      "loss": 0.8385,
+      "step": 755
+    },
+    {
+      "epoch": 0.1829121540312876,
+      "grad_norm": 1.9765625,
+      "learning_rate": 9.984399529452357e-05,
+      "loss": 0.8344,
+      "step": 760
+    },
+    {
+      "epoch": 0.18411552346570398,
+      "grad_norm": 2.125,
+      "learning_rate": 9.983530167382562e-05,
+      "loss": 0.8603,
+      "step": 765
+    },
+    {
+      "epoch": 0.18531889290012032,
+      "grad_norm": 2.421875,
+      "learning_rate": 9.98265489018104e-05,
+      "loss": 0.8683,
+      "step": 770
+    },
+    {
+      "epoch": 0.1865222623345367,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.981773699245579e-05,
+      "loss": 0.8596,
+      "step": 775
+    },
+    {
+      "epoch": 0.18772563176895307,
+      "grad_norm": 2.015625,
+      "learning_rate": 9.980886595983423e-05,
+      "loss": 0.8547,
+      "step": 780
+    },
+    {
+      "epoch": 0.18892900120336945,
+      "grad_norm": 2.0,
+      "learning_rate": 9.979993581811245e-05,
+      "loss": 0.8407,
+      "step": 785
+    },
+    {
+      "epoch": 0.1901323706377858,
+      "grad_norm": 2.0625,
+      "learning_rate": 9.979094658155169e-05,
+      "loss": 0.8383,
+      "step": 790
+    },
+    {
+      "epoch": 0.19133574007220217,
+      "grad_norm": 2.09375,
+      "learning_rate": 9.978189826450748e-05,
+      "loss": 0.8117,
+      "step": 795
+    },
+    {
+      "epoch": 0.19253910950661854,
+      "grad_norm": 2.203125,
+      "learning_rate": 9.977279088142978e-05,
+      "loss": 0.8503,
+      "step": 800
+    },
+    {
+      "epoch": 0.19374247894103488,
+      "grad_norm": 2.28125,
+      "learning_rate": 9.976362444686279e-05,
+      "loss": 0.8837,
+      "step": 805
+    },
+    {
+      "epoch": 0.19494584837545126,
+      "grad_norm": 2.125,
+      "learning_rate": 9.975439897544506e-05,
+      "loss": 0.886,
+      "step": 810
+    },
+    {
+      "epoch": 0.19614921780986763,
+      "grad_norm": 2.125,
+      "learning_rate": 9.974511448190943e-05,
+      "loss": 0.8254,
+      "step": 815
+    },
+    {
+      "epoch": 0.197352587244284,
+      "grad_norm": 2.203125,
+      "learning_rate": 9.973577098108294e-05,
+      "loss": 0.8184,
+      "step": 820
+    },
+    {
+      "epoch": 0.19855595667870035,
+      "grad_norm": 2.203125,
+      "learning_rate": 9.972636848788696e-05,
+      "loss": 0.8493,
+      "step": 825
+    },
+    {
+      "epoch": 0.19975932611311673,
+      "grad_norm": 2.03125,
+      "learning_rate": 9.971690701733699e-05,
+      "loss": 0.8167,
+      "step": 830
+    },
+    {
+      "epoch": 0.2009626955475331,
+      "grad_norm": 2.140625,
+      "learning_rate": 9.970738658454271e-05,
+      "loss": 0.8,
+      "step": 835
+    },
+    {
+      "epoch": 0.20216606498194944,
+      "grad_norm": 2.03125,
+      "learning_rate": 9.969780720470804e-05,
+      "loss": 0.8075,
+      "step": 840
+    },
+    {
+      "epoch": 0.20336943441636582,
+      "grad_norm": 2.3125,
+      "learning_rate": 9.968816889313095e-05,
+      "loss": 0.8419,
+      "step": 845
+    },
+    {
+      "epoch": 0.2045728038507822,
+      "grad_norm": 2.140625,
+      "learning_rate": 9.967847166520357e-05,
+      "loss": 0.8243,
+      "step": 850
+    },
+    {
+      "epoch": 0.20577617328519857,
+      "grad_norm": 2.453125,
+      "learning_rate": 9.966871553641211e-05,
+      "loss": 0.8059,
+      "step": 855
+    },
+    {
+      "epoch": 0.2069795427196149,
+      "grad_norm": 2.0,
+      "learning_rate": 9.965890052233684e-05,
+      "loss": 0.8207,
+      "step": 860
+    },
+    {
+      "epoch": 0.20818291215403129,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.964902663865205e-05,
+      "loss": 0.8258,
+      "step": 865
+    },
+    {
+      "epoch": 0.20938628158844766,
+      "grad_norm": 1.9921875,
+      "learning_rate": 9.963909390112608e-05,
+      "loss": 0.8312,
+      "step": 870
+    },
+    {
+      "epoch": 0.21058965102286403,
+      "grad_norm": 2.140625,
+      "learning_rate": 9.962910232562123e-05,
+      "loss": 0.8332,
+      "step": 875
+    },
+    {
+      "epoch": 0.21179302045728038,
+      "grad_norm": 2.25,
+      "learning_rate": 9.961905192809377e-05,
+      "loss": 0.8314,
+      "step": 880
+    },
+    {
+      "epoch": 0.21299638989169675,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.960894272459392e-05,
+      "loss": 0.8296,
+      "step": 885
+    },
+    {
+      "epoch": 0.21419975932611313,
+      "grad_norm": 2.109375,
+      "learning_rate": 9.959877473126578e-05,
+      "loss": 0.8368,
+      "step": 890
+    },
+    {
+      "epoch": 0.21540312876052947,
+      "grad_norm": 2.0,
+      "learning_rate": 9.958854796434738e-05,
+      "loss": 0.8194,
+      "step": 895
+    },
+    {
+      "epoch": 0.21660649819494585,
+      "grad_norm": 2.203125,
+      "learning_rate": 9.957826244017058e-05,
+      "loss": 0.8188,
+      "step": 900
+    },
+    {
+      "epoch": 0.21780986762936222,
+      "grad_norm": 2.140625,
+      "learning_rate": 9.956791817516107e-05,
+      "loss": 0.7867,
+      "step": 905
+    },
+    {
+      "epoch": 0.2190132370637786,
+      "grad_norm": 1.9921875,
+      "learning_rate": 9.955751518583835e-05,
+      "loss": 0.7898,
+      "step": 910
+    },
+    {
+      "epoch": 0.22021660649819494,
+      "grad_norm": 2.1875,
+      "learning_rate": 9.954705348881573e-05,
+      "loss": 0.8214,
+      "step": 915
+    },
+    {
+      "epoch": 0.2214199759326113,
+      "grad_norm": 2.46875,
+      "learning_rate": 9.953653310080022e-05,
+      "loss": 0.8218,
+      "step": 920
+    },
+    {
+      "epoch": 0.2226233453670277,
+      "grad_norm": 2.21875,
+      "learning_rate": 9.952595403859263e-05,
+      "loss": 0.7663,
+      "step": 925
+    },
+    {
+      "epoch": 0.22382671480144403,
+      "grad_norm": 2.28125,
+      "learning_rate": 9.951531631908739e-05,
+      "loss": 0.7601,
+      "step": 930
+    },
+    {
+      "epoch": 0.2250300842358604,
+      "grad_norm": 2.15625,
+      "learning_rate": 9.95046199592727e-05,
+      "loss": 0.7777,
+      "step": 935
+    },
+    {
+      "epoch": 0.22623345367027678,
+      "grad_norm": 2.265625,
+      "learning_rate": 9.94938649762303e-05,
+      "loss": 0.8109,
+      "step": 940
+    },
+    {
+      "epoch": 0.22743682310469315,
+      "grad_norm": 2.03125,
+      "learning_rate": 9.948305138713562e-05,
+      "loss": 0.8333,
+      "step": 945
+    },
+    {
+      "epoch": 0.2286401925391095,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.947217920925766e-05,
+      "loss": 0.8006,
+      "step": 950
+    },
+    {
+      "epoch": 0.22984356197352587,
+      "grad_norm": 2.21875,
+      "learning_rate": 9.946124845995902e-05,
+      "loss": 0.8408,
+      "step": 955
+    },
+    {
+      "epoch": 0.23104693140794225,
+      "grad_norm": 2.484375,
+      "learning_rate": 9.945025915669577e-05,
+      "loss": 0.798,
+      "step": 960
+    },
+    {
+      "epoch": 0.2322503008423586,
+      "grad_norm": 1.9921875,
+      "learning_rate": 9.943921131701754e-05,
+      "loss": 0.7899,
+      "step": 965
+    },
+    {
+      "epoch": 0.23345367027677497,
+      "grad_norm": 2.015625,
+      "learning_rate": 9.942810495856742e-05,
+      "loss": 0.8247,
+      "step": 970
+    },
+    {
+      "epoch": 0.23465703971119134,
+      "grad_norm": 2.09375,
+      "learning_rate": 9.941694009908194e-05,
+      "loss": 0.7924,
+      "step": 975
+    },
+    {
+      "epoch": 0.2358604091456077,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.940571675639114e-05,
+      "loss": 0.7573,
+      "step": 980
+    },
+    {
+      "epoch": 0.23706377858002406,
+      "grad_norm": 2.09375,
+      "learning_rate": 9.939443494841831e-05,
+      "loss": 0.825,
+      "step": 985
+    },
+    {
+      "epoch": 0.23826714801444043,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.938309469318024e-05,
+      "loss": 0.8057,
+      "step": 990
+    },
+    {
+      "epoch": 0.2394705174488568,
+      "grad_norm": 2.0625,
+      "learning_rate": 9.937169600878699e-05,
+      "loss": 0.7873,
+      "step": 995
+    },
+    {
+      "epoch": 0.24067388688327315,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.936023891344194e-05,
+      "loss": 0.8192,
+      "step": 1000
+    },
+    {
+      "epoch": 0.24067388688327315,
+      "eval_loss": 0.7019978165626526,
+      "eval_runtime": 2.6886,
+      "eval_samples_per_second": 74.389,
+      "eval_steps_per_second": 74.389,
+      "step": 1000
+    },
+    {
+      "epoch": 0.24187725631768953,
+      "grad_norm": 2.21875,
+      "learning_rate": 9.934872342544176e-05,
+      "loss": 0.7758,
+      "step": 1005
+    },
+    {
+      "epoch": 0.2430806257521059,
+      "grad_norm": 2.34375,
+      "learning_rate": 9.933714956317638e-05,
+      "loss": 0.8306,
+      "step": 1010
+    },
+    {
+      "epoch": 0.24428399518652227,
+      "grad_norm": 1.9765625,
+      "learning_rate": 9.932551734512891e-05,
+      "loss": 0.7868,
+      "step": 1015
+    },
+    {
+      "epoch": 0.24548736462093862,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.931382678987572e-05,
+      "loss": 0.7612,
+      "step": 1020
+    },
+    {
+      "epoch": 0.246690734055355,
+      "grad_norm": 1.9765625,
+      "learning_rate": 9.930207791608625e-05,
+      "loss": 0.7586,
+      "step": 1025
+    },
+    {
+      "epoch": 0.24789410348977137,
+      "grad_norm": 2.03125,
+      "learning_rate": 9.92902707425232e-05,
+      "loss": 0.8066,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2490974729241877,
+      "grad_norm": 1.96875,
+      "learning_rate": 9.927840528804226e-05,
+      "loss": 0.8022,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2503008423586041,
+      "grad_norm": 1.9609375,
+      "learning_rate": 9.926648157159222e-05,
+      "loss": 0.7819,
+      "step": 1040
+    },
+    {
+      "epoch": 0.25150421179302046,
+      "grad_norm": 2.0625,
+      "learning_rate": 9.925449961221496e-05,
+      "loss": 0.804,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2527075812274368,
+      "grad_norm": 1.9140625,
+      "learning_rate": 9.924245942904534e-05,
+      "loss": 0.7823,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2539109506618532,
+      "grad_norm": 1.9921875,
+      "learning_rate": 9.923036104131117e-05,
+      "loss": 0.7887,
+      "step": 1055
+    },
+    {
+      "epoch": 0.25511432009626955,
+      "grad_norm": 2.171875,
+      "learning_rate": 9.921820446833328e-05,
+      "loss": 0.8386,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2563176895306859,
+      "grad_norm": 2.15625,
+      "learning_rate": 9.920598972952534e-05,
+      "loss": 0.7996,
+      "step": 1065
+    },
+    {
+      "epoch": 0.2575210589651023,
+      "grad_norm": 1.9140625,
+      "learning_rate": 9.919371684439401e-05,
+      "loss": 0.7612,
+      "step": 1070
+    },
+    {
+      "epoch": 0.25872442839951865,
+      "grad_norm": 2.09375,
+      "learning_rate": 9.918138583253869e-05,
+      "loss": 0.8034,
+      "step": 1075
+    },
+    {
+      "epoch": 0.259927797833935,
+      "grad_norm": 1.859375,
+      "learning_rate": 9.91689967136517e-05,
+      "loss": 0.7952,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2611311672683514,
+      "grad_norm": 2.015625,
+      "learning_rate": 9.915654950751811e-05,
+      "loss": 0.8253,
+      "step": 1085
+    },
+    {
+      "epoch": 0.26233453670276774,
+      "grad_norm": 1.8671875,
+      "learning_rate": 9.914404423401577e-05,
+      "loss": 0.7845,
+      "step": 1090
+    },
+    {
+      "epoch": 0.26353790613718414,
+      "grad_norm": 1.84375,
+      "learning_rate": 9.913148091311526e-05,
+      "loss": 0.7825,
+      "step": 1095
+    },
+    {
+      "epoch": 0.2647412755716005,
+      "grad_norm": 1.9921875,
+      "learning_rate": 9.911885956487988e-05,
+      "loss": 0.767,
+      "step": 1100
+    },
+    {
+      "epoch": 0.26594464500601683,
+      "grad_norm": 2.125,
+      "learning_rate": 9.910618020946552e-05,
+      "loss": 0.8171,
+      "step": 1105
+    },
+    {
+      "epoch": 0.26714801444043323,
+      "grad_norm": 2.109375,
+      "learning_rate": 9.909344286712083e-05,
+      "loss": 0.7493,
+      "step": 1110
+    },
+    {
+      "epoch": 0.2683513838748496,
+      "grad_norm": 1.984375,
+      "learning_rate": 9.908064755818693e-05,
+      "loss": 0.7957,
+      "step": 1115
+    },
+    {
+      "epoch": 0.2695547533092659,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.906779430309763e-05,
+      "loss": 0.7947,
+      "step": 1120
+    },
+    {
+      "epoch": 0.27075812274368233,
+      "grad_norm": 2.125,
+      "learning_rate": 9.90548831223792e-05,
+      "loss": 0.7759,
+      "step": 1125
+    },
+    {
+      "epoch": 0.2719614921780987,
+      "grad_norm": 2.09375,
+      "learning_rate": 9.904191403665043e-05,
+      "loss": 0.8045,
+      "step": 1130
+    },
+    {
+      "epoch": 0.273164861612515,
+      "grad_norm": 2.296875,
+      "learning_rate": 9.902888706662262e-05,
+      "loss": 0.784,
+      "step": 1135
+    },
+    {
+      "epoch": 0.2743682310469314,
+      "grad_norm": 2.0625,
+      "learning_rate": 9.901580223309946e-05,
+      "loss": 0.8168,
+      "step": 1140
+    },
+    {
+      "epoch": 0.27557160048134777,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.900265955697708e-05,
+      "loss": 0.7761,
+      "step": 1145
+    },
+    {
+      "epoch": 0.2767749699157641,
+      "grad_norm": 2.140625,
+      "learning_rate": 9.898945905924397e-05,
+      "loss": 0.7857,
+      "step": 1150
+    },
+    {
+      "epoch": 0.2779783393501805,
+      "grad_norm": 2.359375,
+      "learning_rate": 9.897620076098097e-05,
+      "loss": 0.7853,
+      "step": 1155
+    },
+    {
+      "epoch": 0.27918170878459686,
+      "grad_norm": 2.1875,
+      "learning_rate": 9.89628846833612e-05,
+      "loss": 0.7904,
+      "step": 1160
+    },
+    {
+      "epoch": 0.28038507821901326,
+      "grad_norm": 1.9921875,
+      "learning_rate": 9.894951084765009e-05,
+      "loss": 0.7402,
+      "step": 1165
+    },
+    {
+      "epoch": 0.2815884476534296,
+      "grad_norm": 1.8984375,
+      "learning_rate": 9.893607927520527e-05,
+      "loss": 0.7577,
+      "step": 1170
+    },
+    {
+      "epoch": 0.28279181708784595,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.892258998747662e-05,
+      "loss": 0.8035,
+      "step": 1175
+    },
+    {
+      "epoch": 0.28399518652226236,
+      "grad_norm": 2.015625,
+      "learning_rate": 9.890904300600613e-05,
+      "loss": 0.7901,
+      "step": 1180
+    },
+    {
+      "epoch": 0.2851985559566787,
+      "grad_norm": 1.8203125,
+      "learning_rate": 9.889543835242796e-05,
+      "loss": 0.7655,
+      "step": 1185
+    },
+    {
+      "epoch": 0.28640192539109505,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.888177604846838e-05,
+      "loss": 0.7969,
+      "step": 1190
+    },
+    {
+      "epoch": 0.28760529482551145,
+      "grad_norm": 1.984375,
+      "learning_rate": 9.886805611594573e-05,
+      "loss": 0.7737,
+      "step": 1195
+    },
+    {
+      "epoch": 0.2888086642599278,
+      "grad_norm": 2.015625,
+      "learning_rate": 9.885427857677032e-05,
+      "loss": 0.8114,
+      "step": 1200
+    },
+    {
+      "epoch": 0.29001203369434414,
+      "grad_norm": 2.09375,
+      "learning_rate": 9.884044345294454e-05,
+      "loss": 0.7722,
+      "step": 1205
+    },
+    {
+      "epoch": 0.29121540312876054,
+      "grad_norm": 1.875,
+      "learning_rate": 9.882655076656269e-05,
+      "loss": 0.7765,
+      "step": 1210
+    },
+    {
+      "epoch": 0.2924187725631769,
+      "grad_norm": 1.984375,
+      "learning_rate": 9.8812600539811e-05,
+      "loss": 0.7874,
+      "step": 1215
+    },
+    {
+      "epoch": 0.29362214199759323,
+      "grad_norm": 2.015625,
+      "learning_rate": 9.879859279496761e-05,
+      "loss": 0.7992,
+      "step": 1220
+    },
+    {
+      "epoch": 0.29482551143200963,
+      "grad_norm": 2.015625,
+      "learning_rate": 9.87845275544025e-05,
+      "loss": 0.7625,
+      "step": 1225
+    },
+    {
+      "epoch": 0.296028880866426,
+      "grad_norm": 2.015625,
+      "learning_rate": 9.877040484057748e-05,
+      "loss": 0.8007,
+      "step": 1230
+    },
+    {
+      "epoch": 0.2972322503008424,
+      "grad_norm": 1.953125,
+      "learning_rate": 9.875622467604612e-05,
+      "loss": 0.7735,
+      "step": 1235
+    },
+    {
+      "epoch": 0.29843561973525873,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.874198708345375e-05,
+      "loss": 0.7712,
+      "step": 1240
+    },
+    {
+      "epoch": 0.2996389891696751,
+      "grad_norm": 2.015625,
+      "learning_rate": 9.872769208553744e-05,
+      "loss": 0.7869,
+      "step": 1245
+    },
+    {
+      "epoch": 0.3008423586040915,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.871333970512589e-05,
+      "loss": 0.7851,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3020457280385078,
+      "grad_norm": 2.015625,
+      "learning_rate": 9.869892996513945e-05,
+      "loss": 0.7396,
+      "step": 1255
+    },
+    {
+      "epoch": 0.30324909747292417,
+      "grad_norm": 1.890625,
+      "learning_rate": 9.86844628885901e-05,
+      "loss": 0.8024,
+      "step": 1260
+    },
+    {
+      "epoch": 0.30445246690734057,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.866993849858135e-05,
+      "loss": 0.764,
+      "step": 1265
+    },
+    {
+      "epoch": 0.3056558363417569,
+      "grad_norm": 1.921875,
+      "learning_rate": 9.865535681830825e-05,
+      "loss": 0.7776,
+      "step": 1270
+    },
+    {
+      "epoch": 0.30685920577617326,
+      "grad_norm": 1.9765625,
+      "learning_rate": 9.864071787105735e-05,
+      "loss": 0.7484,
+      "step": 1275
+    },
+    {
+      "epoch": 0.30806257521058966,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.862602168020664e-05,
+      "loss": 0.7562,
+      "step": 1280
+    },
+    {
+      "epoch": 0.309265944645006,
+      "grad_norm": 1.953125,
+      "learning_rate": 9.861126826922553e-05,
+      "loss": 0.7467,
+      "step": 1285
+    },
+    {
+      "epoch": 0.3104693140794224,
+      "grad_norm": 1.9453125,
+      "learning_rate": 9.859645766167482e-05,
+      "loss": 0.7581,
+      "step": 1290
+    },
+    {
+      "epoch": 0.31167268351383876,
+      "grad_norm": 2.21875,
+      "learning_rate": 9.858158988120664e-05,
+      "loss": 0.7602,
+      "step": 1295
+    },
+    {
+      "epoch": 0.3128760529482551,
+      "grad_norm": 1.9140625,
+      "learning_rate": 9.856666495156442e-05,
+      "loss": 0.7553,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3140794223826715,
+      "grad_norm": 2.125,
+      "learning_rate": 9.855168289658285e-05,
+      "loss": 0.7648,
+      "step": 1305
+    },
+    {
+      "epoch": 0.31528279181708785,
+      "grad_norm": 2.09375,
+      "learning_rate": 9.853664374018793e-05,
+      "loss": 0.7684,
+      "step": 1310
+    },
+    {
+      "epoch": 0.3164861612515042,
+      "grad_norm": 2.140625,
+      "learning_rate": 9.852154750639669e-05,
+      "loss": 0.7478,
+      "step": 1315
+    },
+    {
+      "epoch": 0.3176895306859206,
+      "grad_norm": 1.8984375,
+      "learning_rate": 9.850639421931745e-05,
+      "loss": 0.7559,
+      "step": 1320
+    },
+    {
+      "epoch": 0.31889290012033694,
+      "grad_norm": 2.09375,
+      "learning_rate": 9.849118390314957e-05,
+      "loss": 0.7993,
+      "step": 1325
+    },
+    {
+      "epoch": 0.3200962695547533,
+      "grad_norm": 2.03125,
+      "learning_rate": 9.847591658218353e-05,
+      "loss": 0.7609,
+      "step": 1330
+    },
+    {
+      "epoch": 0.3212996389891697,
+      "grad_norm": 2.203125,
+      "learning_rate": 9.846059228080081e-05,
+      "loss": 0.7558,
+      "step": 1335
+    },
+    {
+      "epoch": 0.32250300842358604,
+      "grad_norm": 2.0,
+      "learning_rate": 9.844521102347389e-05,
+      "loss": 0.7756,
+      "step": 1340
+    },
+    {
+      "epoch": 0.3237063778580024,
+      "grad_norm": 2.125,
+      "learning_rate": 9.842977283476621e-05,
+      "loss": 0.778,
+      "step": 1345
+    },
+    {
+      "epoch": 0.3249097472924188,
+      "grad_norm": 2.09375,
+      "learning_rate": 9.841427773933217e-05,
+      "loss": 0.7515,
+      "step": 1350
+    },
+    {
+      "epoch": 0.32611311672683513,
+      "grad_norm": 2.03125,
+      "learning_rate": 9.839872576191697e-05,
+      "loss": 0.7545,
+      "step": 1355
+    },
+    {
+      "epoch": 0.32731648616125153,
+      "grad_norm": 2.015625,
+      "learning_rate": 9.838311692735671e-05,
+      "loss": 0.7668,
+      "step": 1360
+    },
+    {
+      "epoch": 0.3285198555956679,
+      "grad_norm": 2.1875,
+      "learning_rate": 9.836745126057828e-05,
+      "loss": 0.7554,
+      "step": 1365
+    },
+    {
+      "epoch": 0.3297232250300842,
+      "grad_norm": 1.984375,
+      "learning_rate": 9.83517287865993e-05,
+      "loss": 0.7413,
+      "step": 1370
+    },
+    {
+      "epoch": 0.3309265944645006,
+      "grad_norm": 1.8671875,
+      "learning_rate": 9.833594953052811e-05,
+      "loss": 0.7243,
+      "step": 1375
+    },
+    {
+      "epoch": 0.33212996389891697,
+      "grad_norm": 1.9296875,
+      "learning_rate": 9.83201135175638e-05,
+      "loss": 0.7439,
+      "step": 1380
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 2.0625,
+      "learning_rate": 9.830422077299601e-05,
+      "loss": 0.7836,
+      "step": 1385
+    },
+    {
+      "epoch": 0.3345367027677497,
+      "grad_norm": 1.8515625,
+      "learning_rate": 9.828827132220504e-05,
+      "loss": 0.7121,
+      "step": 1390
+    },
+    {
+      "epoch": 0.33574007220216606,
+      "grad_norm": 1.9921875,
+      "learning_rate": 9.827226519066169e-05,
+      "loss": 0.726,
+      "step": 1395
+    },
+    {
+      "epoch": 0.3369434416365824,
+      "grad_norm": 1.875,
+      "learning_rate": 9.825620240392733e-05,
+      "loss": 0.7737,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3381468110709988,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.82400829876538e-05,
+      "loss": 0.7407,
+      "step": 1405
+    },
+    {
+      "epoch": 0.33935018050541516,
+      "grad_norm": 1.734375,
+      "learning_rate": 9.822390696758336e-05,
+      "loss": 0.765,
+      "step": 1410
+    },
+    {
+      "epoch": 0.3405535499398315,
+      "grad_norm": 1.9921875,
+      "learning_rate": 9.820767436954865e-05,
+      "loss": 0.7207,
+      "step": 1415
+    },
+    {
+      "epoch": 0.3417569193742479,
+      "grad_norm": 2.140625,
+      "learning_rate": 9.81913852194727e-05,
+      "loss": 0.742,
+      "step": 1420
+    },
+    {
+      "epoch": 0.34296028880866425,
+      "grad_norm": 1.859375,
+      "learning_rate": 9.817503954336885e-05,
+      "loss": 0.7437,
+      "step": 1425
+    },
+    {
+      "epoch": 0.34416365824308065,
+      "grad_norm": 2.296875,
+      "learning_rate": 9.815863736734066e-05,
+      "loss": 0.7577,
+      "step": 1430
+    },
+    {
+      "epoch": 0.345367027677497,
+      "grad_norm": 2.109375,
+      "learning_rate": 9.814217871758198e-05,
+      "loss": 0.7372,
+      "step": 1435
+    },
+    {
+      "epoch": 0.34657039711191334,
+      "grad_norm": 2.203125,
+      "learning_rate": 9.812566362037682e-05,
+      "loss": 0.7302,
+      "step": 1440
+    },
+    {
+      "epoch": 0.34777376654632974,
+      "grad_norm": 2.1875,
+      "learning_rate": 9.810909210209932e-05,
+      "loss": 0.7469,
+      "step": 1445
+    },
+    {
+      "epoch": 0.3489771359807461,
+      "grad_norm": 2.015625,
+      "learning_rate": 9.809246418921374e-05,
+      "loss": 0.7544,
+      "step": 1450
+    },
+    {
+      "epoch": 0.35018050541516244,
+      "grad_norm": 1.8359375,
+      "learning_rate": 9.807577990827442e-05,
+      "loss": 0.7288,
+      "step": 1455
+    },
+    {
+      "epoch": 0.35138387484957884,
+      "grad_norm": 2.203125,
+      "learning_rate": 9.805903928592567e-05,
+      "loss": 0.7725,
+      "step": 1460
+    },
+    {
+      "epoch": 0.3525872442839952,
+      "grad_norm": 2.15625,
+      "learning_rate": 9.804224234890185e-05,
+      "loss": 0.7269,
+      "step": 1465
+    },
+    {
+      "epoch": 0.35379061371841153,
+      "grad_norm": 1.921875,
+      "learning_rate": 9.802538912402715e-05,
+      "loss": 0.7577,
+      "step": 1470
+    },
+    {
+      "epoch": 0.35499398315282793,
+      "grad_norm": 1.953125,
+      "learning_rate": 9.800847963821573e-05,
+      "loss": 0.7393,
+      "step": 1475
+    },
+    {
+      "epoch": 0.3561973525872443,
+      "grad_norm": 2.140625,
+      "learning_rate": 9.799151391847158e-05,
+      "loss": 0.7159,
+      "step": 1480
+    },
+    {
+      "epoch": 0.3574007220216607,
+      "grad_norm": 1.890625,
+      "learning_rate": 9.79744919918885e-05,
+      "loss": 0.7584,
+      "step": 1485
+    },
+    {
+      "epoch": 0.358604091456077,
+      "grad_norm": 1.875,
+      "learning_rate": 9.795741388565e-05,
+      "loss": 0.7267,
+      "step": 1490
+    },
+    {
+      "epoch": 0.35980746089049337,
+      "grad_norm": 2.03125,
+      "learning_rate": 9.794027962702939e-05,
+      "loss": 0.7622,
+      "step": 1495
+    },
+    {
+      "epoch": 0.36101083032490977,
+      "grad_norm": 1.8671875,
+      "learning_rate": 9.792308924338958e-05,
+      "loss": 0.7363,
+      "step": 1500
+    },
+    {
+      "epoch": 0.36101083032490977,
+      "eval_loss": 0.644199550151825,
+      "eval_runtime": 2.6553,
+      "eval_samples_per_second": 75.32,
+      "eval_steps_per_second": 75.32,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3622141997593261,
+      "grad_norm": 1.9296875,
+      "learning_rate": 9.790584276218317e-05,
+      "loss": 0.7645,
+      "step": 1505
+    },
+    {
+      "epoch": 0.36341756919374246,
+      "grad_norm": 2.03125,
+      "learning_rate": 9.788854021095226e-05,
+      "loss": 0.7213,
+      "step": 1510
+    },
+    {
+      "epoch": 0.36462093862815886,
+      "grad_norm": 1.875,
+      "learning_rate": 9.78711816173286e-05,
+      "loss": 0.7573,
+      "step": 1515
+    },
+    {
+      "epoch": 0.3658243080625752,
+      "grad_norm": 1.96875,
+      "learning_rate": 9.785376700903338e-05,
+      "loss": 0.7674,
+      "step": 1520
+    },
+    {
+      "epoch": 0.36702767749699156,
+      "grad_norm": 1.8984375,
+      "learning_rate": 9.783629641387724e-05,
+      "loss": 0.7761,
+      "step": 1525
+    },
+    {
+      "epoch": 0.36823104693140796,
+      "grad_norm": 1.921875,
+      "learning_rate": 9.781876985976023e-05,
+      "loss": 0.7686,
+      "step": 1530
+    },
+    {
+      "epoch": 0.3694344163658243,
+      "grad_norm": 2.09375,
+      "learning_rate": 9.780118737467178e-05,
+      "loss": 0.7692,
+      "step": 1535
+    },
+    {
+      "epoch": 0.37063778580024065,
+      "grad_norm": 1.9765625,
+      "learning_rate": 9.778354898669067e-05,
+      "loss": 0.7359,
+      "step": 1540
+    },
+    {
+      "epoch": 0.37184115523465705,
+      "grad_norm": 2.0,
+      "learning_rate": 9.776585472398488e-05,
+      "loss": 0.7126,
+      "step": 1545
+    },
+    {
+      "epoch": 0.3730445246690734,
+      "grad_norm": 2.140625,
+      "learning_rate": 9.774810461481165e-05,
+      "loss": 0.7319,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3742478941034898,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.773029868751746e-05,
+      "loss": 0.7645,
+      "step": 1555
+    },
+    {
+      "epoch": 0.37545126353790614,
+      "grad_norm": 1.96875,
+      "learning_rate": 9.771243697053787e-05,
+      "loss": 0.735,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3766546329723225,
+      "grad_norm": 2.234375,
+      "learning_rate": 9.769451949239755e-05,
+      "loss": 0.7707,
+      "step": 1565
+    },
+    {
+      "epoch": 0.3778580024067389,
+      "grad_norm": 1.96875,
+      "learning_rate": 9.767654628171021e-05,
+      "loss": 0.7207,
+      "step": 1570
+    },
+    {
+      "epoch": 0.37906137184115524,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.765851736717858e-05,
+      "loss": 0.7396,
+      "step": 1575
+    },
+    {
+      "epoch": 0.3802647412755716,
+      "grad_norm": 1.890625,
+      "learning_rate": 9.764043277759434e-05,
+      "loss": 0.7359,
+      "step": 1580
+    },
+    {
+      "epoch": 0.381468110709988,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.762229254183808e-05,
+      "loss": 0.7163,
+      "step": 1585
+    },
+    {
+      "epoch": 0.38267148014440433,
+      "grad_norm": 2.015625,
+      "learning_rate": 9.760409668887927e-05,
+      "loss": 0.7657,
+      "step": 1590
+    },
+    {
+      "epoch": 0.3838748495788207,
+      "grad_norm": 1.9609375,
+      "learning_rate": 9.758584524777617e-05,
+      "loss": 0.7268,
+      "step": 1595
+    },
+    {
+      "epoch": 0.3850782190132371,
+      "grad_norm": 1.953125,
+      "learning_rate": 9.756753824767585e-05,
+      "loss": 0.6808,
+      "step": 1600
+    },
+    {
+      "epoch": 0.3862815884476534,
+      "grad_norm": 1.90625,
+      "learning_rate": 9.754917571781406e-05,
+      "loss": 0.7574,
+      "step": 1605
+    },
+    {
+      "epoch": 0.38748495788206977,
+      "grad_norm": 2.09375,
+      "learning_rate": 9.753075768751528e-05,
+      "loss": 0.7114,
+      "step": 1610
+    },
+    {
+      "epoch": 0.38868832731648617,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.751228418619257e-05,
+      "loss": 0.7273,
+      "step": 1615
+    },
+    {
+      "epoch": 0.3898916967509025,
+      "grad_norm": 2.09375,
+      "learning_rate": 9.749375524334765e-05,
+      "loss": 0.7251,
+      "step": 1620
+    },
+    {
+      "epoch": 0.3910950661853189,
+      "grad_norm": 1.921875,
+      "learning_rate": 9.747517088857068e-05,
+      "loss": 0.7132,
+      "step": 1625
+    },
+    {
+      "epoch": 0.39229843561973526,
+      "grad_norm": 1.796875,
+      "learning_rate": 9.745653115154038e-05,
+      "loss": 0.7307,
+      "step": 1630
+    },
+    {
+      "epoch": 0.3935018050541516,
+      "grad_norm": 2.25,
+      "learning_rate": 9.743783606202393e-05,
+      "loss": 0.7415,
+      "step": 1635
+    },
+    {
+      "epoch": 0.394705174488568,
+      "grad_norm": 2.03125,
+      "learning_rate": 9.741908564987682e-05,
+      "loss": 0.7341,
+      "step": 1640
+    },
+    {
+      "epoch": 0.39590854392298436,
+      "grad_norm": 1.8359375,
+      "learning_rate": 9.740027994504299e-05,
+      "loss": 0.7851,
+      "step": 1645
+    },
+    {
+      "epoch": 0.3971119133574007,
+      "grad_norm": 1.890625,
+      "learning_rate": 9.73814189775546e-05,
+      "loss": 0.7394,
+      "step": 1650
+    },
+    {
+      "epoch": 0.3983152827918171,
+      "grad_norm": 1.9296875,
+      "learning_rate": 9.736250277753213e-05,
+      "loss": 0.7116,
+      "step": 1655
+    },
+    {
+      "epoch": 0.39951865222623345,
+      "grad_norm": 1.875,
+      "learning_rate": 9.734353137518419e-05,
+      "loss": 0.7178,
+      "step": 1660
+    },
+    {
+      "epoch": 0.4007220216606498,
+      "grad_norm": 1.96875,
+      "learning_rate": 9.732450480080762e-05,
+      "loss": 0.7381,
+      "step": 1665
+    },
+    {
+      "epoch": 0.4019253910950662,
+      "grad_norm": 2.0625,
+      "learning_rate": 9.730542308478733e-05,
+      "loss": 0.7467,
+      "step": 1670
+    },
+    {
+      "epoch": 0.40312876052948254,
+      "grad_norm": 2.109375,
+      "learning_rate": 9.728628625759628e-05,
+      "loss": 0.7412,
+      "step": 1675
+    },
+    {
+      "epoch": 0.4043321299638989,
+      "grad_norm": 1.921875,
+      "learning_rate": 9.726709434979548e-05,
+      "loss": 0.7245,
+      "step": 1680
+    },
+    {
+      "epoch": 0.4055354993983153,
+      "grad_norm": 2.109375,
+      "learning_rate": 9.724784739203386e-05,
+      "loss": 0.7203,
+      "step": 1685
+    },
+    {
+      "epoch": 0.40673886883273164,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.72285454150483e-05,
+      "loss": 0.735,
+      "step": 1690
+    },
+    {
+      "epoch": 0.40794223826714804,
+      "grad_norm": 1.8828125,
+      "learning_rate": 9.720918844966352e-05,
+      "loss": 0.7185,
+      "step": 1695
+    },
+    {
+      "epoch": 0.4091456077015644,
+      "grad_norm": 1.765625,
+      "learning_rate": 9.718977652679205e-05,
+      "loss": 0.6864,
+      "step": 1700
+    },
+    {
+      "epoch": 0.41034897713598073,
+      "grad_norm": 1.8828125,
+      "learning_rate": 9.71703096774342e-05,
+      "loss": 0.7182,
+      "step": 1705
+    },
+    {
+      "epoch": 0.41155234657039713,
+      "grad_norm": 1.9765625,
+      "learning_rate": 9.7150787932678e-05,
+      "loss": 0.7583,
+      "step": 1710
+    },
+    {
+      "epoch": 0.4127557160048135,
+      "grad_norm": 1.859375,
+      "learning_rate": 9.713121132369914e-05,
+      "loss": 0.7215,
+      "step": 1715
+    },
+    {
+      "epoch": 0.4139590854392298,
+      "grad_norm": 2.125,
+      "learning_rate": 9.711157988176094e-05,
+      "loss": 0.7288,
+      "step": 1720
+    },
+    {
+      "epoch": 0.4151624548736462,
+      "grad_norm": 2.0,
+      "learning_rate": 9.709189363821422e-05,
+      "loss": 0.7336,
+      "step": 1725
+    },
+    {
+      "epoch": 0.41636582430806257,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.707215262449744e-05,
+      "loss": 0.7454,
+      "step": 1730
+    },
+    {
+      "epoch": 0.4175691937424789,
+      "grad_norm": 2.0,
+      "learning_rate": 9.705235687213642e-05,
+      "loss": 0.7579,
+      "step": 1735
+    },
+    {
+      "epoch": 0.4187725631768953,
+      "grad_norm": 1.9140625,
+      "learning_rate": 9.703250641274442e-05,
+      "loss": 0.7328,
+      "step": 1740
+    },
+    {
+      "epoch": 0.41997593261131166,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.701260127802211e-05,
+      "loss": 0.6927,
+      "step": 1745
+    },
+    {
+      "epoch": 0.42117930204572807,
+      "grad_norm": 2.03125,
+      "learning_rate": 9.69926414997574e-05,
+      "loss": 0.6976,
+      "step": 1750
+    },
+    {
+      "epoch": 0.4223826714801444,
+      "grad_norm": 2.0,
+      "learning_rate": 9.697262710982557e-05,
+      "loss": 0.6918,
+      "step": 1755
+    },
+    {
+      "epoch": 0.42358604091456076,
+      "grad_norm": 1.796875,
+      "learning_rate": 9.695255814018899e-05,
+      "loss": 0.7205,
+      "step": 1760
+    },
+    {
+      "epoch": 0.42478941034897716,
+      "grad_norm": 1.875,
+      "learning_rate": 9.693243462289728e-05,
+      "loss": 0.6963,
+      "step": 1765
+    },
+    {
+      "epoch": 0.4259927797833935,
+      "grad_norm": 1.9921875,
+      "learning_rate": 9.691225659008713e-05,
+      "loss": 0.7363,
+      "step": 1770
+    },
+    {
+      "epoch": 0.42719614921780985,
+      "grad_norm": 1.953125,
+      "learning_rate": 9.689202407398234e-05,
+      "loss": 0.715,
+      "step": 1775
+    },
+    {
+      "epoch": 0.42839951865222625,
+      "grad_norm": 1.9453125,
+      "learning_rate": 9.687173710689363e-05,
+      "loss": 0.7121,
+      "step": 1780
+    },
+    {
+      "epoch": 0.4296028880866426,
+      "grad_norm": 1.9453125,
+      "learning_rate": 9.685139572121876e-05,
+      "loss": 0.7253,
+      "step": 1785
+    },
+    {
+      "epoch": 0.43080625752105894,
+      "grad_norm": 1.8046875,
+      "learning_rate": 9.683099994944237e-05,
+      "loss": 0.7242,
+      "step": 1790
+    },
+    {
+      "epoch": 0.43200962695547535,
+      "grad_norm": 1.9375,
+      "learning_rate": 9.681054982413593e-05,
+      "loss": 0.7028,
+      "step": 1795
+    },
+    {
+      "epoch": 0.4332129963898917,
+      "grad_norm": 1.890625,
+      "learning_rate": 9.679004537795773e-05,
+      "loss": 0.7058,
+      "step": 1800
+    },
+    {
+      "epoch": 0.43441636582430804,
+      "grad_norm": 1.921875,
+      "learning_rate": 9.67694866436528e-05,
+      "loss": 0.7455,
+      "step": 1805
+    },
+    {
+      "epoch": 0.43561973525872444,
+      "grad_norm": 1.9609375,
+      "learning_rate": 9.67488736540529e-05,
+      "loss": 0.7209,
+      "step": 1810
+    },
+    {
+      "epoch": 0.4368231046931408,
+      "grad_norm": 2.03125,
+      "learning_rate": 9.672820644207639e-05,
+      "loss": 0.7455,
+      "step": 1815
+    },
+    {
+      "epoch": 0.4380264741275572,
+      "grad_norm": 2.140625,
+      "learning_rate": 9.670748504072822e-05,
+      "loss": 0.7298,
+      "step": 1820
+    },
+    {
+      "epoch": 0.43922984356197353,
+      "grad_norm": 1.8984375,
+      "learning_rate": 9.668670948309992e-05,
+      "loss": 0.7449,
+      "step": 1825
+    },
+    {
+      "epoch": 0.4404332129963899,
+      "grad_norm": 1.8515625,
+      "learning_rate": 9.666587980236952e-05,
+      "loss": 0.7101,
+      "step": 1830
+    },
+    {
+      "epoch": 0.4416365824308063,
+      "grad_norm": 2.171875,
+      "learning_rate": 9.664499603180135e-05,
+      "loss": 0.7173,
+      "step": 1835
+    },
+    {
+      "epoch": 0.4428399518652226,
+      "grad_norm": 1.9921875,
+      "learning_rate": 9.662405820474634e-05,
+      "loss": 0.705,
+      "step": 1840
+    },
+    {
+      "epoch": 0.44404332129963897,
+      "grad_norm": 2.15625,
+      "learning_rate": 9.660306635464152e-05,
+      "loss": 0.7186,
+      "step": 1845
+    },
+    {
+      "epoch": 0.4452466907340554,
+      "grad_norm": 1.9296875,
+      "learning_rate": 9.658202051501034e-05,
+      "loss": 0.7093,
+      "step": 1850
+    },
+    {
+      "epoch": 0.4464500601684717,
+      "grad_norm": 1.9375,
+      "learning_rate": 9.656092071946243e-05,
+      "loss": 0.6771,
+      "step": 1855
+    },
+    {
+      "epoch": 0.44765342960288806,
+      "grad_norm": 1.953125,
+      "learning_rate": 9.65397670016936e-05,
+      "loss": 0.7028,
+      "step": 1860
+    },
+    {
+      "epoch": 0.44885679903730447,
+      "grad_norm": 1.9453125,
+      "learning_rate": 9.65185593954857e-05,
+      "loss": 0.7239,
+      "step": 1865
+    },
+    {
+      "epoch": 0.4500601684717208,
+      "grad_norm": 2.140625,
+      "learning_rate": 9.649729793470675e-05,
+      "loss": 0.718,
+      "step": 1870
+    },
+    {
+      "epoch": 0.45126353790613716,
+      "grad_norm": 1.84375,
+      "learning_rate": 9.64759826533107e-05,
+      "loss": 0.6947,
+      "step": 1875
+    },
+    {
+      "epoch": 0.45246690734055356,
+      "grad_norm": 1.8046875,
+      "learning_rate": 9.645461358533747e-05,
+      "loss": 0.708,
+      "step": 1880
+    },
+    {
+      "epoch": 0.4536702767749699,
+      "grad_norm": 1.859375,
+      "learning_rate": 9.643319076491285e-05,
+      "loss": 0.7103,
+      "step": 1885
+    },
+    {
+      "epoch": 0.4548736462093863,
+      "grad_norm": 1.953125,
+      "learning_rate": 9.641171422624853e-05,
+      "loss": 0.7052,
+      "step": 1890
+    },
+    {
+      "epoch": 0.45607701564380265,
+      "grad_norm": 1.7421875,
+      "learning_rate": 9.639018400364193e-05,
+      "loss": 0.7162,
+      "step": 1895
+    },
+    {
+      "epoch": 0.457280385078219,
+      "grad_norm": 1.8671875,
+      "learning_rate": 9.636860013147622e-05,
+      "loss": 0.7069,
+      "step": 1900
+    },
+    {
+      "epoch": 0.4584837545126354,
+      "grad_norm": 1.96875,
+      "learning_rate": 9.63469626442203e-05,
+      "loss": 0.7118,
+      "step": 1905
+    },
+    {
+      "epoch": 0.45968712394705175,
+      "grad_norm": 2.0,
+      "learning_rate": 9.632527157642855e-05,
+      "loss": 0.7374,
+      "step": 1910
+    },
+    {
+      "epoch": 0.4608904933814681,
+      "grad_norm": 1.875,
+      "learning_rate": 9.630352696274109e-05,
+      "loss": 0.7097,
+      "step": 1915
+    },
+    {
+      "epoch": 0.4620938628158845,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.628172883788343e-05,
+      "loss": 0.7122,
+      "step": 1920
+    },
+    {
+      "epoch": 0.46329723225030084,
+      "grad_norm": 2.015625,
+      "learning_rate": 9.625987723666661e-05,
+      "loss": 0.7131,
+      "step": 1925
+    },
+    {
+      "epoch": 0.4645006016847172,
+      "grad_norm": 2.0,
+      "learning_rate": 9.623797219398698e-05,
+      "loss": 0.7118,
+      "step": 1930
+    },
+    {
+      "epoch": 0.4657039711191336,
+      "grad_norm": 1.7109375,
+      "learning_rate": 9.621601374482635e-05,
+      "loss": 0.7052,
+      "step": 1935
+    },
+    {
+      "epoch": 0.46690734055354993,
+      "grad_norm": 1.921875,
+      "learning_rate": 9.619400192425173e-05,
+      "loss": 0.6992,
+      "step": 1940
+    },
+    {
+      "epoch": 0.4681107099879663,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.61719367674154e-05,
+      "loss": 0.7074,
+      "step": 1945
+    },
+    {
+      "epoch": 0.4693140794223827,
+      "grad_norm": 1.8828125,
+      "learning_rate": 9.614981830955479e-05,
+      "loss": 0.6958,
+      "step": 1950
+    },
+    {
+      "epoch": 0.470517448856799,
+      "grad_norm": 1.9375,
+      "learning_rate": 9.61276465859925e-05,
+      "loss": 0.7268,
+      "step": 1955
+    },
+    {
+      "epoch": 0.4717208182912154,
+      "grad_norm": 2.0625,
+      "learning_rate": 9.610542163213614e-05,
+      "loss": 0.6819,
+      "step": 1960
+    },
+    {
+      "epoch": 0.4729241877256318,
+      "grad_norm": 1.90625,
+      "learning_rate": 9.608314348347836e-05,
+      "loss": 0.7258,
+      "step": 1965
+    },
+    {
+      "epoch": 0.4741275571600481,
+      "grad_norm": 1.6171875,
+      "learning_rate": 9.606081217559674e-05,
+      "loss": 0.7005,
+      "step": 1970
+    },
+    {
+      "epoch": 0.4753309265944645,
+      "grad_norm": 1.7734375,
+      "learning_rate": 9.60384277441538e-05,
+      "loss": 0.6824,
+      "step": 1975
+    },
+    {
+      "epoch": 0.47653429602888087,
+      "grad_norm": 1.8203125,
+      "learning_rate": 9.60159902248968e-05,
+      "loss": 0.6861,
+      "step": 1980
+    },
+    {
+      "epoch": 0.4777376654632972,
+      "grad_norm": 1.859375,
+      "learning_rate": 9.59934996536579e-05,
+      "loss": 0.7052,
+      "step": 1985
+    },
+    {
+      "epoch": 0.4789410348977136,
+      "grad_norm": 1.90625,
+      "learning_rate": 9.59709560663539e-05,
+      "loss": 0.6741,
+      "step": 1990
+    },
+    {
+      "epoch": 0.48014440433212996,
+      "grad_norm": 1.953125,
+      "learning_rate": 9.594835949898627e-05,
+      "loss": 0.6708,
+      "step": 1995
+    },
+    {
+      "epoch": 0.4813477737665463,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.592570998764115e-05,
+      "loss": 0.715,
+      "step": 2000
+    },
+    {
+      "epoch": 0.4813477737665463,
+      "eval_loss": 0.6060333847999573,
+      "eval_runtime": 2.6578,
+      "eval_samples_per_second": 75.249,
+      "eval_steps_per_second": 75.249,
+      "step": 2000
+    },
+    {
+      "epoch": 0.4825511432009627,
+      "grad_norm": 1.8125,
+      "learning_rate": 9.590300756848915e-05,
+      "loss": 0.6681,
+      "step": 2005
+    },
+    {
+      "epoch": 0.48375451263537905,
+      "grad_norm": 1.8515625,
+      "learning_rate": 9.588025227778541e-05,
+      "loss": 0.6976,
+      "step": 2010
+    },
+    {
+      "epoch": 0.48495788206979545,
+      "grad_norm": 2.03125,
+      "learning_rate": 9.585744415186954e-05,
+      "loss": 0.7225,
+      "step": 2015
+    },
+    {
+      "epoch": 0.4861612515042118,
+      "grad_norm": 1.859375,
+      "learning_rate": 9.583458322716543e-05,
+      "loss": 0.6724,
+      "step": 2020
+    },
+    {
+      "epoch": 0.48736462093862815,
+      "grad_norm": 1.9921875,
+      "learning_rate": 9.581166954018142e-05,
+      "loss": 0.6961,
+      "step": 2025
+    },
+    {
+      "epoch": 0.48856799037304455,
+      "grad_norm": 1.90625,
+      "learning_rate": 9.578870312750999e-05,
+      "loss": 0.7048,
+      "step": 2030
+    },
+    {
+      "epoch": 0.4897713598074609,
+      "grad_norm": 1.8046875,
+      "learning_rate": 9.576568402582787e-05,
+      "loss": 0.6904,
+      "step": 2035
+    },
+    {
+      "epoch": 0.49097472924187724,
+      "grad_norm": 1.890625,
+      "learning_rate": 9.574261227189596e-05,
+      "loss": 0.6904,
+      "step": 2040
+    },
+    {
+      "epoch": 0.49217809867629364,
+      "grad_norm": 1.90625,
+      "learning_rate": 9.57194879025592e-05,
+      "loss": 0.714,
+      "step": 2045
+    },
+    {
+      "epoch": 0.49338146811071,
+      "grad_norm": 1.7265625,
+      "learning_rate": 9.569631095474656e-05,
+      "loss": 0.7156,
+      "step": 2050
+    },
+    {
+      "epoch": 0.49458483754512633,
+      "grad_norm": 2.0,
+      "learning_rate": 9.567308146547101e-05,
+      "loss": 0.73,
+      "step": 2055
+    },
+    {
+      "epoch": 0.49578820697954273,
+      "grad_norm": 2.015625,
+      "learning_rate": 9.56497994718294e-05,
+      "loss": 0.7007,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4969915764139591,
+      "grad_norm": 1.828125,
+      "learning_rate": 9.562646501100241e-05,
+      "loss": 0.7158,
+      "step": 2065
+    },
+    {
+      "epoch": 0.4981949458483754,
+      "grad_norm": 2.1875,
+      "learning_rate": 9.560307812025458e-05,
+      "loss": 0.7404,
+      "step": 2070
+    },
+    {
+      "epoch": 0.4993983152827918,
+      "grad_norm": 1.8359375,
+      "learning_rate": 9.55796388369341e-05,
+      "loss": 0.6807,
+      "step": 2075
+    },
+    {
+      "epoch": 0.5006016847172082,
+      "grad_norm": 2.0625,
+      "learning_rate": 9.555614719847286e-05,
+      "loss": 0.7248,
+      "step": 2080
+    },
+    {
+      "epoch": 0.5018050541516246,
+      "grad_norm": 1.9921875,
+      "learning_rate": 9.553260324238636e-05,
+      "loss": 0.708,
+      "step": 2085
+    },
+    {
+      "epoch": 0.5030084235860409,
+      "grad_norm": 1.9296875,
+      "learning_rate": 9.550900700627365e-05,
+      "loss": 0.7364,
+      "step": 2090
+    },
+    {
+      "epoch": 0.5042117930204573,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.54853585278173e-05,
+      "loss": 0.6955,
+      "step": 2095
+    },
+    {
+      "epoch": 0.5054151624548736,
+      "grad_norm": 1.890625,
+      "learning_rate": 9.546165784478324e-05,
+      "loss": 0.6905,
+      "step": 2100
+    },
+    {
+      "epoch": 0.50661853188929,
+      "grad_norm": 1.9140625,
+      "learning_rate": 9.543790499502083e-05,
+      "loss": 0.7117,
+      "step": 2105
+    },
+    {
+      "epoch": 0.5078219013237064,
+      "grad_norm": 1.9375,
+      "learning_rate": 9.541410001646273e-05,
+      "loss": 0.6493,
+      "step": 2110
+    },
+    {
+      "epoch": 0.5090252707581228,
+      "grad_norm": 1.8203125,
+      "learning_rate": 9.539024294712479e-05,
+      "loss": 0.6618,
+      "step": 2115
+    },
+    {
+      "epoch": 0.5102286401925391,
+      "grad_norm": 1.6796875,
+      "learning_rate": 9.536633382510613e-05,
+      "loss": 0.6866,
+      "step": 2120
+    },
+    {
+      "epoch": 0.5114320096269555,
+      "grad_norm": 1.8671875,
+      "learning_rate": 9.534237268858897e-05,
+      "loss": 0.699,
+      "step": 2125
+    },
+    {
+      "epoch": 0.5126353790613718,
+      "grad_norm": 1.734375,
+      "learning_rate": 9.531835957583856e-05,
+      "loss": 0.6887,
+      "step": 2130
+    },
+    {
+      "epoch": 0.5138387484957883,
+      "grad_norm": 1.7578125,
+      "learning_rate": 9.529429452520319e-05,
+      "loss": 0.7027,
+      "step": 2135
+    },
+    {
+      "epoch": 0.5150421179302046,
+      "grad_norm": 2.015625,
+      "learning_rate": 9.527017757511408e-05,
+      "loss": 0.7184,
+      "step": 2140
+    },
+    {
+      "epoch": 0.516245487364621,
+      "grad_norm": 1.9375,
+      "learning_rate": 9.524600876408537e-05,
+      "loss": 0.6526,
+      "step": 2145
+    },
+    {
+      "epoch": 0.5174488567990373,
+      "grad_norm": 1.9375,
+      "learning_rate": 9.522178813071393e-05,
+      "loss": 0.7014,
+      "step": 2150
+    },
+    {
+      "epoch": 0.5186522262334536,
+      "grad_norm": 2.015625,
+      "learning_rate": 9.519751571367947e-05,
+      "loss": 0.6784,
+      "step": 2155
+    },
+    {
+      "epoch": 0.51985559566787,
+      "grad_norm": 1.9921875,
+      "learning_rate": 9.51731915517444e-05,
+      "loss": 0.6842,
+      "step": 2160
+    },
+    {
+      "epoch": 0.5210589651022864,
+      "grad_norm": 2.140625,
+      "learning_rate": 9.51488156837537e-05,
+      "loss": 0.6869,
+      "step": 2165
+    },
+    {
+      "epoch": 0.5222623345367028,
+      "grad_norm": 1.859375,
+      "learning_rate": 9.512438814863498e-05,
+      "loss": 0.7064,
+      "step": 2170
+    },
+    {
+      "epoch": 0.5234657039711191,
+      "grad_norm": 1.8515625,
+      "learning_rate": 9.509990898539835e-05,
+      "loss": 0.689,
+      "step": 2175
+    },
+    {
+      "epoch": 0.5246690734055355,
+      "grad_norm": 1.8359375,
+      "learning_rate": 9.507537823313632e-05,
+      "loss": 0.6771,
+      "step": 2180
+    },
+    {
+      "epoch": 0.5258724428399518,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.505079593102387e-05,
+      "loss": 0.7189,
+      "step": 2185
+    },
+    {
+      "epoch": 0.5270758122743683,
+      "grad_norm": 1.921875,
+      "learning_rate": 9.502616211831825e-05,
+      "loss": 0.6938,
+      "step": 2190
+    },
+    {
+      "epoch": 0.5282791817087846,
+      "grad_norm": 1.859375,
+      "learning_rate": 9.500147683435897e-05,
+      "loss": 0.698,
+      "step": 2195
+    },
+    {
+      "epoch": 0.529482551143201,
+      "grad_norm": 1.9375,
+      "learning_rate": 9.497674011856776e-05,
+      "loss": 0.7275,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5306859205776173,
+      "grad_norm": 1.9296875,
+      "learning_rate": 9.495195201044847e-05,
+      "loss": 0.6597,
+      "step": 2205
+    },
+    {
+      "epoch": 0.5318892900120337,
+      "grad_norm": 1.7734375,
+      "learning_rate": 9.492711254958704e-05,
+      "loss": 0.6686,
+      "step": 2210
+    },
+    {
+      "epoch": 0.53309265944645,
+      "grad_norm": 1.78125,
+      "learning_rate": 9.490222177565137e-05,
+      "loss": 0.6785,
+      "step": 2215
+    },
+    {
+      "epoch": 0.5342960288808665,
+      "grad_norm": 1.953125,
+      "learning_rate": 9.487727972839139e-05,
+      "loss": 0.6563,
+      "step": 2220
+    },
+    {
+      "epoch": 0.5354993983152828,
+      "grad_norm": 2.03125,
+      "learning_rate": 9.485228644763884e-05,
+      "loss": 0.6653,
+      "step": 2225
+    },
+    {
+      "epoch": 0.5367027677496992,
+      "grad_norm": 1.796875,
+      "learning_rate": 9.482724197330728e-05,
+      "loss": 0.6741,
+      "step": 2230
+    },
+    {
+      "epoch": 0.5379061371841155,
+      "grad_norm": 1.8671875,
+      "learning_rate": 9.480214634539208e-05,
+      "loss": 0.6841,
+      "step": 2235
+    },
+    {
+      "epoch": 0.5391095066185319,
+      "grad_norm": 1.8046875,
+      "learning_rate": 9.477699960397028e-05,
+      "loss": 0.6954,
+      "step": 2240
+    },
+    {
+      "epoch": 0.5403128760529483,
+      "grad_norm": 1.953125,
+      "learning_rate": 9.475180178920049e-05,
+      "loss": 0.6455,
+      "step": 2245
+    },
+    {
+      "epoch": 0.5415162454873647,
+      "grad_norm": 1.6875,
+      "learning_rate": 9.472655294132295e-05,
+      "loss": 0.6932,
+      "step": 2250
+    },
+    {
+      "epoch": 0.542719614921781,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.470125310065938e-05,
+      "loss": 0.6943,
+      "step": 2255
+    },
+    {
+      "epoch": 0.5439229843561973,
+      "grad_norm": 1.875,
+      "learning_rate": 9.467590230761291e-05,
+      "loss": 0.6848,
+      "step": 2260
+    },
+    {
+      "epoch": 0.5451263537906137,
+      "grad_norm": 1.875,
+      "learning_rate": 9.46505006026681e-05,
+      "loss": 0.6785,
+      "step": 2265
+    },
+    {
+      "epoch": 0.54632972322503,
+      "grad_norm": 1.8671875,
+      "learning_rate": 9.462504802639071e-05,
+      "loss": 0.6803,
+      "step": 2270
+    },
+    {
+      "epoch": 0.5475330926594465,
+      "grad_norm": 1.9921875,
+      "learning_rate": 9.459954461942788e-05,
+      "loss": 0.7146,
+      "step": 2275
+    },
+    {
+      "epoch": 0.5487364620938628,
+      "grad_norm": 2.0,
+      "learning_rate": 9.45739904225078e-05,
+      "loss": 0.6779,
+      "step": 2280
+    },
+    {
+      "epoch": 0.5499398315282792,
+      "grad_norm": 1.8671875,
+      "learning_rate": 9.454838547643983e-05,
+      "loss": 0.6444,
+      "step": 2285
+    },
+    {
+      "epoch": 0.5511432009626955,
+      "grad_norm": 1.7265625,
+      "learning_rate": 9.452272982211438e-05,
+      "loss": 0.7118,
+      "step": 2290
+    },
+    {
+      "epoch": 0.5523465703971119,
+      "grad_norm": 1.9375,
+      "learning_rate": 9.44970235005028e-05,
+      "loss": 0.6885,
+      "step": 2295
+    },
+    {
+      "epoch": 0.5535499398315282,
+      "grad_norm": 1.859375,
+      "learning_rate": 9.447126655265739e-05,
+      "loss": 0.6662,
+      "step": 2300
+    },
+    {
+      "epoch": 0.5547533092659447,
+      "grad_norm": 1.7421875,
+      "learning_rate": 9.44454590197113e-05,
+      "loss": 0.6727,
+      "step": 2305
+    },
+    {
+      "epoch": 0.555956678700361,
+      "grad_norm": 1.9921875,
+      "learning_rate": 9.441960094287843e-05,
+      "loss": 0.6777,
+      "step": 2310
+    },
+    {
+      "epoch": 0.5571600481347774,
+      "grad_norm": 1.9296875,
+      "learning_rate": 9.439369236345344e-05,
+      "loss": 0.6696,
+      "step": 2315
+    },
+    {
+      "epoch": 0.5583634175691937,
+      "grad_norm": 1.875,
+      "learning_rate": 9.436773332281159e-05,
+      "loss": 0.6575,
+      "step": 2320
+    },
+    {
+      "epoch": 0.5595667870036101,
+      "grad_norm": 1.9921875,
+      "learning_rate": 9.434172386240877e-05,
+      "loss": 0.6939,
+      "step": 2325
+    },
+    {
+      "epoch": 0.5607701564380265,
+      "grad_norm": 2.03125,
+      "learning_rate": 9.431566402378138e-05,
+      "loss": 0.713,
+      "step": 2330
+    },
+    {
+      "epoch": 0.5619735258724429,
+      "grad_norm": 1.890625,
+      "learning_rate": 9.428955384854625e-05,
+      "loss": 0.6941,
+      "step": 2335
+    },
+    {
+      "epoch": 0.5631768953068592,
+      "grad_norm": 1.9375,
+      "learning_rate": 9.426339337840063e-05,
+      "loss": 0.6716,
+      "step": 2340
+    },
+    {
+      "epoch": 0.5643802647412756,
+      "grad_norm": 1.90625,
+      "learning_rate": 9.423718265512205e-05,
+      "loss": 0.7128,
+      "step": 2345
+    },
+    {
+      "epoch": 0.5655836341756919,
+      "grad_norm": 1.8828125,
+      "learning_rate": 9.421092172056834e-05,
+      "loss": 0.7135,
+      "step": 2350
+    },
+    {
+      "epoch": 0.5667870036101083,
+      "grad_norm": 1.8359375,
+      "learning_rate": 9.418461061667748e-05,
+      "loss": 0.6849,
+      "step": 2355
+    },
+    {
+      "epoch": 0.5679903730445247,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.415824938546756e-05,
+      "loss": 0.6792,
+      "step": 2360
+    },
+    {
+      "epoch": 0.5691937424789411,
+      "grad_norm": 1.765625,
+      "learning_rate": 9.413183806903677e-05,
+      "loss": 0.6967,
+      "step": 2365
+    },
+    {
+      "epoch": 0.5703971119133574,
+      "grad_norm": 1.75,
+      "learning_rate": 9.410537670956326e-05,
+      "loss": 0.681,
+      "step": 2370
+    },
+    {
+      "epoch": 0.5716004813477737,
+      "grad_norm": 1.71875,
+      "learning_rate": 9.407886534930506e-05,
+      "loss": 0.6516,
+      "step": 2375
+    },
+    {
+      "epoch": 0.5728038507821901,
+      "grad_norm": 1.9453125,
+      "learning_rate": 9.405230403060008e-05,
+      "loss": 0.7073,
+      "step": 2380
+    },
+    {
+      "epoch": 0.5740072202166066,
+      "grad_norm": 1.90625,
+      "learning_rate": 9.402569279586608e-05,
+      "loss": 0.6888,
+      "step": 2385
+    },
+    {
+      "epoch": 0.5752105896510229,
+      "grad_norm": 1.859375,
+      "learning_rate": 9.39990316876004e-05,
+      "loss": 0.7084,
+      "step": 2390
+    },
+    {
+      "epoch": 0.5764139590854392,
+      "grad_norm": 1.8828125,
+      "learning_rate": 9.397232074838013e-05,
+      "loss": 0.6788,
+      "step": 2395
+    },
+    {
+      "epoch": 0.5776173285198556,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.39455600208619e-05,
+      "loss": 0.6872,
+      "step": 2400
+    },
+    {
+      "epoch": 0.5788206979542719,
+      "grad_norm": 2.0,
+      "learning_rate": 9.391874954778185e-05,
+      "loss": 0.7138,
+      "step": 2405
+    },
+    {
+      "epoch": 0.5800240673886883,
+      "grad_norm": 2.109375,
+      "learning_rate": 9.389188937195558e-05,
+      "loss": 0.6856,
+      "step": 2410
+    },
+    {
+      "epoch": 0.5812274368231047,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.386497953627805e-05,
+      "loss": 0.6685,
+      "step": 2415
+    },
+    {
+      "epoch": 0.5824308062575211,
+      "grad_norm": 1.8671875,
+      "learning_rate": 9.383802008372352e-05,
+      "loss": 0.674,
+      "step": 2420
+    },
+    {
+      "epoch": 0.5836341756919374,
+      "grad_norm": 1.8203125,
+      "learning_rate": 9.38110110573455e-05,
+      "loss": 0.6623,
+      "step": 2425
+    },
+    {
+      "epoch": 0.5848375451263538,
+      "grad_norm": 1.75,
+      "learning_rate": 9.378395250027666e-05,
+      "loss": 0.7277,
+      "step": 2430
+    },
+    {
+      "epoch": 0.5860409145607701,
+      "grad_norm": 1.8359375,
+      "learning_rate": 9.375684445572877e-05,
+      "loss": 0.6575,
+      "step": 2435
+    },
+    {
+      "epoch": 0.5872442839951865,
+      "grad_norm": 1.9375,
+      "learning_rate": 9.372968696699263e-05,
+      "loss": 0.6956,
+      "step": 2440
+    },
+    {
+      "epoch": 0.5884476534296029,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.370248007743801e-05,
+      "loss": 0.6764,
+      "step": 2445
+    },
+    {
+      "epoch": 0.5896510228640193,
+      "grad_norm": 1.765625,
+      "learning_rate": 9.367522383051357e-05,
+      "loss": 0.6936,
+      "step": 2450
+    },
+    {
+      "epoch": 0.5908543922984356,
+      "grad_norm": 1.8515625,
+      "learning_rate": 9.364791826974676e-05,
+      "loss": 0.6705,
+      "step": 2455
+    },
+    {
+      "epoch": 0.592057761732852,
+      "grad_norm": 1.765625,
+      "learning_rate": 9.362056343874385e-05,
+      "loss": 0.6488,
+      "step": 2460
+    },
+    {
+      "epoch": 0.5932611311672683,
+      "grad_norm": 1.9921875,
+      "learning_rate": 9.359315938118973e-05,
+      "loss": 0.6975,
+      "step": 2465
+    },
+    {
+      "epoch": 0.5944645006016848,
+      "grad_norm": 1.984375,
+      "learning_rate": 9.356570614084792e-05,
+      "loss": 0.6532,
+      "step": 2470
+    },
+    {
+      "epoch": 0.5956678700361011,
+      "grad_norm": 1.6953125,
+      "learning_rate": 9.35382037615605e-05,
+      "loss": 0.6788,
+      "step": 2475
+    },
+    {
+      "epoch": 0.5968712394705175,
+      "grad_norm": 2.0,
+      "learning_rate": 9.351065228724801e-05,
+      "loss": 0.6791,
+      "step": 2480
+    },
+    {
+      "epoch": 0.5980746089049338,
+      "grad_norm": 1.890625,
+      "learning_rate": 9.348305176190939e-05,
+      "loss": 0.6726,
+      "step": 2485
+    },
+    {
+      "epoch": 0.5992779783393501,
+      "grad_norm": 1.9453125,
+      "learning_rate": 9.345540222962194e-05,
+      "loss": 0.6567,
+      "step": 2490
+    },
+    {
+      "epoch": 0.6004813477737665,
+      "grad_norm": 2.109375,
+      "learning_rate": 9.342770373454118e-05,
+      "loss": 0.693,
+      "step": 2495
+    },
+    {
+      "epoch": 0.601684717208183,
+      "grad_norm": 1.8828125,
+      "learning_rate": 9.339995632090086e-05,
+      "loss": 0.6774,
+      "step": 2500
+    },
+    {
+      "epoch": 0.601684717208183,
+      "eval_loss": 0.5781615972518921,
+      "eval_runtime": 2.7886,
+      "eval_samples_per_second": 71.721,
+      "eval_steps_per_second": 71.721,
+      "step": 2500
+    },
+    {
+      "epoch": 0.6028880866425993,
+      "grad_norm": 1.96875,
+      "learning_rate": 9.337216003301284e-05,
+      "loss": 0.697,
+      "step": 2505
+    },
+    {
+      "epoch": 0.6040914560770156,
+      "grad_norm": 1.7421875,
+      "learning_rate": 9.334431491526702e-05,
+      "loss": 0.6809,
+      "step": 2510
+    },
+    {
+      "epoch": 0.605294825511432,
+      "grad_norm": 1.78125,
+      "learning_rate": 9.33164210121313e-05,
+      "loss": 0.6707,
+      "step": 2515
+    },
+    {
+      "epoch": 0.6064981949458483,
+      "grad_norm": 1.8515625,
+      "learning_rate": 9.328847836815145e-05,
+      "loss": 0.6736,
+      "step": 2520
+    },
+    {
+      "epoch": 0.6077015643802648,
+      "grad_norm": 1.78125,
+      "learning_rate": 9.326048702795112e-05,
+      "loss": 0.6887,
+      "step": 2525
+    },
+    {
+      "epoch": 0.6089049338146811,
+      "grad_norm": 1.8125,
+      "learning_rate": 9.32324470362317e-05,
+      "loss": 0.6671,
+      "step": 2530
+    },
+    {
+      "epoch": 0.6101083032490975,
+      "grad_norm": 1.8125,
+      "learning_rate": 9.320435843777231e-05,
+      "loss": 0.6665,
+      "step": 2535
+    },
+    {
+      "epoch": 0.6113116726835138,
+      "grad_norm": 1.8984375,
+      "learning_rate": 9.317622127742965e-05,
+      "loss": 0.6633,
+      "step": 2540
+    },
+    {
+      "epoch": 0.6125150421179302,
+      "grad_norm": 1.7734375,
+      "learning_rate": 9.3148035600138e-05,
+      "loss": 0.6628,
+      "step": 2545
+    },
+    {
+      "epoch": 0.6137184115523465,
+      "grad_norm": 1.7421875,
+      "learning_rate": 9.311980145090908e-05,
+      "loss": 0.6491,
+      "step": 2550
+    },
+    {
+      "epoch": 0.614921780986763,
+      "grad_norm": 1.8046875,
+      "learning_rate": 9.30915188748321e-05,
+      "loss": 0.6935,
+      "step": 2555
+    },
+    {
+      "epoch": 0.6161251504211793,
+      "grad_norm": 2.015625,
+      "learning_rate": 9.306318791707354e-05,
+      "loss": 0.683,
+      "step": 2560
+    },
+    {
+      "epoch": 0.6173285198555957,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.303480862287714e-05,
+      "loss": 0.6731,
+      "step": 2565
+    },
+    {
+      "epoch": 0.618531889290012,
+      "grad_norm": 1.9296875,
+      "learning_rate": 9.300638103756387e-05,
+      "loss": 0.6911,
+      "step": 2570
+    },
+    {
+      "epoch": 0.6197352587244284,
+      "grad_norm": 1.859375,
+      "learning_rate": 9.29779052065318e-05,
+      "loss": 0.6096,
+      "step": 2575
+    },
+    {
+      "epoch": 0.6209386281588448,
+      "grad_norm": 1.984375,
+      "learning_rate": 9.294938117525605e-05,
+      "loss": 0.6799,
+      "step": 2580
+    },
+    {
+      "epoch": 0.6221419975932612,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.292080898928872e-05,
+      "loss": 0.6589,
+      "step": 2585
+    },
+    {
+      "epoch": 0.6233453670276775,
+      "grad_norm": 1.8671875,
+      "learning_rate": 9.289218869425879e-05,
+      "loss": 0.6221,
+      "step": 2590
+    },
+    {
+      "epoch": 0.6245487364620939,
+      "grad_norm": 1.7265625,
+      "learning_rate": 9.28635203358721e-05,
+      "loss": 0.6513,
+      "step": 2595
+    },
+    {
+      "epoch": 0.6257521058965102,
+      "grad_norm": 1.84375,
+      "learning_rate": 9.283480395991122e-05,
+      "loss": 0.6326,
+      "step": 2600
+    },
+    {
+      "epoch": 0.6269554753309265,
+      "grad_norm": 1.9140625,
+      "learning_rate": 9.28060396122354e-05,
+      "loss": 0.6824,
+      "step": 2605
+    },
+    {
+      "epoch": 0.628158844765343,
+      "grad_norm": 1.875,
+      "learning_rate": 9.277722733878054e-05,
+      "loss": 0.6777,
+      "step": 2610
+    },
+    {
+      "epoch": 0.6293622141997594,
+      "grad_norm": 1.8046875,
+      "learning_rate": 9.274836718555903e-05,
+      "loss": 0.652,
+      "step": 2615
+    },
+    {
+      "epoch": 0.6305655836341757,
+      "grad_norm": 1.8828125,
+      "learning_rate": 9.271945919865976e-05,
+      "loss": 0.6514,
+      "step": 2620
+    },
+    {
+      "epoch": 0.631768953068592,
+      "grad_norm": 1.953125,
+      "learning_rate": 9.269050342424796e-05,
+      "loss": 0.6998,
+      "step": 2625
+    },
+    {
+      "epoch": 0.6329723225030084,
+      "grad_norm": 1.8671875,
+      "learning_rate": 9.266149990856521e-05,
+      "loss": 0.6756,
+      "step": 2630
+    },
+    {
+      "epoch": 0.6341756919374247,
+      "grad_norm": 1.78125,
+      "learning_rate": 9.263244869792937e-05,
+      "loss": 0.6714,
+      "step": 2635
+    },
+    {
+      "epoch": 0.6353790613718412,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.260334983873439e-05,
+      "loss": 0.6422,
+      "step": 2640
+    },
+    {
+      "epoch": 0.6365824308062575,
+      "grad_norm": 1.6875,
+      "learning_rate": 9.257420337745034e-05,
+      "loss": 0.6618,
+      "step": 2645
+    },
+    {
+      "epoch": 0.6377858002406739,
+      "grad_norm": 1.890625,
+      "learning_rate": 9.254500936062334e-05,
+      "loss": 0.652,
+      "step": 2650
+    },
+    {
+      "epoch": 0.6389891696750902,
+      "grad_norm": 1.7109375,
+      "learning_rate": 9.251576783487541e-05,
+      "loss": 0.6936,
+      "step": 2655
+    },
+    {
+      "epoch": 0.6401925391095066,
+      "grad_norm": 1.7578125,
+      "learning_rate": 9.248647884690448e-05,
+      "loss": 0.6398,
+      "step": 2660
+    },
+    {
+      "epoch": 0.641395908543923,
+      "grad_norm": 1.90625,
+      "learning_rate": 9.245714244348425e-05,
+      "loss": 0.6414,
+      "step": 2665
+    },
+    {
+      "epoch": 0.6425992779783394,
+      "grad_norm": 1.7265625,
+      "learning_rate": 9.242775867146415e-05,
+      "loss": 0.6777,
+      "step": 2670
+    },
+    {
+      "epoch": 0.6438026474127557,
+      "grad_norm": 1.8359375,
+      "learning_rate": 9.239832757776926e-05,
+      "loss": 0.6736,
+      "step": 2675
+    },
+    {
+      "epoch": 0.6450060168471721,
+      "grad_norm": 1.8984375,
+      "learning_rate": 9.236884920940024e-05,
+      "loss": 0.662,
+      "step": 2680
+    },
+    {
+      "epoch": 0.6462093862815884,
+      "grad_norm": 1.8359375,
+      "learning_rate": 9.233932361343321e-05,
+      "loss": 0.6441,
+      "step": 2685
+    },
+    {
+      "epoch": 0.6474127557160048,
+      "grad_norm": 1.9765625,
+      "learning_rate": 9.230975083701976e-05,
+      "loss": 0.6484,
+      "step": 2690
+    },
+    {
+      "epoch": 0.6486161251504212,
+      "grad_norm": 1.8125,
+      "learning_rate": 9.228013092738675e-05,
+      "loss": 0.6511,
+      "step": 2695
+    },
+    {
+      "epoch": 0.6498194945848376,
+      "grad_norm": 1.9609375,
+      "learning_rate": 9.225046393183642e-05,
+      "loss": 0.6311,
+      "step": 2700
+    },
+    {
+      "epoch": 0.6510228640192539,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.222074989774612e-05,
+      "loss": 0.657,
+      "step": 2705
+    },
+    {
+      "epoch": 0.6522262334536703,
+      "grad_norm": 1.8125,
+      "learning_rate": 9.219098887256835e-05,
+      "loss": 0.6849,
+      "step": 2710
+    },
+    {
+      "epoch": 0.6534296028880866,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.216118090383067e-05,
+      "loss": 0.6848,
+      "step": 2715
+    },
+    {
+      "epoch": 0.6546329723225031,
+      "grad_norm": 1.65625,
+      "learning_rate": 9.213132603913553e-05,
+      "loss": 0.6083,
+      "step": 2720
+    },
+    {
+      "epoch": 0.6558363417569194,
+      "grad_norm": 1.734375,
+      "learning_rate": 9.210142432616035e-05,
+      "loss": 0.6691,
+      "step": 2725
+    },
+    {
+      "epoch": 0.6570397111913358,
+      "grad_norm": 1.6875,
+      "learning_rate": 9.207147581265739e-05,
+      "loss": 0.6658,
+      "step": 2730
+    },
+    {
+      "epoch": 0.6582430806257521,
+      "grad_norm": 1.9453125,
+      "learning_rate": 9.204148054645356e-05,
+      "loss": 0.6723,
+      "step": 2735
+    },
+    {
+      "epoch": 0.6594464500601684,
+      "grad_norm": 1.828125,
+      "learning_rate": 9.201143857545048e-05,
+      "loss": 0.6736,
+      "step": 2740
+    },
+    {
+      "epoch": 0.6606498194945848,
+      "grad_norm": 1.890625,
+      "learning_rate": 9.198134994762436e-05,
+      "loss": 0.6836,
+      "step": 2745
+    },
+    {
+      "epoch": 0.6618531889290012,
+      "grad_norm": 1.9609375,
+      "learning_rate": 9.19512147110259e-05,
+      "loss": 0.6888,
+      "step": 2750
+    },
+    {
+      "epoch": 0.6630565583634176,
+      "grad_norm": 1.6875,
+      "learning_rate": 9.192103291378025e-05,
+      "loss": 0.6839,
+      "step": 2755
+    },
+    {
+      "epoch": 0.6642599277978339,
+      "grad_norm": 1.59375,
+      "learning_rate": 9.189080460408692e-05,
+      "loss": 0.6831,
+      "step": 2760
+    },
+    {
+      "epoch": 0.6654632972322503,
+      "grad_norm": 1.84375,
+      "learning_rate": 9.186052983021966e-05,
+      "loss": 0.6923,
+      "step": 2765
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 1.9140625,
+      "learning_rate": 9.18302086405265e-05,
+      "loss": 0.6671,
+      "step": 2770
+    },
+    {
+      "epoch": 0.6678700361010831,
+      "grad_norm": 1.8671875,
+      "learning_rate": 9.17998410834295e-05,
+      "loss": 0.6979,
+      "step": 2775
+    },
+    {
+      "epoch": 0.6690734055354994,
+      "grad_norm": 2.359375,
+      "learning_rate": 9.176942720742483e-05,
+      "loss": 0.6569,
+      "step": 2780
+    },
+    {
+      "epoch": 0.6702767749699158,
+      "grad_norm": 1.8359375,
+      "learning_rate": 9.173896706108264e-05,
+      "loss": 0.6729,
+      "step": 2785
+    },
+    {
+      "epoch": 0.6714801444043321,
+      "grad_norm": 1.90625,
+      "learning_rate": 9.17084606930469e-05,
+      "loss": 0.6415,
+      "step": 2790
+    },
+    {
+      "epoch": 0.6726835138387485,
+      "grad_norm": 1.84375,
+      "learning_rate": 9.167790815203546e-05,
+      "loss": 0.6599,
+      "step": 2795
+    },
+    {
+      "epoch": 0.6738868832731648,
+      "grad_norm": 1.8359375,
+      "learning_rate": 9.164730948683991e-05,
+      "loss": 0.6626,
+      "step": 2800
+    },
+    {
+      "epoch": 0.6750902527075813,
+      "grad_norm": 1.75,
+      "learning_rate": 9.161666474632543e-05,
+      "loss": 0.6445,
+      "step": 2805
+    },
+    {
+      "epoch": 0.6762936221419976,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.158597397943088e-05,
+      "loss": 0.6633,
+      "step": 2810
+    },
+    {
+      "epoch": 0.677496991576414,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.155523723516855e-05,
+      "loss": 0.6755,
+      "step": 2815
+    },
+    {
+      "epoch": 0.6787003610108303,
+      "grad_norm": 1.859375,
+      "learning_rate": 9.152445456262417e-05,
+      "loss": 0.6859,
+      "step": 2820
+    },
+    {
+      "epoch": 0.6799037304452467,
+      "grad_norm": 2.0,
+      "learning_rate": 9.14936260109568e-05,
+      "loss": 0.6821,
+      "step": 2825
+    },
+    {
+      "epoch": 0.681107099879663,
+      "grad_norm": 1.875,
+      "learning_rate": 9.146275162939884e-05,
+      "loss": 0.6637,
+      "step": 2830
+    },
+    {
+      "epoch": 0.6823104693140795,
+      "grad_norm": 1.9921875,
+      "learning_rate": 9.143183146725579e-05,
+      "loss": 0.6739,
+      "step": 2835
+    },
+    {
+      "epoch": 0.6835138387484958,
+      "grad_norm": 2.0625,
+      "learning_rate": 9.140086557390632e-05,
+      "loss": 0.6819,
+      "step": 2840
+    },
+    {
+      "epoch": 0.6847172081829122,
+      "grad_norm": 1.59375,
+      "learning_rate": 9.13698539988021e-05,
+      "loss": 0.6595,
+      "step": 2845
+    },
+    {
+      "epoch": 0.6859205776173285,
+      "grad_norm": 1.96875,
+      "learning_rate": 9.133879679146775e-05,
+      "loss": 0.6268,
+      "step": 2850
+    },
+    {
+      "epoch": 0.6871239470517448,
+      "grad_norm": 1.9765625,
+      "learning_rate": 9.13076940015008e-05,
+      "loss": 0.6373,
+      "step": 2855
+    },
+    {
+      "epoch": 0.6883273164861613,
+      "grad_norm": 1.8203125,
+      "learning_rate": 9.127654567857154e-05,
+      "loss": 0.6741,
+      "step": 2860
+    },
+    {
+      "epoch": 0.6895306859205776,
+      "grad_norm": 1.6484375,
+      "learning_rate": 9.124535187242297e-05,
+      "loss": 0.6556,
+      "step": 2865
+    },
+    {
+      "epoch": 0.690734055354994,
+      "grad_norm": 1.8203125,
+      "learning_rate": 9.121411263287077e-05,
+      "loss": 0.6569,
+      "step": 2870
+    },
+    {
+      "epoch": 0.6919374247894103,
+      "grad_norm": 1.765625,
+      "learning_rate": 9.118282800980315e-05,
+      "loss": 0.6308,
+      "step": 2875
+    },
+    {
+      "epoch": 0.6931407942238267,
+      "grad_norm": 1.6640625,
+      "learning_rate": 9.115149805318075e-05,
+      "loss": 0.6613,
+      "step": 2880
+    },
+    {
+      "epoch": 0.694344163658243,
+      "grad_norm": 1.7421875,
+      "learning_rate": 9.11201228130367e-05,
+      "loss": 0.6141,
+      "step": 2885
+    },
+    {
+      "epoch": 0.6955475330926595,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.108870233947634e-05,
+      "loss": 0.6304,
+      "step": 2890
+    },
+    {
+      "epoch": 0.6967509025270758,
+      "grad_norm": 1.859375,
+      "learning_rate": 9.105723668267736e-05,
+      "loss": 0.6662,
+      "step": 2895
+    },
+    {
+      "epoch": 0.6979542719614922,
+      "grad_norm": 1.828125,
+      "learning_rate": 9.10257258928895e-05,
+      "loss": 0.6322,
+      "step": 2900
+    },
+    {
+      "epoch": 0.6991576413959085,
+      "grad_norm": 1.8515625,
+      "learning_rate": 9.099417002043466e-05,
+      "loss": 0.6472,
+      "step": 2905
+    },
+    {
+      "epoch": 0.7003610108303249,
+      "grad_norm": 1.7578125,
+      "learning_rate": 9.09625691157067e-05,
+      "loss": 0.6505,
+      "step": 2910
+    },
+    {
+      "epoch": 0.7015643802647413,
+      "grad_norm": 1.921875,
+      "learning_rate": 9.093092322917136e-05,
+      "loss": 0.6569,
+      "step": 2915
+    },
+    {
+      "epoch": 0.7027677496991577,
+      "grad_norm": 1.875,
+      "learning_rate": 9.089923241136629e-05,
+      "loss": 0.6569,
+      "step": 2920
+    },
+    {
+      "epoch": 0.703971119133574,
+      "grad_norm": 1.796875,
+      "learning_rate": 9.086749671290083e-05,
+      "loss": 0.6523,
+      "step": 2925
+    },
+    {
+      "epoch": 0.7051744885679904,
+      "grad_norm": 1.8125,
+      "learning_rate": 9.083571618445603e-05,
+      "loss": 0.6619,
+      "step": 2930
+    },
+    {
+      "epoch": 0.7063778580024067,
+      "grad_norm": 1.796875,
+      "learning_rate": 9.080389087678451e-05,
+      "loss": 0.65,
+      "step": 2935
+    },
+    {
+      "epoch": 0.7075812274368231,
+      "grad_norm": 1.890625,
+      "learning_rate": 9.077202084071043e-05,
+      "loss": 0.6475,
+      "step": 2940
+    },
+    {
+      "epoch": 0.7087845968712395,
+      "grad_norm": 1.9296875,
+      "learning_rate": 9.074010612712936e-05,
+      "loss": 0.6487,
+      "step": 2945
+    },
+    {
+      "epoch": 0.7099879663056559,
+      "grad_norm": 1.8125,
+      "learning_rate": 9.070814678700821e-05,
+      "loss": 0.6291,
+      "step": 2950
+    },
+    {
+      "epoch": 0.7111913357400722,
+      "grad_norm": 1.9140625,
+      "learning_rate": 9.067614287138514e-05,
+      "loss": 0.6742,
+      "step": 2955
+    },
+    {
+      "epoch": 0.7123947051744886,
+      "grad_norm": 1.6796875,
+      "learning_rate": 9.064409443136955e-05,
+      "loss": 0.6316,
+      "step": 2960
+    },
+    {
+      "epoch": 0.7135980746089049,
+      "grad_norm": 1.9453125,
+      "learning_rate": 9.061200151814195e-05,
+      "loss": 0.6678,
+      "step": 2965
+    },
+    {
+      "epoch": 0.7148014440433214,
+      "grad_norm": 1.8671875,
+      "learning_rate": 9.057986418295378e-05,
+      "loss": 0.6628,
+      "step": 2970
+    },
+    {
+      "epoch": 0.7160048134777377,
+      "grad_norm": 1.8359375,
+      "learning_rate": 9.054768247712753e-05,
+      "loss": 0.6297,
+      "step": 2975
+    },
+    {
+      "epoch": 0.717208182912154,
+      "grad_norm": 1.8046875,
+      "learning_rate": 9.051545645205646e-05,
+      "loss": 0.6502,
+      "step": 2980
+    },
+    {
+      "epoch": 0.7184115523465704,
+      "grad_norm": 1.859375,
+      "learning_rate": 9.048318615920468e-05,
+      "loss": 0.6704,
+      "step": 2985
+    },
+    {
+      "epoch": 0.7196149217809867,
+      "grad_norm": 1.7109375,
+      "learning_rate": 9.045087165010694e-05,
+      "loss": 0.6473,
+      "step": 2990
+    },
+    {
+      "epoch": 0.7208182912154031,
+      "grad_norm": 1.734375,
+      "learning_rate": 9.041851297636862e-05,
+      "loss": 0.6508,
+      "step": 2995
+    },
+    {
+      "epoch": 0.7220216606498195,
+      "grad_norm": 1.7734375,
+      "learning_rate": 9.038611018966564e-05,
+      "loss": 0.6492,
+      "step": 3000
+    },
+    {
+      "epoch": 0.7220216606498195,
+      "eval_loss": 0.5566111207008362,
+      "eval_runtime": 2.6561,
+      "eval_samples_per_second": 75.299,
+      "eval_steps_per_second": 75.299,
+      "step": 3000
+    },
+    {
+      "epoch": 0.7232250300842359,
+      "grad_norm": 1.984375,
+      "learning_rate": 9.035366334174436e-05,
+      "loss": 0.6277,
+      "step": 3005
+    },
+    {
+      "epoch": 0.7244283995186522,
+      "grad_norm": 1.71875,
+      "learning_rate": 9.032117248442153e-05,
+      "loss": 0.649,
+      "step": 3010
+    },
+    {
+      "epoch": 0.7256317689530686,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.02886376695841e-05,
+      "loss": 0.6538,
+      "step": 3015
+    },
+    {
+      "epoch": 0.7268351383874849,
+      "grad_norm": 1.8984375,
+      "learning_rate": 9.02560589491893e-05,
+      "loss": 0.635,
+      "step": 3020
+    },
+    {
+      "epoch": 0.7280385078219013,
+      "grad_norm": 1.8125,
+      "learning_rate": 9.022343637526446e-05,
+      "loss": 0.6399,
+      "step": 3025
+    },
+    {
+      "epoch": 0.7292418772563177,
+      "grad_norm": 1.953125,
+      "learning_rate": 9.019076999990694e-05,
+      "loss": 0.6522,
+      "step": 3030
+    },
+    {
+      "epoch": 0.7304452466907341,
+      "grad_norm": 1.9765625,
+      "learning_rate": 9.015805987528402e-05,
+      "loss": 0.6566,
+      "step": 3035
+    },
+    {
+      "epoch": 0.7316486161251504,
+      "grad_norm": 1.9609375,
+      "learning_rate": 9.012530605363289e-05,
+      "loss": 0.6781,
+      "step": 3040
+    },
+    {
+      "epoch": 0.7328519855595668,
+      "grad_norm": 1.8046875,
+      "learning_rate": 9.009250858726047e-05,
+      "loss": 0.6405,
+      "step": 3045
+    },
+    {
+      "epoch": 0.7340553549939831,
+      "grad_norm": 1.890625,
+      "learning_rate": 9.005966752854345e-05,
+      "loss": 0.6467,
+      "step": 3050
+    },
+    {
+      "epoch": 0.7352587244283996,
+      "grad_norm": 1.8046875,
+      "learning_rate": 9.002678292992809e-05,
+      "loss": 0.6345,
+      "step": 3055
+    },
+    {
+      "epoch": 0.7364620938628159,
+      "grad_norm": 1.75,
+      "learning_rate": 8.999385484393018e-05,
+      "loss": 0.6224,
+      "step": 3060
+    },
+    {
+      "epoch": 0.7376654632972323,
+      "grad_norm": 1.8671875,
+      "learning_rate": 8.996088332313497e-05,
+      "loss": 0.6236,
+      "step": 3065
+    },
+    {
+      "epoch": 0.7388688327316486,
+      "grad_norm": 1.9296875,
+      "learning_rate": 8.992786842019708e-05,
+      "loss": 0.6464,
+      "step": 3070
+    },
+    {
+      "epoch": 0.740072202166065,
+      "grad_norm": 1.8671875,
+      "learning_rate": 8.989481018784036e-05,
+      "loss": 0.6277,
+      "step": 3075
+    },
+    {
+      "epoch": 0.7412755716004813,
+      "grad_norm": 1.7734375,
+      "learning_rate": 8.986170867885797e-05,
+      "loss": 0.6739,
+      "step": 3080
+    },
+    {
+      "epoch": 0.7424789410348978,
+      "grad_norm": 1.7890625,
+      "learning_rate": 8.982856394611208e-05,
+      "loss": 0.6344,
+      "step": 3085
+    },
+    {
+      "epoch": 0.7436823104693141,
+      "grad_norm": 1.8046875,
+      "learning_rate": 8.97953760425339e-05,
+      "loss": 0.6574,
+      "step": 3090
+    },
+    {
+      "epoch": 0.7448856799037304,
+      "grad_norm": 1.8203125,
+      "learning_rate": 8.976214502112361e-05,
+      "loss": 0.6485,
+      "step": 3095
+    },
+    {
+      "epoch": 0.7460890493381468,
+      "grad_norm": 1.828125,
+      "learning_rate": 8.972887093495022e-05,
+      "loss": 0.6452,
+      "step": 3100
+    },
+    {
+      "epoch": 0.7472924187725631,
+      "grad_norm": 1.984375,
+      "learning_rate": 8.969555383715156e-05,
+      "loss": 0.6481,
+      "step": 3105
+    },
+    {
+      "epoch": 0.7484957882069796,
+      "grad_norm": 1.84375,
+      "learning_rate": 8.966219378093411e-05,
+      "loss": 0.6169,
+      "step": 3110
+    },
+    {
+      "epoch": 0.7496991576413959,
+      "grad_norm": 2.09375,
+      "learning_rate": 8.962879081957296e-05,
+      "loss": 0.6774,
+      "step": 3115
+    },
+    {
+      "epoch": 0.7509025270758123,
+      "grad_norm": 1.6953125,
+      "learning_rate": 8.95953450064117e-05,
+      "loss": 0.6313,
+      "step": 3120
+    },
+    {
+      "epoch": 0.7521058965102286,
+      "grad_norm": 1.8046875,
+      "learning_rate": 8.956185639486242e-05,
+      "loss": 0.6508,
+      "step": 3125
+    },
+    {
+      "epoch": 0.753309265944645,
+      "grad_norm": 1.71875,
+      "learning_rate": 8.952832503840548e-05,
+      "loss": 0.6782,
+      "step": 3130
+    },
+    {
+      "epoch": 0.7545126353790613,
+      "grad_norm": 1.890625,
+      "learning_rate": 8.949475099058955e-05,
+      "loss": 0.681,
+      "step": 3135
+    },
+    {
+      "epoch": 0.7557160048134778,
+      "grad_norm": 1.84375,
+      "learning_rate": 8.946113430503143e-05,
+      "loss": 0.6357,
+      "step": 3140
+    },
+    {
+      "epoch": 0.7569193742478941,
+      "grad_norm": 1.9140625,
+      "learning_rate": 8.942747503541607e-05,
+      "loss": 0.67,
+      "step": 3145
+    },
+    {
+      "epoch": 0.7581227436823105,
+      "grad_norm": 1.8203125,
+      "learning_rate": 8.93937732354964e-05,
+      "loss": 0.6323,
+      "step": 3150
+    },
+    {
+      "epoch": 0.7593261131167268,
+      "grad_norm": 1.75,
+      "learning_rate": 8.936002895909326e-05,
+      "loss": 0.6367,
+      "step": 3155
+    },
+    {
+      "epoch": 0.7605294825511432,
+      "grad_norm": 1.875,
+      "learning_rate": 8.932624226009533e-05,
+      "loss": 0.6521,
+      "step": 3160
+    },
+    {
+      "epoch": 0.7617328519855595,
+      "grad_norm": 1.8359375,
+      "learning_rate": 8.929241319245903e-05,
+      "loss": 0.6794,
+      "step": 3165
+    },
+    {
+      "epoch": 0.762936221419976,
+      "grad_norm": 1.75,
+      "learning_rate": 8.925854181020849e-05,
+      "loss": 0.6673,
+      "step": 3170
+    },
+    {
+      "epoch": 0.7641395908543923,
+      "grad_norm": 1.7265625,
+      "learning_rate": 8.922462816743533e-05,
+      "loss": 0.6203,
+      "step": 3175
+    },
+    {
+      "epoch": 0.7653429602888087,
+      "grad_norm": 1.796875,
+      "learning_rate": 8.919067231829874e-05,
+      "loss": 0.6765,
+      "step": 3180
+    },
+    {
+      "epoch": 0.766546329723225,
+      "grad_norm": 1.6953125,
+      "learning_rate": 8.915667431702524e-05,
+      "loss": 0.6379,
+      "step": 3185
+    },
+    {
+      "epoch": 0.7677496991576414,
+      "grad_norm": 1.7890625,
+      "learning_rate": 8.912263421790873e-05,
+      "loss": 0.6319,
+      "step": 3190
+    },
+    {
+      "epoch": 0.7689530685920578,
+      "grad_norm": 1.828125,
+      "learning_rate": 8.908855207531031e-05,
+      "loss": 0.6547,
+      "step": 3195
+    },
+    {
+      "epoch": 0.7701564380264742,
+      "grad_norm": 1.8984375,
+      "learning_rate": 8.905442794365822e-05,
+      "loss": 0.6284,
+      "step": 3200
+    },
+    {
+      "epoch": 0.7713598074608905,
+      "grad_norm": 1.75,
+      "learning_rate": 8.902026187744776e-05,
+      "loss": 0.6478,
+      "step": 3205
+    },
+    {
+      "epoch": 0.7725631768953068,
+      "grad_norm": 1.8828125,
+      "learning_rate": 8.89860539312412e-05,
+      "loss": 0.6863,
+      "step": 3210
+    },
+    {
+      "epoch": 0.7737665463297232,
+      "grad_norm": 1.75,
+      "learning_rate": 8.89518041596677e-05,
+      "loss": 0.6777,
+      "step": 3215
+    },
+    {
+      "epoch": 0.7749699157641395,
+      "grad_norm": 1.703125,
+      "learning_rate": 8.891751261742318e-05,
+      "loss": 0.6713,
+      "step": 3220
+    },
+    {
+      "epoch": 0.776173285198556,
+      "grad_norm": 1.859375,
+      "learning_rate": 8.88831793592703e-05,
+      "loss": 0.6486,
+      "step": 3225
+    },
+    {
+      "epoch": 0.7773766546329723,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.884880444003835e-05,
+      "loss": 0.6512,
+      "step": 3230
+    },
+    {
+      "epoch": 0.7785800240673887,
+      "grad_norm": 2.046875,
+      "learning_rate": 8.881438791462308e-05,
+      "loss": 0.6519,
+      "step": 3235
+    },
+    {
+      "epoch": 0.779783393501805,
+      "grad_norm": 1.6796875,
+      "learning_rate": 8.877992983798678e-05,
+      "loss": 0.6361,
+      "step": 3240
+    },
+    {
+      "epoch": 0.7809867629362214,
+      "grad_norm": 1.671875,
+      "learning_rate": 8.874543026515806e-05,
+      "loss": 0.6329,
+      "step": 3245
+    },
+    {
+      "epoch": 0.7821901323706378,
+      "grad_norm": 1.6796875,
+      "learning_rate": 8.871088925123174e-05,
+      "loss": 0.6481,
+      "step": 3250
+    },
+    {
+      "epoch": 0.7833935018050542,
+      "grad_norm": 1.765625,
+      "learning_rate": 8.867630685136893e-05,
+      "loss": 0.6521,
+      "step": 3255
+    },
+    {
+      "epoch": 0.7845968712394705,
+      "grad_norm": 1.7734375,
+      "learning_rate": 8.864168312079671e-05,
+      "loss": 0.6221,
+      "step": 3260
+    },
+    {
+      "epoch": 0.7858002406738869,
+      "grad_norm": 1.8671875,
+      "learning_rate": 8.860701811480828e-05,
+      "loss": 0.6471,
+      "step": 3265
+    },
+    {
+      "epoch": 0.7870036101083032,
+      "grad_norm": 1.7265625,
+      "learning_rate": 8.857231188876265e-05,
+      "loss": 0.6588,
+      "step": 3270
+    },
+    {
+      "epoch": 0.7882069795427196,
+      "grad_norm": 1.7578125,
+      "learning_rate": 8.853756449808476e-05,
+      "loss": 0.6249,
+      "step": 3275
+    },
+    {
+      "epoch": 0.789410348977136,
+      "grad_norm": 1.9453125,
+      "learning_rate": 8.85027759982652e-05,
+      "loss": 0.6372,
+      "step": 3280
+    },
+    {
+      "epoch": 0.7906137184115524,
+      "grad_norm": 1.9921875,
+      "learning_rate": 8.846794644486026e-05,
+      "loss": 0.6523,
+      "step": 3285
+    },
+    {
+      "epoch": 0.7918170878459687,
+      "grad_norm": 1.9375,
+      "learning_rate": 8.843307589349178e-05,
+      "loss": 0.6329,
+      "step": 3290
+    },
+    {
+      "epoch": 0.7930204572803851,
+      "grad_norm": 1.8125,
+      "learning_rate": 8.839816439984706e-05,
+      "loss": 0.6525,
+      "step": 3295
+    },
+    {
+      "epoch": 0.7942238267148014,
+      "grad_norm": 1.796875,
+      "learning_rate": 8.836321201967884e-05,
+      "loss": 0.6191,
+      "step": 3300
+    },
+    {
+      "epoch": 0.7954271961492179,
+      "grad_norm": 1.6953125,
+      "learning_rate": 8.832821880880504e-05,
+      "loss": 0.5898,
+      "step": 3305
+    },
+    {
+      "epoch": 0.7966305655836342,
+      "grad_norm": 1.90625,
+      "learning_rate": 8.829318482310891e-05,
+      "loss": 0.6426,
+      "step": 3310
+    },
+    {
+      "epoch": 0.7978339350180506,
+      "grad_norm": 1.7265625,
+      "learning_rate": 8.825811011853874e-05,
+      "loss": 0.6301,
+      "step": 3315
+    },
+    {
+      "epoch": 0.7990373044524669,
+      "grad_norm": 1.8671875,
+      "learning_rate": 8.822299475110787e-05,
+      "loss": 0.6321,
+      "step": 3320
+    },
+    {
+      "epoch": 0.8002406738868832,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.81878387768946e-05,
+      "loss": 0.6336,
+      "step": 3325
+    },
+    {
+      "epoch": 0.8014440433212996,
+      "grad_norm": 1.7734375,
+      "learning_rate": 8.815264225204201e-05,
+      "loss": 0.6566,
+      "step": 3330
+    },
+    {
+      "epoch": 0.802647412755716,
+      "grad_norm": 1.8828125,
+      "learning_rate": 8.811740523275802e-05,
+      "loss": 0.6386,
+      "step": 3335
+    },
+    {
+      "epoch": 0.8038507821901324,
+      "grad_norm": 1.84375,
+      "learning_rate": 8.808212777531515e-05,
+      "loss": 0.6428,
+      "step": 3340
+    },
+    {
+      "epoch": 0.8050541516245487,
+      "grad_norm": 1.796875,
+      "learning_rate": 8.804680993605056e-05,
+      "loss": 0.648,
+      "step": 3345
+    },
+    {
+      "epoch": 0.8062575210589651,
+      "grad_norm": 1.9921875,
+      "learning_rate": 8.801145177136586e-05,
+      "loss": 0.6724,
+      "step": 3350
+    },
+    {
+      "epoch": 0.8074608904933814,
+      "grad_norm": 1.7890625,
+      "learning_rate": 8.797605333772706e-05,
+      "loss": 0.6142,
+      "step": 3355
+    },
+    {
+      "epoch": 0.8086642599277978,
+      "grad_norm": 1.765625,
+      "learning_rate": 8.794061469166449e-05,
+      "loss": 0.6161,
+      "step": 3360
+    },
+    {
+      "epoch": 0.8098676293622142,
+      "grad_norm": 1.890625,
+      "learning_rate": 8.790513588977268e-05,
+      "loss": 0.6351,
+      "step": 3365
+    },
+    {
+      "epoch": 0.8110709987966306,
+      "grad_norm": 1.796875,
+      "learning_rate": 8.786961698871031e-05,
+      "loss": 0.6748,
+      "step": 3370
+    },
+    {
+      "epoch": 0.8122743682310469,
+      "grad_norm": 1.8515625,
+      "learning_rate": 8.78340580452001e-05,
+      "loss": 0.6414,
+      "step": 3375
+    },
+    {
+      "epoch": 0.8134777376654633,
+      "grad_norm": 1.8046875,
+      "learning_rate": 8.779845911602868e-05,
+      "loss": 0.6533,
+      "step": 3380
+    },
+    {
+      "epoch": 0.8146811070998796,
+      "grad_norm": 1.6171875,
+      "learning_rate": 8.776282025804659e-05,
+      "loss": 0.6725,
+      "step": 3385
+    },
+    {
+      "epoch": 0.8158844765342961,
+      "grad_norm": 1.828125,
+      "learning_rate": 8.77271415281681e-05,
+      "loss": 0.6398,
+      "step": 3390
+    },
+    {
+      "epoch": 0.8170878459687124,
+      "grad_norm": 1.7734375,
+      "learning_rate": 8.769142298337113e-05,
+      "loss": 0.6317,
+      "step": 3395
+    },
+    {
+      "epoch": 0.8182912154031288,
+      "grad_norm": 1.953125,
+      "learning_rate": 8.765566468069726e-05,
+      "loss": 0.6587,
+      "step": 3400
+    },
+    {
+      "epoch": 0.8194945848375451,
+      "grad_norm": 1.6953125,
+      "learning_rate": 8.761986667725147e-05,
+      "loss": 0.6201,
+      "step": 3405
+    },
+    {
+      "epoch": 0.8206979542719615,
+      "grad_norm": 1.7265625,
+      "learning_rate": 8.758402903020221e-05,
+      "loss": 0.5976,
+      "step": 3410
+    },
+    {
+      "epoch": 0.8219013237063778,
+      "grad_norm": 1.796875,
+      "learning_rate": 8.754815179678124e-05,
+      "loss": 0.6232,
+      "step": 3415
+    },
+    {
+      "epoch": 0.8231046931407943,
+      "grad_norm": 1.8671875,
+      "learning_rate": 8.751223503428348e-05,
+      "loss": 0.6512,
+      "step": 3420
+    },
+    {
+      "epoch": 0.8243080625752106,
+      "grad_norm": 1.71875,
+      "learning_rate": 8.747627880006703e-05,
+      "loss": 0.6592,
+      "step": 3425
+    },
+    {
+      "epoch": 0.825511432009627,
+      "grad_norm": 1.875,
+      "learning_rate": 8.7440283151553e-05,
+      "loss": 0.6428,
+      "step": 3430
+    },
+    {
+      "epoch": 0.8267148014440433,
+      "grad_norm": 1.6640625,
+      "learning_rate": 8.740424814622546e-05,
+      "loss": 0.574,
+      "step": 3435
+    },
+    {
+      "epoch": 0.8279181708784596,
+      "grad_norm": 1.7265625,
+      "learning_rate": 8.736817384163131e-05,
+      "loss": 0.614,
+      "step": 3440
+    },
+    {
+      "epoch": 0.8291215403128761,
+      "grad_norm": 1.75,
+      "learning_rate": 8.733206029538023e-05,
+      "loss": 0.6405,
+      "step": 3445
+    },
+    {
+      "epoch": 0.8303249097472925,
+      "grad_norm": 1.859375,
+      "learning_rate": 8.729590756514455e-05,
+      "loss": 0.6592,
+      "step": 3450
+    },
+    {
+      "epoch": 0.8315282791817088,
+      "grad_norm": 1.859375,
+      "learning_rate": 8.725971570865921e-05,
+      "loss": 0.6584,
+      "step": 3455
+    },
+    {
+      "epoch": 0.8327316486161251,
+      "grad_norm": 1.7578125,
+      "learning_rate": 8.722348478372159e-05,
+      "loss": 0.6305,
+      "step": 3460
+    },
+    {
+      "epoch": 0.8339350180505415,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.718721484819147e-05,
+      "loss": 0.6425,
+      "step": 3465
+    },
+    {
+      "epoch": 0.8351383874849578,
+      "grad_norm": 1.6953125,
+      "learning_rate": 8.715090595999096e-05,
+      "loss": 0.6302,
+      "step": 3470
+    },
+    {
+      "epoch": 0.8363417569193743,
+      "grad_norm": 1.765625,
+      "learning_rate": 8.711455817710437e-05,
+      "loss": 0.6188,
+      "step": 3475
+    },
+    {
+      "epoch": 0.8375451263537906,
+      "grad_norm": 1.7265625,
+      "learning_rate": 8.70781715575781e-05,
+      "loss": 0.6559,
+      "step": 3480
+    },
+    {
+      "epoch": 0.838748495788207,
+      "grad_norm": 2.15625,
+      "learning_rate": 8.704174615952055e-05,
+      "loss": 0.6032,
+      "step": 3485
+    },
+    {
+      "epoch": 0.8399518652226233,
+      "grad_norm": 1.7265625,
+      "learning_rate": 8.70052820411021e-05,
+      "loss": 0.6567,
+      "step": 3490
+    },
+    {
+      "epoch": 0.8411552346570397,
+      "grad_norm": 1.6796875,
+      "learning_rate": 8.696877926055497e-05,
+      "loss": 0.5912,
+      "step": 3495
+    },
+    {
+      "epoch": 0.8423586040914561,
+      "grad_norm": 1.8359375,
+      "learning_rate": 8.693223787617304e-05,
+      "loss": 0.6628,
+      "step": 3500
+    },
+    {
+      "epoch": 0.8423586040914561,
+      "eval_loss": 0.5347813367843628,
+      "eval_runtime": 2.6685,
+      "eval_samples_per_second": 74.949,
+      "eval_steps_per_second": 74.949,
+      "step": 3500
+    },
+    {
+      "epoch": 0.8435619735258725,
+      "grad_norm": 1.828125,
+      "learning_rate": 8.689565794631195e-05,
+      "loss": 0.6222,
+      "step": 3505
+    },
+    {
+      "epoch": 0.8447653429602888,
+      "grad_norm": 1.6484375,
+      "learning_rate": 8.685903952938881e-05,
+      "loss": 0.618,
+      "step": 3510
+    },
+    {
+      "epoch": 0.8459687123947052,
+      "grad_norm": 2.5625,
+      "learning_rate": 8.682238268388225e-05,
+      "loss": 0.6677,
+      "step": 3515
+    },
+    {
+      "epoch": 0.8471720818291215,
+      "grad_norm": 1.640625,
+      "learning_rate": 8.678568746833222e-05,
+      "loss": 0.6229,
+      "step": 3520
+    },
+    {
+      "epoch": 0.8483754512635379,
+      "grad_norm": 1.625,
+      "learning_rate": 8.674895394134e-05,
+      "loss": 0.5955,
+      "step": 3525
+    },
+    {
+      "epoch": 0.8495788206979543,
+      "grad_norm": 1.765625,
+      "learning_rate": 8.6712182161568e-05,
+      "loss": 0.6069,
+      "step": 3530
+    },
+    {
+      "epoch": 0.8507821901323707,
+      "grad_norm": 1.6328125,
+      "learning_rate": 8.667537218773973e-05,
+      "loss": 0.6031,
+      "step": 3535
+    },
+    {
+      "epoch": 0.851985559566787,
+      "grad_norm": 1.6953125,
+      "learning_rate": 8.663852407863973e-05,
+      "loss": 0.6408,
+      "step": 3540
+    },
+    {
+      "epoch": 0.8531889290012034,
+      "grad_norm": 1.765625,
+      "learning_rate": 8.66016378931134e-05,
+      "loss": 0.625,
+      "step": 3545
+    },
+    {
+      "epoch": 0.8543922984356197,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.656471369006697e-05,
+      "loss": 0.5847,
+      "step": 3550
+    },
+    {
+      "epoch": 0.855595667870036,
+      "grad_norm": 1.9921875,
+      "learning_rate": 8.65277515284674e-05,
+      "loss": 0.6638,
+      "step": 3555
+    },
+    {
+      "epoch": 0.8567990373044525,
+      "grad_norm": 1.828125,
+      "learning_rate": 8.649075146734222e-05,
+      "loss": 0.6436,
+      "step": 3560
+    },
+    {
+      "epoch": 0.8580024067388689,
+      "grad_norm": 1.9765625,
+      "learning_rate": 8.64537135657795e-05,
+      "loss": 0.6382,
+      "step": 3565
+    },
+    {
+      "epoch": 0.8592057761732852,
+      "grad_norm": 1.7890625,
+      "learning_rate": 8.641663788292779e-05,
+      "loss": 0.6395,
+      "step": 3570
+    },
+    {
+      "epoch": 0.8604091456077015,
+      "grad_norm": 1.8515625,
+      "learning_rate": 8.637952447799589e-05,
+      "loss": 0.6353,
+      "step": 3575
+    },
+    {
+      "epoch": 0.8616125150421179,
+      "grad_norm": 1.75,
+      "learning_rate": 8.634237341025292e-05,
+      "loss": 0.6124,
+      "step": 3580
+    },
+    {
+      "epoch": 0.8628158844765343,
+      "grad_norm": 1.796875,
+      "learning_rate": 8.63051847390281e-05,
+      "loss": 0.6215,
+      "step": 3585
+    },
+    {
+      "epoch": 0.8640192539109507,
+      "grad_norm": 1.9375,
+      "learning_rate": 8.626795852371071e-05,
+      "loss": 0.6341,
+      "step": 3590
+    },
+    {
+      "epoch": 0.865222623345367,
+      "grad_norm": 1.90625,
+      "learning_rate": 8.623069482375002e-05,
+      "loss": 0.6319,
+      "step": 3595
+    },
+    {
+      "epoch": 0.8664259927797834,
+      "grad_norm": 1.796875,
+      "learning_rate": 8.61933936986551e-05,
+      "loss": 0.6488,
+      "step": 3600
+    },
+    {
+      "epoch": 0.8676293622141997,
+      "grad_norm": 1.90625,
+      "learning_rate": 8.615605520799485e-05,
+      "loss": 0.6515,
+      "step": 3605
+    },
+    {
+      "epoch": 0.8688327316486161,
+      "grad_norm": 1.7890625,
+      "learning_rate": 8.611867941139779e-05,
+      "loss": 0.629,
+      "step": 3610
+    },
+    {
+      "epoch": 0.8700361010830325,
+      "grad_norm": 1.765625,
+      "learning_rate": 8.608126636855206e-05,
+      "loss": 0.6395,
+      "step": 3615
+    },
+    {
+      "epoch": 0.8712394705174489,
+      "grad_norm": 1.8515625,
+      "learning_rate": 8.604381613920524e-05,
+      "loss": 0.6356,
+      "step": 3620
+    },
+    {
+      "epoch": 0.8724428399518652,
+      "grad_norm": 1.84375,
+      "learning_rate": 8.600632878316433e-05,
+      "loss": 0.6264,
+      "step": 3625
+    },
+    {
+      "epoch": 0.8736462093862816,
+      "grad_norm": 1.6875,
+      "learning_rate": 8.596880436029559e-05,
+      "loss": 0.5993,
+      "step": 3630
+    },
+    {
+      "epoch": 0.8748495788206979,
+      "grad_norm": 1.7578125,
+      "learning_rate": 8.593124293052452e-05,
+      "loss": 0.6338,
+      "step": 3635
+    },
+    {
+      "epoch": 0.8760529482551144,
+      "grad_norm": 1.9375,
+      "learning_rate": 8.589364455383568e-05,
+      "loss": 0.6467,
+      "step": 3640
+    },
+    {
+      "epoch": 0.8772563176895307,
+      "grad_norm": 1.7578125,
+      "learning_rate": 8.585600929027262e-05,
+      "loss": 0.625,
+      "step": 3645
+    },
+    {
+      "epoch": 0.8784596871239471,
+      "grad_norm": 1.671875,
+      "learning_rate": 8.581833719993785e-05,
+      "loss": 0.6371,
+      "step": 3650
+    },
+    {
+      "epoch": 0.8796630565583634,
+      "grad_norm": 1.6640625,
+      "learning_rate": 8.578062834299264e-05,
+      "loss": 0.6239,
+      "step": 3655
+    },
+    {
+      "epoch": 0.8808664259927798,
+      "grad_norm": 1.9375,
+      "learning_rate": 8.574288277965703e-05,
+      "loss": 0.6019,
+      "step": 3660
+    },
+    {
+      "epoch": 0.8820697954271961,
+      "grad_norm": 1.9296875,
+      "learning_rate": 8.570510057020962e-05,
+      "loss": 0.6434,
+      "step": 3665
+    },
+    {
+      "epoch": 0.8832731648616126,
+      "grad_norm": 1.6875,
+      "learning_rate": 8.566728177498757e-05,
+      "loss": 0.623,
+      "step": 3670
+    },
+    {
+      "epoch": 0.8844765342960289,
+      "grad_norm": 1.96875,
+      "learning_rate": 8.562942645438646e-05,
+      "loss": 0.6125,
+      "step": 3675
+    },
+    {
+      "epoch": 0.8856799037304453,
+      "grad_norm": 1.6875,
+      "learning_rate": 8.559153466886017e-05,
+      "loss": 0.5848,
+      "step": 3680
+    },
+    {
+      "epoch": 0.8868832731648616,
+      "grad_norm": 1.703125,
+      "learning_rate": 8.555360647892087e-05,
+      "loss": 0.6241,
+      "step": 3685
+    },
+    {
+      "epoch": 0.8880866425992779,
+      "grad_norm": 1.5703125,
+      "learning_rate": 8.551564194513882e-05,
+      "loss": 0.6044,
+      "step": 3690
+    },
+    {
+      "epoch": 0.8892900120336944,
+      "grad_norm": 1.6328125,
+      "learning_rate": 8.547764112814234e-05,
+      "loss": 0.6086,
+      "step": 3695
+    },
+    {
+      "epoch": 0.8904933814681107,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.54396040886177e-05,
+      "loss": 0.5887,
+      "step": 3700
+    },
+    {
+      "epoch": 0.8916967509025271,
+      "grad_norm": 1.8203125,
+      "learning_rate": 8.5401530887309e-05,
+      "loss": 0.6021,
+      "step": 3705
+    },
+    {
+      "epoch": 0.8929001203369434,
+      "grad_norm": 1.8671875,
+      "learning_rate": 8.536342158501808e-05,
+      "loss": 0.6041,
+      "step": 3710
+    },
+    {
+      "epoch": 0.8941034897713598,
+      "grad_norm": 1.9296875,
+      "learning_rate": 8.532527624260448e-05,
+      "loss": 0.6172,
+      "step": 3715
+    },
+    {
+      "epoch": 0.8953068592057761,
+      "grad_norm": 1.8125,
+      "learning_rate": 8.528709492098524e-05,
+      "loss": 0.6071,
+      "step": 3720
+    },
+    {
+      "epoch": 0.8965102286401926,
+      "grad_norm": 1.6875,
+      "learning_rate": 8.524887768113489e-05,
+      "loss": 0.6526,
+      "step": 3725
+    },
+    {
+      "epoch": 0.8977135980746089,
+      "grad_norm": 1.9453125,
+      "learning_rate": 8.521062458408529e-05,
+      "loss": 0.6289,
+      "step": 3730
+    },
+    {
+      "epoch": 0.8989169675090253,
+      "grad_norm": 1.6484375,
+      "learning_rate": 8.517233569092561e-05,
+      "loss": 0.6516,
+      "step": 3735
+    },
+    {
+      "epoch": 0.9001203369434416,
+      "grad_norm": 1.96875,
+      "learning_rate": 8.513401106280215e-05,
+      "loss": 0.6479,
+      "step": 3740
+    },
+    {
+      "epoch": 0.901323706377858,
+      "grad_norm": 1.6484375,
+      "learning_rate": 8.50956507609183e-05,
+      "loss": 0.6149,
+      "step": 3745
+    },
+    {
+      "epoch": 0.9025270758122743,
+      "grad_norm": 1.671875,
+      "learning_rate": 8.505725484653437e-05,
+      "loss": 0.6184,
+      "step": 3750
+    },
+    {
+      "epoch": 0.9037304452466908,
+      "grad_norm": 1.7734375,
+      "learning_rate": 8.501882338096763e-05,
+      "loss": 0.627,
+      "step": 3755
+    },
+    {
+      "epoch": 0.9049338146811071,
+      "grad_norm": 1.8359375,
+      "learning_rate": 8.498035642559203e-05,
+      "loss": 0.6689,
+      "step": 3760
+    },
+    {
+      "epoch": 0.9061371841155235,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.494185404183828e-05,
+      "loss": 0.633,
+      "step": 3765
+    },
+    {
+      "epoch": 0.9073405535499398,
+      "grad_norm": 1.765625,
+      "learning_rate": 8.49033162911936e-05,
+      "loss": 0.5911,
+      "step": 3770
+    },
+    {
+      "epoch": 0.9085439229843562,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.486474323520172e-05,
+      "loss": 0.6439,
+      "step": 3775
+    },
+    {
+      "epoch": 0.9097472924187726,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.482613493546275e-05,
+      "loss": 0.6373,
+      "step": 3780
+    },
+    {
+      "epoch": 0.910950661853189,
+      "grad_norm": 1.6875,
+      "learning_rate": 8.47874914536331e-05,
+      "loss": 0.6325,
+      "step": 3785
+    },
+    {
+      "epoch": 0.9121540312876053,
+      "grad_norm": 1.71875,
+      "learning_rate": 8.474881285142532e-05,
+      "loss": 0.6449,
+      "step": 3790
+    },
+    {
+      "epoch": 0.9133574007220217,
+      "grad_norm": 1.6328125,
+      "learning_rate": 8.471009919060811e-05,
+      "loss": 0.6369,
+      "step": 3795
+    },
+    {
+      "epoch": 0.914560770156438,
+      "grad_norm": 1.734375,
+      "learning_rate": 8.46713505330061e-05,
+      "loss": 0.633,
+      "step": 3800
+    },
+    {
+      "epoch": 0.9157641395908543,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.463256694049982e-05,
+      "loss": 0.6176,
+      "step": 3805
+    },
+    {
+      "epoch": 0.9169675090252708,
+      "grad_norm": 1.8828125,
+      "learning_rate": 8.459374847502562e-05,
+      "loss": 0.6279,
+      "step": 3810
+    },
+    {
+      "epoch": 0.9181708784596871,
+      "grad_norm": 2.140625,
+      "learning_rate": 8.455489519857552e-05,
+      "loss": 0.613,
+      "step": 3815
+    },
+    {
+      "epoch": 0.9193742478941035,
+      "grad_norm": 1.671875,
+      "learning_rate": 8.451600717319714e-05,
+      "loss": 0.6297,
+      "step": 3820
+    },
+    {
+      "epoch": 0.9205776173285198,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.447708446099358e-05,
+      "loss": 0.6043,
+      "step": 3825
+    },
+    {
+      "epoch": 0.9217809867629362,
+      "grad_norm": 1.734375,
+      "learning_rate": 8.443812712412331e-05,
+      "loss": 0.6347,
+      "step": 3830
+    },
+    {
+      "epoch": 0.9229843561973526,
+      "grad_norm": 1.765625,
+      "learning_rate": 8.439913522480016e-05,
+      "loss": 0.6038,
+      "step": 3835
+    },
+    {
+      "epoch": 0.924187725631769,
+      "grad_norm": 1.9140625,
+      "learning_rate": 8.436010882529314e-05,
+      "loss": 0.6332,
+      "step": 3840
+    },
+    {
+      "epoch": 0.9253910950661853,
+      "grad_norm": 1.7265625,
+      "learning_rate": 8.432104798792632e-05,
+      "loss": 0.617,
+      "step": 3845
+    },
+    {
+      "epoch": 0.9265944645006017,
+      "grad_norm": 1.875,
+      "learning_rate": 8.428195277507874e-05,
+      "loss": 0.6533,
+      "step": 3850
+    },
+    {
+      "epoch": 0.927797833935018,
+      "grad_norm": 1.828125,
+      "learning_rate": 8.424282324918444e-05,
+      "loss": 0.6323,
+      "step": 3855
+    },
+    {
+      "epoch": 0.9290012033694344,
+      "grad_norm": 1.640625,
+      "learning_rate": 8.420365947273217e-05,
+      "loss": 0.6411,
+      "step": 3860
+    },
+    {
+      "epoch": 0.9302045728038508,
+      "grad_norm": 1.7578125,
+      "learning_rate": 8.41644615082654e-05,
+      "loss": 0.6448,
+      "step": 3865
+    },
+    {
+      "epoch": 0.9314079422382672,
+      "grad_norm": 1.734375,
+      "learning_rate": 8.412522941838221e-05,
+      "loss": 0.6081,
+      "step": 3870
+    },
+    {
+      "epoch": 0.9326113116726835,
+      "grad_norm": 1.7890625,
+      "learning_rate": 8.408596326573518e-05,
+      "loss": 0.5949,
+      "step": 3875
+    },
+    {
+      "epoch": 0.9338146811070999,
+      "grad_norm": 1.90625,
+      "learning_rate": 8.404666311303126e-05,
+      "loss": 0.6219,
+      "step": 3880
+    },
+    {
+      "epoch": 0.9350180505415162,
+      "grad_norm": 1.734375,
+      "learning_rate": 8.400732902303172e-05,
+      "loss": 0.6101,
+      "step": 3885
+    },
+    {
+      "epoch": 0.9362214199759326,
+      "grad_norm": 1.546875,
+      "learning_rate": 8.3967961058552e-05,
+      "loss": 0.6045,
+      "step": 3890
+    },
+    {
+      "epoch": 0.937424789410349,
+      "grad_norm": 1.6328125,
+      "learning_rate": 8.392855928246169e-05,
+      "loss": 0.6527,
+      "step": 3895
+    },
+    {
+      "epoch": 0.9386281588447654,
+      "grad_norm": 1.625,
+      "learning_rate": 8.388912375768433e-05,
+      "loss": 0.5743,
+      "step": 3900
+    },
+    {
+      "epoch": 0.9398315282791817,
+      "grad_norm": 1.640625,
+      "learning_rate": 8.384965454719734e-05,
+      "loss": 0.6105,
+      "step": 3905
+    },
+    {
+      "epoch": 0.941034897713598,
+      "grad_norm": 1.84375,
+      "learning_rate": 8.381015171403202e-05,
+      "loss": 0.6586,
+      "step": 3910
+    },
+    {
+      "epoch": 0.9422382671480144,
+      "grad_norm": 1.7734375,
+      "learning_rate": 8.377061532127327e-05,
+      "loss": 0.608,
+      "step": 3915
+    },
+    {
+      "epoch": 0.9434416365824309,
+      "grad_norm": 1.5390625,
+      "learning_rate": 8.373104543205963e-05,
+      "loss": 0.5946,
+      "step": 3920
+    },
+    {
+      "epoch": 0.9446450060168472,
+      "grad_norm": 1.609375,
+      "learning_rate": 8.369144210958312e-05,
+      "loss": 0.5954,
+      "step": 3925
+    },
+    {
+      "epoch": 0.9458483754512635,
+      "grad_norm": 1.875,
+      "learning_rate": 8.365180541708917e-05,
+      "loss": 0.631,
+      "step": 3930
+    },
+    {
+      "epoch": 0.9470517448856799,
+      "grad_norm": 1.7578125,
+      "learning_rate": 8.361213541787648e-05,
+      "loss": 0.596,
+      "step": 3935
+    },
+    {
+      "epoch": 0.9482551143200962,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.357243217529694e-05,
+      "loss": 0.6246,
+      "step": 3940
+    },
+    {
+      "epoch": 0.9494584837545126,
+      "grad_norm": 1.71875,
+      "learning_rate": 8.353269575275555e-05,
+      "loss": 0.6135,
+      "step": 3945
+    },
+    {
+      "epoch": 0.950661853188929,
+      "grad_norm": 1.8828125,
+      "learning_rate": 8.34929262137103e-05,
+      "loss": 0.6014,
+      "step": 3950
+    },
+    {
+      "epoch": 0.9518652226233454,
+      "grad_norm": 1.8203125,
+      "learning_rate": 8.3453123621672e-05,
+      "loss": 0.6129,
+      "step": 3955
+    },
+    {
+      "epoch": 0.9530685920577617,
+      "grad_norm": 1.7578125,
+      "learning_rate": 8.341328804020435e-05,
+      "loss": 0.6022,
+      "step": 3960
+    },
+    {
+      "epoch": 0.9542719614921781,
+      "grad_norm": 1.796875,
+      "learning_rate": 8.337341953292365e-05,
+      "loss": 0.5934,
+      "step": 3965
+    },
+    {
+      "epoch": 0.9554753309265944,
+      "grad_norm": 1.6328125,
+      "learning_rate": 8.33335181634988e-05,
+      "loss": 0.6331,
+      "step": 3970
+    },
+    {
+      "epoch": 0.9566787003610109,
+      "grad_norm": 1.796875,
+      "learning_rate": 8.329358399565123e-05,
+      "loss": 0.6013,
+      "step": 3975
+    },
+    {
+      "epoch": 0.9578820697954272,
+      "grad_norm": 1.625,
+      "learning_rate": 8.325361709315466e-05,
+      "loss": 0.5836,
+      "step": 3980
+    },
+    {
+      "epoch": 0.9590854392298436,
+      "grad_norm": 1.640625,
+      "learning_rate": 8.321361751983516e-05,
+      "loss": 0.6138,
+      "step": 3985
+    },
+    {
+      "epoch": 0.9602888086642599,
+      "grad_norm": 1.875,
+      "learning_rate": 8.317358533957095e-05,
+      "loss": 0.6239,
+      "step": 3990
+    },
+    {
+      "epoch": 0.9614921780986763,
+      "grad_norm": 1.6484375,
+      "learning_rate": 8.313352061629231e-05,
+      "loss": 0.6073,
+      "step": 3995
+    },
+    {
+      "epoch": 0.9626955475330926,
+      "grad_norm": 1.734375,
+      "learning_rate": 8.309342341398151e-05,
+      "loss": 0.6274,
+      "step": 4000
+    },
+    {
+      "epoch": 0.9626955475330926,
+      "eval_loss": 0.5121676325798035,
+      "eval_runtime": 2.6684,
+      "eval_samples_per_second": 74.95,
+      "eval_steps_per_second": 74.95,
+      "step": 4000
+    },
+    {
+      "epoch": 0.9638989169675091,
+      "grad_norm": 1.6640625,
+      "learning_rate": 8.305329379667268e-05,
+      "loss": 0.6214,
+      "step": 4005
+    },
+    {
+      "epoch": 0.9651022864019254,
+      "grad_norm": 1.75,
+      "learning_rate": 8.30131318284517e-05,
+      "loss": 0.6042,
+      "step": 4010
+    },
+    {
+      "epoch": 0.9663056558363418,
+      "grad_norm": 1.6015625,
+      "learning_rate": 8.297293757345617e-05,
+      "loss": 0.5803,
+      "step": 4015
+    },
+    {
+      "epoch": 0.9675090252707581,
+      "grad_norm": 1.796875,
+      "learning_rate": 8.293271109587516e-05,
+      "loss": 0.6131,
+      "step": 4020
+    },
+    {
+      "epoch": 0.9687123947051745,
+      "grad_norm": 1.8046875,
+      "learning_rate": 8.289245245994927e-05,
+      "loss": 0.6206,
+      "step": 4025
+    },
+    {
+      "epoch": 0.9699157641395909,
+      "grad_norm": 1.8203125,
+      "learning_rate": 8.285216172997044e-05,
+      "loss": 0.6494,
+      "step": 4030
+    },
+    {
+      "epoch": 0.9711191335740073,
+      "grad_norm": 1.9140625,
+      "learning_rate": 8.281183897028185e-05,
+      "loss": 0.6191,
+      "step": 4035
+    },
+    {
+      "epoch": 0.9723225030084236,
+      "grad_norm": 1.6796875,
+      "learning_rate": 8.277148424527785e-05,
+      "loss": 0.6054,
+      "step": 4040
+    },
+    {
+      "epoch": 0.97352587244284,
+      "grad_norm": 1.6640625,
+      "learning_rate": 8.273109761940382e-05,
+      "loss": 0.5866,
+      "step": 4045
+    },
+    {
+      "epoch": 0.9747292418772563,
+      "grad_norm": 1.734375,
+      "learning_rate": 8.269067915715609e-05,
+      "loss": 0.6342,
+      "step": 4050
+    },
+    {
+      "epoch": 0.9759326113116726,
+      "grad_norm": 1.7265625,
+      "learning_rate": 8.265022892308182e-05,
+      "loss": 0.6076,
+      "step": 4055
+    },
+    {
+      "epoch": 0.9771359807460891,
+      "grad_norm": 1.8515625,
+      "learning_rate": 8.260974698177894e-05,
+      "loss": 0.6061,
+      "step": 4060
+    },
+    {
+      "epoch": 0.9783393501805054,
+      "grad_norm": 1.71875,
+      "learning_rate": 8.2569233397896e-05,
+      "loss": 0.6105,
+      "step": 4065
+    },
+    {
+      "epoch": 0.9795427196149218,
+      "grad_norm": 1.8359375,
+      "learning_rate": 8.252868823613206e-05,
+      "loss": 0.6044,
+      "step": 4070
+    },
+    {
+      "epoch": 0.9807460890493381,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.248811156123662e-05,
+      "loss": 0.6238,
+      "step": 4075
+    },
+    {
+      "epoch": 0.9819494584837545,
+      "grad_norm": 1.6796875,
+      "learning_rate": 8.244750343800957e-05,
+      "loss": 0.5803,
+      "step": 4080
+    },
+    {
+      "epoch": 0.9831528279181708,
+      "grad_norm": 1.7265625,
+      "learning_rate": 8.240686393130091e-05,
+      "loss": 0.6073,
+      "step": 4085
+    },
+    {
+      "epoch": 0.9843561973525873,
+      "grad_norm": 1.75,
+      "learning_rate": 8.236619310601085e-05,
+      "loss": 0.6112,
+      "step": 4090
+    },
+    {
+      "epoch": 0.9855595667870036,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.232549102708955e-05,
+      "loss": 0.6062,
+      "step": 4095
+    },
+    {
+      "epoch": 0.98676293622142,
+      "grad_norm": 1.71875,
+      "learning_rate": 8.228475775953712e-05,
+      "loss": 0.6052,
+      "step": 4100
+    },
+    {
+      "epoch": 0.9879663056558363,
+      "grad_norm": 1.625,
+      "learning_rate": 8.224399336840348e-05,
+      "loss": 0.6065,
+      "step": 4105
+    },
+    {
+      "epoch": 0.9891696750902527,
+      "grad_norm": 1.8359375,
+      "learning_rate": 8.220319791878824e-05,
+      "loss": 0.6078,
+      "step": 4110
+    },
+    {
+      "epoch": 0.9903730445246691,
+      "grad_norm": 1.7265625,
+      "learning_rate": 8.21623714758406e-05,
+      "loss": 0.6207,
+      "step": 4115
+    },
+    {
+      "epoch": 0.9915764139590855,
+      "grad_norm": 1.8515625,
+      "learning_rate": 8.212151410475929e-05,
+      "loss": 0.6269,
+      "step": 4120
+    },
+    {
+      "epoch": 0.9927797833935018,
+      "grad_norm": 1.6640625,
+      "learning_rate": 8.208062587079237e-05,
+      "loss": 0.5992,
+      "step": 4125
+    },
+    {
+      "epoch": 0.9939831528279182,
+      "grad_norm": 1.7109375,
+      "learning_rate": 8.203970683923724e-05,
+      "loss": 0.6045,
+      "step": 4130
+    },
+    {
+      "epoch": 0.9951865222623345,
+      "grad_norm": 1.6328125,
+      "learning_rate": 8.199875707544049e-05,
+      "loss": 0.6244,
+      "step": 4135
+    },
+    {
+      "epoch": 0.9963898916967509,
+      "grad_norm": 1.578125,
+      "learning_rate": 8.195777664479775e-05,
+      "loss": 0.6142,
+      "step": 4140
+    },
+    {
+      "epoch": 0.9975932611311673,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.191676561275365e-05,
+      "loss": 0.6293,
+      "step": 4145
+    },
+    {
+      "epoch": 0.9987966305655837,
+      "grad_norm": 1.7109375,
+      "learning_rate": 8.187572404480168e-05,
+      "loss": 0.6087,
+      "step": 4150
+    },
+    {
+      "epoch": 0.9997593261131167,
+      "eval_loss": 0.5122986435890198,
+      "eval_runtime": 2.6521,
+      "eval_samples_per_second": 75.413,
+      "eval_steps_per_second": 75.413,
+      "step": 4154
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.8671875,
+      "learning_rate": 8.183465200648407e-05,
+      "loss": 0.5928,
+      "step": 4155
+    },
+    {
+      "epoch": 1.0012033694344165,
+      "grad_norm": 1.6953125,
+      "learning_rate": 8.179354956339176e-05,
+      "loss": 0.5627,
+      "step": 4160
+    },
+    {
+      "epoch": 1.0024067388688327,
+      "grad_norm": 1.71875,
+      "learning_rate": 8.175241678116423e-05,
+      "loss": 0.5571,
+      "step": 4165
+    },
+    {
+      "epoch": 1.0036101083032491,
+      "grad_norm": 1.8359375,
+      "learning_rate": 8.171125372548937e-05,
+      "loss": 0.6396,
+      "step": 4170
+    },
+    {
+      "epoch": 1.0048134777376654,
+      "grad_norm": 1.8125,
+      "learning_rate": 8.167006046210347e-05,
+      "loss": 0.6025,
+      "step": 4175
+    },
+    {
+      "epoch": 1.0060168471720818,
+      "grad_norm": 1.6640625,
+      "learning_rate": 8.162883705679107e-05,
+      "loss": 0.6016,
+      "step": 4180
+    },
+    {
+      "epoch": 1.0072202166064983,
+      "grad_norm": 1.7578125,
+      "learning_rate": 8.158758357538474e-05,
+      "loss": 0.5501,
+      "step": 4185
+    },
+    {
+      "epoch": 1.0084235860409145,
+      "grad_norm": 1.7890625,
+      "learning_rate": 8.154630008376522e-05,
+      "loss": 0.548,
+      "step": 4190
+    },
+    {
+      "epoch": 1.009626955475331,
+      "grad_norm": 1.6328125,
+      "learning_rate": 8.150498664786107e-05,
+      "loss": 0.5533,
+      "step": 4195
+    },
+    {
+      "epoch": 1.0108303249097472,
+      "grad_norm": 1.671875,
+      "learning_rate": 8.146364333364874e-05,
+      "loss": 0.5879,
+      "step": 4200
+    },
+    {
+      "epoch": 1.0120336943441637,
+      "grad_norm": 1.671875,
+      "learning_rate": 8.142227020715236e-05,
+      "loss": 0.5898,
+      "step": 4205
+    },
+    {
+      "epoch": 1.01323706377858,
+      "grad_norm": 1.7578125,
+      "learning_rate": 8.138086733444366e-05,
+      "loss": 0.5648,
+      "step": 4210
+    },
+    {
+      "epoch": 1.0144404332129964,
+      "grad_norm": 1.8359375,
+      "learning_rate": 8.13394347816419e-05,
+      "loss": 0.5852,
+      "step": 4215
+    },
+    {
+      "epoch": 1.0156438026474128,
+      "grad_norm": 1.75,
+      "learning_rate": 8.129797261491373e-05,
+      "loss": 0.5723,
+      "step": 4220
+    },
+    {
+      "epoch": 1.016847172081829,
+      "grad_norm": 1.921875,
+      "learning_rate": 8.125648090047308e-05,
+      "loss": 0.5741,
+      "step": 4225
+    },
+    {
+      "epoch": 1.0180505415162455,
+      "grad_norm": 1.765625,
+      "learning_rate": 8.12149597045811e-05,
+      "loss": 0.578,
+      "step": 4230
+    },
+    {
+      "epoch": 1.0192539109506618,
+      "grad_norm": 1.8203125,
+      "learning_rate": 8.117340909354598e-05,
+      "loss": 0.5999,
+      "step": 4235
+    },
+    {
+      "epoch": 1.0204572803850782,
+      "grad_norm": 1.8984375,
+      "learning_rate": 8.11318291337229e-05,
+      "loss": 0.6071,
+      "step": 4240
+    },
+    {
+      "epoch": 1.0216606498194947,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.10902198915139e-05,
+      "loss": 0.6093,
+      "step": 4245
+    },
+    {
+      "epoch": 1.022864019253911,
+      "grad_norm": 1.796875,
+      "learning_rate": 8.104858143336784e-05,
+      "loss": 0.5742,
+      "step": 4250
+    },
+    {
+      "epoch": 1.0240673886883274,
+      "grad_norm": 1.75,
+      "learning_rate": 8.100691382578017e-05,
+      "loss": 0.5686,
+      "step": 4255
+    },
+    {
+      "epoch": 1.0252707581227436,
+      "grad_norm": 1.8046875,
+      "learning_rate": 8.096521713529291e-05,
+      "loss": 0.5793,
+      "step": 4260
+    },
+    {
+      "epoch": 1.02647412755716,
+      "grad_norm": 1.640625,
+      "learning_rate": 8.092349142849451e-05,
+      "loss": 0.5686,
+      "step": 4265
+    },
+    {
+      "epoch": 1.0276774969915765,
+      "grad_norm": 1.859375,
+      "learning_rate": 8.088173677201983e-05,
+      "loss": 0.5715,
+      "step": 4270
+    },
+    {
+      "epoch": 1.0288808664259927,
+      "grad_norm": 1.640625,
+      "learning_rate": 8.083995323254985e-05,
+      "loss": 0.5892,
+      "step": 4275
+    },
+    {
+      "epoch": 1.0300842358604092,
+      "grad_norm": 1.6171875,
+      "learning_rate": 8.079814087681177e-05,
+      "loss": 0.5761,
+      "step": 4280
+    },
+    {
+      "epoch": 1.0312876052948254,
+      "grad_norm": 1.59375,
+      "learning_rate": 8.075629977157875e-05,
+      "loss": 0.5844,
+      "step": 4285
+    },
+    {
+      "epoch": 1.032490974729242,
+      "grad_norm": 1.8515625,
+      "learning_rate": 8.071442998366989e-05,
+      "loss": 0.5738,
+      "step": 4290
+    },
+    {
+      "epoch": 1.0336943441636581,
+      "grad_norm": 1.6484375,
+      "learning_rate": 8.06725315799501e-05,
+      "loss": 0.5608,
+      "step": 4295
+    },
+    {
+      "epoch": 1.0348977135980746,
+      "grad_norm": 1.5625,
+      "learning_rate": 8.063060462732998e-05,
+      "loss": 0.5486,
+      "step": 4300
+    },
+    {
+      "epoch": 1.036101083032491,
+      "grad_norm": 1.6484375,
+      "learning_rate": 8.058864919276573e-05,
+      "loss": 0.5528,
+      "step": 4305
+    },
+    {
+      "epoch": 1.0373044524669073,
+      "grad_norm": 1.828125,
+      "learning_rate": 8.054666534325897e-05,
+      "loss": 0.5759,
+      "step": 4310
+    },
+    {
+      "epoch": 1.0385078219013237,
+      "grad_norm": 1.8828125,
+      "learning_rate": 8.050465314585683e-05,
+      "loss": 0.5518,
+      "step": 4315
+    },
+    {
+      "epoch": 1.03971119133574,
+      "grad_norm": 1.9453125,
+      "learning_rate": 8.046261266765159e-05,
+      "loss": 0.5691,
+      "step": 4320
+    },
+    {
+      "epoch": 1.0409145607701564,
+      "grad_norm": 1.703125,
+      "learning_rate": 8.042054397578074e-05,
+      "loss": 0.5697,
+      "step": 4325
+    },
+    {
+      "epoch": 1.0421179302045729,
+      "grad_norm": 1.7890625,
+      "learning_rate": 8.037844713742681e-05,
+      "loss": 0.5935,
+      "step": 4330
+    },
+    {
+      "epoch": 1.0433212996389891,
+      "grad_norm": 1.6171875,
+      "learning_rate": 8.033632221981734e-05,
+      "loss": 0.55,
+      "step": 4335
+    },
+    {
+      "epoch": 1.0445246690734056,
+      "grad_norm": 1.640625,
+      "learning_rate": 8.029416929022463e-05,
+      "loss": 0.5731,
+      "step": 4340
+    },
+    {
+      "epoch": 1.0457280385078218,
+      "grad_norm": 1.8515625,
+      "learning_rate": 8.025198841596576e-05,
+      "loss": 0.5923,
+      "step": 4345
+    },
+    {
+      "epoch": 1.0469314079422383,
+      "grad_norm": 1.8046875,
+      "learning_rate": 8.020977966440242e-05,
+      "loss": 0.5673,
+      "step": 4350
+    },
+    {
+      "epoch": 1.0481347773766547,
+      "grad_norm": 1.7265625,
+      "learning_rate": 8.016754310294083e-05,
+      "loss": 0.564,
+      "step": 4355
+    },
+    {
+      "epoch": 1.049338146811071,
+      "grad_norm": 1.8515625,
+      "learning_rate": 8.012527879903161e-05,
+      "loss": 0.5489,
+      "step": 4360
+    },
+    {
+      "epoch": 1.0505415162454874,
+      "grad_norm": 1.671875,
+      "learning_rate": 8.008298682016972e-05,
+      "loss": 0.5378,
+      "step": 4365
+    },
+    {
+      "epoch": 1.0517448856799037,
+      "grad_norm": 1.7109375,
+      "learning_rate": 8.004066723389425e-05,
+      "loss": 0.5723,
+      "step": 4370
+    },
+    {
+      "epoch": 1.05294825511432,
+      "grad_norm": 1.7421875,
+      "learning_rate": 7.999832010778844e-05,
+      "loss": 0.5554,
+      "step": 4375
+    },
+    {
+      "epoch": 1.0541516245487366,
+      "grad_norm": 1.6484375,
+      "learning_rate": 7.995594550947946e-05,
+      "loss": 0.5525,
+      "step": 4380
+    },
+    {
+      "epoch": 1.0553549939831528,
+      "grad_norm": 1.7421875,
+      "learning_rate": 7.991354350663841e-05,
+      "loss": 0.5611,
+      "step": 4385
+    },
+    {
+      "epoch": 1.0565583634175693,
+      "grad_norm": 1.859375,
+      "learning_rate": 7.98711141669801e-05,
+      "loss": 0.5853,
+      "step": 4390
+    },
+    {
+      "epoch": 1.0577617328519855,
+      "grad_norm": 1.9375,
+      "learning_rate": 7.982865755826304e-05,
+      "loss": 0.5756,
+      "step": 4395
+    },
+    {
+      "epoch": 1.058965102286402,
+      "grad_norm": 1.75,
+      "learning_rate": 7.978617374828925e-05,
+      "loss": 0.6022,
+      "step": 4400
+    },
+    {
+      "epoch": 1.0601684717208182,
+      "grad_norm": 1.6796875,
+      "learning_rate": 7.974366280490421e-05,
+      "loss": 0.5918,
+      "step": 4405
+    },
+    {
+      "epoch": 1.0613718411552346,
+      "grad_norm": 1.8203125,
+      "learning_rate": 7.970112479599673e-05,
+      "loss": 0.5539,
+      "step": 4410
+    },
+    {
+      "epoch": 1.062575210589651,
+      "grad_norm": 1.5859375,
+      "learning_rate": 7.965855978949883e-05,
+      "loss": 0.5474,
+      "step": 4415
+    },
+    {
+      "epoch": 1.0637785800240673,
+      "grad_norm": 1.703125,
+      "learning_rate": 7.961596785338567e-05,
+      "loss": 0.5954,
+      "step": 4420
+    },
+    {
+      "epoch": 1.0649819494584838,
+      "grad_norm": 1.7265625,
+      "learning_rate": 7.95733490556754e-05,
+      "loss": 0.5742,
+      "step": 4425
+    },
+    {
+      "epoch": 1.0661853188929,
+      "grad_norm": 1.6953125,
+      "learning_rate": 7.953070346442906e-05,
+      "loss": 0.5809,
+      "step": 4430
+    },
+    {
+      "epoch": 1.0673886883273165,
+      "grad_norm": 1.78125,
+      "learning_rate": 7.948803114775048e-05,
+      "loss": 0.5991,
+      "step": 4435
+    },
+    {
+      "epoch": 1.068592057761733,
+      "grad_norm": 1.8515625,
+      "learning_rate": 7.944533217378621e-05,
+      "loss": 0.5433,
+      "step": 4440
+    },
+    {
+      "epoch": 1.0697954271961492,
+      "grad_norm": 1.765625,
+      "learning_rate": 7.940260661072528e-05,
+      "loss": 0.5351,
+      "step": 4445
+    },
+    {
+      "epoch": 1.0709987966305656,
+      "grad_norm": 1.7734375,
+      "learning_rate": 7.935985452679931e-05,
+      "loss": 0.5339,
+      "step": 4450
+    },
+    {
+      "epoch": 1.0722021660649819,
+      "grad_norm": 1.78125,
+      "learning_rate": 7.931707599028214e-05,
+      "loss": 0.5568,
+      "step": 4455
+    },
+    {
+      "epoch": 1.0734055354993983,
+      "grad_norm": 1.5625,
+      "learning_rate": 7.927427106948996e-05,
+      "loss": 0.5831,
+      "step": 4460
+    },
+    {
+      "epoch": 1.0746089049338148,
+      "grad_norm": 1.7578125,
+      "learning_rate": 7.923143983278104e-05,
+      "loss": 0.5555,
+      "step": 4465
+    },
+    {
+      "epoch": 1.075812274368231,
+      "grad_norm": 1.703125,
+      "learning_rate": 7.918858234855566e-05,
+      "loss": 0.5828,
+      "step": 4470
+    },
+    {
+      "epoch": 1.0770156438026475,
+      "grad_norm": 1.6875,
+      "learning_rate": 7.914569868525606e-05,
+      "loss": 0.5506,
+      "step": 4475
+    },
+    {
+      "epoch": 1.0782190132370637,
+      "grad_norm": 1.828125,
+      "learning_rate": 7.910278891136629e-05,
+      "loss": 0.5747,
+      "step": 4480
+    },
+    {
+      "epoch": 1.0794223826714802,
+      "grad_norm": 1.7734375,
+      "learning_rate": 7.905985309541206e-05,
+      "loss": 0.5624,
+      "step": 4485
+    },
+    {
+      "epoch": 1.0806257521058966,
+      "grad_norm": 1.6328125,
+      "learning_rate": 7.90168913059607e-05,
+      "loss": 0.5785,
+      "step": 4490
+    },
+    {
+      "epoch": 1.0818291215403129,
+      "grad_norm": 1.7109375,
+      "learning_rate": 7.897390361162101e-05,
+      "loss": 0.5106,
+      "step": 4495
+    },
+    {
+      "epoch": 1.0830324909747293,
+      "grad_norm": 1.921875,
+      "learning_rate": 7.893089008104314e-05,
+      "loss": 0.5917,
+      "step": 4500
+    },
+    {
+      "epoch": 1.0830324909747293,
+      "eval_loss": 0.5059241652488708,
+      "eval_runtime": 2.6442,
+      "eval_samples_per_second": 75.636,
+      "eval_steps_per_second": 75.636,
+      "step": 4500
+    },
+    {
+      "epoch": 1.0842358604091455,
+      "grad_norm": 1.6328125,
+      "learning_rate": 7.88878507829185e-05,
+      "loss": 0.5805,
+      "step": 4505
+    },
+    {
+      "epoch": 1.085439229843562,
+      "grad_norm": 1.6640625,
+      "learning_rate": 7.884478578597971e-05,
+      "loss": 0.567,
+      "step": 4510
+    },
+    {
+      "epoch": 1.0866425992779782,
+      "grad_norm": 1.828125,
+      "learning_rate": 7.880169515900036e-05,
+      "loss": 0.5463,
+      "step": 4515
+    },
+    {
+      "epoch": 1.0878459687123947,
+      "grad_norm": 1.671875,
+      "learning_rate": 7.875857897079499e-05,
+      "loss": 0.5636,
+      "step": 4520
+    },
+    {
+      "epoch": 1.0890493381468112,
+      "grad_norm": 1.75,
+      "learning_rate": 7.871543729021899e-05,
+      "loss": 0.5626,
+      "step": 4525
+    },
+    {
+      "epoch": 1.0902527075812274,
+      "grad_norm": 1.7890625,
+      "learning_rate": 7.867227018616841e-05,
+      "loss": 0.5819,
+      "step": 4530
+    },
+    {
+      "epoch": 1.0914560770156438,
+      "grad_norm": 1.6640625,
+      "learning_rate": 7.862907772757996e-05,
+      "loss": 0.5692,
+      "step": 4535
+    },
+    {
+      "epoch": 1.09265944645006,
+      "grad_norm": 1.6875,
+      "learning_rate": 7.858585998343077e-05,
+      "loss": 0.5725,
+      "step": 4540
+    },
+    {
+      "epoch": 1.0938628158844765,
+      "grad_norm": 1.640625,
+      "learning_rate": 7.854261702273843e-05,
+      "loss": 0.5476,
+      "step": 4545
+    },
+    {
+      "epoch": 1.095066185318893,
+      "grad_norm": 1.796875,
+      "learning_rate": 7.849934891456073e-05,
+      "loss": 0.554,
+      "step": 4550
+    },
+    {
+      "epoch": 1.0962695547533092,
+      "grad_norm": 1.6796875,
+      "learning_rate": 7.845605572799565e-05,
+      "loss": 0.5864,
+      "step": 4555
+    },
+    {
+      "epoch": 1.0974729241877257,
+      "grad_norm": 1.8125,
+      "learning_rate": 7.841273753218124e-05,
+      "loss": 0.5642,
+      "step": 4560
+    },
+    {
+      "epoch": 1.098676293622142,
+      "grad_norm": 1.6875,
+      "learning_rate": 7.836939439629546e-05,
+      "loss": 0.5659,
+      "step": 4565
+    },
+    {
+      "epoch": 1.0998796630565584,
+      "grad_norm": 1.6484375,
+      "learning_rate": 7.832602638955609e-05,
+      "loss": 0.5449,
+      "step": 4570
+    },
+    {
+      "epoch": 1.1010830324909748,
+      "grad_norm": 1.59375,
+      "learning_rate": 7.828263358122067e-05,
+      "loss": 0.571,
+      "step": 4575
+    },
+    {
+      "epoch": 1.102286401925391,
+      "grad_norm": 1.6796875,
+      "learning_rate": 7.82392160405863e-05,
+      "loss": 0.5436,
+      "step": 4580
+    },
+    {
+      "epoch": 1.1034897713598075,
+      "grad_norm": 1.6796875,
+      "learning_rate": 7.81957738369896e-05,
+      "loss": 0.5785,
+      "step": 4585
+    },
+    {
+      "epoch": 1.1046931407942238,
+      "grad_norm": 1.7734375,
+      "learning_rate": 7.815230703980658e-05,
+      "loss": 0.5879,
+      "step": 4590
+    },
+    {
+      "epoch": 1.1058965102286402,
+      "grad_norm": 1.796875,
+      "learning_rate": 7.810881571845253e-05,
+      "loss": 0.5882,
+      "step": 4595
+    },
+    {
+      "epoch": 1.1070998796630565,
+      "grad_norm": 1.71875,
+      "learning_rate": 7.806529994238188e-05,
+      "loss": 0.5838,
+      "step": 4600
+    },
+    {
+      "epoch": 1.108303249097473,
+      "grad_norm": 1.640625,
+      "learning_rate": 7.802175978108813e-05,
+      "loss": 0.5813,
+      "step": 4605
+    },
+    {
+      "epoch": 1.1095066185318894,
+      "grad_norm": 1.8125,
+      "learning_rate": 7.797819530410373e-05,
+      "loss": 0.5836,
+      "step": 4610
+    },
+    {
+      "epoch": 1.1107099879663056,
+      "grad_norm": 1.7109375,
+      "learning_rate": 7.793460658099995e-05,
+      "loss": 0.6057,
+      "step": 4615
+    },
+    {
+      "epoch": 1.111913357400722,
+      "grad_norm": 1.7265625,
+      "learning_rate": 7.789099368138678e-05,
+      "loss": 0.5564,
+      "step": 4620
+    },
+    {
+      "epoch": 1.1131167268351383,
+      "grad_norm": 1.5546875,
+      "learning_rate": 7.78473566749128e-05,
+      "loss": 0.5527,
+      "step": 4625
+    },
+    {
+      "epoch": 1.1143200962695547,
+      "grad_norm": 1.875,
+      "learning_rate": 7.780369563126516e-05,
+      "loss": 0.5814,
+      "step": 4630
+    },
+    {
+      "epoch": 1.1155234657039712,
+      "grad_norm": 1.625,
+      "learning_rate": 7.77600106201693e-05,
+      "loss": 0.5483,
+      "step": 4635
+    },
+    {
+      "epoch": 1.1167268351383874,
+      "grad_norm": 1.6640625,
+      "learning_rate": 7.7716301711389e-05,
+      "loss": 0.5661,
+      "step": 4640
+    },
+    {
+      "epoch": 1.117930204572804,
+      "grad_norm": 1.8125,
+      "learning_rate": 7.767256897472616e-05,
+      "loss": 0.5879,
+      "step": 4645
+    },
+    {
+      "epoch": 1.1191335740072201,
+      "grad_norm": 1.9140625,
+      "learning_rate": 7.76288124800208e-05,
+      "loss": 0.5954,
+      "step": 4650
+    },
+    {
+      "epoch": 1.1203369434416366,
+      "grad_norm": 1.828125,
+      "learning_rate": 7.758503229715083e-05,
+      "loss": 0.5783,
+      "step": 4655
+    },
+    {
+      "epoch": 1.121540312876053,
+      "grad_norm": 1.765625,
+      "learning_rate": 7.754122849603197e-05,
+      "loss": 0.5783,
+      "step": 4660
+    },
+    {
+      "epoch": 1.1227436823104693,
+      "grad_norm": 1.7421875,
+      "learning_rate": 7.749740114661771e-05,
+      "loss": 0.5272,
+      "step": 4665
+    },
+    {
+      "epoch": 1.1239470517448857,
+      "grad_norm": 1.578125,
+      "learning_rate": 7.745355031889913e-05,
+      "loss": 0.5758,
+      "step": 4670
+    },
+    {
+      "epoch": 1.125150421179302,
+      "grad_norm": 1.6015625,
+      "learning_rate": 7.740967608290477e-05,
+      "loss": 0.534,
+      "step": 4675
+    },
+    {
+      "epoch": 1.1263537906137184,
+      "grad_norm": 1.8671875,
+      "learning_rate": 7.736577850870058e-05,
+      "loss": 0.5413,
+      "step": 4680
+    },
+    {
+      "epoch": 1.1275571600481347,
+      "grad_norm": 1.6484375,
+      "learning_rate": 7.73218576663898e-05,
+      "loss": 0.5879,
+      "step": 4685
+    },
+    {
+      "epoch": 1.1287605294825511,
+      "grad_norm": 1.8203125,
+      "learning_rate": 7.727791362611279e-05,
+      "loss": 0.5373,
+      "step": 4690
+    },
+    {
+      "epoch": 1.1299638989169676,
+      "grad_norm": 1.7421875,
+      "learning_rate": 7.7233946458047e-05,
+      "loss": 0.5613,
+      "step": 4695
+    },
+    {
+      "epoch": 1.1311672683513838,
+      "grad_norm": 1.8984375,
+      "learning_rate": 7.718995623240674e-05,
+      "loss": 0.592,
+      "step": 4700
+    },
+    {
+      "epoch": 1.1323706377858003,
+      "grad_norm": 1.703125,
+      "learning_rate": 7.714594301944324e-05,
+      "loss": 0.5852,
+      "step": 4705
+    },
+    {
+      "epoch": 1.1335740072202167,
+      "grad_norm": 1.7890625,
+      "learning_rate": 7.710190688944438e-05,
+      "loss": 0.5835,
+      "step": 4710
+    },
+    {
+      "epoch": 1.134777376654633,
+      "grad_norm": 1.8671875,
+      "learning_rate": 7.705784791273463e-05,
+      "loss": 0.5649,
+      "step": 4715
+    },
+    {
+      "epoch": 1.1359807460890494,
+      "grad_norm": 1.6484375,
+      "learning_rate": 7.701376615967497e-05,
+      "loss": 0.5628,
+      "step": 4720
+    },
+    {
+      "epoch": 1.1371841155234657,
+      "grad_norm": 1.7421875,
+      "learning_rate": 7.696966170066276e-05,
+      "loss": 0.5844,
+      "step": 4725
+    },
+    {
+      "epoch": 1.1383874849578821,
+      "grad_norm": 1.671875,
+      "learning_rate": 7.69255346061316e-05,
+      "loss": 0.5349,
+      "step": 4730
+    },
+    {
+      "epoch": 1.1395908543922983,
+      "grad_norm": 1.8125,
+      "learning_rate": 7.688138494655125e-05,
+      "loss": 0.5429,
+      "step": 4735
+    },
+    {
+      "epoch": 1.1407942238267148,
+      "grad_norm": 1.7265625,
+      "learning_rate": 7.683721279242749e-05,
+      "loss": 0.5762,
+      "step": 4740
+    },
+    {
+      "epoch": 1.1419975932611313,
+      "grad_norm": 1.6875,
+      "learning_rate": 7.679301821430205e-05,
+      "loss": 0.5778,
+      "step": 4745
+    },
+    {
+      "epoch": 1.1432009626955475,
+      "grad_norm": 1.671875,
+      "learning_rate": 7.674880128275246e-05,
+      "loss": 0.5823,
+      "step": 4750
+    },
+    {
+      "epoch": 1.144404332129964,
+      "grad_norm": 1.5703125,
+      "learning_rate": 7.670456206839193e-05,
+      "loss": 0.5535,
+      "step": 4755
+    },
+    {
+      "epoch": 1.1456077015643802,
+      "grad_norm": 1.609375,
+      "learning_rate": 7.666030064186928e-05,
+      "loss": 0.5651,
+      "step": 4760
+    },
+    {
+      "epoch": 1.1468110709987966,
+      "grad_norm": 1.6015625,
+      "learning_rate": 7.661601707386877e-05,
+      "loss": 0.5744,
+      "step": 4765
+    },
+    {
+      "epoch": 1.1480144404332129,
+      "grad_norm": 1.6875,
+      "learning_rate": 7.657171143511008e-05,
+      "loss": 0.5506,
+      "step": 4770
+    },
+    {
+      "epoch": 1.1492178098676293,
+      "grad_norm": 1.84375,
+      "learning_rate": 7.652738379634806e-05,
+      "loss": 0.5681,
+      "step": 4775
+    },
+    {
+      "epoch": 1.1504211793020458,
+      "grad_norm": 1.5703125,
+      "learning_rate": 7.648303422837275e-05,
+      "loss": 0.5271,
+      "step": 4780
+    },
+    {
+      "epoch": 1.151624548736462,
+      "grad_norm": 1.703125,
+      "learning_rate": 7.643866280200916e-05,
+      "loss": 0.5712,
+      "step": 4785
+    },
+    {
+      "epoch": 1.1528279181708785,
+      "grad_norm": 1.5390625,
+      "learning_rate": 7.639426958811728e-05,
+      "loss": 0.5676,
+      "step": 4790
+    },
+    {
+      "epoch": 1.154031287605295,
+      "grad_norm": 1.84375,
+      "learning_rate": 7.634985465759185e-05,
+      "loss": 0.5726,
+      "step": 4795
+    },
+    {
+      "epoch": 1.1552346570397112,
+      "grad_norm": 1.78125,
+      "learning_rate": 7.630541808136224e-05,
+      "loss": 0.5879,
+      "step": 4800
+    },
+    {
+      "epoch": 1.1564380264741276,
+      "grad_norm": 1.6875,
+      "learning_rate": 7.626095993039252e-05,
+      "loss": 0.5786,
+      "step": 4805
+    },
+    {
+      "epoch": 1.1576413959085439,
+      "grad_norm": 1.7734375,
+      "learning_rate": 7.621648027568106e-05,
+      "loss": 0.5461,
+      "step": 4810
+    },
+    {
+      "epoch": 1.1588447653429603,
+      "grad_norm": 1.796875,
+      "learning_rate": 7.61719791882607e-05,
+      "loss": 0.5605,
+      "step": 4815
+    },
+    {
+      "epoch": 1.1600481347773766,
+      "grad_norm": 2.0,
+      "learning_rate": 7.612745673919841e-05,
+      "loss": 0.5982,
+      "step": 4820
+    },
+    {
+      "epoch": 1.161251504211793,
+      "grad_norm": 1.90625,
+      "learning_rate": 7.608291299959535e-05,
+      "loss": 0.5808,
+      "step": 4825
+    },
+    {
+      "epoch": 1.1624548736462095,
+      "grad_norm": 1.7578125,
+      "learning_rate": 7.603834804058663e-05,
+      "loss": 0.5782,
+      "step": 4830
+    },
+    {
+      "epoch": 1.1636582430806257,
+      "grad_norm": 1.7109375,
+      "learning_rate": 7.599376193334128e-05,
+      "loss": 0.5563,
+      "step": 4835
+    },
+    {
+      "epoch": 1.1648616125150422,
+      "grad_norm": 1.8125,
+      "learning_rate": 7.594915474906208e-05,
+      "loss": 0.5864,
+      "step": 4840
+    },
+    {
+      "epoch": 1.1660649819494584,
+      "grad_norm": 1.71875,
+      "learning_rate": 7.590452655898546e-05,
+      "loss": 0.5598,
+      "step": 4845
+    },
+    {
+      "epoch": 1.1672683513838749,
+      "grad_norm": 1.8515625,
+      "learning_rate": 7.585987743438143e-05,
+      "loss": 0.5418,
+      "step": 4850
+    },
+    {
+      "epoch": 1.168471720818291,
+      "grad_norm": 1.75,
+      "learning_rate": 7.58152074465534e-05,
+      "loss": 0.5851,
+      "step": 4855
+    },
+    {
+      "epoch": 1.1696750902527075,
+      "grad_norm": 1.8359375,
+      "learning_rate": 7.577051666683814e-05,
+      "loss": 0.5671,
+      "step": 4860
+    },
+    {
+      "epoch": 1.170878459687124,
+      "grad_norm": 1.6875,
+      "learning_rate": 7.572580516660559e-05,
+      "loss": 0.5606,
+      "step": 4865
+    },
+    {
+      "epoch": 1.1720818291215402,
+      "grad_norm": 2.015625,
+      "learning_rate": 7.568107301725875e-05,
+      "loss": 0.5843,
+      "step": 4870
+    },
+    {
+      "epoch": 1.1732851985559567,
+      "grad_norm": 1.8515625,
+      "learning_rate": 7.563632029023366e-05,
+      "loss": 0.5452,
+      "step": 4875
+    },
+    {
+      "epoch": 1.1744885679903732,
+      "grad_norm": 1.984375,
+      "learning_rate": 7.55915470569992e-05,
+      "loss": 0.5551,
+      "step": 4880
+    },
+    {
+      "epoch": 1.1756919374247894,
+      "grad_norm": 1.765625,
+      "learning_rate": 7.554675338905699e-05,
+      "loss": 0.585,
+      "step": 4885
+    },
+    {
+      "epoch": 1.1768953068592058,
+      "grad_norm": 1.6328125,
+      "learning_rate": 7.550193935794128e-05,
+      "loss": 0.5754,
+      "step": 4890
+    },
+    {
+      "epoch": 1.178098676293622,
+      "grad_norm": 1.734375,
+      "learning_rate": 7.545710503521885e-05,
+      "loss": 0.5483,
+      "step": 4895
+    },
+    {
+      "epoch": 1.1793020457280385,
+      "grad_norm": 1.7265625,
+      "learning_rate": 7.541225049248888e-05,
+      "loss": 0.5752,
+      "step": 4900
+    },
+    {
+      "epoch": 1.1805054151624548,
+      "grad_norm": 1.671875,
+      "learning_rate": 7.536737580138282e-05,
+      "loss": 0.5567,
+      "step": 4905
+    },
+    {
+      "epoch": 1.1817087845968712,
+      "grad_norm": 1.8046875,
+      "learning_rate": 7.532248103356436e-05,
+      "loss": 0.5805,
+      "step": 4910
+    },
+    {
+      "epoch": 1.1829121540312877,
+      "grad_norm": 1.65625,
+      "learning_rate": 7.527756626072919e-05,
+      "loss": 0.5726,
+      "step": 4915
+    },
+    {
+      "epoch": 1.184115523465704,
+      "grad_norm": 1.859375,
+      "learning_rate": 7.523263155460495e-05,
+      "loss": 0.5953,
+      "step": 4920
+    },
+    {
+      "epoch": 1.1853188929001204,
+      "grad_norm": 1.6328125,
+      "learning_rate": 7.518767698695114e-05,
+      "loss": 0.5617,
+      "step": 4925
+    },
+    {
+      "epoch": 1.1865222623345366,
+      "grad_norm": 1.65625,
+      "learning_rate": 7.514270262955898e-05,
+      "loss": 0.5497,
+      "step": 4930
+    },
+    {
+      "epoch": 1.187725631768953,
+      "grad_norm": 1.7109375,
+      "learning_rate": 7.509770855425127e-05,
+      "loss": 0.5711,
+      "step": 4935
+    },
+    {
+      "epoch": 1.1889290012033695,
+      "grad_norm": 1.7421875,
+      "learning_rate": 7.505269483288231e-05,
+      "loss": 0.5499,
+      "step": 4940
+    },
+    {
+      "epoch": 1.1901323706377858,
+      "grad_norm": 1.90625,
+      "learning_rate": 7.50076615373378e-05,
+      "loss": 0.592,
+      "step": 4945
+    },
+    {
+      "epoch": 1.1913357400722022,
+      "grad_norm": 1.6796875,
+      "learning_rate": 7.496260873953466e-05,
+      "loss": 0.5446,
+      "step": 4950
+    },
+    {
+      "epoch": 1.1925391095066185,
+      "grad_norm": 1.65625,
+      "learning_rate": 7.491753651142096e-05,
+      "loss": 0.5606,
+      "step": 4955
+    },
+    {
+      "epoch": 1.193742478941035,
+      "grad_norm": 1.8125,
+      "learning_rate": 7.48724449249758e-05,
+      "loss": 0.5566,
+      "step": 4960
+    },
+    {
+      "epoch": 1.1949458483754514,
+      "grad_norm": 1.734375,
+      "learning_rate": 7.482733405220924e-05,
+      "loss": 0.544,
+      "step": 4965
+    },
+    {
+      "epoch": 1.1961492178098676,
+      "grad_norm": 1.7109375,
+      "learning_rate": 7.478220396516206e-05,
+      "loss": 0.5476,
+      "step": 4970
+    },
+    {
+      "epoch": 1.197352587244284,
+      "grad_norm": 1.578125,
+      "learning_rate": 7.473705473590582e-05,
+      "loss": 0.556,
+      "step": 4975
+    },
+    {
+      "epoch": 1.1985559566787003,
+      "grad_norm": 1.6015625,
+      "learning_rate": 7.469188643654254e-05,
+      "loss": 0.5527,
+      "step": 4980
+    },
+    {
+      "epoch": 1.1997593261131168,
+      "grad_norm": 1.703125,
+      "learning_rate": 7.46466991392048e-05,
+      "loss": 0.5428,
+      "step": 4985
+    },
+    {
+      "epoch": 1.200962695547533,
+      "grad_norm": 1.71875,
+      "learning_rate": 7.460149291605544e-05,
+      "loss": 0.5524,
+      "step": 4990
+    },
+    {
+      "epoch": 1.2021660649819494,
+      "grad_norm": 1.78125,
+      "learning_rate": 7.455626783928757e-05,
+      "loss": 0.597,
+      "step": 4995
+    },
+    {
+      "epoch": 1.203369434416366,
+      "grad_norm": 1.6796875,
+      "learning_rate": 7.45110239811244e-05,
+      "loss": 0.5684,
+      "step": 5000
+    },
+    {
+      "epoch": 1.203369434416366,
+      "eval_loss": 0.49116280674934387,
+      "eval_runtime": 2.6476,
+      "eval_samples_per_second": 75.54,
+      "eval_steps_per_second": 75.54,
+      "step": 5000
+    },
+    {
+      "epoch": 1.2045728038507821,
+      "grad_norm": 1.6015625,
+      "learning_rate": 7.446576141381909e-05,
+      "loss": 0.5554,
+      "step": 5005
+    },
+    {
+      "epoch": 1.2057761732851986,
+      "grad_norm": 1.703125,
+      "learning_rate": 7.442048020965478e-05,
+      "loss": 0.5728,
+      "step": 5010
+    },
+    {
+      "epoch": 1.2069795427196148,
+      "grad_norm": 1.6875,
+      "learning_rate": 7.437518044094426e-05,
+      "loss": 0.5353,
+      "step": 5015
+    },
+    {
+      "epoch": 1.2081829121540313,
+      "grad_norm": 1.6171875,
+      "learning_rate": 7.432986218003002e-05,
+      "loss": 0.5448,
+      "step": 5020
+    },
+    {
+      "epoch": 1.2093862815884477,
+      "grad_norm": 1.8046875,
+      "learning_rate": 7.428452549928408e-05,
+      "loss": 0.5824,
+      "step": 5025
+    },
+    {
+      "epoch": 1.210589651022864,
+      "grad_norm": 1.65625,
+      "learning_rate": 7.423917047110788e-05,
+      "loss": 0.5641,
+      "step": 5030
+    },
+    {
+      "epoch": 1.2117930204572804,
+      "grad_norm": 1.921875,
+      "learning_rate": 7.419379716793217e-05,
+      "loss": 0.5362,
+      "step": 5035
+    },
+    {
+      "epoch": 1.2129963898916967,
+      "grad_norm": 1.65625,
+      "learning_rate": 7.414840566221683e-05,
+      "loss": 0.5717,
+      "step": 5040
+    },
+    {
+      "epoch": 1.2141997593261131,
+      "grad_norm": 1.734375,
+      "learning_rate": 7.410299602645087e-05,
+      "loss": 0.59,
+      "step": 5045
+    },
+    {
+      "epoch": 1.2154031287605296,
+      "grad_norm": 1.7578125,
+      "learning_rate": 7.405756833315221e-05,
+      "loss": 0.5442,
+      "step": 5050
+    },
+    {
+      "epoch": 1.2166064981949458,
+      "grad_norm": 1.8984375,
+      "learning_rate": 7.401212265486765e-05,
+      "loss": 0.5929,
+      "step": 5055
+    },
+    {
+      "epoch": 1.2178098676293623,
+      "grad_norm": 1.6640625,
+      "learning_rate": 7.396665906417269e-05,
+      "loss": 0.5278,
+      "step": 5060
+    },
+    {
+      "epoch": 1.2190132370637785,
+      "grad_norm": 1.609375,
+      "learning_rate": 7.392117763367142e-05,
+      "loss": 0.5568,
+      "step": 5065
+    },
+    {
+      "epoch": 1.220216606498195,
+      "grad_norm": 1.7578125,
+      "learning_rate": 7.387567843599643e-05,
+      "loss": 0.5513,
+      "step": 5070
+    },
+    {
+      "epoch": 1.2214199759326112,
+      "grad_norm": 1.8359375,
+      "learning_rate": 7.383016154380869e-05,
+      "loss": 0.5313,
+      "step": 5075
+    },
+    {
+      "epoch": 1.2226233453670277,
+      "grad_norm": 1.8203125,
+      "learning_rate": 7.378462702979744e-05,
+      "loss": 0.5842,
+      "step": 5080
+    },
+    {
+      "epoch": 1.2238267148014441,
+      "grad_norm": 1.703125,
+      "learning_rate": 7.373907496668005e-05,
+      "loss": 0.561,
+      "step": 5085
+    },
+    {
+      "epoch": 1.2250300842358604,
+      "grad_norm": 1.78125,
+      "learning_rate": 7.369350542720187e-05,
+      "loss": 0.5642,
+      "step": 5090
+    },
+    {
+      "epoch": 1.2262334536702768,
+      "grad_norm": 1.59375,
+      "learning_rate": 7.364791848413625e-05,
+      "loss": 0.5696,
+      "step": 5095
+    },
+    {
+      "epoch": 1.2274368231046933,
+      "grad_norm": 1.84375,
+      "learning_rate": 7.360231421028426e-05,
+      "loss": 0.5507,
+      "step": 5100
+    },
+    {
+      "epoch": 1.2286401925391095,
+      "grad_norm": 1.5625,
+      "learning_rate": 7.355669267847469e-05,
+      "loss": 0.5596,
+      "step": 5105
+    },
+    {
+      "epoch": 1.229843561973526,
+      "grad_norm": 1.734375,
+      "learning_rate": 7.351105396156385e-05,
+      "loss": 0.5707,
+      "step": 5110
+    },
+    {
+      "epoch": 1.2310469314079422,
+      "grad_norm": 1.90625,
+      "learning_rate": 7.346539813243554e-05,
+      "loss": 0.5737,
+      "step": 5115
+    },
+    {
+      "epoch": 1.2322503008423586,
+      "grad_norm": 1.8046875,
+      "learning_rate": 7.341972526400083e-05,
+      "loss": 0.5665,
+      "step": 5120
+    },
+    {
+      "epoch": 1.2334536702767749,
+      "grad_norm": 1.6015625,
+      "learning_rate": 7.337403542919807e-05,
+      "loss": 0.5499,
+      "step": 5125
+    },
+    {
+      "epoch": 1.2346570397111913,
+      "grad_norm": 1.6796875,
+      "learning_rate": 7.332832870099264e-05,
+      "loss": 0.5377,
+      "step": 5130
+    },
+    {
+      "epoch": 1.2358604091456078,
+      "grad_norm": 1.796875,
+      "learning_rate": 7.328260515237696e-05,
+      "loss": 0.5678,
+      "step": 5135
+    },
+    {
+      "epoch": 1.237063778580024,
+      "grad_norm": 1.8046875,
+      "learning_rate": 7.323686485637024e-05,
+      "loss": 0.5411,
+      "step": 5140
+    },
+    {
+      "epoch": 1.2382671480144405,
+      "grad_norm": 1.8046875,
+      "learning_rate": 7.319110788601853e-05,
+      "loss": 0.5587,
+      "step": 5145
+    },
+    {
+      "epoch": 1.2394705174488567,
+      "grad_norm": 1.796875,
+      "learning_rate": 7.31453343143944e-05,
+      "loss": 0.5671,
+      "step": 5150
+    },
+    {
+      "epoch": 1.2406738868832732,
+      "grad_norm": 1.7734375,
+      "learning_rate": 7.3099544214597e-05,
+      "loss": 0.5675,
+      "step": 5155
+    },
+    {
+      "epoch": 1.2418772563176894,
+      "grad_norm": 1.890625,
+      "learning_rate": 7.305373765975188e-05,
+      "loss": 0.552,
+      "step": 5160
+    },
+    {
+      "epoch": 1.2430806257521059,
+      "grad_norm": 1.6484375,
+      "learning_rate": 7.300791472301084e-05,
+      "loss": 0.5643,
+      "step": 5165
+    },
+    {
+      "epoch": 1.2442839951865223,
+      "grad_norm": 1.7734375,
+      "learning_rate": 7.296207547755183e-05,
+      "loss": 0.5496,
+      "step": 5170
+    },
+    {
+      "epoch": 1.2454873646209386,
+      "grad_norm": 1.7421875,
+      "learning_rate": 7.29162199965789e-05,
+      "loss": 0.5535,
+      "step": 5175
+    },
+    {
+      "epoch": 1.246690734055355,
+      "grad_norm": 1.765625,
+      "learning_rate": 7.287034835332196e-05,
+      "loss": 0.5823,
+      "step": 5180
+    },
+    {
+      "epoch": 1.2478941034897715,
+      "grad_norm": 1.7890625,
+      "learning_rate": 7.282446062103678e-05,
+      "loss": 0.5633,
+      "step": 5185
+    },
+    {
+      "epoch": 1.2490974729241877,
+      "grad_norm": 1.640625,
+      "learning_rate": 7.277855687300481e-05,
+      "loss": 0.5696,
+      "step": 5190
+    },
+    {
+      "epoch": 1.2503008423586042,
+      "grad_norm": 1.8046875,
+      "learning_rate": 7.273263718253305e-05,
+      "loss": 0.5769,
+      "step": 5195
+    },
+    {
+      "epoch": 1.2515042117930204,
+      "grad_norm": 1.7265625,
+      "learning_rate": 7.268670162295402e-05,
+      "loss": 0.5457,
+      "step": 5200
+    },
+    {
+      "epoch": 1.2527075812274369,
+      "grad_norm": 1.71875,
+      "learning_rate": 7.26407502676255e-05,
+      "loss": 0.5766,
+      "step": 5205
+    },
+    {
+      "epoch": 1.253910950661853,
+      "grad_norm": 1.578125,
+      "learning_rate": 7.259478318993058e-05,
+      "loss": 0.5354,
+      "step": 5210
+    },
+    {
+      "epoch": 1.2551143200962696,
+      "grad_norm": 1.734375,
+      "learning_rate": 7.25488004632774e-05,
+      "loss": 0.5561,
+      "step": 5215
+    },
+    {
+      "epoch": 1.256317689530686,
+      "grad_norm": 1.6484375,
+      "learning_rate": 7.250280216109913e-05,
+      "loss": 0.5684,
+      "step": 5220
+    },
+    {
+      "epoch": 1.2575210589651022,
+      "grad_norm": 1.8046875,
+      "learning_rate": 7.245678835685378e-05,
+      "loss": 0.5453,
+      "step": 5225
+    },
+    {
+      "epoch": 1.2587244283995187,
+      "grad_norm": 1.6171875,
+      "learning_rate": 7.24107591240241e-05,
+      "loss": 0.5364,
+      "step": 5230
+    },
+    {
+      "epoch": 1.259927797833935,
+      "grad_norm": 1.609375,
+      "learning_rate": 7.236471453611757e-05,
+      "loss": 0.5611,
+      "step": 5235
+    },
+    {
+      "epoch": 1.2611311672683514,
+      "grad_norm": 1.734375,
+      "learning_rate": 7.231865466666609e-05,
+      "loss": 0.5558,
+      "step": 5240
+    },
+    {
+      "epoch": 1.2623345367027676,
+      "grad_norm": 1.6484375,
+      "learning_rate": 7.227257958922601e-05,
+      "loss": 0.6038,
+      "step": 5245
+    },
+    {
+      "epoch": 1.263537906137184,
+      "grad_norm": 1.546875,
+      "learning_rate": 7.222648937737797e-05,
+      "loss": 0.5456,
+      "step": 5250
+    },
+    {
+      "epoch": 1.2647412755716005,
+      "grad_norm": 1.7734375,
+      "learning_rate": 7.218038410472676e-05,
+      "loss": 0.553,
+      "step": 5255
+    },
+    {
+      "epoch": 1.2659446450060168,
+      "grad_norm": 1.65625,
+      "learning_rate": 7.213426384490126e-05,
+      "loss": 0.556,
+      "step": 5260
+    },
+    {
+      "epoch": 1.2671480144404332,
+      "grad_norm": 1.5859375,
+      "learning_rate": 7.208812867155422e-05,
+      "loss": 0.5676,
+      "step": 5265
+    },
+    {
+      "epoch": 1.2683513838748497,
+      "grad_norm": 1.703125,
+      "learning_rate": 7.204197865836226e-05,
+      "loss": 0.5484,
+      "step": 5270
+    },
+    {
+      "epoch": 1.269554753309266,
+      "grad_norm": 1.8125,
+      "learning_rate": 7.199581387902569e-05,
+      "loss": 0.5779,
+      "step": 5275
+    },
+    {
+      "epoch": 1.2707581227436824,
+      "grad_norm": 1.6953125,
+      "learning_rate": 7.194963440726839e-05,
+      "loss": 0.5296,
+      "step": 5280
+    },
+    {
+      "epoch": 1.2719614921780986,
+      "grad_norm": 1.7890625,
+      "learning_rate": 7.19034403168377e-05,
+      "loss": 0.5792,
+      "step": 5285
+    },
+    {
+      "epoch": 1.273164861612515,
+      "grad_norm": 1.8515625,
+      "learning_rate": 7.185723168150432e-05,
+      "loss": 0.5887,
+      "step": 5290
+    },
+    {
+      "epoch": 1.2743682310469313,
+      "grad_norm": 1.7890625,
+      "learning_rate": 7.181100857506215e-05,
+      "loss": 0.556,
+      "step": 5295
+    },
+    {
+      "epoch": 1.2755716004813478,
+      "grad_norm": 1.7109375,
+      "learning_rate": 7.176477107132825e-05,
+      "loss": 0.5639,
+      "step": 5300
+    },
+    {
+      "epoch": 1.2767749699157642,
+      "grad_norm": 1.6484375,
+      "learning_rate": 7.171851924414262e-05,
+      "loss": 0.551,
+      "step": 5305
+    },
+    {
+      "epoch": 1.2779783393501805,
+      "grad_norm": 1.7421875,
+      "learning_rate": 7.167225316736816e-05,
+      "loss": 0.5926,
+      "step": 5310
+    },
+    {
+      "epoch": 1.279181708784597,
+      "grad_norm": 1.703125,
+      "learning_rate": 7.162597291489054e-05,
+      "loss": 0.5579,
+      "step": 5315
+    },
+    {
+      "epoch": 1.2803850782190134,
+      "grad_norm": 1.7109375,
+      "learning_rate": 7.157967856061803e-05,
+      "loss": 0.5578,
+      "step": 5320
+    },
+    {
+      "epoch": 1.2815884476534296,
+      "grad_norm": 1.578125,
+      "learning_rate": 7.153337017848145e-05,
+      "loss": 0.5501,
+      "step": 5325
+    },
+    {
+      "epoch": 1.2827918170878458,
+      "grad_norm": 1.6875,
+      "learning_rate": 7.148704784243401e-05,
+      "loss": 0.5606,
+      "step": 5330
+    },
+    {
+      "epoch": 1.2839951865222623,
+      "grad_norm": 1.6875,
+      "learning_rate": 7.14407116264512e-05,
+      "loss": 0.5626,
+      "step": 5335
+    },
+    {
+      "epoch": 1.2851985559566788,
+      "grad_norm": 1.6015625,
+      "learning_rate": 7.13943616045307e-05,
+      "loss": 0.5929,
+      "step": 5340
+    },
+    {
+      "epoch": 1.286401925391095,
+      "grad_norm": 1.953125,
+      "learning_rate": 7.134799785069222e-05,
+      "loss": 0.5574,
+      "step": 5345
+    },
+    {
+      "epoch": 1.2876052948255114,
+      "grad_norm": 1.9609375,
+      "learning_rate": 7.130162043897738e-05,
+      "loss": 0.5547,
+      "step": 5350
+    },
+    {
+      "epoch": 1.288808664259928,
+      "grad_norm": 1.7109375,
+      "learning_rate": 7.125522944344964e-05,
+      "loss": 0.5545,
+      "step": 5355
+    },
+    {
+      "epoch": 1.2900120336943441,
+      "grad_norm": 1.8125,
+      "learning_rate": 7.120882493819416e-05,
+      "loss": 0.5794,
+      "step": 5360
+    },
+    {
+      "epoch": 1.2912154031287606,
+      "grad_norm": 1.7734375,
+      "learning_rate": 7.116240699731764e-05,
+      "loss": 0.5531,
+      "step": 5365
+    },
+    {
+      "epoch": 1.2924187725631768,
+      "grad_norm": 1.78125,
+      "learning_rate": 7.111597569494826e-05,
+      "loss": 0.5526,
+      "step": 5370
+    },
+    {
+      "epoch": 1.2936221419975933,
+      "grad_norm": 1.6875,
+      "learning_rate": 7.106953110523552e-05,
+      "loss": 0.5128,
+      "step": 5375
+    },
+    {
+      "epoch": 1.2948255114320095,
+      "grad_norm": 1.6171875,
+      "learning_rate": 7.102307330235018e-05,
+      "loss": 0.5917,
+      "step": 5380
+    },
+    {
+      "epoch": 1.296028880866426,
+      "grad_norm": 1.703125,
+      "learning_rate": 7.097660236048408e-05,
+      "loss": 0.5499,
+      "step": 5385
+    },
+    {
+      "epoch": 1.2972322503008424,
+      "grad_norm": 1.796875,
+      "learning_rate": 7.093011835385e-05,
+      "loss": 0.5861,
+      "step": 5390
+    },
+    {
+      "epoch": 1.2984356197352587,
+      "grad_norm": 1.78125,
+      "learning_rate": 7.088362135668165e-05,
+      "loss": 0.5668,
+      "step": 5395
+    },
+    {
+      "epoch": 1.2996389891696751,
+      "grad_norm": 1.7578125,
+      "learning_rate": 7.083711144323343e-05,
+      "loss": 0.5407,
+      "step": 5400
+    },
+    {
+      "epoch": 1.3008423586040916,
+      "grad_norm": 1.5703125,
+      "learning_rate": 7.079058868778041e-05,
+      "loss": 0.5213,
+      "step": 5405
+    },
+    {
+      "epoch": 1.3020457280385078,
+      "grad_norm": 2.0,
+      "learning_rate": 7.074405316461816e-05,
+      "loss": 0.5457,
+      "step": 5410
+    },
+    {
+      "epoch": 1.303249097472924,
+      "grad_norm": 1.75,
+      "learning_rate": 7.06975049480626e-05,
+      "loss": 0.544,
+      "step": 5415
+    },
+    {
+      "epoch": 1.3044524669073405,
+      "grad_norm": 1.6796875,
+      "learning_rate": 7.065094411245e-05,
+      "loss": 0.581,
+      "step": 5420
+    },
+    {
+      "epoch": 1.305655836341757,
+      "grad_norm": 1.796875,
+      "learning_rate": 7.060437073213668e-05,
+      "loss": 0.5678,
+      "step": 5425
+    },
+    {
+      "epoch": 1.3068592057761732,
+      "grad_norm": 1.6328125,
+      "learning_rate": 7.05577848814991e-05,
+      "loss": 0.5917,
+      "step": 5430
+    },
+    {
+      "epoch": 1.3080625752105897,
+      "grad_norm": 1.6796875,
+      "learning_rate": 7.051118663493353e-05,
+      "loss": 0.5271,
+      "step": 5435
+    },
+    {
+      "epoch": 1.3092659446450061,
+      "grad_norm": 1.71875,
+      "learning_rate": 7.046457606685615e-05,
+      "loss": 0.5609,
+      "step": 5440
+    },
+    {
+      "epoch": 1.3104693140794224,
+      "grad_norm": 1.7421875,
+      "learning_rate": 7.041795325170268e-05,
+      "loss": 0.5716,
+      "step": 5445
+    },
+    {
+      "epoch": 1.3116726835138388,
+      "grad_norm": 1.671875,
+      "learning_rate": 7.037131826392854e-05,
+      "loss": 0.5357,
+      "step": 5450
+    },
+    {
+      "epoch": 1.312876052948255,
+      "grad_norm": 1.6328125,
+      "learning_rate": 7.032467117800849e-05,
+      "loss": 0.5427,
+      "step": 5455
+    },
+    {
+      "epoch": 1.3140794223826715,
+      "grad_norm": 1.8203125,
+      "learning_rate": 7.027801206843664e-05,
+      "loss": 0.5346,
+      "step": 5460
+    },
+    {
+      "epoch": 1.3152827918170877,
+      "grad_norm": 1.578125,
+      "learning_rate": 7.023134100972636e-05,
+      "loss": 0.5622,
+      "step": 5465
+    },
+    {
+      "epoch": 1.3164861612515042,
+      "grad_norm": 1.6171875,
+      "learning_rate": 7.018465807640995e-05,
+      "loss": 0.5627,
+      "step": 5470
+    },
+    {
+      "epoch": 1.3176895306859207,
+      "grad_norm": 1.734375,
+      "learning_rate": 7.013796334303887e-05,
+      "loss": 0.5483,
+      "step": 5475
+    },
+    {
+      "epoch": 1.3188929001203369,
+      "grad_norm": 1.6875,
+      "learning_rate": 7.009125688418325e-05,
+      "loss": 0.5764,
+      "step": 5480
+    },
+    {
+      "epoch": 1.3200962695547533,
+      "grad_norm": 1.671875,
+      "learning_rate": 7.004453877443206e-05,
+      "loss": 0.5626,
+      "step": 5485
+    },
+    {
+      "epoch": 1.3212996389891698,
+      "grad_norm": 1.8203125,
+      "learning_rate": 6.999780908839284e-05,
+      "loss": 0.5367,
+      "step": 5490
+    },
+    {
+      "epoch": 1.322503008423586,
+      "grad_norm": 1.984375,
+      "learning_rate": 6.995106790069157e-05,
+      "loss": 0.5612,
+      "step": 5495
+    },
+    {
+      "epoch": 1.3237063778580023,
+      "grad_norm": 1.671875,
+      "learning_rate": 6.99043152859727e-05,
+      "loss": 0.5511,
+      "step": 5500
+    },
+    {
+      "epoch": 1.3237063778580023,
+      "eval_loss": 0.48222118616104126,
+      "eval_runtime": 2.6547,
+      "eval_samples_per_second": 75.338,
+      "eval_steps_per_second": 75.338,
+      "step": 5500
+    },
+    {
+      "epoch": 1.3249097472924187,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.985755131889885e-05,
+      "loss": 0.56,
+      "step": 5505
+    },
+    {
+      "epoch": 1.3261131167268352,
+      "grad_norm": 1.5703125,
+      "learning_rate": 6.981077607415076e-05,
+      "loss": 0.5547,
+      "step": 5510
+    },
+    {
+      "epoch": 1.3273164861612514,
+      "grad_norm": 1.7265625,
+      "learning_rate": 6.976398962642725e-05,
+      "loss": 0.5537,
+      "step": 5515
+    },
+    {
+      "epoch": 1.3285198555956679,
+      "grad_norm": 1.6328125,
+      "learning_rate": 6.971719205044497e-05,
+      "loss": 0.5344,
+      "step": 5520
+    },
+    {
+      "epoch": 1.3297232250300843,
+      "grad_norm": 1.7734375,
+      "learning_rate": 6.967038342093837e-05,
+      "loss": 0.5497,
+      "step": 5525
+    },
+    {
+      "epoch": 1.3309265944645006,
+      "grad_norm": 1.65625,
+      "learning_rate": 6.962356381265953e-05,
+      "loss": 0.5408,
+      "step": 5530
+    },
+    {
+      "epoch": 1.332129963898917,
+      "grad_norm": 1.7421875,
+      "learning_rate": 6.957673330037809e-05,
+      "loss": 0.5532,
+      "step": 5535
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 1.859375,
+      "learning_rate": 6.952989195888108e-05,
+      "loss": 0.5074,
+      "step": 5540
+    },
+    {
+      "epoch": 1.3345367027677497,
+      "grad_norm": 1.6328125,
+      "learning_rate": 6.948303986297285e-05,
+      "loss": 0.5585,
+      "step": 5545
+    },
+    {
+      "epoch": 1.335740072202166,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.943617708747491e-05,
+      "loss": 0.5719,
+      "step": 5550
+    },
+    {
+      "epoch": 1.3369434416365824,
+      "grad_norm": 1.734375,
+      "learning_rate": 6.938930370722581e-05,
+      "loss": 0.5581,
+      "step": 5555
+    },
+    {
+      "epoch": 1.3381468110709989,
+      "grad_norm": 1.78125,
+      "learning_rate": 6.934241979708105e-05,
+      "loss": 0.5387,
+      "step": 5560
+    },
+    {
+      "epoch": 1.339350180505415,
+      "grad_norm": 1.609375,
+      "learning_rate": 6.929552543191295e-05,
+      "loss": 0.5287,
+      "step": 5565
+    },
+    {
+      "epoch": 1.3405535499398316,
+      "grad_norm": 1.6171875,
+      "learning_rate": 6.924862068661052e-05,
+      "loss": 0.5346,
+      "step": 5570
+    },
+    {
+      "epoch": 1.341756919374248,
+      "grad_norm": 1.578125,
+      "learning_rate": 6.920170563607937e-05,
+      "loss": 0.5506,
+      "step": 5575
+    },
+    {
+      "epoch": 1.3429602888086642,
+      "grad_norm": 1.625,
+      "learning_rate": 6.915478035524151e-05,
+      "loss": 0.5401,
+      "step": 5580
+    },
+    {
+      "epoch": 1.3441636582430807,
+      "grad_norm": 1.765625,
+      "learning_rate": 6.910784491903533e-05,
+      "loss": 0.5626,
+      "step": 5585
+    },
+    {
+      "epoch": 1.345367027677497,
+      "grad_norm": 1.75,
+      "learning_rate": 6.906089940241545e-05,
+      "loss": 0.5523,
+      "step": 5590
+    },
+    {
+      "epoch": 1.3465703971119134,
+      "grad_norm": 1.65625,
+      "learning_rate": 6.901394388035256e-05,
+      "loss": 0.5049,
+      "step": 5595
+    },
+    {
+      "epoch": 1.3477737665463296,
+      "grad_norm": 1.671875,
+      "learning_rate": 6.896697842783335e-05,
+      "loss": 0.5747,
+      "step": 5600
+    },
+    {
+      "epoch": 1.348977135980746,
+      "grad_norm": 1.8515625,
+      "learning_rate": 6.892000311986032e-05,
+      "loss": 0.5706,
+      "step": 5605
+    },
+    {
+      "epoch": 1.3501805054151625,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.88730180314518e-05,
+      "loss": 0.5127,
+      "step": 5610
+    },
+    {
+      "epoch": 1.3513838748495788,
+      "grad_norm": 1.7421875,
+      "learning_rate": 6.882602323764164e-05,
+      "loss": 0.5712,
+      "step": 5615
+    },
+    {
+      "epoch": 1.3525872442839952,
+      "grad_norm": 1.8359375,
+      "learning_rate": 6.877901881347923e-05,
+      "loss": 0.5524,
+      "step": 5620
+    },
+    {
+      "epoch": 1.3537906137184115,
+      "grad_norm": 1.6328125,
+      "learning_rate": 6.873200483402938e-05,
+      "loss": 0.5465,
+      "step": 5625
+    },
+    {
+      "epoch": 1.354993983152828,
+      "grad_norm": 1.7578125,
+      "learning_rate": 6.868498137437209e-05,
+      "loss": 0.5542,
+      "step": 5630
+    },
+    {
+      "epoch": 1.3561973525872442,
+      "grad_norm": 1.9140625,
+      "learning_rate": 6.863794850960255e-05,
+      "loss": 0.5428,
+      "step": 5635
+    },
+    {
+      "epoch": 1.3574007220216606,
+      "grad_norm": 1.734375,
+      "learning_rate": 6.859090631483094e-05,
+      "loss": 0.5673,
+      "step": 5640
+    },
+    {
+      "epoch": 1.358604091456077,
+      "grad_norm": 1.5859375,
+      "learning_rate": 6.854385486518237e-05,
+      "loss": 0.5339,
+      "step": 5645
+    },
+    {
+      "epoch": 1.3598074608904933,
+      "grad_norm": 1.6171875,
+      "learning_rate": 6.849679423579671e-05,
+      "loss": 0.5295,
+      "step": 5650
+    },
+    {
+      "epoch": 1.3610108303249098,
+      "grad_norm": 1.6953125,
+      "learning_rate": 6.844972450182847e-05,
+      "loss": 0.5426,
+      "step": 5655
+    },
+    {
+      "epoch": 1.3622141997593262,
+      "grad_norm": 1.78125,
+      "learning_rate": 6.840264573844672e-05,
+      "loss": 0.5092,
+      "step": 5660
+    },
+    {
+      "epoch": 1.3634175691937425,
+      "grad_norm": 1.671875,
+      "learning_rate": 6.835555802083496e-05,
+      "loss": 0.5254,
+      "step": 5665
+    },
+    {
+      "epoch": 1.364620938628159,
+      "grad_norm": 1.640625,
+      "learning_rate": 6.830846142419101e-05,
+      "loss": 0.5454,
+      "step": 5670
+    },
+    {
+      "epoch": 1.3658243080625752,
+      "grad_norm": 1.5859375,
+      "learning_rate": 6.82613560237268e-05,
+      "loss": 0.5416,
+      "step": 5675
+    },
+    {
+      "epoch": 1.3670276774969916,
+      "grad_norm": 1.703125,
+      "learning_rate": 6.821424189466835e-05,
+      "loss": 0.5677,
+      "step": 5680
+    },
+    {
+      "epoch": 1.3682310469314078,
+      "grad_norm": 1.8828125,
+      "learning_rate": 6.816711911225568e-05,
+      "loss": 0.5598,
+      "step": 5685
+    },
+    {
+      "epoch": 1.3694344163658243,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.811998775174254e-05,
+      "loss": 0.5408,
+      "step": 5690
+    },
+    {
+      "epoch": 1.3706377858002408,
+      "grad_norm": 1.78125,
+      "learning_rate": 6.807284788839642e-05,
+      "loss": 0.5276,
+      "step": 5695
+    },
+    {
+      "epoch": 1.371841155234657,
+      "grad_norm": 1.703125,
+      "learning_rate": 6.80256995974984e-05,
+      "loss": 0.5043,
+      "step": 5700
+    },
+    {
+      "epoch": 1.3730445246690735,
+      "grad_norm": 1.90625,
+      "learning_rate": 6.797854295434299e-05,
+      "loss": 0.5536,
+      "step": 5705
+    },
+    {
+      "epoch": 1.37424789410349,
+      "grad_norm": 1.671875,
+      "learning_rate": 6.793137803423807e-05,
+      "loss": 0.5534,
+      "step": 5710
+    },
+    {
+      "epoch": 1.3754512635379061,
+      "grad_norm": 1.625,
+      "learning_rate": 6.788420491250469e-05,
+      "loss": 0.5142,
+      "step": 5715
+    },
+    {
+      "epoch": 1.3766546329723224,
+      "grad_norm": 1.6875,
+      "learning_rate": 6.783702366447707e-05,
+      "loss": 0.5742,
+      "step": 5720
+    },
+    {
+      "epoch": 1.3778580024067388,
+      "grad_norm": 1.5703125,
+      "learning_rate": 6.77898343655023e-05,
+      "loss": 0.5231,
+      "step": 5725
+    },
+    {
+      "epoch": 1.3790613718411553,
+      "grad_norm": 1.640625,
+      "learning_rate": 6.774263709094048e-05,
+      "loss": 0.5276,
+      "step": 5730
+    },
+    {
+      "epoch": 1.3802647412755715,
+      "grad_norm": 1.8359375,
+      "learning_rate": 6.76954319161643e-05,
+      "loss": 0.5268,
+      "step": 5735
+    },
+    {
+      "epoch": 1.381468110709988,
+      "grad_norm": 1.5546875,
+      "learning_rate": 6.764821891655914e-05,
+      "loss": 0.5439,
+      "step": 5740
+    },
+    {
+      "epoch": 1.3826714801444044,
+      "grad_norm": 1.8125,
+      "learning_rate": 6.760099816752288e-05,
+      "loss": 0.5383,
+      "step": 5745
+    },
+    {
+      "epoch": 1.3838748495788207,
+      "grad_norm": 1.734375,
+      "learning_rate": 6.755376974446573e-05,
+      "loss": 0.5373,
+      "step": 5750
+    },
+    {
+      "epoch": 1.3850782190132371,
+      "grad_norm": 1.671875,
+      "learning_rate": 6.750653372281018e-05,
+      "loss": 0.5448,
+      "step": 5755
+    },
+    {
+      "epoch": 1.3862815884476534,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6.745929017799091e-05,
+      "loss": 0.5449,
+      "step": 5760
+    },
+    {
+      "epoch": 1.3874849578820698,
+      "grad_norm": 1.7734375,
+      "learning_rate": 6.74120391854545e-05,
+      "loss": 0.5297,
+      "step": 5765
+    },
+    {
+      "epoch": 1.388688327316486,
+      "grad_norm": 1.65625,
+      "learning_rate": 6.73647808206595e-05,
+      "loss": 0.5672,
+      "step": 5770
+    },
+    {
+      "epoch": 1.3898916967509025,
+      "grad_norm": 1.6875,
+      "learning_rate": 6.731751515907624e-05,
+      "loss": 0.5419,
+      "step": 5775
+    },
+    {
+      "epoch": 1.391095066185319,
+      "grad_norm": 1.6875,
+      "learning_rate": 6.727024227618667e-05,
+      "loss": 0.5312,
+      "step": 5780
+    },
+    {
+      "epoch": 1.3922984356197352,
+      "grad_norm": 1.8125,
+      "learning_rate": 6.72229622474843e-05,
+      "loss": 0.5624,
+      "step": 5785
+    },
+    {
+      "epoch": 1.3935018050541517,
+      "grad_norm": 1.6796875,
+      "learning_rate": 6.717567514847401e-05,
+      "loss": 0.5465,
+      "step": 5790
+    },
+    {
+      "epoch": 1.3947051744885681,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.712838105467204e-05,
+      "loss": 0.5549,
+      "step": 5795
+    },
+    {
+      "epoch": 1.3959085439229844,
+      "grad_norm": 1.5859375,
+      "learning_rate": 6.70810800416057e-05,
+      "loss": 0.537,
+      "step": 5800
+    },
+    {
+      "epoch": 1.3971119133574006,
+      "grad_norm": 1.6953125,
+      "learning_rate": 6.703377218481343e-05,
+      "loss": 0.529,
+      "step": 5805
+    },
+    {
+      "epoch": 1.398315282791817,
+      "grad_norm": 1.59375,
+      "learning_rate": 6.698645755984457e-05,
+      "loss": 0.5774,
+      "step": 5810
+    },
+    {
+      "epoch": 1.3995186522262335,
+      "grad_norm": 1.734375,
+      "learning_rate": 6.693913624225931e-05,
+      "loss": 0.5552,
+      "step": 5815
+    },
+    {
+      "epoch": 1.4007220216606497,
+      "grad_norm": 1.6328125,
+      "learning_rate": 6.689180830762847e-05,
+      "loss": 0.5401,
+      "step": 5820
+    },
+    {
+      "epoch": 1.4019253910950662,
+      "grad_norm": 1.671875,
+      "learning_rate": 6.684447383153344e-05,
+      "loss": 0.5649,
+      "step": 5825
+    },
+    {
+      "epoch": 1.4031287605294827,
+      "grad_norm": 1.734375,
+      "learning_rate": 6.679713288956611e-05,
+      "loss": 0.536,
+      "step": 5830
+    },
+    {
+      "epoch": 1.404332129963899,
+      "grad_norm": 1.734375,
+      "learning_rate": 6.674978555732864e-05,
+      "loss": 0.5612,
+      "step": 5835
+    },
+    {
+      "epoch": 1.4055354993983153,
+      "grad_norm": 1.78125,
+      "learning_rate": 6.670243191043344e-05,
+      "loss": 0.5487,
+      "step": 5840
+    },
+    {
+      "epoch": 1.4067388688327316,
+      "grad_norm": 1.5703125,
+      "learning_rate": 6.665507202450296e-05,
+      "loss": 0.5507,
+      "step": 5845
+    },
+    {
+      "epoch": 1.407942238267148,
+      "grad_norm": 1.53125,
+      "learning_rate": 6.660770597516963e-05,
+      "loss": 0.547,
+      "step": 5850
+    },
+    {
+      "epoch": 1.4091456077015643,
+      "grad_norm": 1.8046875,
+      "learning_rate": 6.656033383807577e-05,
+      "loss": 0.557,
+      "step": 5855
+    },
+    {
+      "epoch": 1.4103489771359807,
+      "grad_norm": 1.5625,
+      "learning_rate": 6.651295568887334e-05,
+      "loss": 0.5393,
+      "step": 5860
+    },
+    {
+      "epoch": 1.4115523465703972,
+      "grad_norm": 1.6796875,
+      "learning_rate": 6.646557160322395e-05,
+      "loss": 0.5572,
+      "step": 5865
+    },
+    {
+      "epoch": 1.4127557160048134,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.641818165679871e-05,
+      "loss": 0.5423,
+      "step": 5870
+    },
+    {
+      "epoch": 1.4139590854392299,
+      "grad_norm": 1.8671875,
+      "learning_rate": 6.637078592527801e-05,
+      "loss": 0.561,
+      "step": 5875
+    },
+    {
+      "epoch": 1.4151624548736463,
+      "grad_norm": 1.8515625,
+      "learning_rate": 6.632338448435156e-05,
+      "loss": 0.5859,
+      "step": 5880
+    },
+    {
+      "epoch": 1.4163658243080626,
+      "grad_norm": 1.7578125,
+      "learning_rate": 6.627597740971817e-05,
+      "loss": 0.5415,
+      "step": 5885
+    },
+    {
+      "epoch": 1.4175691937424788,
+      "grad_norm": 1.65625,
+      "learning_rate": 6.622856477708562e-05,
+      "loss": 0.5513,
+      "step": 5890
+    },
+    {
+      "epoch": 1.4187725631768953,
+      "grad_norm": 1.640625,
+      "learning_rate": 6.618114666217058e-05,
+      "loss": 0.5319,
+      "step": 5895
+    },
+    {
+      "epoch": 1.4199759326113117,
+      "grad_norm": 1.9375,
+      "learning_rate": 6.613372314069847e-05,
+      "loss": 0.5853,
+      "step": 5900
+    },
+    {
+      "epoch": 1.421179302045728,
+      "grad_norm": 1.5078125,
+      "learning_rate": 6.608629428840336e-05,
+      "loss": 0.5123,
+      "step": 5905
+    },
+    {
+      "epoch": 1.4223826714801444,
+      "grad_norm": 1.7890625,
+      "learning_rate": 6.603886018102781e-05,
+      "loss": 0.5672,
+      "step": 5910
+    },
+    {
+      "epoch": 1.4235860409145609,
+      "grad_norm": 1.6796875,
+      "learning_rate": 6.599142089432282e-05,
+      "loss": 0.5437,
+      "step": 5915
+    },
+    {
+      "epoch": 1.424789410348977,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.594397650404759e-05,
+      "loss": 0.5595,
+      "step": 5920
+    },
+    {
+      "epoch": 1.4259927797833936,
+      "grad_norm": 1.6328125,
+      "learning_rate": 6.589652708596952e-05,
+      "loss": 0.5239,
+      "step": 5925
+    },
+    {
+      "epoch": 1.4271961492178098,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.584907271586402e-05,
+      "loss": 0.5637,
+      "step": 5930
+    },
+    {
+      "epoch": 1.4283995186522263,
+      "grad_norm": 1.8828125,
+      "learning_rate": 6.580161346951442e-05,
+      "loss": 0.5361,
+      "step": 5935
+    },
+    {
+      "epoch": 1.4296028880866425,
+      "grad_norm": 1.7890625,
+      "learning_rate": 6.575414942271184e-05,
+      "loss": 0.5286,
+      "step": 5940
+    },
+    {
+      "epoch": 1.430806257521059,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.570668065125504e-05,
+      "loss": 0.5967,
+      "step": 5945
+    },
+    {
+      "epoch": 1.4320096269554754,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.565920723095035e-05,
+      "loss": 0.536,
+      "step": 5950
+    },
+    {
+      "epoch": 1.4332129963898916,
+      "grad_norm": 1.796875,
+      "learning_rate": 6.561172923761153e-05,
+      "loss": 0.5393,
+      "step": 5955
+    },
+    {
+      "epoch": 1.434416365824308,
+      "grad_norm": 1.7734375,
+      "learning_rate": 6.556424674705962e-05,
+      "loss": 0.5651,
+      "step": 5960
+    },
+    {
+      "epoch": 1.4356197352587245,
+      "grad_norm": 1.78125,
+      "learning_rate": 6.551675983512283e-05,
+      "loss": 0.5558,
+      "step": 5965
+    },
+    {
+      "epoch": 1.4368231046931408,
+      "grad_norm": 1.7890625,
+      "learning_rate": 6.546926857763648e-05,
+      "loss": 0.5897,
+      "step": 5970
+    },
+    {
+      "epoch": 1.4380264741275572,
+      "grad_norm": 1.6484375,
+      "learning_rate": 6.54217730504428e-05,
+      "loss": 0.5656,
+      "step": 5975
+    },
+    {
+      "epoch": 1.4392298435619735,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.537427332939083e-05,
+      "loss": 0.515,
+      "step": 5980
+    },
+    {
+      "epoch": 1.44043321299639,
+      "grad_norm": 1.6875,
+      "learning_rate": 6.532676949033634e-05,
+      "loss": 0.5269,
+      "step": 5985
+    },
+    {
+      "epoch": 1.4416365824308062,
+      "grad_norm": 1.671875,
+      "learning_rate": 6.527926160914163e-05,
+      "loss": 0.5277,
+      "step": 5990
+    },
+    {
+      "epoch": 1.4428399518652226,
+      "grad_norm": 1.5859375,
+      "learning_rate": 6.523174976167547e-05,
+      "loss": 0.5545,
+      "step": 5995
+    },
+    {
+      "epoch": 1.444043321299639,
+      "grad_norm": 1.6484375,
+      "learning_rate": 6.518423402381303e-05,
+      "loss": 0.5319,
+      "step": 6000
+    },
+    {
+      "epoch": 1.444043321299639,
+      "eval_loss": 0.46789681911468506,
+      "eval_runtime": 2.6551,
+      "eval_samples_per_second": 75.327,
+      "eval_steps_per_second": 75.327,
+      "step": 6000
+    },
+    {
+      "epoch": 1.4452466907340553,
+      "grad_norm": 1.6484375,
+      "learning_rate": 6.513671447143558e-05,
+      "loss": 0.5483,
+      "step": 6005
+    },
+    {
+      "epoch": 1.4464500601684718,
+      "grad_norm": 1.640625,
+      "learning_rate": 6.508919118043056e-05,
+      "loss": 0.5712,
+      "step": 6010
+    },
+    {
+      "epoch": 1.447653429602888,
+      "grad_norm": 1.7890625,
+      "learning_rate": 6.504166422669137e-05,
+      "loss": 0.5172,
+      "step": 6015
+    },
+    {
+      "epoch": 1.4488567990373045,
+      "grad_norm": 1.625,
+      "learning_rate": 6.499413368611722e-05,
+      "loss": 0.5448,
+      "step": 6020
+    },
+    {
+      "epoch": 1.4500601684717207,
+      "grad_norm": 1.765625,
+      "learning_rate": 6.49465996346131e-05,
+      "loss": 0.5728,
+      "step": 6025
+    },
+    {
+      "epoch": 1.4512635379061372,
+      "grad_norm": 1.671875,
+      "learning_rate": 6.489906214808959e-05,
+      "loss": 0.5545,
+      "step": 6030
+    },
+    {
+      "epoch": 1.4524669073405536,
+      "grad_norm": 1.6015625,
+      "learning_rate": 6.485152130246271e-05,
+      "loss": 0.5488,
+      "step": 6035
+    },
+    {
+      "epoch": 1.4536702767749698,
+      "grad_norm": 1.6015625,
+      "learning_rate": 6.480397717365393e-05,
+      "loss": 0.5566,
+      "step": 6040
+    },
+    {
+      "epoch": 1.4548736462093863,
+      "grad_norm": 1.7265625,
+      "learning_rate": 6.47564298375899e-05,
+      "loss": 0.563,
+      "step": 6045
+    },
+    {
+      "epoch": 1.4560770156438028,
+      "grad_norm": 1.734375,
+      "learning_rate": 6.470887937020238e-05,
+      "loss": 0.5657,
+      "step": 6050
+    },
+    {
+      "epoch": 1.457280385078219,
+      "grad_norm": 1.609375,
+      "learning_rate": 6.466132584742819e-05,
+      "loss": 0.5372,
+      "step": 6055
+    },
+    {
+      "epoch": 1.4584837545126355,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.4613769345209e-05,
+      "loss": 0.5713,
+      "step": 6060
+    },
+    {
+      "epoch": 1.4596871239470517,
+      "grad_norm": 1.671875,
+      "learning_rate": 6.456620993949122e-05,
+      "loss": 0.5335,
+      "step": 6065
+    },
+    {
+      "epoch": 1.4608904933814681,
+      "grad_norm": 1.6328125,
+      "learning_rate": 6.451864770622593e-05,
+      "loss": 0.5202,
+      "step": 6070
+    },
+    {
+      "epoch": 1.4620938628158844,
+      "grad_norm": 1.8203125,
+      "learning_rate": 6.44710827213687e-05,
+      "loss": 0.5149,
+      "step": 6075
+    },
+    {
+      "epoch": 1.4632972322503008,
+      "grad_norm": 1.5859375,
+      "learning_rate": 6.442351506087949e-05,
+      "loss": 0.5188,
+      "step": 6080
+    },
+    {
+      "epoch": 1.4645006016847173,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.437594480072255e-05,
+      "loss": 0.5472,
+      "step": 6085
+    },
+    {
+      "epoch": 1.4657039711191335,
+      "grad_norm": 1.6875,
+      "learning_rate": 6.432837201686631e-05,
+      "loss": 0.5382,
+      "step": 6090
+    },
+    {
+      "epoch": 1.46690734055355,
+      "grad_norm": 1.9609375,
+      "learning_rate": 6.428079678528317e-05,
+      "loss": 0.5424,
+      "step": 6095
+    },
+    {
+      "epoch": 1.4681107099879662,
+      "grad_norm": 1.6328125,
+      "learning_rate": 6.423321918194947e-05,
+      "loss": 0.5614,
+      "step": 6100
+    },
+    {
+      "epoch": 1.4693140794223827,
+      "grad_norm": 1.71875,
+      "learning_rate": 6.418563928284533e-05,
+      "loss": 0.5205,
+      "step": 6105
+    },
+    {
+      "epoch": 1.470517448856799,
+      "grad_norm": 1.7734375,
+      "learning_rate": 6.413805716395453e-05,
+      "loss": 0.5253,
+      "step": 6110
+    },
+    {
+      "epoch": 1.4717208182912154,
+      "grad_norm": 1.5546875,
+      "learning_rate": 6.409047290126444e-05,
+      "loss": 0.5575,
+      "step": 6115
+    },
+    {
+      "epoch": 1.4729241877256318,
+      "grad_norm": 1.7265625,
+      "learning_rate": 6.404288657076581e-05,
+      "loss": 0.5642,
+      "step": 6120
+    },
+    {
+      "epoch": 1.474127557160048,
+      "grad_norm": 1.6796875,
+      "learning_rate": 6.399529824845267e-05,
+      "loss": 0.5647,
+      "step": 6125
+    },
+    {
+      "epoch": 1.4753309265944645,
+      "grad_norm": 1.640625,
+      "learning_rate": 6.39477080103223e-05,
+      "loss": 0.5488,
+      "step": 6130
+    },
+    {
+      "epoch": 1.476534296028881,
+      "grad_norm": 1.65625,
+      "learning_rate": 6.390011593237498e-05,
+      "loss": 0.5288,
+      "step": 6135
+    },
+    {
+      "epoch": 1.4777376654632972,
+      "grad_norm": 1.6328125,
+      "learning_rate": 6.385252209061396e-05,
+      "loss": 0.5336,
+      "step": 6140
+    },
+    {
+      "epoch": 1.4789410348977137,
+      "grad_norm": 1.578125,
+      "learning_rate": 6.380492656104528e-05,
+      "loss": 0.5341,
+      "step": 6145
+    },
+    {
+      "epoch": 1.48014440433213,
+      "grad_norm": 1.6484375,
+      "learning_rate": 6.375732941967771e-05,
+      "loss": 0.5528,
+      "step": 6150
+    },
+    {
+      "epoch": 1.4813477737665464,
+      "grad_norm": 1.8125,
+      "learning_rate": 6.370973074252259e-05,
+      "loss": 0.5265,
+      "step": 6155
+    },
+    {
+      "epoch": 1.4825511432009626,
+      "grad_norm": 1.71875,
+      "learning_rate": 6.366213060559366e-05,
+      "loss": 0.5381,
+      "step": 6160
+    },
+    {
+      "epoch": 1.483754512635379,
+      "grad_norm": 1.65625,
+      "learning_rate": 6.361452908490706e-05,
+      "loss": 0.5573,
+      "step": 6165
+    },
+    {
+      "epoch": 1.4849578820697955,
+      "grad_norm": 1.6796875,
+      "learning_rate": 6.356692625648108e-05,
+      "loss": 0.5663,
+      "step": 6170
+    },
+    {
+      "epoch": 1.4861612515042117,
+      "grad_norm": 1.640625,
+      "learning_rate": 6.351932219633617e-05,
+      "loss": 0.5554,
+      "step": 6175
+    },
+    {
+      "epoch": 1.4873646209386282,
+      "grad_norm": 1.5546875,
+      "learning_rate": 6.347171698049466e-05,
+      "loss": 0.54,
+      "step": 6180
+    },
+    {
+      "epoch": 1.4885679903730447,
+      "grad_norm": 1.671875,
+      "learning_rate": 6.34241106849808e-05,
+      "loss": 0.5344,
+      "step": 6185
+    },
+    {
+      "epoch": 1.489771359807461,
+      "grad_norm": 1.640625,
+      "learning_rate": 6.337650338582055e-05,
+      "loss": 0.552,
+      "step": 6190
+    },
+    {
+      "epoch": 1.4909747292418771,
+      "grad_norm": 1.609375,
+      "learning_rate": 6.33288951590414e-05,
+      "loss": 0.5318,
+      "step": 6195
+    },
+    {
+      "epoch": 1.4921780986762936,
+      "grad_norm": 1.78125,
+      "learning_rate": 6.328128608067245e-05,
+      "loss": 0.5585,
+      "step": 6200
+    },
+    {
+      "epoch": 1.49338146811071,
+      "grad_norm": 1.8125,
+      "learning_rate": 6.323367622674404e-05,
+      "loss": 0.5456,
+      "step": 6205
+    },
+    {
+      "epoch": 1.4945848375451263,
+      "grad_norm": 1.546875,
+      "learning_rate": 6.318606567328783e-05,
+      "loss": 0.5315,
+      "step": 6210
+    },
+    {
+      "epoch": 1.4957882069795427,
+      "grad_norm": 2.28125,
+      "learning_rate": 6.313845449633651e-05,
+      "loss": 0.5548,
+      "step": 6215
+    },
+    {
+      "epoch": 1.4969915764139592,
+      "grad_norm": 1.71875,
+      "learning_rate": 6.309084277192387e-05,
+      "loss": 0.5095,
+      "step": 6220
+    },
+    {
+      "epoch": 1.4981949458483754,
+      "grad_norm": 1.671875,
+      "learning_rate": 6.304323057608452e-05,
+      "loss": 0.5339,
+      "step": 6225
+    },
+    {
+      "epoch": 1.4993983152827919,
+      "grad_norm": 1.6484375,
+      "learning_rate": 6.299561798485382e-05,
+      "loss": 0.5374,
+      "step": 6230
+    },
+    {
+      "epoch": 1.5006016847172083,
+      "grad_norm": 1.7265625,
+      "learning_rate": 6.294800507426777e-05,
+      "loss": 0.5234,
+      "step": 6235
+    },
+    {
+      "epoch": 1.5018050541516246,
+      "grad_norm": 1.75,
+      "learning_rate": 6.290039192036287e-05,
+      "loss": 0.5835,
+      "step": 6240
+    },
+    {
+      "epoch": 1.5030084235860408,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6.285277859917603e-05,
+      "loss": 0.5291,
+      "step": 6245
+    },
+    {
+      "epoch": 1.5042117930204573,
+      "grad_norm": 1.59375,
+      "learning_rate": 6.280516518674442e-05,
+      "loss": 0.5387,
+      "step": 6250
+    },
+    {
+      "epoch": 1.5054151624548737,
+      "grad_norm": 1.78125,
+      "learning_rate": 6.275755175910535e-05,
+      "loss": 0.5424,
+      "step": 6255
+    },
+    {
+      "epoch": 1.50661853188929,
+      "grad_norm": 1.7734375,
+      "learning_rate": 6.27099383922961e-05,
+      "loss": 0.5523,
+      "step": 6260
+    },
+    {
+      "epoch": 1.5078219013237064,
+      "grad_norm": 1.71875,
+      "learning_rate": 6.266232516235398e-05,
+      "loss": 0.5244,
+      "step": 6265
+    },
+    {
+      "epoch": 1.5090252707581229,
+      "grad_norm": 1.6015625,
+      "learning_rate": 6.2614712145316e-05,
+      "loss": 0.5452,
+      "step": 6270
+    },
+    {
+      "epoch": 1.510228640192539,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.25670994172188e-05,
+      "loss": 0.5606,
+      "step": 6275
+    },
+    {
+      "epoch": 1.5114320096269553,
+      "grad_norm": 1.671875,
+      "learning_rate": 6.251948705409863e-05,
+      "loss": 0.542,
+      "step": 6280
+    },
+    {
+      "epoch": 1.5126353790613718,
+      "grad_norm": 1.734375,
+      "learning_rate": 6.24718751319911e-05,
+      "loss": 0.5373,
+      "step": 6285
+    },
+    {
+      "epoch": 1.5138387484957883,
+      "grad_norm": 1.75,
+      "learning_rate": 6.242426372693119e-05,
+      "loss": 0.5523,
+      "step": 6290
+    },
+    {
+      "epoch": 1.5150421179302045,
+      "grad_norm": 1.90625,
+      "learning_rate": 6.237665291495294e-05,
+      "loss": 0.5456,
+      "step": 6295
+    },
+    {
+      "epoch": 1.516245487364621,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.232904277208953e-05,
+      "loss": 0.5438,
+      "step": 6300
+    },
+    {
+      "epoch": 1.5174488567990374,
+      "grad_norm": 1.7578125,
+      "learning_rate": 6.228143337437309e-05,
+      "loss": 0.573,
+      "step": 6305
+    },
+    {
+      "epoch": 1.5186522262334536,
+      "grad_norm": 1.7734375,
+      "learning_rate": 6.223382479783445e-05,
+      "loss": 0.5395,
+      "step": 6310
+    },
+    {
+      "epoch": 1.5198555956678699,
+      "grad_norm": 1.78125,
+      "learning_rate": 6.218621711850323e-05,
+      "loss": 0.551,
+      "step": 6315
+    },
+    {
+      "epoch": 1.5210589651022866,
+      "grad_norm": 1.796875,
+      "learning_rate": 6.213861041240763e-05,
+      "loss": 0.5512,
+      "step": 6320
+    },
+    {
+      "epoch": 1.5222623345367028,
+      "grad_norm": 1.8359375,
+      "learning_rate": 6.209100475557418e-05,
+      "loss": 0.5605,
+      "step": 6325
+    },
+    {
+      "epoch": 1.523465703971119,
+      "grad_norm": 1.703125,
+      "learning_rate": 6.204340022402783e-05,
+      "loss": 0.5225,
+      "step": 6330
+    },
+    {
+      "epoch": 1.5246690734055355,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6.199579689379172e-05,
+      "loss": 0.5727,
+      "step": 6335
+    },
+    {
+      "epoch": 1.525872442839952,
+      "grad_norm": 1.703125,
+      "learning_rate": 6.194819484088706e-05,
+      "loss": 0.5537,
+      "step": 6340
+    },
+    {
+      "epoch": 1.5270758122743682,
+      "grad_norm": 1.7734375,
+      "learning_rate": 6.1900594141333e-05,
+      "loss": 0.5679,
+      "step": 6345
+    },
+    {
+      "epoch": 1.5282791817087846,
+      "grad_norm": 1.671875,
+      "learning_rate": 6.185299487114657e-05,
+      "loss": 0.5396,
+      "step": 6350
+    },
+    {
+      "epoch": 1.529482551143201,
+      "grad_norm": 1.8671875,
+      "learning_rate": 6.180539710634246e-05,
+      "loss": 0.5241,
+      "step": 6355
+    },
+    {
+      "epoch": 1.5306859205776173,
+      "grad_norm": 1.6171875,
+      "learning_rate": 6.175780092293305e-05,
+      "loss": 0.5411,
+      "step": 6360
+    },
+    {
+      "epoch": 1.5318892900120336,
+      "grad_norm": 1.703125,
+      "learning_rate": 6.171020639692808e-05,
+      "loss": 0.5657,
+      "step": 6365
+    },
+    {
+      "epoch": 1.53309265944645,
+      "grad_norm": 1.625,
+      "learning_rate": 6.166261360433472e-05,
+      "loss": 0.519,
+      "step": 6370
+    },
+    {
+      "epoch": 1.5342960288808665,
+      "grad_norm": 1.7265625,
+      "learning_rate": 6.161502262115736e-05,
+      "loss": 0.5226,
+      "step": 6375
+    },
+    {
+      "epoch": 1.5354993983152827,
+      "grad_norm": 1.625,
+      "learning_rate": 6.15674335233975e-05,
+      "loss": 0.5691,
+      "step": 6380
+    },
+    {
+      "epoch": 1.5367027677496992,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6.151984638705359e-05,
+      "loss": 0.5337,
+      "step": 6385
+    },
+    {
+      "epoch": 1.5379061371841156,
+      "grad_norm": 1.625,
+      "learning_rate": 6.147226128812099e-05,
+      "loss": 0.522,
+      "step": 6390
+    },
+    {
+      "epoch": 1.5391095066185319,
+      "grad_norm": 1.6015625,
+      "learning_rate": 6.142467830259183e-05,
+      "loss": 0.549,
+      "step": 6395
+    },
+    {
+      "epoch": 1.5403128760529483,
+      "grad_norm": 1.6875,
+      "learning_rate": 6.137709750645475e-05,
+      "loss": 0.5509,
+      "step": 6400
+    },
+    {
+      "epoch": 1.5415162454873648,
+      "grad_norm": 1.671875,
+      "learning_rate": 6.132951897569506e-05,
+      "loss": 0.524,
+      "step": 6405
+    },
+    {
+      "epoch": 1.542719614921781,
+      "grad_norm": 1.65625,
+      "learning_rate": 6.128194278629432e-05,
+      "loss": 0.5388,
+      "step": 6410
+    },
+    {
+      "epoch": 1.5439229843561972,
+      "grad_norm": 1.78125,
+      "learning_rate": 6.12343690142304e-05,
+      "loss": 0.531,
+      "step": 6415
+    },
+    {
+      "epoch": 1.5451263537906137,
+      "grad_norm": 1.7265625,
+      "learning_rate": 6.11867977354773e-05,
+      "loss": 0.5653,
+      "step": 6420
+    },
+    {
+      "epoch": 1.5463297232250302,
+      "grad_norm": 1.6484375,
+      "learning_rate": 6.113922902600507e-05,
+      "loss": 0.57,
+      "step": 6425
+    },
+    {
+      "epoch": 1.5475330926594464,
+      "grad_norm": 1.609375,
+      "learning_rate": 6.10916629617796e-05,
+      "loss": 0.5529,
+      "step": 6430
+    },
+    {
+      "epoch": 1.5487364620938628,
+      "grad_norm": 1.546875,
+      "learning_rate": 6.104409961876263e-05,
+      "loss": 0.5196,
+      "step": 6435
+    },
+    {
+      "epoch": 1.5499398315282793,
+      "grad_norm": 1.8203125,
+      "learning_rate": 6.0996539072911456e-05,
+      "loss": 0.555,
+      "step": 6440
+    },
+    {
+      "epoch": 1.5511432009626955,
+      "grad_norm": 1.6328125,
+      "learning_rate": 6.0948981400178985e-05,
+      "loss": 0.511,
+      "step": 6445
+    },
+    {
+      "epoch": 1.5523465703971118,
+      "grad_norm": 1.71875,
+      "learning_rate": 6.090142667651353e-05,
+      "loss": 0.5294,
+      "step": 6450
+    },
+    {
+      "epoch": 1.5535499398315282,
+      "grad_norm": 1.6328125,
+      "learning_rate": 6.085387497785866e-05,
+      "loss": 0.5526,
+      "step": 6455
+    },
+    {
+      "epoch": 1.5547533092659447,
+      "grad_norm": 1.78125,
+      "learning_rate": 6.080632638015313e-05,
+      "loss": 0.5515,
+      "step": 6460
+    },
+    {
+      "epoch": 1.555956678700361,
+      "grad_norm": 1.7734375,
+      "learning_rate": 6.0758780959330756e-05,
+      "loss": 0.5441,
+      "step": 6465
+    },
+    {
+      "epoch": 1.5571600481347774,
+      "grad_norm": 1.7578125,
+      "learning_rate": 6.0711238791320256e-05,
+      "loss": 0.5897,
+      "step": 6470
+    },
+    {
+      "epoch": 1.5583634175691938,
+      "grad_norm": 1.75,
+      "learning_rate": 6.066369995204517e-05,
+      "loss": 0.5655,
+      "step": 6475
+    },
+    {
+      "epoch": 1.55956678700361,
+      "grad_norm": 1.734375,
+      "learning_rate": 6.061616451742371e-05,
+      "loss": 0.5299,
+      "step": 6480
+    },
+    {
+      "epoch": 1.5607701564380265,
+      "grad_norm": 1.6328125,
+      "learning_rate": 6.056863256336866e-05,
+      "loss": 0.5418,
+      "step": 6485
+    },
+    {
+      "epoch": 1.561973525872443,
+      "grad_norm": 1.75,
+      "learning_rate": 6.0521104165787255e-05,
+      "loss": 0.5359,
+      "step": 6490
+    },
+    {
+      "epoch": 1.5631768953068592,
+      "grad_norm": 1.734375,
+      "learning_rate": 6.047357940058102e-05,
+      "loss": 0.5489,
+      "step": 6495
+    },
+    {
+      "epoch": 1.5643802647412755,
+      "grad_norm": 1.8203125,
+      "learning_rate": 6.042605834364572e-05,
+      "loss": 0.5236,
+      "step": 6500
+    },
+    {
+      "epoch": 1.5643802647412755,
+      "eval_loss": 0.46134811639785767,
+      "eval_runtime": 2.6539,
+      "eval_samples_per_second": 75.362,
+      "eval_steps_per_second": 75.362,
+      "step": 6500
+    },
+    {
+      "epoch": 1.565583634175692,
+      "grad_norm": 1.640625,
+      "learning_rate": 6.037854107087116e-05,
+      "loss": 0.5509,
+      "step": 6505
+    },
+    {
+      "epoch": 1.5667870036101084,
+      "grad_norm": 1.71875,
+      "learning_rate": 6.033102765814113e-05,
+      "loss": 0.5156,
+      "step": 6510
+    },
+    {
+      "epoch": 1.5679903730445246,
+      "grad_norm": 1.703125,
+      "learning_rate": 6.028351818133324e-05,
+      "loss": 0.5783,
+      "step": 6515
+    },
+    {
+      "epoch": 1.569193742478941,
+      "grad_norm": 1.8203125,
+      "learning_rate": 6.023601271631882e-05,
+      "loss": 0.5482,
+      "step": 6520
+    },
+    {
+      "epoch": 1.5703971119133575,
+      "grad_norm": 1.6171875,
+      "learning_rate": 6.018851133896279e-05,
+      "loss": 0.5538,
+      "step": 6525
+    },
+    {
+      "epoch": 1.5716004813477737,
+      "grad_norm": 1.75,
+      "learning_rate": 6.014101412512352e-05,
+      "loss": 0.5254,
+      "step": 6530
+    },
+    {
+      "epoch": 1.57280385078219,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.009352115065279e-05,
+      "loss": 0.5402,
+      "step": 6535
+    },
+    {
+      "epoch": 1.5740072202166067,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.004603249139556e-05,
+      "loss": 0.518,
+      "step": 6540
+    },
+    {
+      "epoch": 1.575210589651023,
+      "grad_norm": 1.671875,
+      "learning_rate": 5.999854822318991e-05,
+      "loss": 0.5331,
+      "step": 6545
+    },
+    {
+      "epoch": 1.5764139590854391,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.995106842186692e-05,
+      "loss": 0.5737,
+      "step": 6550
+    },
+    {
+      "epoch": 1.5776173285198556,
+      "grad_norm": 1.515625,
+      "learning_rate": 5.9903593163250524e-05,
+      "loss": 0.5486,
+      "step": 6555
+    },
+    {
+      "epoch": 1.578820697954272,
+      "grad_norm": 1.5859375,
+      "learning_rate": 5.985612252315738e-05,
+      "loss": 0.5431,
+      "step": 6560
+    },
+    {
+      "epoch": 1.5800240673886883,
+      "grad_norm": 1.8125,
+      "learning_rate": 5.980865657739683e-05,
+      "loss": 0.5495,
+      "step": 6565
+    },
+    {
+      "epoch": 1.5812274368231047,
+      "grad_norm": 1.7265625,
+      "learning_rate": 5.976119540177067e-05,
+      "loss": 0.5227,
+      "step": 6570
+    },
+    {
+      "epoch": 1.5824308062575212,
+      "grad_norm": 1.65625,
+      "learning_rate": 5.9713739072073096e-05,
+      "loss": 0.5472,
+      "step": 6575
+    },
+    {
+      "epoch": 1.5836341756919374,
+      "grad_norm": 1.6171875,
+      "learning_rate": 5.966628766409056e-05,
+      "loss": 0.5873,
+      "step": 6580
+    },
+    {
+      "epoch": 1.5848375451263537,
+      "grad_norm": 1.703125,
+      "learning_rate": 5.961884125360166e-05,
+      "loss": 0.5349,
+      "step": 6585
+    },
+    {
+      "epoch": 1.5860409145607701,
+      "grad_norm": 1.46875,
+      "learning_rate": 5.957139991637701e-05,
+      "loss": 0.5293,
+      "step": 6590
+    },
+    {
+      "epoch": 1.5872442839951866,
+      "grad_norm": 1.7265625,
+      "learning_rate": 5.9523963728179093e-05,
+      "loss": 0.5599,
+      "step": 6595
+    },
+    {
+      "epoch": 1.5884476534296028,
+      "grad_norm": 1.71875,
+      "learning_rate": 5.947653276476223e-05,
+      "loss": 0.5388,
+      "step": 6600
+    },
+    {
+      "epoch": 1.5896510228640193,
+      "grad_norm": 1.8203125,
+      "learning_rate": 5.942910710187236e-05,
+      "loss": 0.5362,
+      "step": 6605
+    },
+    {
+      "epoch": 1.5908543922984357,
+      "grad_norm": 1.671875,
+      "learning_rate": 5.9381686815246954e-05,
+      "loss": 0.496,
+      "step": 6610
+    },
+    {
+      "epoch": 1.592057761732852,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.933427198061491e-05,
+      "loss": 0.5428,
+      "step": 6615
+    },
+    {
+      "epoch": 1.5932611311672682,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.9286862673696416e-05,
+      "loss": 0.5177,
+      "step": 6620
+    },
+    {
+      "epoch": 1.5944645006016849,
+      "grad_norm": 1.8359375,
+      "learning_rate": 5.9239458970202826e-05,
+      "loss": 0.5348,
+      "step": 6625
+    },
+    {
+      "epoch": 1.595667870036101,
+      "grad_norm": 1.6953125,
+      "learning_rate": 5.9192060945836565e-05,
+      "loss": 0.4923,
+      "step": 6630
+    },
+    {
+      "epoch": 1.5968712394705173,
+      "grad_norm": 1.65625,
+      "learning_rate": 5.9144668676290955e-05,
+      "loss": 0.5447,
+      "step": 6635
+    },
+    {
+      "epoch": 1.5980746089049338,
+      "grad_norm": 1.625,
+      "learning_rate": 5.909728223725017e-05,
+      "loss": 0.5258,
+      "step": 6640
+    },
+    {
+      "epoch": 1.5992779783393503,
+      "grad_norm": 1.6015625,
+      "learning_rate": 5.9049901704389036e-05,
+      "loss": 0.5539,
+      "step": 6645
+    },
+    {
+      "epoch": 1.6004813477737665,
+      "grad_norm": 1.8359375,
+      "learning_rate": 5.9002527153372964e-05,
+      "loss": 0.5423,
+      "step": 6650
+    },
+    {
+      "epoch": 1.601684717208183,
+      "grad_norm": 1.71875,
+      "learning_rate": 5.8955158659857795e-05,
+      "loss": 0.5508,
+      "step": 6655
+    },
+    {
+      "epoch": 1.6028880866425994,
+      "grad_norm": 1.734375,
+      "learning_rate": 5.8907796299489736e-05,
+      "loss": 0.5579,
+      "step": 6660
+    },
+    {
+      "epoch": 1.6040914560770156,
+      "grad_norm": 1.5625,
+      "learning_rate": 5.886044014790516e-05,
+      "loss": 0.5439,
+      "step": 6665
+    },
+    {
+      "epoch": 1.6052948255114319,
+      "grad_norm": 1.7734375,
+      "learning_rate": 5.8813090280730544e-05,
+      "loss": 0.5171,
+      "step": 6670
+    },
+    {
+      "epoch": 1.6064981949458483,
+      "grad_norm": 1.6171875,
+      "learning_rate": 5.876574677358232e-05,
+      "loss": 0.5517,
+      "step": 6675
+    },
+    {
+      "epoch": 1.6077015643802648,
+      "grad_norm": 1.8046875,
+      "learning_rate": 5.871840970206677e-05,
+      "loss": 0.5603,
+      "step": 6680
+    },
+    {
+      "epoch": 1.608904933814681,
+      "grad_norm": 1.71875,
+      "learning_rate": 5.8671079141779896e-05,
+      "loss": 0.5159,
+      "step": 6685
+    },
+    {
+      "epoch": 1.6101083032490975,
+      "grad_norm": 1.6953125,
+      "learning_rate": 5.8623755168307306e-05,
+      "loss": 0.5102,
+      "step": 6690
+    },
+    {
+      "epoch": 1.611311672683514,
+      "grad_norm": 1.6953125,
+      "learning_rate": 5.857643785722407e-05,
+      "loss": 0.5499,
+      "step": 6695
+    },
+    {
+      "epoch": 1.6125150421179302,
+      "grad_norm": 1.8125,
+      "learning_rate": 5.8529127284094636e-05,
+      "loss": 0.5489,
+      "step": 6700
+    },
+    {
+      "epoch": 1.6137184115523464,
+      "grad_norm": 1.6171875,
+      "learning_rate": 5.848182352447271e-05,
+      "loss": 0.5285,
+      "step": 6705
+    },
+    {
+      "epoch": 1.614921780986763,
+      "grad_norm": 1.7109375,
+      "learning_rate": 5.843452665390107e-05,
+      "loss": 0.5249,
+      "step": 6710
+    },
+    {
+      "epoch": 1.6161251504211793,
+      "grad_norm": 1.6953125,
+      "learning_rate": 5.838723674791153e-05,
+      "loss": 0.53,
+      "step": 6715
+    },
+    {
+      "epoch": 1.6173285198555956,
+      "grad_norm": 1.625,
+      "learning_rate": 5.833995388202477e-05,
+      "loss": 0.5461,
+      "step": 6720
+    },
+    {
+      "epoch": 1.618531889290012,
+      "grad_norm": 1.640625,
+      "learning_rate": 5.829267813175021e-05,
+      "loss": 0.5346,
+      "step": 6725
+    },
+    {
+      "epoch": 1.6197352587244285,
+      "grad_norm": 1.65625,
+      "learning_rate": 5.8245409572585945e-05,
+      "loss": 0.5415,
+      "step": 6730
+    },
+    {
+      "epoch": 1.6209386281588447,
+      "grad_norm": 1.640625,
+      "learning_rate": 5.819814828001857e-05,
+      "loss": 0.5267,
+      "step": 6735
+    },
+    {
+      "epoch": 1.6221419975932612,
+      "grad_norm": 1.6875,
+      "learning_rate": 5.815089432952304e-05,
+      "loss": 0.5565,
+      "step": 6740
+    },
+    {
+      "epoch": 1.6233453670276776,
+      "grad_norm": 1.703125,
+      "learning_rate": 5.810364779656263e-05,
+      "loss": 0.5342,
+      "step": 6745
+    },
+    {
+      "epoch": 1.6245487364620939,
+      "grad_norm": 1.7578125,
+      "learning_rate": 5.8056408756588765e-05,
+      "loss": 0.5535,
+      "step": 6750
+    },
+    {
+      "epoch": 1.62575210589651,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.800917728504089e-05,
+      "loss": 0.5675,
+      "step": 6755
+    },
+    {
+      "epoch": 1.6269554753309265,
+      "grad_norm": 1.828125,
+      "learning_rate": 5.796195345734634e-05,
+      "loss": 0.5713,
+      "step": 6760
+    },
+    {
+      "epoch": 1.628158844765343,
+      "grad_norm": 1.7265625,
+      "learning_rate": 5.7914737348920306e-05,
+      "loss": 0.5212,
+      "step": 6765
+    },
+    {
+      "epoch": 1.6293622141997592,
+      "grad_norm": 1.6640625,
+      "learning_rate": 5.786752903516559e-05,
+      "loss": 0.5071,
+      "step": 6770
+    },
+    {
+      "epoch": 1.6305655836341757,
+      "grad_norm": 1.703125,
+      "learning_rate": 5.782032859147257e-05,
+      "loss": 0.5464,
+      "step": 6775
+    },
+    {
+      "epoch": 1.6317689530685922,
+      "grad_norm": 1.8828125,
+      "learning_rate": 5.777313609321908e-05,
+      "loss": 0.5345,
+      "step": 6780
+    },
+    {
+      "epoch": 1.6329723225030084,
+      "grad_norm": 1.734375,
+      "learning_rate": 5.772595161577023e-05,
+      "loss": 0.5479,
+      "step": 6785
+    },
+    {
+      "epoch": 1.6341756919374246,
+      "grad_norm": 1.7109375,
+      "learning_rate": 5.767877523447831e-05,
+      "loss": 0.5244,
+      "step": 6790
+    },
+    {
+      "epoch": 1.6353790613718413,
+      "grad_norm": 1.71875,
+      "learning_rate": 5.7631607024682735e-05,
+      "loss": 0.5496,
+      "step": 6795
+    },
+    {
+      "epoch": 1.6365824308062575,
+      "grad_norm": 1.6484375,
+      "learning_rate": 5.758444706170984e-05,
+      "loss": 0.5478,
+      "step": 6800
+    },
+    {
+      "epoch": 1.6377858002406738,
+      "grad_norm": 1.59375,
+      "learning_rate": 5.7537295420872764e-05,
+      "loss": 0.5553,
+      "step": 6805
+    },
+    {
+      "epoch": 1.6389891696750902,
+      "grad_norm": 1.515625,
+      "learning_rate": 5.7490152177471396e-05,
+      "loss": 0.506,
+      "step": 6810
+    },
+    {
+      "epoch": 1.6401925391095067,
+      "grad_norm": 1.6484375,
+      "learning_rate": 5.744301740679222e-05,
+      "loss": 0.5151,
+      "step": 6815
+    },
+    {
+      "epoch": 1.641395908543923,
+      "grad_norm": 1.5,
+      "learning_rate": 5.739589118410812e-05,
+      "loss": 0.5418,
+      "step": 6820
+    },
+    {
+      "epoch": 1.6425992779783394,
+      "grad_norm": 1.640625,
+      "learning_rate": 5.734877358467841e-05,
+      "loss": 0.5289,
+      "step": 6825
+    },
+    {
+      "epoch": 1.6438026474127558,
+      "grad_norm": 1.7734375,
+      "learning_rate": 5.730166468374861e-05,
+      "loss": 0.5453,
+      "step": 6830
+    },
+    {
+      "epoch": 1.645006016847172,
+      "grad_norm": 1.703125,
+      "learning_rate": 5.725456455655032e-05,
+      "loss": 0.5122,
+      "step": 6835
+    },
+    {
+      "epoch": 1.6462093862815883,
+      "grad_norm": 1.5859375,
+      "learning_rate": 5.7207473278301156e-05,
+      "loss": 0.5027,
+      "step": 6840
+    },
+    {
+      "epoch": 1.6474127557160048,
+      "grad_norm": 1.640625,
+      "learning_rate": 5.716039092420459e-05,
+      "loss": 0.5548,
+      "step": 6845
+    },
+    {
+      "epoch": 1.6486161251504212,
+      "grad_norm": 1.6484375,
+      "learning_rate": 5.711331756944984e-05,
+      "loss": 0.5339,
+      "step": 6850
+    },
+    {
+      "epoch": 1.6498194945848375,
+      "grad_norm": 1.8203125,
+      "learning_rate": 5.706625328921179e-05,
+      "loss": 0.5223,
+      "step": 6855
+    },
+    {
+      "epoch": 1.651022864019254,
+      "grad_norm": 1.625,
+      "learning_rate": 5.701919815865077e-05,
+      "loss": 0.5125,
+      "step": 6860
+    },
+    {
+      "epoch": 1.6522262334536704,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.697215225291253e-05,
+      "loss": 0.5413,
+      "step": 6865
+    },
+    {
+      "epoch": 1.6534296028880866,
+      "grad_norm": 1.7265625,
+      "learning_rate": 5.692511564712811e-05,
+      "loss": 0.5466,
+      "step": 6870
+    },
+    {
+      "epoch": 1.654632972322503,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.687808841641366e-05,
+      "loss": 0.5303,
+      "step": 6875
+    },
+    {
+      "epoch": 1.6558363417569195,
+      "grad_norm": 1.796875,
+      "learning_rate": 5.683107063587036e-05,
+      "loss": 0.5052,
+      "step": 6880
+    },
+    {
+      "epoch": 1.6570397111913358,
+      "grad_norm": 1.625,
+      "learning_rate": 5.678406238058436e-05,
+      "loss": 0.5444,
+      "step": 6885
+    },
+    {
+      "epoch": 1.658243080625752,
+      "grad_norm": 1.7890625,
+      "learning_rate": 5.673706372562651e-05,
+      "loss": 0.5216,
+      "step": 6890
+    },
+    {
+      "epoch": 1.6594464500601684,
+      "grad_norm": 1.6875,
+      "learning_rate": 5.6690074746052387e-05,
+      "loss": 0.4973,
+      "step": 6895
+    },
+    {
+      "epoch": 1.660649819494585,
+      "grad_norm": 1.5546875,
+      "learning_rate": 5.6643095516902086e-05,
+      "loss": 0.5067,
+      "step": 6900
+    },
+    {
+      "epoch": 1.6618531889290011,
+      "grad_norm": 1.6171875,
+      "learning_rate": 5.659612611320016e-05,
+      "loss": 0.513,
+      "step": 6905
+    },
+    {
+      "epoch": 1.6630565583634176,
+      "grad_norm": 1.609375,
+      "learning_rate": 5.654916660995545e-05,
+      "loss": 0.5462,
+      "step": 6910
+    },
+    {
+      "epoch": 1.664259927797834,
+      "grad_norm": 1.6484375,
+      "learning_rate": 5.6502217082160985e-05,
+      "loss": 0.5546,
+      "step": 6915
+    },
+    {
+      "epoch": 1.6654632972322503,
+      "grad_norm": 1.765625,
+      "learning_rate": 5.645527760479389e-05,
+      "loss": 0.5366,
+      "step": 6920
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 1.6484375,
+      "learning_rate": 5.6408348252815175e-05,
+      "loss": 0.5309,
+      "step": 6925
+    },
+    {
+      "epoch": 1.6678700361010832,
+      "grad_norm": 1.671875,
+      "learning_rate": 5.636142910116977e-05,
+      "loss": 0.5298,
+      "step": 6930
+    },
+    {
+      "epoch": 1.6690734055354994,
+      "grad_norm": 1.7734375,
+      "learning_rate": 5.6314520224786246e-05,
+      "loss": 0.5296,
+      "step": 6935
+    },
+    {
+      "epoch": 1.6702767749699157,
+      "grad_norm": 1.5703125,
+      "learning_rate": 5.626762169857681e-05,
+      "loss": 0.5437,
+      "step": 6940
+    },
+    {
+      "epoch": 1.6714801444043321,
+      "grad_norm": 1.671875,
+      "learning_rate": 5.6220733597437104e-05,
+      "loss": 0.5303,
+      "step": 6945
+    },
+    {
+      "epoch": 1.6726835138387486,
+      "grad_norm": 1.5859375,
+      "learning_rate": 5.6173855996246134e-05,
+      "loss": 0.5361,
+      "step": 6950
+    },
+    {
+      "epoch": 1.6738868832731648,
+      "grad_norm": 1.6953125,
+      "learning_rate": 5.6126988969866155e-05,
+      "loss": 0.5326,
+      "step": 6955
+    },
+    {
+      "epoch": 1.6750902527075813,
+      "grad_norm": 1.5234375,
+      "learning_rate": 5.6080132593142505e-05,
+      "loss": 0.5333,
+      "step": 6960
+    },
+    {
+      "epoch": 1.6762936221419977,
+      "grad_norm": 1.6484375,
+      "learning_rate": 5.6033286940903516e-05,
+      "loss": 0.5072,
+      "step": 6965
+    },
+    {
+      "epoch": 1.677496991576414,
+      "grad_norm": 1.8671875,
+      "learning_rate": 5.598645208796045e-05,
+      "loss": 0.5403,
+      "step": 6970
+    },
+    {
+      "epoch": 1.6787003610108302,
+      "grad_norm": 1.7265625,
+      "learning_rate": 5.593962810910722e-05,
+      "loss": 0.5101,
+      "step": 6975
+    },
+    {
+      "epoch": 1.6799037304452467,
+      "grad_norm": 1.734375,
+      "learning_rate": 5.5892815079120496e-05,
+      "loss": 0.5334,
+      "step": 6980
+    },
+    {
+      "epoch": 1.6811070998796631,
+      "grad_norm": 1.65625,
+      "learning_rate": 5.5846013072759353e-05,
+      "loss": 0.5519,
+      "step": 6985
+    },
+    {
+      "epoch": 1.6823104693140793,
+      "grad_norm": 1.609375,
+      "learning_rate": 5.5799222164765336e-05,
+      "loss": 0.5024,
+      "step": 6990
+    },
+    {
+      "epoch": 1.6835138387484958,
+      "grad_norm": 1.6640625,
+      "learning_rate": 5.575244242986223e-05,
+      "loss": 0.5289,
+      "step": 6995
+    },
+    {
+      "epoch": 1.6847172081829123,
+      "grad_norm": 1.671875,
+      "learning_rate": 5.570567394275598e-05,
+      "loss": 0.505,
+      "step": 7000
+    },
+    {
+      "epoch": 1.6847172081829123,
+      "eval_loss": 0.4532296061515808,
+      "eval_runtime": 2.6618,
+      "eval_samples_per_second": 75.138,
+      "eval_steps_per_second": 75.138,
+      "step": 7000
+    },
+    {
+      "epoch": 1.6859205776173285,
+      "grad_norm": 1.75,
+      "learning_rate": 5.565891677813456e-05,
+      "loss": 0.5362,
+      "step": 7005
+    },
+    {
+      "epoch": 1.6871239470517447,
+      "grad_norm": 1.796875,
+      "learning_rate": 5.561217101066792e-05,
+      "loss": 0.5461,
+      "step": 7010
+    },
+    {
+      "epoch": 1.6883273164861614,
+      "grad_norm": 1.7578125,
+      "learning_rate": 5.5565436715007724e-05,
+      "loss": 0.5384,
+      "step": 7015
+    },
+    {
+      "epoch": 1.6895306859205776,
+      "grad_norm": 1.859375,
+      "learning_rate": 5.5518713965787376e-05,
+      "loss": 0.5458,
+      "step": 7020
+    },
+    {
+      "epoch": 1.6907340553549939,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.547200283762182e-05,
+      "loss": 0.5111,
+      "step": 7025
+    },
+    {
+      "epoch": 1.6919374247894103,
+      "grad_norm": 1.65625,
+      "learning_rate": 5.542530340510744e-05,
+      "loss": 0.509,
+      "step": 7030
+    },
+    {
+      "epoch": 1.6931407942238268,
+      "grad_norm": 1.6015625,
+      "learning_rate": 5.537861574282195e-05,
+      "loss": 0.5409,
+      "step": 7035
+    },
+    {
+      "epoch": 1.694344163658243,
+      "grad_norm": 1.578125,
+      "learning_rate": 5.533193992532426e-05,
+      "loss": 0.5218,
+      "step": 7040
+    },
+    {
+      "epoch": 1.6955475330926595,
+      "grad_norm": 1.796875,
+      "learning_rate": 5.5285276027154364e-05,
+      "loss": 0.5498,
+      "step": 7045
+    },
+    {
+      "epoch": 1.696750902527076,
+      "grad_norm": 1.546875,
+      "learning_rate": 5.523862412283323e-05,
+      "loss": 0.5424,
+      "step": 7050
+    },
+    {
+      "epoch": 1.6979542719614922,
+      "grad_norm": 1.9140625,
+      "learning_rate": 5.519198428686266e-05,
+      "loss": 0.5261,
+      "step": 7055
+    },
+    {
+      "epoch": 1.6991576413959084,
+      "grad_norm": 1.6875,
+      "learning_rate": 5.5145356593725205e-05,
+      "loss": 0.5015,
+      "step": 7060
+    },
+    {
+      "epoch": 1.7003610108303249,
+      "grad_norm": 1.65625,
+      "learning_rate": 5.5098741117883974e-05,
+      "loss": 0.5267,
+      "step": 7065
+    },
+    {
+      "epoch": 1.7015643802647413,
+      "grad_norm": 1.7109375,
+      "learning_rate": 5.505213793378264e-05,
+      "loss": 0.5124,
+      "step": 7070
+    },
+    {
+      "epoch": 1.7027677496991576,
+      "grad_norm": 1.890625,
+      "learning_rate": 5.5005547115845195e-05,
+      "loss": 0.5388,
+      "step": 7075
+    },
+    {
+      "epoch": 1.703971119133574,
+      "grad_norm": 1.6171875,
+      "learning_rate": 5.495896873847589e-05,
+      "loss": 0.5185,
+      "step": 7080
+    },
+    {
+      "epoch": 1.7051744885679905,
+      "grad_norm": 1.65625,
+      "learning_rate": 5.491240287605911e-05,
+      "loss": 0.5264,
+      "step": 7085
+    },
+    {
+      "epoch": 1.7063778580024067,
+      "grad_norm": 1.6328125,
+      "learning_rate": 5.486584960295926e-05,
+      "loss": 0.4885,
+      "step": 7090
+    },
+    {
+      "epoch": 1.707581227436823,
+      "grad_norm": 1.96875,
+      "learning_rate": 5.4819308993520635e-05,
+      "loss": 0.5509,
+      "step": 7095
+    },
+    {
+      "epoch": 1.7087845968712396,
+      "grad_norm": 1.703125,
+      "learning_rate": 5.477278112206731e-05,
+      "loss": 0.5354,
+      "step": 7100
+    },
+    {
+      "epoch": 1.7099879663056559,
+      "grad_norm": 1.6875,
+      "learning_rate": 5.4726266062903016e-05,
+      "loss": 0.526,
+      "step": 7105
+    },
+    {
+      "epoch": 1.711191335740072,
+      "grad_norm": 1.734375,
+      "learning_rate": 5.467976389031103e-05,
+      "loss": 0.5211,
+      "step": 7110
+    },
+    {
+      "epoch": 1.7123947051744886,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.463327467855402e-05,
+      "loss": 0.4991,
+      "step": 7115
+    },
+    {
+      "epoch": 1.713598074608905,
+      "grad_norm": 1.7421875,
+      "learning_rate": 5.458679850187402e-05,
+      "loss": 0.5524,
+      "step": 7120
+    },
+    {
+      "epoch": 1.7148014440433212,
+      "grad_norm": 1.640625,
+      "learning_rate": 5.454033543449218e-05,
+      "loss": 0.5212,
+      "step": 7125
+    },
+    {
+      "epoch": 1.7160048134777377,
+      "grad_norm": 1.65625,
+      "learning_rate": 5.449388555060873e-05,
+      "loss": 0.5653,
+      "step": 7130
+    },
+    {
+      "epoch": 1.7172081829121542,
+      "grad_norm": 1.5625,
+      "learning_rate": 5.444744892440289e-05,
+      "loss": 0.5258,
+      "step": 7135
+    },
+    {
+      "epoch": 1.7184115523465704,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.4401025630032645e-05,
+      "loss": 0.5132,
+      "step": 7140
+    },
+    {
+      "epoch": 1.7196149217809866,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.435461574163473e-05,
+      "loss": 0.5255,
+      "step": 7145
+    },
+    {
+      "epoch": 1.720818291215403,
+      "grad_norm": 1.6328125,
+      "learning_rate": 5.4308219333324475e-05,
+      "loss": 0.5403,
+      "step": 7150
+    },
+    {
+      "epoch": 1.7220216606498195,
+      "grad_norm": 1.640625,
+      "learning_rate": 5.426183647919564e-05,
+      "loss": 0.569,
+      "step": 7155
+    },
+    {
+      "epoch": 1.7232250300842358,
+      "grad_norm": 1.5859375,
+      "learning_rate": 5.421546725332038e-05,
+      "loss": 0.5447,
+      "step": 7160
+    },
+    {
+      "epoch": 1.7244283995186522,
+      "grad_norm": 1.6875,
+      "learning_rate": 5.416911172974909e-05,
+      "loss": 0.5287,
+      "step": 7165
+    },
+    {
+      "epoch": 1.7256317689530687,
+      "grad_norm": 1.6171875,
+      "learning_rate": 5.412276998251026e-05,
+      "loss": 0.5299,
+      "step": 7170
+    },
+    {
+      "epoch": 1.726835138387485,
+      "grad_norm": 1.703125,
+      "learning_rate": 5.407644208561036e-05,
+      "loss": 0.5364,
+      "step": 7175
+    },
+    {
+      "epoch": 1.7280385078219012,
+      "grad_norm": 1.6484375,
+      "learning_rate": 5.40301281130338e-05,
+      "loss": 0.5338,
+      "step": 7180
+    },
+    {
+      "epoch": 1.7292418772563178,
+      "grad_norm": 1.6328125,
+      "learning_rate": 5.398382813874271e-05,
+      "loss": 0.5242,
+      "step": 7185
+    },
+    {
+      "epoch": 1.730445246690734,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.393754223667687e-05,
+      "loss": 0.5531,
+      "step": 7190
+    },
+    {
+      "epoch": 1.7316486161251503,
+      "grad_norm": 1.796875,
+      "learning_rate": 5.3891270480753595e-05,
+      "loss": 0.4976,
+      "step": 7195
+    },
+    {
+      "epoch": 1.7328519855595668,
+      "grad_norm": 1.609375,
+      "learning_rate": 5.3845012944867614e-05,
+      "loss": 0.5231,
+      "step": 7200
+    },
+    {
+      "epoch": 1.7340553549939832,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.379876970289094e-05,
+      "loss": 0.502,
+      "step": 7205
+    },
+    {
+      "epoch": 1.7352587244283995,
+      "grad_norm": 1.8046875,
+      "learning_rate": 5.3752540828672736e-05,
+      "loss": 0.5223,
+      "step": 7210
+    },
+    {
+      "epoch": 1.736462093862816,
+      "grad_norm": 1.5546875,
+      "learning_rate": 5.3706326396039276e-05,
+      "loss": 0.5151,
+      "step": 7215
+    },
+    {
+      "epoch": 1.7376654632972324,
+      "grad_norm": 1.8984375,
+      "learning_rate": 5.3660126478793724e-05,
+      "loss": 0.5491,
+      "step": 7220
+    },
+    {
+      "epoch": 1.7388688327316486,
+      "grad_norm": 1.625,
+      "learning_rate": 5.361394115071608e-05,
+      "loss": 0.5299,
+      "step": 7225
+    },
+    {
+      "epoch": 1.7400722021660648,
+      "grad_norm": 1.96875,
+      "learning_rate": 5.356777048556303e-05,
+      "loss": 0.5394,
+      "step": 7230
+    },
+    {
+      "epoch": 1.7412755716004813,
+      "grad_norm": 1.65625,
+      "learning_rate": 5.352161455706787e-05,
+      "loss": 0.5529,
+      "step": 7235
+    },
+    {
+      "epoch": 1.7424789410348978,
+      "grad_norm": 1.7578125,
+      "learning_rate": 5.347547343894035e-05,
+      "loss": 0.5515,
+      "step": 7240
+    },
+    {
+      "epoch": 1.743682310469314,
+      "grad_norm": 1.609375,
+      "learning_rate": 5.3429347204866555e-05,
+      "loss": 0.5264,
+      "step": 7245
+    },
+    {
+      "epoch": 1.7448856799037304,
+      "grad_norm": 1.8359375,
+      "learning_rate": 5.338323592850881e-05,
+      "loss": 0.5158,
+      "step": 7250
+    },
+    {
+      "epoch": 1.746089049338147,
+      "grad_norm": 1.640625,
+      "learning_rate": 5.333713968350558e-05,
+      "loss": 0.5633,
+      "step": 7255
+    },
+    {
+      "epoch": 1.7472924187725631,
+      "grad_norm": 1.8125,
+      "learning_rate": 5.329105854347129e-05,
+      "loss": 0.5356,
+      "step": 7260
+    },
+    {
+      "epoch": 1.7484957882069796,
+      "grad_norm": 1.7265625,
+      "learning_rate": 5.324499258199625e-05,
+      "loss": 0.5117,
+      "step": 7265
+    },
+    {
+      "epoch": 1.749699157641396,
+      "grad_norm": 1.7265625,
+      "learning_rate": 5.319894187264653e-05,
+      "loss": 0.5246,
+      "step": 7270
+    },
+    {
+      "epoch": 1.7509025270758123,
+      "grad_norm": 1.671875,
+      "learning_rate": 5.315290648896386e-05,
+      "loss": 0.5215,
+      "step": 7275
+    },
+    {
+      "epoch": 1.7521058965102285,
+      "grad_norm": 1.6875,
+      "learning_rate": 5.310688650446546e-05,
+      "loss": 0.5411,
+      "step": 7280
+    },
+    {
+      "epoch": 1.753309265944645,
+      "grad_norm": 1.703125,
+      "learning_rate": 5.3060881992644e-05,
+      "loss": 0.4988,
+      "step": 7285
+    },
+    {
+      "epoch": 1.7545126353790614,
+      "grad_norm": 1.625,
+      "learning_rate": 5.301489302696741e-05,
+      "loss": 0.5333,
+      "step": 7290
+    },
+    {
+      "epoch": 1.7557160048134777,
+      "grad_norm": 1.5703125,
+      "learning_rate": 5.29689196808788e-05,
+      "loss": 0.5175,
+      "step": 7295
+    },
+    {
+      "epoch": 1.7569193742478941,
+      "grad_norm": 1.703125,
+      "learning_rate": 5.292296202779636e-05,
+      "loss": 0.5333,
+      "step": 7300
+    },
+    {
+      "epoch": 1.7581227436823106,
+      "grad_norm": 1.6484375,
+      "learning_rate": 5.287702014111317e-05,
+      "loss": 0.563,
+      "step": 7305
+    },
+    {
+      "epoch": 1.7593261131167268,
+      "grad_norm": 1.6171875,
+      "learning_rate": 5.2831094094197186e-05,
+      "loss": 0.5025,
+      "step": 7310
+    },
+    {
+      "epoch": 1.760529482551143,
+      "grad_norm": 1.453125,
+      "learning_rate": 5.2785183960391025e-05,
+      "loss": 0.519,
+      "step": 7315
+    },
+    {
+      "epoch": 1.7617328519855595,
+      "grad_norm": 1.5625,
+      "learning_rate": 5.2739289813011925e-05,
+      "loss": 0.4948,
+      "step": 7320
+    },
+    {
+      "epoch": 1.762936221419976,
+      "grad_norm": 1.734375,
+      "learning_rate": 5.269341172535156e-05,
+      "loss": 0.5202,
+      "step": 7325
+    },
+    {
+      "epoch": 1.7641395908543922,
+      "grad_norm": 1.6953125,
+      "learning_rate": 5.2647549770675984e-05,
+      "loss": 0.4981,
+      "step": 7330
+    },
+    {
+      "epoch": 1.7653429602888087,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.2601704022225466e-05,
+      "loss": 0.5138,
+      "step": 7335
+    },
+    {
+      "epoch": 1.7665463297232251,
+      "grad_norm": 1.7421875,
+      "learning_rate": 5.2555874553214414e-05,
+      "loss": 0.5303,
+      "step": 7340
+    },
+    {
+      "epoch": 1.7677496991576414,
+      "grad_norm": 1.6484375,
+      "learning_rate": 5.251006143683123e-05,
+      "loss": 0.5314,
+      "step": 7345
+    },
+    {
+      "epoch": 1.7689530685920578,
+      "grad_norm": 1.671875,
+      "learning_rate": 5.2464264746238215e-05,
+      "loss": 0.5064,
+      "step": 7350
+    },
+    {
+      "epoch": 1.7701564380264743,
+      "grad_norm": 1.4921875,
+      "learning_rate": 5.241848455457141e-05,
+      "loss": 0.5114,
+      "step": 7355
+    },
+    {
+      "epoch": 1.7713598074608905,
+      "grad_norm": 1.5546875,
+      "learning_rate": 5.237272093494052e-05,
+      "loss": 0.5072,
+      "step": 7360
+    },
+    {
+      "epoch": 1.7725631768953067,
+      "grad_norm": 1.7265625,
+      "learning_rate": 5.2326973960428794e-05,
+      "loss": 0.534,
+      "step": 7365
+    },
+    {
+      "epoch": 1.7737665463297232,
+      "grad_norm": 1.5234375,
+      "learning_rate": 5.2281243704092886e-05,
+      "loss": 0.5141,
+      "step": 7370
+    },
+    {
+      "epoch": 1.7749699157641396,
+      "grad_norm": 1.65625,
+      "learning_rate": 5.2235530238962774e-05,
+      "loss": 0.5247,
+      "step": 7375
+    },
+    {
+      "epoch": 1.7761732851985559,
+      "grad_norm": 1.71875,
+      "learning_rate": 5.218983363804158e-05,
+      "loss": 0.5001,
+      "step": 7380
+    },
+    {
+      "epoch": 1.7773766546329723,
+      "grad_norm": 1.640625,
+      "learning_rate": 5.214415397430552e-05,
+      "loss": 0.5389,
+      "step": 7385
+    },
+    {
+      "epoch": 1.7785800240673888,
+      "grad_norm": 1.6484375,
+      "learning_rate": 5.2098491320703795e-05,
+      "loss": 0.5073,
+      "step": 7390
+    },
+    {
+      "epoch": 1.779783393501805,
+      "grad_norm": 1.796875,
+      "learning_rate": 5.205284575015837e-05,
+      "loss": 0.5679,
+      "step": 7395
+    },
+    {
+      "epoch": 1.7809867629362213,
+      "grad_norm": 1.8359375,
+      "learning_rate": 5.200721733556399e-05,
+      "loss": 0.5513,
+      "step": 7400
+    },
+    {
+      "epoch": 1.782190132370638,
+      "grad_norm": 1.59375,
+      "learning_rate": 5.196160614978798e-05,
+      "loss": 0.5526,
+      "step": 7405
+    },
+    {
+      "epoch": 1.7833935018050542,
+      "grad_norm": 1.7421875,
+      "learning_rate": 5.191601226567014e-05,
+      "loss": 0.5022,
+      "step": 7410
+    },
+    {
+      "epoch": 1.7845968712394704,
+      "grad_norm": 1.8125,
+      "learning_rate": 5.187043575602264e-05,
+      "loss": 0.5309,
+      "step": 7415
+    },
+    {
+      "epoch": 1.7858002406738869,
+      "grad_norm": 1.640625,
+      "learning_rate": 5.182487669362992e-05,
+      "loss": 0.5247,
+      "step": 7420
+    },
+    {
+      "epoch": 1.7870036101083033,
+      "grad_norm": 1.71875,
+      "learning_rate": 5.177933515124856e-05,
+      "loss": 0.5395,
+      "step": 7425
+    },
+    {
+      "epoch": 1.7882069795427196,
+      "grad_norm": 1.59375,
+      "learning_rate": 5.173381120160712e-05,
+      "loss": 0.5064,
+      "step": 7430
+    },
+    {
+      "epoch": 1.789410348977136,
+      "grad_norm": 1.796875,
+      "learning_rate": 5.1688304917406134e-05,
+      "loss": 0.5329,
+      "step": 7435
+    },
+    {
+      "epoch": 1.7906137184115525,
+      "grad_norm": 1.640625,
+      "learning_rate": 5.1642816371317875e-05,
+      "loss": 0.5505,
+      "step": 7440
+    },
+    {
+      "epoch": 1.7918170878459687,
+      "grad_norm": 1.6484375,
+      "learning_rate": 5.159734563598629e-05,
+      "loss": 0.5252,
+      "step": 7445
+    },
+    {
+      "epoch": 1.793020457280385,
+      "grad_norm": 1.7265625,
+      "learning_rate": 5.1551892784026905e-05,
+      "loss": 0.5287,
+      "step": 7450
+    },
+    {
+      "epoch": 1.7942238267148014,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.150645788802666e-05,
+      "loss": 0.5489,
+      "step": 7455
+    },
+    {
+      "epoch": 1.7954271961492179,
+      "grad_norm": 1.6640625,
+      "learning_rate": 5.1461041020543855e-05,
+      "loss": 0.529,
+      "step": 7460
+    },
+    {
+      "epoch": 1.796630565583634,
+      "grad_norm": 1.65625,
+      "learning_rate": 5.141564225410797e-05,
+      "loss": 0.5412,
+      "step": 7465
+    },
+    {
+      "epoch": 1.7978339350180506,
+      "grad_norm": 1.6875,
+      "learning_rate": 5.137026166121958e-05,
+      "loss": 0.5298,
+      "step": 7470
+    },
+    {
+      "epoch": 1.799037304452467,
+      "grad_norm": 1.609375,
+      "learning_rate": 5.1324899314350255e-05,
+      "loss": 0.5254,
+      "step": 7475
+    },
+    {
+      "epoch": 1.8002406738868832,
+      "grad_norm": 1.5859375,
+      "learning_rate": 5.12795552859424e-05,
+      "loss": 0.4987,
+      "step": 7480
+    },
+    {
+      "epoch": 1.8014440433212995,
+      "grad_norm": 1.71875,
+      "learning_rate": 5.1234229648409194e-05,
+      "loss": 0.5129,
+      "step": 7485
+    },
+    {
+      "epoch": 1.8026474127557162,
+      "grad_norm": 1.5703125,
+      "learning_rate": 5.118892247413443e-05,
+      "loss": 0.5365,
+      "step": 7490
+    },
+    {
+      "epoch": 1.8038507821901324,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.1143633835472424e-05,
+      "loss": 0.5648,
+      "step": 7495
+    },
+    {
+      "epoch": 1.8050541516245486,
+      "grad_norm": 1.65625,
+      "learning_rate": 5.1098363804747874e-05,
+      "loss": 0.5456,
+      "step": 7500
+    },
+    {
+      "epoch": 1.8050541516245486,
+      "eval_loss": 0.44866788387298584,
+      "eval_runtime": 2.6548,
+      "eval_samples_per_second": 75.335,
+      "eval_steps_per_second": 75.335,
+      "step": 7500
+    },
+    {
+      "epoch": 1.806257521058965,
+      "grad_norm": 1.6015625,
+      "learning_rate": 5.105311245425579e-05,
+      "loss": 0.5104,
+      "step": 7505
+    },
+    {
+      "epoch": 1.8074608904933815,
+      "grad_norm": 1.6484375,
+      "learning_rate": 5.100787985626133e-05,
+      "loss": 0.4927,
+      "step": 7510
+    },
+    {
+      "epoch": 1.8086642599277978,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.0962666082999714e-05,
+      "loss": 0.5157,
+      "step": 7515
+    },
+    {
+      "epoch": 1.8098676293622142,
+      "grad_norm": 1.65625,
+      "learning_rate": 5.0917471206676075e-05,
+      "loss": 0.5282,
+      "step": 7520
+    },
+    {
+      "epoch": 1.8110709987966307,
+      "grad_norm": 1.5859375,
+      "learning_rate": 5.087229529946542e-05,
+      "loss": 0.5211,
+      "step": 7525
+    },
+    {
+      "epoch": 1.812274368231047,
+      "grad_norm": 1.765625,
+      "learning_rate": 5.08271384335124e-05,
+      "loss": 0.5446,
+      "step": 7530
+    },
+    {
+      "epoch": 1.8134777376654632,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.07820006809313e-05,
+      "loss": 0.5155,
+      "step": 7535
+    },
+    {
+      "epoch": 1.8146811070998796,
+      "grad_norm": 1.5625,
+      "learning_rate": 5.073688211380587e-05,
+      "loss": 0.5065,
+      "step": 7540
+    },
+    {
+      "epoch": 1.815884476534296,
+      "grad_norm": 1.7109375,
+      "learning_rate": 5.069178280418923e-05,
+      "loss": 0.508,
+      "step": 7545
+    },
+    {
+      "epoch": 1.8170878459687123,
+      "grad_norm": 1.8046875,
+      "learning_rate": 5.064670282410371e-05,
+      "loss": 0.5329,
+      "step": 7550
+    },
+    {
+      "epoch": 1.8182912154031288,
+      "grad_norm": 1.46875,
+      "learning_rate": 5.0601642245540826e-05,
+      "loss": 0.5221,
+      "step": 7555
+    },
+    {
+      "epoch": 1.8194945848375452,
+      "grad_norm": 1.7109375,
+      "learning_rate": 5.055660114046108e-05,
+      "loss": 0.5261,
+      "step": 7560
+    },
+    {
+      "epoch": 1.8206979542719615,
+      "grad_norm": 1.6953125,
+      "learning_rate": 5.0511579580793865e-05,
+      "loss": 0.5041,
+      "step": 7565
+    },
+    {
+      "epoch": 1.8219013237063777,
+      "grad_norm": 1.6484375,
+      "learning_rate": 5.046657763843739e-05,
+      "loss": 0.4916,
+      "step": 7570
+    },
+    {
+      "epoch": 1.8231046931407944,
+      "grad_norm": 1.6953125,
+      "learning_rate": 5.0421595385258514e-05,
+      "loss": 0.5279,
+      "step": 7575
+    },
+    {
+      "epoch": 1.8243080625752106,
+      "grad_norm": 1.578125,
+      "learning_rate": 5.037663289309264e-05,
+      "loss": 0.5219,
+      "step": 7580
+    },
+    {
+      "epoch": 1.8255114320096268,
+      "grad_norm": 1.7421875,
+      "learning_rate": 5.0331690233743665e-05,
+      "loss": 0.5261,
+      "step": 7585
+    },
+    {
+      "epoch": 1.8267148014440433,
+      "grad_norm": 1.703125,
+      "learning_rate": 5.028676747898376e-05,
+      "loss": 0.5183,
+      "step": 7590
+    },
+    {
+      "epoch": 1.8279181708784598,
+      "grad_norm": 1.5546875,
+      "learning_rate": 5.024186470055332e-05,
+      "loss": 0.4969,
+      "step": 7595
+    },
+    {
+      "epoch": 1.829121540312876,
+      "grad_norm": 1.90625,
+      "learning_rate": 5.019698197016088e-05,
+      "loss": 0.5332,
+      "step": 7600
+    },
+    {
+      "epoch": 1.8303249097472925,
+      "grad_norm": 1.7421875,
+      "learning_rate": 5.0152119359482886e-05,
+      "loss": 0.5203,
+      "step": 7605
+    },
+    {
+      "epoch": 1.831528279181709,
+      "grad_norm": 1.796875,
+      "learning_rate": 5.010727694016373e-05,
+      "loss": 0.5195,
+      "step": 7610
+    },
+    {
+      "epoch": 1.8327316486161251,
+      "grad_norm": 1.5234375,
+      "learning_rate": 5.00624547838155e-05,
+      "loss": 0.5075,
+      "step": 7615
+    },
+    {
+      "epoch": 1.8339350180505414,
+      "grad_norm": 1.6640625,
+      "learning_rate": 5.001765296201796e-05,
+      "loss": 0.5253,
+      "step": 7620
+    },
+    {
+      "epoch": 1.8351383874849578,
+      "grad_norm": 1.5625,
+      "learning_rate": 4.997287154631837e-05,
+      "loss": 0.5408,
+      "step": 7625
+    },
+    {
+      "epoch": 1.8363417569193743,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.992811060823143e-05,
+      "loss": 0.5472,
+      "step": 7630
+    },
+    {
+      "epoch": 1.8375451263537905,
+      "grad_norm": 1.7109375,
+      "learning_rate": 4.9883370219239146e-05,
+      "loss": 0.5116,
+      "step": 7635
+    },
+    {
+      "epoch": 1.838748495788207,
+      "grad_norm": 1.8359375,
+      "learning_rate": 4.983865045079065e-05,
+      "loss": 0.5468,
+      "step": 7640
+    },
+    {
+      "epoch": 1.8399518652226234,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.979395137430222e-05,
+      "loss": 0.527,
+      "step": 7645
+    },
+    {
+      "epoch": 1.8411552346570397,
+      "grad_norm": 1.6953125,
+      "learning_rate": 4.974927306115703e-05,
+      "loss": 0.5331,
+      "step": 7650
+    },
+    {
+      "epoch": 1.8423586040914561,
+      "grad_norm": 1.75,
+      "learning_rate": 4.970461558270513e-05,
+      "loss": 0.5134,
+      "step": 7655
+    },
+    {
+      "epoch": 1.8435619735258726,
+      "grad_norm": 1.671875,
+      "learning_rate": 4.965997901026327e-05,
+      "loss": 0.5049,
+      "step": 7660
+    },
+    {
+      "epoch": 1.8447653429602888,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.961536341511485e-05,
+      "loss": 0.546,
+      "step": 7665
+    },
+    {
+      "epoch": 1.845968712394705,
+      "grad_norm": 1.6796875,
+      "learning_rate": 4.957076886850973e-05,
+      "loss": 0.523,
+      "step": 7670
+    },
+    {
+      "epoch": 1.8471720818291215,
+      "grad_norm": 1.6484375,
+      "learning_rate": 4.952619544166418e-05,
+      "loss": 0.5506,
+      "step": 7675
+    },
+    {
+      "epoch": 1.848375451263538,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.9481643205760744e-05,
+      "loss": 0.5393,
+      "step": 7680
+    },
+    {
+      "epoch": 1.8495788206979542,
+      "grad_norm": 1.75,
+      "learning_rate": 4.94371122319481e-05,
+      "loss": 0.521,
+      "step": 7685
+    },
+    {
+      "epoch": 1.8507821901323707,
+      "grad_norm": 1.671875,
+      "learning_rate": 4.939260259134102e-05,
+      "loss": 0.5014,
+      "step": 7690
+    },
+    {
+      "epoch": 1.8519855595667871,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.9348114355020146e-05,
+      "loss": 0.5539,
+      "step": 7695
+    },
+    {
+      "epoch": 1.8531889290012034,
+      "grad_norm": 1.6171875,
+      "learning_rate": 4.930364759403198e-05,
+      "loss": 0.5332,
+      "step": 7700
+    },
+    {
+      "epoch": 1.8543922984356196,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.925920237938872e-05,
+      "loss": 0.527,
+      "step": 7705
+    },
+    {
+      "epoch": 1.855595667870036,
+      "grad_norm": 1.5859375,
+      "learning_rate": 4.921477878206815e-05,
+      "loss": 0.5315,
+      "step": 7710
+    },
+    {
+      "epoch": 1.8567990373044525,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.9170376873013535e-05,
+      "loss": 0.5518,
+      "step": 7715
+    },
+    {
+      "epoch": 1.8580024067388687,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.91259967231335e-05,
+      "loss": 0.4976,
+      "step": 7720
+    },
+    {
+      "epoch": 1.8592057761732852,
+      "grad_norm": 1.8671875,
+      "learning_rate": 4.908163840330192e-05,
+      "loss": 0.5039,
+      "step": 7725
+    },
+    {
+      "epoch": 1.8604091456077017,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.9037301984357806e-05,
+      "loss": 0.5119,
+      "step": 7730
+    },
+    {
+      "epoch": 1.8616125150421179,
+      "grad_norm": 1.6171875,
+      "learning_rate": 4.899298753710522e-05,
+      "loss": 0.476,
+      "step": 7735
+    },
+    {
+      "epoch": 1.8628158844765343,
+      "grad_norm": 1.7265625,
+      "learning_rate": 4.894869513231311e-05,
+      "loss": 0.5074,
+      "step": 7740
+    },
+    {
+      "epoch": 1.8640192539109508,
+      "grad_norm": 1.671875,
+      "learning_rate": 4.890442484071522e-05,
+      "loss": 0.5264,
+      "step": 7745
+    },
+    {
+      "epoch": 1.865222623345367,
+      "grad_norm": 1.8125,
+      "learning_rate": 4.886017673301e-05,
+      "loss": 0.5583,
+      "step": 7750
+    },
+    {
+      "epoch": 1.8664259927797833,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.8815950879860446e-05,
+      "loss": 0.5001,
+      "step": 7755
+    },
+    {
+      "epoch": 1.8676293622141997,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.877174735189404e-05,
+      "loss": 0.5219,
+      "step": 7760
+    },
+    {
+      "epoch": 1.8688327316486162,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.87275662197026e-05,
+      "loss": 0.5202,
+      "step": 7765
+    },
+    {
+      "epoch": 1.8700361010830324,
+      "grad_norm": 1.7109375,
+      "learning_rate": 4.868340755384218e-05,
+      "loss": 0.5163,
+      "step": 7770
+    },
+    {
+      "epoch": 1.8712394705174489,
+      "grad_norm": 1.625,
+      "learning_rate": 4.863927142483294e-05,
+      "loss": 0.5207,
+      "step": 7775
+    },
+    {
+      "epoch": 1.8724428399518653,
+      "grad_norm": 1.6640625,
+      "learning_rate": 4.859515790315909e-05,
+      "loss": 0.5159,
+      "step": 7780
+    },
+    {
+      "epoch": 1.8736462093862816,
+      "grad_norm": 1.6484375,
+      "learning_rate": 4.8551067059268704e-05,
+      "loss": 0.5157,
+      "step": 7785
+    },
+    {
+      "epoch": 1.8748495788206978,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.850699896357363e-05,
+      "loss": 0.5055,
+      "step": 7790
+    },
+    {
+      "epoch": 1.8760529482551145,
+      "grad_norm": 1.4765625,
+      "learning_rate": 4.8462953686449424e-05,
+      "loss": 0.5044,
+      "step": 7795
+    },
+    {
+      "epoch": 1.8772563176895307,
+      "grad_norm": 1.6171875,
+      "learning_rate": 4.8418931298235174e-05,
+      "loss": 0.5022,
+      "step": 7800
+    },
+    {
+      "epoch": 1.878459687123947,
+      "grad_norm": 1.640625,
+      "learning_rate": 4.8374931869233425e-05,
+      "loss": 0.5135,
+      "step": 7805
+    },
+    {
+      "epoch": 1.8796630565583634,
+      "grad_norm": 1.6640625,
+      "learning_rate": 4.8330955469710065e-05,
+      "loss": 0.5238,
+      "step": 7810
+    },
+    {
+      "epoch": 1.8808664259927799,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.828700216989418e-05,
+      "loss": 0.5563,
+      "step": 7815
+    },
+    {
+      "epoch": 1.882069795427196,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.8243072039977986e-05,
+      "loss": 0.5009,
+      "step": 7820
+    },
+    {
+      "epoch": 1.8832731648616126,
+      "grad_norm": 1.640625,
+      "learning_rate": 4.819916515011669e-05,
+      "loss": 0.4995,
+      "step": 7825
+    },
+    {
+      "epoch": 1.884476534296029,
+      "grad_norm": 1.765625,
+      "learning_rate": 4.815528157042841e-05,
+      "loss": 0.5511,
+      "step": 7830
+    },
+    {
+      "epoch": 1.8856799037304453,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.811142137099398e-05,
+      "loss": 0.5293,
+      "step": 7835
+    },
+    {
+      "epoch": 1.8868832731648615,
+      "grad_norm": 1.6171875,
+      "learning_rate": 4.806758462185695e-05,
+      "loss": 0.5794,
+      "step": 7840
+    },
+    {
+      "epoch": 1.888086642599278,
+      "grad_norm": 1.7109375,
+      "learning_rate": 4.802377139302342e-05,
+      "loss": 0.5428,
+      "step": 7845
+    },
+    {
+      "epoch": 1.8892900120336944,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.797998175446188e-05,
+      "loss": 0.4973,
+      "step": 7850
+    },
+    {
+      "epoch": 1.8904933814681106,
+      "grad_norm": 1.6640625,
+      "learning_rate": 4.79362157761032e-05,
+      "loss": 0.5191,
+      "step": 7855
+    },
+    {
+      "epoch": 1.891696750902527,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.789247352784044e-05,
+      "loss": 0.511,
+      "step": 7860
+    },
+    {
+      "epoch": 1.8929001203369435,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.784875507952876e-05,
+      "loss": 0.5398,
+      "step": 7865
+    },
+    {
+      "epoch": 1.8941034897713598,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.780506050098533e-05,
+      "loss": 0.5216,
+      "step": 7870
+    },
+    {
+      "epoch": 1.895306859205776,
+      "grad_norm": 1.6484375,
+      "learning_rate": 4.7761389861989194e-05,
+      "loss": 0.5211,
+      "step": 7875
+    },
+    {
+      "epoch": 1.8965102286401927,
+      "grad_norm": 1.734375,
+      "learning_rate": 4.7717743232281144e-05,
+      "loss": 0.5146,
+      "step": 7880
+    },
+    {
+      "epoch": 1.897713598074609,
+      "grad_norm": 1.6171875,
+      "learning_rate": 4.7674120681563665e-05,
+      "loss": 0.5227,
+      "step": 7885
+    },
+    {
+      "epoch": 1.8989169675090252,
+      "grad_norm": 1.5859375,
+      "learning_rate": 4.763052227950076e-05,
+      "loss": 0.4982,
+      "step": 7890
+    },
+    {
+      "epoch": 1.9001203369434416,
+      "grad_norm": 1.7265625,
+      "learning_rate": 4.758694809571788e-05,
+      "loss": 0.5442,
+      "step": 7895
+    },
+    {
+      "epoch": 1.901323706377858,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.754339819980181e-05,
+      "loss": 0.5375,
+      "step": 7900
+    },
+    {
+      "epoch": 1.9025270758122743,
+      "grad_norm": 1.4921875,
+      "learning_rate": 4.749987266130052e-05,
+      "loss": 0.543,
+      "step": 7905
+    },
+    {
+      "epoch": 1.9037304452466908,
+      "grad_norm": 1.6171875,
+      "learning_rate": 4.7456371549723105e-05,
+      "loss": 0.5202,
+      "step": 7910
+    },
+    {
+      "epoch": 1.9049338146811072,
+      "grad_norm": 1.7109375,
+      "learning_rate": 4.741289493453965e-05,
+      "loss": 0.5017,
+      "step": 7915
+    },
+    {
+      "epoch": 1.9061371841155235,
+      "grad_norm": 1.7109375,
+      "learning_rate": 4.73694428851811e-05,
+      "loss": 0.5265,
+      "step": 7920
+    },
+    {
+      "epoch": 1.9073405535499397,
+      "grad_norm": 1.7890625,
+      "learning_rate": 4.73260154710392e-05,
+      "loss": 0.5084,
+      "step": 7925
+    },
+    {
+      "epoch": 1.9085439229843562,
+      "grad_norm": 1.7734375,
+      "learning_rate": 4.728261276146632e-05,
+      "loss": 0.5465,
+      "step": 7930
+    },
+    {
+      "epoch": 1.9097472924187726,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.723923482577541e-05,
+      "loss": 0.5287,
+      "step": 7935
+    },
+    {
+      "epoch": 1.9109506618531888,
+      "grad_norm": 1.671875,
+      "learning_rate": 4.719588173323983e-05,
+      "loss": 0.4913,
+      "step": 7940
+    },
+    {
+      "epoch": 1.9121540312876053,
+      "grad_norm": 1.6484375,
+      "learning_rate": 4.715255355309328e-05,
+      "loss": 0.5275,
+      "step": 7945
+    },
+    {
+      "epoch": 1.9133574007220218,
+      "grad_norm": 1.5234375,
+      "learning_rate": 4.710925035452967e-05,
+      "loss": 0.4948,
+      "step": 7950
+    },
+    {
+      "epoch": 1.914560770156438,
+      "grad_norm": 1.640625,
+      "learning_rate": 4.706597220670303e-05,
+      "loss": 0.5135,
+      "step": 7955
+    },
+    {
+      "epoch": 1.9157641395908542,
+      "grad_norm": 1.6640625,
+      "learning_rate": 4.702271917872734e-05,
+      "loss": 0.5358,
+      "step": 7960
+    },
+    {
+      "epoch": 1.916967509025271,
+      "grad_norm": 1.734375,
+      "learning_rate": 4.697949133967653e-05,
+      "loss": 0.5001,
+      "step": 7965
+    },
+    {
+      "epoch": 1.9181708784596871,
+      "grad_norm": 1.5859375,
+      "learning_rate": 4.6936288758584256e-05,
+      "loss": 0.5063,
+      "step": 7970
+    },
+    {
+      "epoch": 1.9193742478941034,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.689311150444387e-05,
+      "loss": 0.5036,
+      "step": 7975
+    },
+    {
+      "epoch": 1.9205776173285198,
+      "grad_norm": 1.5625,
+      "learning_rate": 4.684995964620821e-05,
+      "loss": 0.5218,
+      "step": 7980
+    },
+    {
+      "epoch": 1.9217809867629363,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.680683325278967e-05,
+      "loss": 0.5123,
+      "step": 7985
+    },
+    {
+      "epoch": 1.9229843561973525,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.6763732393059874e-05,
+      "loss": 0.5022,
+      "step": 7990
+    },
+    {
+      "epoch": 1.924187725631769,
+      "grad_norm": 1.640625,
+      "learning_rate": 4.6720657135849716e-05,
+      "loss": 0.4931,
+      "step": 7995
+    },
+    {
+      "epoch": 1.9253910950661854,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.667760754994923e-05,
+      "loss": 0.5287,
+      "step": 8000
+    },
+    {
+      "epoch": 1.9253910950661854,
+      "eval_loss": 0.44210097193717957,
+      "eval_runtime": 2.6601,
+      "eval_samples_per_second": 75.184,
+      "eval_steps_per_second": 75.184,
+      "step": 8000
+    },
+    {
+      "epoch": 1.9265944645006017,
+      "grad_norm": 1.671875,
+      "learning_rate": 4.663458370410737e-05,
+      "loss": 0.5263,
+      "step": 8005
+    },
+    {
+      "epoch": 1.927797833935018,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.6591585667032065e-05,
+      "loss": 0.5124,
+      "step": 8010
+    },
+    {
+      "epoch": 1.9290012033694344,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.6548613507389994e-05,
+      "loss": 0.519,
+      "step": 8015
+    },
+    {
+      "epoch": 1.9302045728038508,
+      "grad_norm": 1.671875,
+      "learning_rate": 4.650566729380652e-05,
+      "loss": 0.5118,
+      "step": 8020
+    },
+    {
+      "epoch": 1.931407942238267,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.646274709486554e-05,
+      "loss": 0.5414,
+      "step": 8025
+    },
+    {
+      "epoch": 1.9326113116726835,
+      "grad_norm": 1.671875,
+      "learning_rate": 4.641985297910945e-05,
+      "loss": 0.5142,
+      "step": 8030
+    },
+    {
+      "epoch": 1.9338146811071,
+      "grad_norm": 1.5859375,
+      "learning_rate": 4.637698501503899e-05,
+      "loss": 0.5272,
+      "step": 8035
+    },
+    {
+      "epoch": 1.9350180505415162,
+      "grad_norm": 1.4921875,
+      "learning_rate": 4.6334143271113086e-05,
+      "loss": 0.4837,
+      "step": 8040
+    },
+    {
+      "epoch": 1.9362214199759324,
+      "grad_norm": 1.7734375,
+      "learning_rate": 4.629132781574886e-05,
+      "loss": 0.5211,
+      "step": 8045
+    },
+    {
+      "epoch": 1.9374247894103491,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.6248538717321385e-05,
+      "loss": 0.5349,
+      "step": 8050
+    },
+    {
+      "epoch": 1.9386281588447654,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.620577604416369e-05,
+      "loss": 0.5315,
+      "step": 8055
+    },
+    {
+      "epoch": 1.9398315282791816,
+      "grad_norm": 1.5625,
+      "learning_rate": 4.616303986456659e-05,
+      "loss": 0.5028,
+      "step": 8060
+    },
+    {
+      "epoch": 1.941034897713598,
+      "grad_norm": 1.7109375,
+      "learning_rate": 4.612033024677859e-05,
+      "loss": 0.5402,
+      "step": 8065
+    },
+    {
+      "epoch": 1.9422382671480145,
+      "grad_norm": 1.7109375,
+      "learning_rate": 4.607764725900576e-05,
+      "loss": 0.5104,
+      "step": 8070
+    },
+    {
+      "epoch": 1.9434416365824307,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.603499096941167e-05,
+      "loss": 0.5094,
+      "step": 8075
+    },
+    {
+      "epoch": 1.9446450060168472,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.5992361446117254e-05,
+      "loss": 0.5097,
+      "step": 8080
+    },
+    {
+      "epoch": 1.9458483754512637,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.5949758757200655e-05,
+      "loss": 0.485,
+      "step": 8085
+    },
+    {
+      "epoch": 1.94705174488568,
+      "grad_norm": 1.546875,
+      "learning_rate": 4.590718297069724e-05,
+      "loss": 0.5589,
+      "step": 8090
+    },
+    {
+      "epoch": 1.9482551143200961,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.586463415459933e-05,
+      "loss": 0.5094,
+      "step": 8095
+    },
+    {
+      "epoch": 1.9494584837545126,
+      "grad_norm": 1.625,
+      "learning_rate": 4.5822112376856236e-05,
+      "loss": 0.5018,
+      "step": 8100
+    },
+    {
+      "epoch": 1.950661853188929,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.577961770537406e-05,
+      "loss": 0.5215,
+      "step": 8105
+    },
+    {
+      "epoch": 1.9518652226233453,
+      "grad_norm": 2.78125,
+      "learning_rate": 4.573715020801563e-05,
+      "loss": 0.5018,
+      "step": 8110
+    },
+    {
+      "epoch": 1.9530685920577617,
+      "grad_norm": 1.671875,
+      "learning_rate": 4.569470995260035e-05,
+      "loss": 0.5245,
+      "step": 8115
+    },
+    {
+      "epoch": 1.9542719614921782,
+      "grad_norm": 1.671875,
+      "learning_rate": 4.565229700690417e-05,
+      "loss": 0.5267,
+      "step": 8120
+    },
+    {
+      "epoch": 1.9554753309265944,
+      "grad_norm": 1.6171875,
+      "learning_rate": 4.560991143865939e-05,
+      "loss": 0.5192,
+      "step": 8125
+    },
+    {
+      "epoch": 1.9566787003610109,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.556755331555459e-05,
+      "loss": 0.5226,
+      "step": 8130
+    },
+    {
+      "epoch": 1.9578820697954273,
+      "grad_norm": 1.640625,
+      "learning_rate": 4.552522270523454e-05,
+      "loss": 0.5142,
+      "step": 8135
+    },
+    {
+      "epoch": 1.9590854392298436,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.5482919675300036e-05,
+      "loss": 0.5439,
+      "step": 8140
+    },
+    {
+      "epoch": 1.9602888086642598,
+      "grad_norm": 1.7421875,
+      "learning_rate": 4.544064429330788e-05,
+      "loss": 0.5377,
+      "step": 8145
+    },
+    {
+      "epoch": 1.9614921780986763,
+      "grad_norm": 1.546875,
+      "learning_rate": 4.5398396626770674e-05,
+      "loss": 0.4933,
+      "step": 8150
+    },
+    {
+      "epoch": 1.9626955475330927,
+      "grad_norm": 1.640625,
+      "learning_rate": 4.535617674315682e-05,
+      "loss": 0.5361,
+      "step": 8155
+    },
+    {
+      "epoch": 1.963898916967509,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.5313984709890246e-05,
+      "loss": 0.4983,
+      "step": 8160
+    },
+    {
+      "epoch": 1.9651022864019254,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.5271820594350514e-05,
+      "loss": 0.4995,
+      "step": 8165
+    },
+    {
+      "epoch": 1.9663056558363419,
+      "grad_norm": 1.7109375,
+      "learning_rate": 4.522968446387257e-05,
+      "loss": 0.5564,
+      "step": 8170
+    },
+    {
+      "epoch": 1.967509025270758,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.5187576385746606e-05,
+      "loss": 0.5333,
+      "step": 8175
+    },
+    {
+      "epoch": 1.9687123947051743,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.5145496427218115e-05,
+      "loss": 0.5434,
+      "step": 8180
+    },
+    {
+      "epoch": 1.969915764139591,
+      "grad_norm": 1.7109375,
+      "learning_rate": 4.51034446554876e-05,
+      "loss": 0.5388,
+      "step": 8185
+    },
+    {
+      "epoch": 1.9711191335740073,
+      "grad_norm": 1.5859375,
+      "learning_rate": 4.50614211377106e-05,
+      "loss": 0.5048,
+      "step": 8190
+    },
+    {
+      "epoch": 1.9723225030084235,
+      "grad_norm": 1.7421875,
+      "learning_rate": 4.501942594099751e-05,
+      "loss": 0.5497,
+      "step": 8195
+    },
+    {
+      "epoch": 1.97352587244284,
+      "grad_norm": 1.546875,
+      "learning_rate": 4.497745913241351e-05,
+      "loss": 0.5155,
+      "step": 8200
+    },
+    {
+      "epoch": 1.9747292418772564,
+      "grad_norm": 1.53125,
+      "learning_rate": 4.4935520778978436e-05,
+      "loss": 0.5105,
+      "step": 8205
+    },
+    {
+      "epoch": 1.9759326113116726,
+      "grad_norm": 1.8125,
+      "learning_rate": 4.489361094766668e-05,
+      "loss": 0.5334,
+      "step": 8210
+    },
+    {
+      "epoch": 1.977135980746089,
+      "grad_norm": 1.6640625,
+      "learning_rate": 4.4851729705407116e-05,
+      "loss": 0.5078,
+      "step": 8215
+    },
+    {
+      "epoch": 1.9783393501805056,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.4809877119082895e-05,
+      "loss": 0.5246,
+      "step": 8220
+    },
+    {
+      "epoch": 1.9795427196149218,
+      "grad_norm": 1.6484375,
+      "learning_rate": 4.476805325553149e-05,
+      "loss": 0.5111,
+      "step": 8225
+    },
+    {
+      "epoch": 1.980746089049338,
+      "grad_norm": 1.5859375,
+      "learning_rate": 4.472625818154443e-05,
+      "loss": 0.5518,
+      "step": 8230
+    },
+    {
+      "epoch": 1.9819494584837545,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.4684491963867316e-05,
+      "loss": 0.4914,
+      "step": 8235
+    },
+    {
+      "epoch": 1.983152827918171,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.464275466919963e-05,
+      "loss": 0.4908,
+      "step": 8240
+    },
+    {
+      "epoch": 1.9843561973525872,
+      "grad_norm": 1.734375,
+      "learning_rate": 4.460104636419471e-05,
+      "loss": 0.5182,
+      "step": 8245
+    },
+    {
+      "epoch": 1.9855595667870036,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.455936711545955e-05,
+      "loss": 0.5369,
+      "step": 8250
+    },
+    {
+      "epoch": 1.98676293622142,
+      "grad_norm": 1.765625,
+      "learning_rate": 4.4517716989554765e-05,
+      "loss": 0.5488,
+      "step": 8255
+    },
+    {
+      "epoch": 1.9879663056558363,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.447609605299448e-05,
+      "loss": 0.5152,
+      "step": 8260
+    },
+    {
+      "epoch": 1.9891696750902526,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.443450437224617e-05,
+      "loss": 0.5413,
+      "step": 8265
+    },
+    {
+      "epoch": 1.9903730445246692,
+      "grad_norm": 1.6484375,
+      "learning_rate": 4.4392942013730636e-05,
+      "loss": 0.5233,
+      "step": 8270
+    },
+    {
+      "epoch": 1.9915764139590855,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.4351409043821766e-05,
+      "loss": 0.5414,
+      "step": 8275
+    },
+    {
+      "epoch": 1.9927797833935017,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.4309905528846614e-05,
+      "loss": 0.5151,
+      "step": 8280
+    },
+    {
+      "epoch": 1.9939831528279182,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.4268431535085135e-05,
+      "loss": 0.5228,
+      "step": 8285
+    },
+    {
+      "epoch": 1.9951865222623346,
+      "grad_norm": 1.5234375,
+      "learning_rate": 4.422698712877017e-05,
+      "loss": 0.5155,
+      "step": 8290
+    },
+    {
+      "epoch": 1.9963898916967509,
+      "grad_norm": 1.640625,
+      "learning_rate": 4.418557237608729e-05,
+      "loss": 0.5104,
+      "step": 8295
+    },
+    {
+      "epoch": 1.9975932611311673,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.41441873431747e-05,
+      "loss": 0.5153,
+      "step": 8300
+    },
+    {
+      "epoch": 1.9987966305655838,
+      "grad_norm": 1.6953125,
+      "learning_rate": 4.410283209612319e-05,
+      "loss": 0.5152,
+      "step": 8305
+    },
+    {
+      "epoch": 1.9995186522262336,
+      "eval_loss": 0.4374042749404907,
+      "eval_runtime": 2.6528,
+      "eval_samples_per_second": 75.393,
+      "eval_steps_per_second": 75.393,
+      "step": 8308
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.406150670097592e-05,
+      "loss": 0.5064,
+      "step": 8310
+    },
+    {
+      "epoch": 2.0012033694344162,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.4020211223728445e-05,
+      "loss": 0.4974,
+      "step": 8315
+    },
+    {
+      "epoch": 2.002406738868833,
+      "grad_norm": 1.671875,
+      "learning_rate": 4.3978945730328466e-05,
+      "loss": 0.4635,
+      "step": 8320
+    },
+    {
+      "epoch": 2.003610108303249,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.393771028667588e-05,
+      "loss": 0.487,
+      "step": 8325
+    },
+    {
+      "epoch": 2.0048134777376654,
+      "grad_norm": 1.640625,
+      "learning_rate": 4.389650495862253e-05,
+      "loss": 0.496,
+      "step": 8330
+    },
+    {
+      "epoch": 2.0060168471720816,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.38553298119722e-05,
+      "loss": 0.493,
+      "step": 8335
+    },
+    {
+      "epoch": 2.0072202166064983,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.381418491248049e-05,
+      "loss": 0.4847,
+      "step": 8340
+    },
+    {
+      "epoch": 2.0084235860409145,
+      "grad_norm": 1.515625,
+      "learning_rate": 4.377307032585463e-05,
+      "loss": 0.489,
+      "step": 8345
+    },
+    {
+      "epoch": 2.0096269554753308,
+      "grad_norm": 1.6171875,
+      "learning_rate": 4.373198611775352e-05,
+      "loss": 0.4824,
+      "step": 8350
+    },
+    {
+      "epoch": 2.0108303249097474,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.3690932353787495e-05,
+      "loss": 0.5218,
+      "step": 8355
+    },
+    {
+      "epoch": 2.0120336943441637,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.364990909951829e-05,
+      "loss": 0.4906,
+      "step": 8360
+    },
+    {
+      "epoch": 2.01323706377858,
+      "grad_norm": 1.7890625,
+      "learning_rate": 4.36089164204589e-05,
+      "loss": 0.5099,
+      "step": 8365
+    },
+    {
+      "epoch": 2.0144404332129966,
+      "grad_norm": 1.5546875,
+      "learning_rate": 4.3567954382073545e-05,
+      "loss": 0.4783,
+      "step": 8370
+    },
+    {
+      "epoch": 2.015643802647413,
+      "grad_norm": 1.546875,
+      "learning_rate": 4.352702304977743e-05,
+      "loss": 0.4765,
+      "step": 8375
+    },
+    {
+      "epoch": 2.016847172081829,
+      "grad_norm": 1.7421875,
+      "learning_rate": 4.348612248893679e-05,
+      "loss": 0.503,
+      "step": 8380
+    },
+    {
+      "epoch": 2.0180505415162453,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.3445252764868696e-05,
+      "loss": 0.4434,
+      "step": 8385
+    },
+    {
+      "epoch": 2.019253910950662,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.340441394284097e-05,
+      "loss": 0.4639,
+      "step": 8390
+    },
+    {
+      "epoch": 2.020457280385078,
+      "grad_norm": 1.4921875,
+      "learning_rate": 4.3363606088072096e-05,
+      "loss": 0.4673,
+      "step": 8395
+    },
+    {
+      "epoch": 2.0216606498194944,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.332282926573107e-05,
+      "loss": 0.4869,
+      "step": 8400
+    },
+    {
+      "epoch": 2.022864019253911,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.3282083540937395e-05,
+      "loss": 0.4653,
+      "step": 8405
+    },
+    {
+      "epoch": 2.0240673886883274,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.324136897876084e-05,
+      "loss": 0.4799,
+      "step": 8410
+    },
+    {
+      "epoch": 2.0252707581227436,
+      "grad_norm": 1.640625,
+      "learning_rate": 4.320068564422146e-05,
+      "loss": 0.5021,
+      "step": 8415
+    },
+    {
+      "epoch": 2.02647412755716,
+      "grad_norm": 1.625,
+      "learning_rate": 4.316003360228941e-05,
+      "loss": 0.4804,
+      "step": 8420
+    },
+    {
+      "epoch": 2.0276774969915765,
+      "grad_norm": 1.8203125,
+      "learning_rate": 4.311941291788489e-05,
+      "loss": 0.5143,
+      "step": 8425
+    },
+    {
+      "epoch": 2.0288808664259927,
+      "grad_norm": 1.5078125,
+      "learning_rate": 4.307882365587802e-05,
+      "loss": 0.501,
+      "step": 8430
+    },
+    {
+      "epoch": 2.030084235860409,
+      "grad_norm": 1.6796875,
+      "learning_rate": 4.303826588108873e-05,
+      "loss": 0.5066,
+      "step": 8435
+    },
+    {
+      "epoch": 2.0312876052948257,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.299773965828667e-05,
+      "loss": 0.497,
+      "step": 8440
+    },
+    {
+      "epoch": 2.032490974729242,
+      "grad_norm": 1.6171875,
+      "learning_rate": 4.295724505219109e-05,
+      "loss": 0.4801,
+      "step": 8445
+    },
+    {
+      "epoch": 2.033694344163658,
+      "grad_norm": 1.765625,
+      "learning_rate": 4.291678212747081e-05,
+      "loss": 0.4918,
+      "step": 8450
+    },
+    {
+      "epoch": 2.034897713598075,
+      "grad_norm": 1.5234375,
+      "learning_rate": 4.287635094874395e-05,
+      "loss": 0.4965,
+      "step": 8455
+    },
+    {
+      "epoch": 2.036101083032491,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.283595158057803e-05,
+      "loss": 0.507,
+      "step": 8460
+    },
+    {
+      "epoch": 2.0373044524669073,
+      "grad_norm": 1.4921875,
+      "learning_rate": 4.2795584087489705e-05,
+      "loss": 0.5007,
+      "step": 8465
+    },
+    {
+      "epoch": 2.0385078219013235,
+      "grad_norm": 1.7265625,
+      "learning_rate": 4.275524853394477e-05,
+      "loss": 0.498,
+      "step": 8470
+    },
+    {
+      "epoch": 2.03971119133574,
+      "grad_norm": 1.6484375,
+      "learning_rate": 4.2714944984357995e-05,
+      "loss": 0.4973,
+      "step": 8475
+    },
+    {
+      "epoch": 2.0409145607701564,
+      "grad_norm": 1.625,
+      "learning_rate": 4.267467350309302e-05,
+      "loss": 0.4778,
+      "step": 8480
+    },
+    {
+      "epoch": 2.0421179302045727,
+      "grad_norm": 1.671875,
+      "learning_rate": 4.2634434154462325e-05,
+      "loss": 0.4889,
+      "step": 8485
+    },
+    {
+      "epoch": 2.0433212996389893,
+      "grad_norm": 1.4765625,
+      "learning_rate": 4.259422700272701e-05,
+      "loss": 0.4903,
+      "step": 8490
+    },
+    {
+      "epoch": 2.0445246690734056,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.255405211209682e-05,
+      "loss": 0.5066,
+      "step": 8495
+    },
+    {
+      "epoch": 2.045728038507822,
+      "grad_norm": 1.75,
+      "learning_rate": 4.251390954672993e-05,
+      "loss": 0.4997,
+      "step": 8500
+    },
+    {
+      "epoch": 2.045728038507822,
+      "eval_loss": 0.4380384087562561,
+      "eval_runtime": 2.6444,
+      "eval_samples_per_second": 75.632,
+      "eval_steps_per_second": 75.632,
+      "step": 8500
+    },
+    {
+      "epoch": 2.046931407942238,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.247379937073293e-05,
+      "loss": 0.5082,
+      "step": 8505
+    },
+    {
+      "epoch": 2.0481347773766547,
+      "grad_norm": 1.765625,
+      "learning_rate": 4.243372164816064e-05,
+      "loss": 0.4997,
+      "step": 8510
+    },
+    {
+      "epoch": 2.049338146811071,
+      "grad_norm": 1.734375,
+      "learning_rate": 4.23936764430161e-05,
+      "loss": 0.5169,
+      "step": 8515
+    },
+    {
+      "epoch": 2.050541516245487,
+      "grad_norm": 1.75,
+      "learning_rate": 4.235366381925042e-05,
+      "loss": 0.4879,
+      "step": 8520
+    },
+    {
+      "epoch": 2.051744885679904,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.231368384076263e-05,
+      "loss": 0.4754,
+      "step": 8525
+    },
+    {
+      "epoch": 2.05294825511432,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.227373657139968e-05,
+      "loss": 0.5083,
+      "step": 8530
+    },
+    {
+      "epoch": 2.0541516245487363,
+      "grad_norm": 1.8203125,
+      "learning_rate": 4.223382207495623e-05,
+      "loss": 0.4871,
+      "step": 8535
+    },
+    {
+      "epoch": 2.055354993983153,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.219394041517467e-05,
+      "loss": 0.4533,
+      "step": 8540
+    },
+    {
+      "epoch": 2.0565583634175693,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.215409165574487e-05,
+      "loss": 0.4723,
+      "step": 8545
+    },
+    {
+      "epoch": 2.0577617328519855,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.211427586030426e-05,
+      "loss": 0.4635,
+      "step": 8550
+    },
+    {
+      "epoch": 2.0589651022864017,
+      "grad_norm": 1.5859375,
+      "learning_rate": 4.207449309243752e-05,
+      "loss": 0.4732,
+      "step": 8555
+    },
+    {
+      "epoch": 2.0601684717208184,
+      "grad_norm": 1.7734375,
+      "learning_rate": 4.203474341567665e-05,
+      "loss": 0.5234,
+      "step": 8560
+    },
+    {
+      "epoch": 2.0613718411552346,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.199502689350083e-05,
+      "loss": 0.487,
+      "step": 8565
+    },
+    {
+      "epoch": 2.062575210589651,
+      "grad_norm": 1.6640625,
+      "learning_rate": 4.195534358933619e-05,
+      "loss": 0.4711,
+      "step": 8570
+    },
+    {
+      "epoch": 2.0637785800240676,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.191569356655592e-05,
+      "loss": 0.4706,
+      "step": 8575
+    },
+    {
+      "epoch": 2.064981949458484,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.1876076888479995e-05,
+      "loss": 0.5129,
+      "step": 8580
+    },
+    {
+      "epoch": 2.0661853188929,
+      "grad_norm": 1.6796875,
+      "learning_rate": 4.183649361837518e-05,
+      "loss": 0.4672,
+      "step": 8585
+    },
+    {
+      "epoch": 2.0673886883273163,
+      "grad_norm": 1.546875,
+      "learning_rate": 4.179694381945485e-05,
+      "loss": 0.4796,
+      "step": 8590
+    },
+    {
+      "epoch": 2.068592057761733,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.175742755487896e-05,
+      "loss": 0.484,
+      "step": 8595
+    },
+    {
+      "epoch": 2.069795427196149,
+      "grad_norm": 1.6328125,
+      "learning_rate": 4.171794488775388e-05,
+      "loss": 0.5078,
+      "step": 8600
+    },
+    {
+      "epoch": 2.0709987966305654,
+      "grad_norm": 1.6953125,
+      "learning_rate": 4.167849588113236e-05,
+      "loss": 0.4734,
+      "step": 8605
+    },
+    {
+      "epoch": 2.072202166064982,
+      "grad_norm": 1.6484375,
+      "learning_rate": 4.1639080598013376e-05,
+      "loss": 0.4798,
+      "step": 8610
+    },
+    {
+      "epoch": 2.0734055354993983,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.1599699101342056e-05,
+      "loss": 0.4717,
+      "step": 8615
+    },
+    {
+      "epoch": 2.0746089049338146,
+      "grad_norm": 1.8359375,
+      "learning_rate": 4.156035145400956e-05,
+      "loss": 0.5044,
+      "step": 8620
+    },
+    {
+      "epoch": 2.0758122743682312,
+      "grad_norm": 1.7578125,
+      "learning_rate": 4.1521037718852995e-05,
+      "loss": 0.5029,
+      "step": 8625
+    },
+    {
+      "epoch": 2.0770156438026475,
+      "grad_norm": 2.1875,
+      "learning_rate": 4.148175795865533e-05,
+      "loss": 0.4681,
+      "step": 8630
+    },
+    {
+      "epoch": 2.0782190132370637,
+      "grad_norm": 1.578125,
+      "learning_rate": 4.1442512236145245e-05,
+      "loss": 0.4573,
+      "step": 8635
+    },
+    {
+      "epoch": 2.07942238267148,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.140330061399709e-05,
+      "loss": 0.4982,
+      "step": 8640
+    },
+    {
+      "epoch": 2.0806257521058966,
+      "grad_norm": 1.625,
+      "learning_rate": 4.136412315483074e-05,
+      "loss": 0.5154,
+      "step": 8645
+    },
+    {
+      "epoch": 2.081829121540313,
+      "grad_norm": 1.5234375,
+      "learning_rate": 4.132497992121152e-05,
+      "loss": 0.4675,
+      "step": 8650
+    },
+    {
+      "epoch": 2.083032490974729,
+      "grad_norm": 1.75,
+      "learning_rate": 4.1285870975650125e-05,
+      "loss": 0.5419,
+      "step": 8655
+    },
+    {
+      "epoch": 2.0842358604091458,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.1246796380602425e-05,
+      "loss": 0.4852,
+      "step": 8660
+    },
+    {
+      "epoch": 2.085439229843562,
+      "grad_norm": 1.59375,
+      "learning_rate": 4.12077561984695e-05,
+      "loss": 0.4931,
+      "step": 8665
+    },
+    {
+      "epoch": 2.0866425992779782,
+      "grad_norm": 1.8125,
+      "learning_rate": 4.116875049159743e-05,
+      "loss": 0.4868,
+      "step": 8670
+    },
+    {
+      "epoch": 2.087845968712395,
+      "grad_norm": 1.609375,
+      "learning_rate": 4.112977932227729e-05,
+      "loss": 0.4852,
+      "step": 8675
+    },
+    {
+      "epoch": 2.089049338146811,
+      "grad_norm": 1.7109375,
+      "learning_rate": 4.109084275274492e-05,
+      "loss": 0.4568,
+      "step": 8680
+    },
+    {
+      "epoch": 2.0902527075812274,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.105194084518098e-05,
+      "loss": 0.507,
+      "step": 8685
+    },
+    {
+      "epoch": 2.0914560770156436,
+      "grad_norm": 1.640625,
+      "learning_rate": 4.101307366171073e-05,
+      "loss": 0.4767,
+      "step": 8690
+    },
+    {
+      "epoch": 2.0926594464500603,
+      "grad_norm": 1.640625,
+      "learning_rate": 4.0974241264404015e-05,
+      "loss": 0.4906,
+      "step": 8695
+    },
+    {
+      "epoch": 2.0938628158844765,
+      "grad_norm": 1.75,
+      "learning_rate": 4.09354437152751e-05,
+      "loss": 0.4946,
+      "step": 8700
+    },
+    {
+      "epoch": 2.0950661853188928,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.089668107628259e-05,
+      "loss": 0.523,
+      "step": 8705
+    },
+    {
+      "epoch": 2.0962695547533094,
+      "grad_norm": 1.7734375,
+      "learning_rate": 4.085795340932937e-05,
+      "loss": 0.4806,
+      "step": 8710
+    },
+    {
+      "epoch": 2.0974729241877257,
+      "grad_norm": 1.6796875,
+      "learning_rate": 4.081926077626244e-05,
+      "loss": 0.4834,
+      "step": 8715
+    },
+    {
+      "epoch": 2.098676293622142,
+      "grad_norm": 1.6484375,
+      "learning_rate": 4.0780603238872895e-05,
+      "loss": 0.488,
+      "step": 8720
+    },
+    {
+      "epoch": 2.099879663056558,
+      "grad_norm": 1.6171875,
+      "learning_rate": 4.0741980858895734e-05,
+      "loss": 0.4758,
+      "step": 8725
+    },
+    {
+      "epoch": 2.101083032490975,
+      "grad_norm": 1.6484375,
+      "learning_rate": 4.0703393698009855e-05,
+      "loss": 0.4907,
+      "step": 8730
+    },
+    {
+      "epoch": 2.102286401925391,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.066484181783788e-05,
+      "loss": 0.4809,
+      "step": 8735
+    },
+    {
+      "epoch": 2.1034897713598073,
+      "grad_norm": 1.7578125,
+      "learning_rate": 4.062632527994611e-05,
+      "loss": 0.5182,
+      "step": 8740
+    },
+    {
+      "epoch": 2.104693140794224,
+      "grad_norm": 1.5859375,
+      "learning_rate": 4.058784414584441e-05,
+      "loss": 0.4739,
+      "step": 8745
+    },
+    {
+      "epoch": 2.10589651022864,
+      "grad_norm": 1.6796875,
+      "learning_rate": 4.054939847698606e-05,
+      "loss": 0.5135,
+      "step": 8750
+    },
+    {
+      "epoch": 2.1070998796630565,
+      "grad_norm": 1.6015625,
+      "learning_rate": 4.0510988334767776e-05,
+      "loss": 0.514,
+      "step": 8755
+    },
+    {
+      "epoch": 2.108303249097473,
+      "grad_norm": 1.6640625,
+      "learning_rate": 4.047261378052946e-05,
+      "loss": 0.4837,
+      "step": 8760
+    },
+    {
+      "epoch": 2.1095066185318894,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.043427487555425e-05,
+      "loss": 0.5302,
+      "step": 8765
+    },
+    {
+      "epoch": 2.1107099879663056,
+      "grad_norm": 1.765625,
+      "learning_rate": 4.03959716810683e-05,
+      "loss": 0.5048,
+      "step": 8770
+    },
+    {
+      "epoch": 2.111913357400722,
+      "grad_norm": 1.6796875,
+      "learning_rate": 4.0357704258240775e-05,
+      "loss": 0.4944,
+      "step": 8775
+    },
+    {
+      "epoch": 2.1131167268351385,
+      "grad_norm": 1.546875,
+      "learning_rate": 4.0319472668183685e-05,
+      "loss": 0.4866,
+      "step": 8780
+    },
+    {
+      "epoch": 2.1143200962695547,
+      "grad_norm": 1.5546875,
+      "learning_rate": 4.028127697195183e-05,
+      "loss": 0.4931,
+      "step": 8785
+    },
+    {
+      "epoch": 2.115523465703971,
+      "grad_norm": 1.5859375,
+      "learning_rate": 4.024311723054271e-05,
+      "loss": 0.4933,
+      "step": 8790
+    },
+    {
+      "epoch": 2.1167268351383877,
+      "grad_norm": 1.6796875,
+      "learning_rate": 4.020499350489635e-05,
+      "loss": 0.4959,
+      "step": 8795
+    },
+    {
+      "epoch": 2.117930204572804,
+      "grad_norm": 1.5703125,
+      "learning_rate": 4.016690585589532e-05,
+      "loss": 0.4659,
+      "step": 8800
+    },
+    {
+      "epoch": 2.11913357400722,
+      "grad_norm": 1.6796875,
+      "learning_rate": 4.012885434436451e-05,
+      "loss": 0.4954,
+      "step": 8805
+    },
+    {
+      "epoch": 2.1203369434416364,
+      "grad_norm": 1.671875,
+      "learning_rate": 4.009083903107119e-05,
+      "loss": 0.4986,
+      "step": 8810
+    },
+    {
+      "epoch": 2.121540312876053,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.005285997672472e-05,
+      "loss": 0.4771,
+      "step": 8815
+    },
+    {
+      "epoch": 2.1227436823104693,
+      "grad_norm": 1.765625,
+      "learning_rate": 4.001491724197664e-05,
+      "loss": 0.4917,
+      "step": 8820
+    },
+    {
+      "epoch": 2.1239470517448855,
+      "grad_norm": 1.765625,
+      "learning_rate": 3.9977010887420435e-05,
+      "loss": 0.4628,
+      "step": 8825
+    },
+    {
+      "epoch": 2.125150421179302,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.993914097359151e-05,
+      "loss": 0.4825,
+      "step": 8830
+    },
+    {
+      "epoch": 2.1263537906137184,
+      "grad_norm": 1.6328125,
+      "learning_rate": 3.99013075609671e-05,
+      "loss": 0.488,
+      "step": 8835
+    },
+    {
+      "epoch": 2.1275571600481347,
+      "grad_norm": 1.703125,
+      "learning_rate": 3.98635107099661e-05,
+      "loss": 0.4911,
+      "step": 8840
+    },
+    {
+      "epoch": 2.1287605294825513,
+      "grad_norm": 1.8359375,
+      "learning_rate": 3.982575048094906e-05,
+      "loss": 0.5006,
+      "step": 8845
+    },
+    {
+      "epoch": 2.1299638989169676,
+      "grad_norm": 1.6171875,
+      "learning_rate": 3.9788026934218e-05,
+      "loss": 0.4646,
+      "step": 8850
+    },
+    {
+      "epoch": 2.131167268351384,
+      "grad_norm": 1.6796875,
+      "learning_rate": 3.975034013001642e-05,
+      "loss": 0.4789,
+      "step": 8855
+    },
+    {
+      "epoch": 2.1323706377858,
+      "grad_norm": 1.5859375,
+      "learning_rate": 3.971269012852908e-05,
+      "loss": 0.5018,
+      "step": 8860
+    },
+    {
+      "epoch": 2.1335740072202167,
+      "grad_norm": 1.75,
+      "learning_rate": 3.967507698988203e-05,
+      "loss": 0.4741,
+      "step": 8865
+    },
+    {
+      "epoch": 2.134777376654633,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.963750077414238e-05,
+      "loss": 0.4832,
+      "step": 8870
+    },
+    {
+      "epoch": 2.135980746089049,
+      "grad_norm": 1.625,
+      "learning_rate": 3.959996154131834e-05,
+      "loss": 0.4968,
+      "step": 8875
+    },
+    {
+      "epoch": 2.137184115523466,
+      "grad_norm": 1.7890625,
+      "learning_rate": 3.9562459351359045e-05,
+      "loss": 0.4943,
+      "step": 8880
+    },
+    {
+      "epoch": 2.138387484957882,
+      "grad_norm": 1.78125,
+      "learning_rate": 3.9524994264154416e-05,
+      "loss": 0.4864,
+      "step": 8885
+    },
+    {
+      "epoch": 2.1395908543922983,
+      "grad_norm": 1.625,
+      "learning_rate": 3.948756633953521e-05,
+      "loss": 0.4893,
+      "step": 8890
+    },
+    {
+      "epoch": 2.140794223826715,
+      "grad_norm": 1.5625,
+      "learning_rate": 3.945017563727277e-05,
+      "loss": 0.4954,
+      "step": 8895
+    },
+    {
+      "epoch": 2.1419975932611313,
+      "grad_norm": 1.5546875,
+      "learning_rate": 3.941282221707903e-05,
+      "loss": 0.482,
+      "step": 8900
+    },
+    {
+      "epoch": 2.1432009626955475,
+      "grad_norm": 1.6015625,
+      "learning_rate": 3.937550613860637e-05,
+      "loss": 0.4891,
+      "step": 8905
+    },
+    {
+      "epoch": 2.1444043321299637,
+      "grad_norm": 1.703125,
+      "learning_rate": 3.933822746144755e-05,
+      "loss": 0.4836,
+      "step": 8910
+    },
+    {
+      "epoch": 2.1456077015643804,
+      "grad_norm": 1.6328125,
+      "learning_rate": 3.9300986245135574e-05,
+      "loss": 0.4893,
+      "step": 8915
+    },
+    {
+      "epoch": 2.1468110709987966,
+      "grad_norm": 1.734375,
+      "learning_rate": 3.926378254914365e-05,
+      "loss": 0.5184,
+      "step": 8920
+    },
+    {
+      "epoch": 2.148014440433213,
+      "grad_norm": 1.8515625,
+      "learning_rate": 3.922661643288507e-05,
+      "loss": 0.4951,
+      "step": 8925
+    },
+    {
+      "epoch": 2.1492178098676296,
+      "grad_norm": 1.71875,
+      "learning_rate": 3.918948795571309e-05,
+      "loss": 0.5026,
+      "step": 8930
+    },
+    {
+      "epoch": 2.150421179302046,
+      "grad_norm": 1.6171875,
+      "learning_rate": 3.9152397176920856e-05,
+      "loss": 0.4973,
+      "step": 8935
+    },
+    {
+      "epoch": 2.151624548736462,
+      "grad_norm": 1.921875,
+      "learning_rate": 3.911534415574134e-05,
+      "loss": 0.4689,
+      "step": 8940
+    },
+    {
+      "epoch": 2.1528279181708783,
+      "grad_norm": 1.75,
+      "learning_rate": 3.9078328951347194e-05,
+      "loss": 0.4896,
+      "step": 8945
+    },
+    {
+      "epoch": 2.154031287605295,
+      "grad_norm": 1.6796875,
+      "learning_rate": 3.904135162285066e-05,
+      "loss": 0.5008,
+      "step": 8950
+    },
+    {
+      "epoch": 2.155234657039711,
+      "grad_norm": 1.578125,
+      "learning_rate": 3.9004412229303564e-05,
+      "loss": 0.4997,
+      "step": 8955
+    },
+    {
+      "epoch": 2.1564380264741274,
+      "grad_norm": 1.8046875,
+      "learning_rate": 3.8967510829697046e-05,
+      "loss": 0.5027,
+      "step": 8960
+    },
+    {
+      "epoch": 2.157641395908544,
+      "grad_norm": 1.8125,
+      "learning_rate": 3.8930647482961664e-05,
+      "loss": 0.4757,
+      "step": 8965
+    },
+    {
+      "epoch": 2.1588447653429603,
+      "grad_norm": 1.5625,
+      "learning_rate": 3.8893822247967144e-05,
+      "loss": 0.4581,
+      "step": 8970
+    },
+    {
+      "epoch": 2.1600481347773766,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.885703518352239e-05,
+      "loss": 0.4969,
+      "step": 8975
+    },
+    {
+      "epoch": 2.1612515042117932,
+      "grad_norm": 1.7890625,
+      "learning_rate": 3.8820286348375334e-05,
+      "loss": 0.5136,
+      "step": 8980
+    },
+    {
+      "epoch": 2.1624548736462095,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.8783575801212846e-05,
+      "loss": 0.471,
+      "step": 8985
+    },
+    {
+      "epoch": 2.1636582430806257,
+      "grad_norm": 1.625,
+      "learning_rate": 3.874690360066068e-05,
+      "loss": 0.4926,
+      "step": 8990
+    },
+    {
+      "epoch": 2.164861612515042,
+      "grad_norm": 1.5390625,
+      "learning_rate": 3.871026980528329e-05,
+      "loss": 0.476,
+      "step": 8995
+    },
+    {
+      "epoch": 2.1660649819494586,
+      "grad_norm": 1.5390625,
+      "learning_rate": 3.867367447358389e-05,
+      "loss": 0.4717,
+      "step": 9000
+    },
+    {
+      "epoch": 2.1660649819494586,
+      "eval_loss": 0.4340422749519348,
+      "eval_runtime": 2.659,
+      "eval_samples_per_second": 75.217,
+      "eval_steps_per_second": 75.217,
+      "step": 9000
+    },
+    {
+      "epoch": 2.167268351383875,
+      "grad_norm": 1.578125,
+      "learning_rate": 3.863711766400418e-05,
+      "loss": 0.4822,
+      "step": 9005
+    },
+    {
+      "epoch": 2.168471720818291,
+      "grad_norm": 1.7890625,
+      "learning_rate": 3.860059943492441e-05,
+      "loss": 0.4784,
+      "step": 9010
+    },
+    {
+      "epoch": 2.1696750902527078,
+      "grad_norm": 1.796875,
+      "learning_rate": 3.856411984466316e-05,
+      "loss": 0.4664,
+      "step": 9015
+    },
+    {
+      "epoch": 2.170878459687124,
+      "grad_norm": 1.765625,
+      "learning_rate": 3.8527678951477334e-05,
+      "loss": 0.4541,
+      "step": 9020
+    },
+    {
+      "epoch": 2.1720818291215402,
+      "grad_norm": 1.5703125,
+      "learning_rate": 3.849127681356208e-05,
+      "loss": 0.4829,
+      "step": 9025
+    },
+    {
+      "epoch": 2.1732851985559565,
+      "grad_norm": 1.625,
+      "learning_rate": 3.8454913489050544e-05,
+      "loss": 0.4587,
+      "step": 9030
+    },
+    {
+      "epoch": 2.174488567990373,
+      "grad_norm": 1.5546875,
+      "learning_rate": 3.841858903601402e-05,
+      "loss": 0.4861,
+      "step": 9035
+    },
+    {
+      "epoch": 2.1756919374247894,
+      "grad_norm": 1.7890625,
+      "learning_rate": 3.8382303512461616e-05,
+      "loss": 0.4912,
+      "step": 9040
+    },
+    {
+      "epoch": 2.1768953068592056,
+      "grad_norm": 1.7109375,
+      "learning_rate": 3.8346056976340335e-05,
+      "loss": 0.4779,
+      "step": 9045
+    },
+    {
+      "epoch": 2.1780986762936223,
+      "grad_norm": 1.6328125,
+      "learning_rate": 3.830984948553489e-05,
+      "loss": 0.4761,
+      "step": 9050
+    },
+    {
+      "epoch": 2.1793020457280385,
+      "grad_norm": 1.890625,
+      "learning_rate": 3.827368109786767e-05,
+      "loss": 0.4887,
+      "step": 9055
+    },
+    {
+      "epoch": 2.1805054151624548,
+      "grad_norm": 1.546875,
+      "learning_rate": 3.8237551871098557e-05,
+      "loss": 0.4917,
+      "step": 9060
+    },
+    {
+      "epoch": 2.1817087845968715,
+      "grad_norm": 1.7109375,
+      "learning_rate": 3.8201461862924956e-05,
+      "loss": 0.4808,
+      "step": 9065
+    },
+    {
+      "epoch": 2.1829121540312877,
+      "grad_norm": 1.640625,
+      "learning_rate": 3.8165411130981627e-05,
+      "loss": 0.4828,
+      "step": 9070
+    },
+    {
+      "epoch": 2.184115523465704,
+      "grad_norm": 1.6640625,
+      "learning_rate": 3.812939973284058e-05,
+      "loss": 0.4741,
+      "step": 9075
+    },
+    {
+      "epoch": 2.18531889290012,
+      "grad_norm": 1.6484375,
+      "learning_rate": 3.809342772601104e-05,
+      "loss": 0.4735,
+      "step": 9080
+    },
+    {
+      "epoch": 2.186522262334537,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.80574951679393e-05,
+      "loss": 0.4672,
+      "step": 9085
+    },
+    {
+      "epoch": 2.187725631768953,
+      "grad_norm": 1.6875,
+      "learning_rate": 3.802160211600868e-05,
+      "loss": 0.4813,
+      "step": 9090
+    },
+    {
+      "epoch": 2.1889290012033693,
+      "grad_norm": 1.6328125,
+      "learning_rate": 3.798574862753938e-05,
+      "loss": 0.5285,
+      "step": 9095
+    },
+    {
+      "epoch": 2.190132370637786,
+      "grad_norm": 1.71875,
+      "learning_rate": 3.7949934759788466e-05,
+      "loss": 0.4922,
+      "step": 9100
+    },
+    {
+      "epoch": 2.191335740072202,
+      "grad_norm": 1.6796875,
+      "learning_rate": 3.791416056994966e-05,
+      "loss": 0.4752,
+      "step": 9105
+    },
+    {
+      "epoch": 2.1925391095066185,
+      "grad_norm": 1.65625,
+      "learning_rate": 3.787842611515339e-05,
+      "loss": 0.4723,
+      "step": 9110
+    },
+    {
+      "epoch": 2.1937424789410347,
+      "grad_norm": 1.578125,
+      "learning_rate": 3.784273145246659e-05,
+      "loss": 0.5092,
+      "step": 9115
+    },
+    {
+      "epoch": 2.1949458483754514,
+      "grad_norm": 1.6171875,
+      "learning_rate": 3.7807076638892646e-05,
+      "loss": 0.4837,
+      "step": 9120
+    },
+    {
+      "epoch": 2.1961492178098676,
+      "grad_norm": 1.546875,
+      "learning_rate": 3.777146173137134e-05,
+      "loss": 0.4488,
+      "step": 9125
+    },
+    {
+      "epoch": 2.197352587244284,
+      "grad_norm": 1.796875,
+      "learning_rate": 3.773588678677867e-05,
+      "loss": 0.4671,
+      "step": 9130
+    },
+    {
+      "epoch": 2.1985559566787005,
+      "grad_norm": 1.7578125,
+      "learning_rate": 3.770035186192685e-05,
+      "loss": 0.4949,
+      "step": 9135
+    },
+    {
+      "epoch": 2.1997593261131168,
+      "grad_norm": 1.53125,
+      "learning_rate": 3.766485701356419e-05,
+      "loss": 0.4647,
+      "step": 9140
+    },
+    {
+      "epoch": 2.200962695547533,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.7629402298374994e-05,
+      "loss": 0.4678,
+      "step": 9145
+    },
+    {
+      "epoch": 2.2021660649819497,
+      "grad_norm": 1.75,
+      "learning_rate": 3.7593987772979436e-05,
+      "loss": 0.4933,
+      "step": 9150
+    },
+    {
+      "epoch": 2.203369434416366,
+      "grad_norm": 1.625,
+      "learning_rate": 3.755861349393356e-05,
+      "loss": 0.4879,
+      "step": 9155
+    },
+    {
+      "epoch": 2.204572803850782,
+      "grad_norm": 1.59375,
+      "learning_rate": 3.752327951772913e-05,
+      "loss": 0.4716,
+      "step": 9160
+    },
+    {
+      "epoch": 2.2057761732851984,
+      "grad_norm": 1.59375,
+      "learning_rate": 3.74879859007935e-05,
+      "loss": 0.5021,
+      "step": 9165
+    },
+    {
+      "epoch": 2.206979542719615,
+      "grad_norm": 1.65625,
+      "learning_rate": 3.745273269948964e-05,
+      "loss": 0.4915,
+      "step": 9170
+    },
+    {
+      "epoch": 2.2081829121540313,
+      "grad_norm": 1.5859375,
+      "learning_rate": 3.7417519970115926e-05,
+      "loss": 0.4831,
+      "step": 9175
+    },
+    {
+      "epoch": 2.2093862815884475,
+      "grad_norm": 1.5390625,
+      "learning_rate": 3.738234776890612e-05,
+      "loss": 0.4543,
+      "step": 9180
+    },
+    {
+      "epoch": 2.210589651022864,
+      "grad_norm": 1.640625,
+      "learning_rate": 3.734721615202927e-05,
+      "loss": 0.4941,
+      "step": 9185
+    },
+    {
+      "epoch": 2.2117930204572804,
+      "grad_norm": 1.703125,
+      "learning_rate": 3.731212517558958e-05,
+      "loss": 0.485,
+      "step": 9190
+    },
+    {
+      "epoch": 2.2129963898916967,
+      "grad_norm": 1.59375,
+      "learning_rate": 3.72770748956264e-05,
+      "loss": 0.4662,
+      "step": 9195
+    },
+    {
+      "epoch": 2.214199759326113,
+      "grad_norm": 1.71875,
+      "learning_rate": 3.724206536811404e-05,
+      "loss": 0.4902,
+      "step": 9200
+    },
+    {
+      "epoch": 2.2154031287605296,
+      "grad_norm": 1.6875,
+      "learning_rate": 3.7207096648961774e-05,
+      "loss": 0.4659,
+      "step": 9205
+    },
+    {
+      "epoch": 2.216606498194946,
+      "grad_norm": 1.6640625,
+      "learning_rate": 3.717216879401367e-05,
+      "loss": 0.4956,
+      "step": 9210
+    },
+    {
+      "epoch": 2.217809867629362,
+      "grad_norm": 1.6171875,
+      "learning_rate": 3.713728185904856e-05,
+      "loss": 0.4943,
+      "step": 9215
+    },
+    {
+      "epoch": 2.2190132370637787,
+      "grad_norm": 1.5625,
+      "learning_rate": 3.710243589977991e-05,
+      "loss": 0.4873,
+      "step": 9220
+    },
+    {
+      "epoch": 2.220216606498195,
+      "grad_norm": 1.5703125,
+      "learning_rate": 3.706763097185577e-05,
+      "loss": 0.4603,
+      "step": 9225
+    },
+    {
+      "epoch": 2.221419975932611,
+      "grad_norm": 1.8515625,
+      "learning_rate": 3.7032867130858654e-05,
+      "loss": 0.4683,
+      "step": 9230
+    },
+    {
+      "epoch": 2.222623345367028,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.6998144432305444e-05,
+      "loss": 0.4932,
+      "step": 9235
+    },
+    {
+      "epoch": 2.223826714801444,
+      "grad_norm": 1.6328125,
+      "learning_rate": 3.696346293164737e-05,
+      "loss": 0.4947,
+      "step": 9240
+    },
+    {
+      "epoch": 2.2250300842358604,
+      "grad_norm": 1.6328125,
+      "learning_rate": 3.692882268426979e-05,
+      "loss": 0.4836,
+      "step": 9245
+    },
+    {
+      "epoch": 2.2262334536702766,
+      "grad_norm": 1.5390625,
+      "learning_rate": 3.6894223745492257e-05,
+      "loss": 0.4734,
+      "step": 9250
+    },
+    {
+      "epoch": 2.2274368231046933,
+      "grad_norm": 1.8046875,
+      "learning_rate": 3.685966617056831e-05,
+      "loss": 0.4968,
+      "step": 9255
+    },
+    {
+      "epoch": 2.2286401925391095,
+      "grad_norm": 1.640625,
+      "learning_rate": 3.682515001468546e-05,
+      "loss": 0.5096,
+      "step": 9260
+    },
+    {
+      "epoch": 2.2298435619735257,
+      "grad_norm": 1.53125,
+      "learning_rate": 3.6790675332965045e-05,
+      "loss": 0.4882,
+      "step": 9265
+    },
+    {
+      "epoch": 2.2310469314079424,
+      "grad_norm": 1.71875,
+      "learning_rate": 3.675624218046218e-05,
+      "loss": 0.4786,
+      "step": 9270
+    },
+    {
+      "epoch": 2.2322503008423586,
+      "grad_norm": 1.59375,
+      "learning_rate": 3.6721850612165695e-05,
+      "loss": 0.4842,
+      "step": 9275
+    },
+    {
+      "epoch": 2.233453670276775,
+      "grad_norm": 1.5859375,
+      "learning_rate": 3.668750068299794e-05,
+      "loss": 0.5154,
+      "step": 9280
+    },
+    {
+      "epoch": 2.234657039711191,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.6653192447814835e-05,
+      "loss": 0.4666,
+      "step": 9285
+    },
+    {
+      "epoch": 2.235860409145608,
+      "grad_norm": 1.5390625,
+      "learning_rate": 3.661892596140566e-05,
+      "loss": 0.5,
+      "step": 9290
+    },
+    {
+      "epoch": 2.237063778580024,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.6584701278493095e-05,
+      "loss": 0.4798,
+      "step": 9295
+    },
+    {
+      "epoch": 2.2382671480144403,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3.655051845373297e-05,
+      "loss": 0.449,
+      "step": 9300
+    },
+    {
+      "epoch": 2.239470517448857,
+      "grad_norm": 1.7109375,
+      "learning_rate": 3.651637754171435e-05,
+      "loss": 0.4917,
+      "step": 9305
+    },
+    {
+      "epoch": 2.240673886883273,
+      "grad_norm": 1.5859375,
+      "learning_rate": 3.648227859695931e-05,
+      "loss": 0.4574,
+      "step": 9310
+    },
+    {
+      "epoch": 2.2418772563176894,
+      "grad_norm": 1.7421875,
+      "learning_rate": 3.644822167392293e-05,
+      "loss": 0.5086,
+      "step": 9315
+    },
+    {
+      "epoch": 2.243080625752106,
+      "grad_norm": 1.765625,
+      "learning_rate": 3.6414206826993216e-05,
+      "loss": 0.4691,
+      "step": 9320
+    },
+    {
+      "epoch": 2.2442839951865223,
+      "grad_norm": 1.5,
+      "learning_rate": 3.638023411049089e-05,
+      "loss": 0.462,
+      "step": 9325
+    },
+    {
+      "epoch": 2.2454873646209386,
+      "grad_norm": 1.5859375,
+      "learning_rate": 3.634630357866948e-05,
+      "loss": 0.4666,
+      "step": 9330
+    },
+    {
+      "epoch": 2.246690734055355,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.6312415285715105e-05,
+      "loss": 0.511,
+      "step": 9335
+    },
+    {
+      "epoch": 2.2478941034897715,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.627856928574643e-05,
+      "loss": 0.4759,
+      "step": 9340
+    },
+    {
+      "epoch": 2.2490974729241877,
+      "grad_norm": 1.6328125,
+      "learning_rate": 3.6244765632814595e-05,
+      "loss": 0.4943,
+      "step": 9345
+    },
+    {
+      "epoch": 2.250300842358604,
+      "grad_norm": 1.546875,
+      "learning_rate": 3.621100438090311e-05,
+      "loss": 0.451,
+      "step": 9350
+    },
+    {
+      "epoch": 2.2515042117930206,
+      "grad_norm": 1.65625,
+      "learning_rate": 3.6177285583927745e-05,
+      "loss": 0.4786,
+      "step": 9355
+    },
+    {
+      "epoch": 2.252707581227437,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.6143609295736494e-05,
+      "loss": 0.4712,
+      "step": 9360
+    },
+    {
+      "epoch": 2.253910950661853,
+      "grad_norm": 1.640625,
+      "learning_rate": 3.6109975570109494e-05,
+      "loss": 0.4952,
+      "step": 9365
+    },
+    {
+      "epoch": 2.2551143200962693,
+      "grad_norm": 1.703125,
+      "learning_rate": 3.607638446075884e-05,
+      "loss": 0.4684,
+      "step": 9370
+    },
+    {
+      "epoch": 2.256317689530686,
+      "grad_norm": 1.625,
+      "learning_rate": 3.604283602132865e-05,
+      "loss": 0.5315,
+      "step": 9375
+    },
+    {
+      "epoch": 2.2575210589651022,
+      "grad_norm": 1.7109375,
+      "learning_rate": 3.6009330305394825e-05,
+      "loss": 0.457,
+      "step": 9380
+    },
+    {
+      "epoch": 2.2587244283995185,
+      "grad_norm": 1.703125,
+      "learning_rate": 3.59758673664651e-05,
+      "loss": 0.466,
+      "step": 9385
+    },
+    {
+      "epoch": 2.259927797833935,
+      "grad_norm": 1.546875,
+      "learning_rate": 3.594244725797883e-05,
+      "loss": 0.4609,
+      "step": 9390
+    },
+    {
+      "epoch": 2.2611311672683514,
+      "grad_norm": 1.59375,
+      "learning_rate": 3.590907003330705e-05,
+      "loss": 0.4598,
+      "step": 9395
+    },
+    {
+      "epoch": 2.2623345367027676,
+      "grad_norm": 1.8671875,
+      "learning_rate": 3.587573574575226e-05,
+      "loss": 0.481,
+      "step": 9400
+    },
+    {
+      "epoch": 2.2635379061371843,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.5842444448548365e-05,
+      "loss": 0.4712,
+      "step": 9405
+    },
+    {
+      "epoch": 2.2647412755716005,
+      "grad_norm": 1.7578125,
+      "learning_rate": 3.5809196194860714e-05,
+      "loss": 0.4952,
+      "step": 9410
+    },
+    {
+      "epoch": 2.2659446450060168,
+      "grad_norm": 1.828125,
+      "learning_rate": 3.577599103778581e-05,
+      "loss": 0.5135,
+      "step": 9415
+    },
+    {
+      "epoch": 2.2671480144404335,
+      "grad_norm": 1.71875,
+      "learning_rate": 3.574282903035138e-05,
+      "loss": 0.4821,
+      "step": 9420
+    },
+    {
+      "epoch": 2.2683513838748497,
+      "grad_norm": 1.6328125,
+      "learning_rate": 3.570971022551624e-05,
+      "loss": 0.4877,
+      "step": 9425
+    },
+    {
+      "epoch": 2.269554753309266,
+      "grad_norm": 1.71875,
+      "learning_rate": 3.567663467617023e-05,
+      "loss": 0.4727,
+      "step": 9430
+    },
+    {
+      "epoch": 2.270758122743682,
+      "grad_norm": 1.640625,
+      "learning_rate": 3.564360243513407e-05,
+      "loss": 0.4715,
+      "step": 9435
+    },
+    {
+      "epoch": 2.271961492178099,
+      "grad_norm": 1.7109375,
+      "learning_rate": 3.561061355515935e-05,
+      "loss": 0.4631,
+      "step": 9440
+    },
+    {
+      "epoch": 2.273164861612515,
+      "grad_norm": 1.7578125,
+      "learning_rate": 3.55776680889284e-05,
+      "loss": 0.4907,
+      "step": 9445
+    },
+    {
+      "epoch": 2.2743682310469313,
+      "grad_norm": 1.75,
+      "learning_rate": 3.5544766089054224e-05,
+      "loss": 0.5336,
+      "step": 9450
+    },
+    {
+      "epoch": 2.2755716004813475,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.551190760808043e-05,
+      "loss": 0.4999,
+      "step": 9455
+    },
+    {
+      "epoch": 2.2767749699157642,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3.547909269848108e-05,
+      "loss": 0.51,
+      "step": 9460
+    },
+    {
+      "epoch": 2.2779783393501805,
+      "grad_norm": 1.6171875,
+      "learning_rate": 3.54463214126607e-05,
+      "loss": 0.4699,
+      "step": 9465
+    },
+    {
+      "epoch": 2.2791817087845967,
+      "grad_norm": 1.6328125,
+      "learning_rate": 3.5413593802954115e-05,
+      "loss": 0.4887,
+      "step": 9470
+    },
+    {
+      "epoch": 2.2803850782190134,
+      "grad_norm": 1.65625,
+      "learning_rate": 3.5380909921626446e-05,
+      "loss": 0.4792,
+      "step": 9475
+    },
+    {
+      "epoch": 2.2815884476534296,
+      "grad_norm": 1.6640625,
+      "learning_rate": 3.534826982087293e-05,
+      "loss": 0.482,
+      "step": 9480
+    },
+    {
+      "epoch": 2.282791817087846,
+      "grad_norm": 1.578125,
+      "learning_rate": 3.531567355281892e-05,
+      "loss": 0.4671,
+      "step": 9485
+    },
+    {
+      "epoch": 2.2839951865222625,
+      "grad_norm": 1.546875,
+      "learning_rate": 3.528312116951975e-05,
+      "loss": 0.4872,
+      "step": 9490
+    },
+    {
+      "epoch": 2.2851985559566788,
+      "grad_norm": 1.640625,
+      "learning_rate": 3.525061272296069e-05,
+      "loss": 0.514,
+      "step": 9495
+    },
+    {
+      "epoch": 2.286401925391095,
+      "grad_norm": 1.6875,
+      "learning_rate": 3.521814826505684e-05,
+      "loss": 0.4797,
+      "step": 9500
+    },
+    {
+      "epoch": 2.286401925391095,
+      "eval_loss": 0.4312799870967865,
+      "eval_runtime": 2.6559,
+      "eval_samples_per_second": 75.304,
+      "eval_steps_per_second": 75.304,
+      "step": 9500
+    },
+    {
+      "epoch": 2.2876052948255117,
+      "grad_norm": 1.671875,
+      "learning_rate": 3.518572784765304e-05,
+      "loss": 0.4765,
+      "step": 9505
+    },
+    {
+      "epoch": 2.288808664259928,
+      "grad_norm": 1.6171875,
+      "learning_rate": 3.515335152252381e-05,
+      "loss": 0.4641,
+      "step": 9510
+    },
+    {
+      "epoch": 2.290012033694344,
+      "grad_norm": 1.6015625,
+      "learning_rate": 3.5121019341373234e-05,
+      "loss": 0.5206,
+      "step": 9515
+    },
+    {
+      "epoch": 2.2912154031287604,
+      "grad_norm": 1.703125,
+      "learning_rate": 3.508873135583495e-05,
+      "loss": 0.4597,
+      "step": 9520
+    },
+    {
+      "epoch": 2.292418772563177,
+      "grad_norm": 1.71875,
+      "learning_rate": 3.505648761747194e-05,
+      "loss": 0.4892,
+      "step": 9525
+    },
+    {
+      "epoch": 2.2936221419975933,
+      "grad_norm": 1.578125,
+      "learning_rate": 3.5024288177776606e-05,
+      "loss": 0.5179,
+      "step": 9530
+    },
+    {
+      "epoch": 2.2948255114320095,
+      "grad_norm": 1.640625,
+      "learning_rate": 3.4992133088170536e-05,
+      "loss": 0.4698,
+      "step": 9535
+    },
+    {
+      "epoch": 2.2960288808664258,
+      "grad_norm": 1.6171875,
+      "learning_rate": 3.496002240000455e-05,
+      "loss": 0.5085,
+      "step": 9540
+    },
+    {
+      "epoch": 2.2972322503008424,
+      "grad_norm": 1.546875,
+      "learning_rate": 3.492795616455852e-05,
+      "loss": 0.4844,
+      "step": 9545
+    },
+    {
+      "epoch": 2.2984356197352587,
+      "grad_norm": 1.734375,
+      "learning_rate": 3.4895934433041344e-05,
+      "loss": 0.4827,
+      "step": 9550
+    },
+    {
+      "epoch": 2.299638989169675,
+      "grad_norm": 1.671875,
+      "learning_rate": 3.4863957256590857e-05,
+      "loss": 0.4983,
+      "step": 9555
+    },
+    {
+      "epoch": 2.3008423586040916,
+      "grad_norm": 1.6015625,
+      "learning_rate": 3.4832024686273724e-05,
+      "loss": 0.4688,
+      "step": 9560
+    },
+    {
+      "epoch": 2.302045728038508,
+      "grad_norm": 1.6171875,
+      "learning_rate": 3.4800136773085376e-05,
+      "loss": 0.4444,
+      "step": 9565
+    },
+    {
+      "epoch": 2.303249097472924,
+      "grad_norm": 1.765625,
+      "learning_rate": 3.476829356794993e-05,
+      "loss": 0.4913,
+      "step": 9570
+    },
+    {
+      "epoch": 2.3044524669073407,
+      "grad_norm": 1.515625,
+      "learning_rate": 3.473649512172013e-05,
+      "loss": 0.5131,
+      "step": 9575
+    },
+    {
+      "epoch": 2.305655836341757,
+      "grad_norm": 1.59375,
+      "learning_rate": 3.4704741485177205e-05,
+      "loss": 0.455,
+      "step": 9580
+    },
+    {
+      "epoch": 2.306859205776173,
+      "grad_norm": 2.484375,
+      "learning_rate": 3.4673032709030835e-05,
+      "loss": 0.4867,
+      "step": 9585
+    },
+    {
+      "epoch": 2.30806257521059,
+      "grad_norm": 1.6015625,
+      "learning_rate": 3.4641368843919074e-05,
+      "loss": 0.4971,
+      "step": 9590
+    },
+    {
+      "epoch": 2.309265944645006,
+      "grad_norm": 1.6171875,
+      "learning_rate": 3.4609749940408235e-05,
+      "loss": 0.4625,
+      "step": 9595
+    },
+    {
+      "epoch": 2.3104693140794224,
+      "grad_norm": 1.640625,
+      "learning_rate": 3.4578176048992855e-05,
+      "loss": 0.5039,
+      "step": 9600
+    },
+    {
+      "epoch": 2.3116726835138386,
+      "grad_norm": 1.6875,
+      "learning_rate": 3.454664722009554e-05,
+      "loss": 0.4597,
+      "step": 9605
+    },
+    {
+      "epoch": 2.3128760529482553,
+      "grad_norm": 1.71875,
+      "learning_rate": 3.4515163504067006e-05,
+      "loss": 0.4524,
+      "step": 9610
+    },
+    {
+      "epoch": 2.3140794223826715,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.4483724951185835e-05,
+      "loss": 0.4841,
+      "step": 9615
+    },
+    {
+      "epoch": 2.3152827918170877,
+      "grad_norm": 1.5859375,
+      "learning_rate": 3.445233161165858e-05,
+      "loss": 0.4709,
+      "step": 9620
+    },
+    {
+      "epoch": 2.316486161251504,
+      "grad_norm": 1.6640625,
+      "learning_rate": 3.442098353561948e-05,
+      "loss": 0.4543,
+      "step": 9625
+    },
+    {
+      "epoch": 2.3176895306859207,
+      "grad_norm": 1.6015625,
+      "learning_rate": 3.4389680773130616e-05,
+      "loss": 0.4861,
+      "step": 9630
+    },
+    {
+      "epoch": 2.318892900120337,
+      "grad_norm": 1.6640625,
+      "learning_rate": 3.435842337418159e-05,
+      "loss": 0.4898,
+      "step": 9635
+    },
+    {
+      "epoch": 2.320096269554753,
+      "grad_norm": 1.640625,
+      "learning_rate": 3.432721138868963e-05,
+      "loss": 0.4986,
+      "step": 9640
+    },
+    {
+      "epoch": 2.32129963898917,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.429604486649943e-05,
+      "loss": 0.482,
+      "step": 9645
+    },
+    {
+      "epoch": 2.322503008423586,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.426492385738303e-05,
+      "loss": 0.4934,
+      "step": 9650
+    },
+    {
+      "epoch": 2.3237063778580023,
+      "grad_norm": 1.59375,
+      "learning_rate": 3.423384841103988e-05,
+      "loss": 0.4967,
+      "step": 9655
+    },
+    {
+      "epoch": 2.324909747292419,
+      "grad_norm": 1.6015625,
+      "learning_rate": 3.4202818577096564e-05,
+      "loss": 0.4717,
+      "step": 9660
+    },
+    {
+      "epoch": 2.326113116726835,
+      "grad_norm": 1.6875,
+      "learning_rate": 3.41718344051069e-05,
+      "loss": 0.4812,
+      "step": 9665
+    },
+    {
+      "epoch": 2.3273164861612514,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.414089594455175e-05,
+      "loss": 0.4733,
+      "step": 9670
+    },
+    {
+      "epoch": 2.328519855595668,
+      "grad_norm": 1.7578125,
+      "learning_rate": 3.411000324483899e-05,
+      "loss": 0.4884,
+      "step": 9675
+    },
+    {
+      "epoch": 2.3297232250300843,
+      "grad_norm": 1.6640625,
+      "learning_rate": 3.407915635530341e-05,
+      "loss": 0.4813,
+      "step": 9680
+    },
+    {
+      "epoch": 2.3309265944645006,
+      "grad_norm": 1.6796875,
+      "learning_rate": 3.404835532520663e-05,
+      "loss": 0.4684,
+      "step": 9685
+    },
+    {
+      "epoch": 2.332129963898917,
+      "grad_norm": 1.7421875,
+      "learning_rate": 3.401760020373708e-05,
+      "loss": 0.4846,
+      "step": 9690
+    },
+    {
+      "epoch": 2.3333333333333335,
+      "grad_norm": 1.703125,
+      "learning_rate": 3.3986891040009825e-05,
+      "loss": 0.4763,
+      "step": 9695
+    },
+    {
+      "epoch": 2.3345367027677497,
+      "grad_norm": 1.8671875,
+      "learning_rate": 3.3956227883066555e-05,
+      "loss": 0.4765,
+      "step": 9700
+    },
+    {
+      "epoch": 2.335740072202166,
+      "grad_norm": 1.5859375,
+      "learning_rate": 3.392561078187548e-05,
+      "loss": 0.4773,
+      "step": 9705
+    },
+    {
+      "epoch": 2.336943441636582,
+      "grad_norm": 1.6484375,
+      "learning_rate": 3.389503978533128e-05,
+      "loss": 0.4516,
+      "step": 9710
+    },
+    {
+      "epoch": 2.338146811070999,
+      "grad_norm": 1.6640625,
+      "learning_rate": 3.3864514942254994e-05,
+      "loss": 0.4536,
+      "step": 9715
+    },
+    {
+      "epoch": 2.339350180505415,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.383403630139397e-05,
+      "loss": 0.4585,
+      "step": 9720
+    },
+    {
+      "epoch": 2.3405535499398313,
+      "grad_norm": 1.59375,
+      "learning_rate": 3.3803603911421726e-05,
+      "loss": 0.4775,
+      "step": 9725
+    },
+    {
+      "epoch": 2.341756919374248,
+      "grad_norm": 1.5546875,
+      "learning_rate": 3.377321782093798e-05,
+      "loss": 0.4656,
+      "step": 9730
+    },
+    {
+      "epoch": 2.3429602888086642,
+      "grad_norm": 1.6875,
+      "learning_rate": 3.374287807846848e-05,
+      "loss": 0.4747,
+      "step": 9735
+    },
+    {
+      "epoch": 2.3441636582430805,
+      "grad_norm": 1.875,
+      "learning_rate": 3.371258473246495e-05,
+      "loss": 0.4847,
+      "step": 9740
+    },
+    {
+      "epoch": 2.345367027677497,
+      "grad_norm": 1.6640625,
+      "learning_rate": 3.368233783130504e-05,
+      "loss": 0.4841,
+      "step": 9745
+    },
+    {
+      "epoch": 2.3465703971119134,
+      "grad_norm": 1.6875,
+      "learning_rate": 3.365213742329222e-05,
+      "loss": 0.5139,
+      "step": 9750
+    },
+    {
+      "epoch": 2.3477737665463296,
+      "grad_norm": 1.65625,
+      "learning_rate": 3.3621983556655705e-05,
+      "loss": 0.4983,
+      "step": 9755
+    },
+    {
+      "epoch": 2.3489771359807463,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.3591876279550384e-05,
+      "loss": 0.5193,
+      "step": 9760
+    },
+    {
+      "epoch": 2.3501805054151625,
+      "grad_norm": 1.8125,
+      "learning_rate": 3.3561815640056776e-05,
+      "loss": 0.4765,
+      "step": 9765
+    },
+    {
+      "epoch": 2.351383874849579,
+      "grad_norm": 1.7421875,
+      "learning_rate": 3.353180168618087e-05,
+      "loss": 0.5217,
+      "step": 9770
+    },
+    {
+      "epoch": 2.352587244283995,
+      "grad_norm": 1.515625,
+      "learning_rate": 3.3501834465854126e-05,
+      "loss": 0.4547,
+      "step": 9775
+    },
+    {
+      "epoch": 2.3537906137184117,
+      "grad_norm": 1.796875,
+      "learning_rate": 3.34719140269334e-05,
+      "loss": 0.458,
+      "step": 9780
+    },
+    {
+      "epoch": 2.354993983152828,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.344204041720077e-05,
+      "loss": 0.4832,
+      "step": 9785
+    },
+    {
+      "epoch": 2.356197352587244,
+      "grad_norm": 1.6875,
+      "learning_rate": 3.341221368436361e-05,
+      "loss": 0.4849,
+      "step": 9790
+    },
+    {
+      "epoch": 2.357400722021661,
+      "grad_norm": 1.6640625,
+      "learning_rate": 3.3382433876054354e-05,
+      "loss": 0.4725,
+      "step": 9795
+    },
+    {
+      "epoch": 2.358604091456077,
+      "grad_norm": 1.7109375,
+      "learning_rate": 3.335270103983057e-05,
+      "loss": 0.456,
+      "step": 9800
+    },
+    {
+      "epoch": 2.3598074608904933,
+      "grad_norm": 1.6328125,
+      "learning_rate": 3.332301522317476e-05,
+      "loss": 0.499,
+      "step": 9805
+    },
+    {
+      "epoch": 2.3610108303249095,
+      "grad_norm": 1.734375,
+      "learning_rate": 3.3293376473494355e-05,
+      "loss": 0.4877,
+      "step": 9810
+    },
+    {
+      "epoch": 2.3622141997593262,
+      "grad_norm": 1.6015625,
+      "learning_rate": 3.326378483812163e-05,
+      "loss": 0.4875,
+      "step": 9815
+    },
+    {
+      "epoch": 2.3634175691937425,
+      "grad_norm": 1.8203125,
+      "learning_rate": 3.323424036431362e-05,
+      "loss": 0.4932,
+      "step": 9820
+    },
+    {
+      "epoch": 2.3646209386281587,
+      "grad_norm": 1.5859375,
+      "learning_rate": 3.3204743099252036e-05,
+      "loss": 0.5022,
+      "step": 9825
+    },
+    {
+      "epoch": 2.3658243080625754,
+      "grad_norm": 1.5234375,
+      "learning_rate": 3.3175293090043175e-05,
+      "loss": 0.5026,
+      "step": 9830
+    },
+    {
+      "epoch": 2.3670276774969916,
+      "grad_norm": 1.5703125,
+      "learning_rate": 3.314589038371793e-05,
+      "loss": 0.5034,
+      "step": 9835
+    },
+    {
+      "epoch": 2.368231046931408,
+      "grad_norm": 1.7578125,
+      "learning_rate": 3.3116535027231584e-05,
+      "loss": 0.4961,
+      "step": 9840
+    },
+    {
+      "epoch": 2.3694344163658245,
+      "grad_norm": 1.625,
+      "learning_rate": 3.3087227067463855e-05,
+      "loss": 0.4831,
+      "step": 9845
+    },
+    {
+      "epoch": 2.3706377858002408,
+      "grad_norm": 1.7109375,
+      "learning_rate": 3.305796655121872e-05,
+      "loss": 0.5074,
+      "step": 9850
+    },
+    {
+      "epoch": 2.371841155234657,
+      "grad_norm": 1.640625,
+      "learning_rate": 3.302875352522446e-05,
+      "loss": 0.5183,
+      "step": 9855
+    },
+    {
+      "epoch": 2.3730445246690732,
+      "grad_norm": 1.5546875,
+      "learning_rate": 3.299958803613344e-05,
+      "loss": 0.4634,
+      "step": 9860
+    },
+    {
+      "epoch": 2.37424789410349,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.2970470130522156e-05,
+      "loss": 0.4769,
+      "step": 9865
+    },
+    {
+      "epoch": 2.375451263537906,
+      "grad_norm": 1.84375,
+      "learning_rate": 3.2941399854891116e-05,
+      "loss": 0.4786,
+      "step": 9870
+    },
+    {
+      "epoch": 2.3766546329723224,
+      "grad_norm": 1.6640625,
+      "learning_rate": 3.291237725566472e-05,
+      "loss": 0.5156,
+      "step": 9875
+    },
+    {
+      "epoch": 2.377858002406739,
+      "grad_norm": 1.71875,
+      "learning_rate": 3.2883402379191304e-05,
+      "loss": 0.5056,
+      "step": 9880
+    },
+    {
+      "epoch": 2.3790613718411553,
+      "grad_norm": 1.671875,
+      "learning_rate": 3.28544752717429e-05,
+      "loss": 0.4833,
+      "step": 9885
+    },
+    {
+      "epoch": 2.3802647412755715,
+      "grad_norm": 1.6171875,
+      "learning_rate": 3.2825595979515344e-05,
+      "loss": 0.4869,
+      "step": 9890
+    },
+    {
+      "epoch": 2.3814681107099878,
+      "grad_norm": 1.6796875,
+      "learning_rate": 3.279676454862805e-05,
+      "loss": 0.4726,
+      "step": 9895
+    },
+    {
+      "epoch": 2.3826714801444044,
+      "grad_norm": 1.6796875,
+      "learning_rate": 3.2767981025124024e-05,
+      "loss": 0.4983,
+      "step": 9900
+    },
+    {
+      "epoch": 2.3838748495788207,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.2739245454969764e-05,
+      "loss": 0.4839,
+      "step": 9905
+    },
+    {
+      "epoch": 2.385078219013237,
+      "grad_norm": 1.5625,
+      "learning_rate": 3.271055788405518e-05,
+      "loss": 0.5325,
+      "step": 9910
+    },
+    {
+      "epoch": 2.3862815884476536,
+      "grad_norm": 1.65625,
+      "learning_rate": 3.2681918358193556e-05,
+      "loss": 0.4825,
+      "step": 9915
+    },
+    {
+      "epoch": 2.38748495788207,
+      "grad_norm": 1.828125,
+      "learning_rate": 3.265332692312139e-05,
+      "loss": 0.4817,
+      "step": 9920
+    },
+    {
+      "epoch": 2.388688327316486,
+      "grad_norm": 1.5546875,
+      "learning_rate": 3.262478362449845e-05,
+      "loss": 0.4546,
+      "step": 9925
+    },
+    {
+      "epoch": 2.3898916967509027,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.2596288507907585e-05,
+      "loss": 0.5006,
+      "step": 9930
+    },
+    {
+      "epoch": 2.391095066185319,
+      "grad_norm": 1.6640625,
+      "learning_rate": 3.2567841618854715e-05,
+      "loss": 0.4897,
+      "step": 9935
+    },
+    {
+      "epoch": 2.392298435619735,
+      "grad_norm": 1.6640625,
+      "learning_rate": 3.2539443002768734e-05,
+      "loss": 0.4854,
+      "step": 9940
+    },
+    {
+      "epoch": 2.3935018050541514,
+      "grad_norm": 1.7109375,
+      "learning_rate": 3.2511092705001466e-05,
+      "loss": 0.4524,
+      "step": 9945
+    },
+    {
+      "epoch": 2.394705174488568,
+      "grad_norm": 1.65625,
+      "learning_rate": 3.248279077082754e-05,
+      "loss": 0.4999,
+      "step": 9950
+    },
+    {
+      "epoch": 2.3959085439229844,
+      "grad_norm": 1.8515625,
+      "learning_rate": 3.245453724544437e-05,
+      "loss": 0.5016,
+      "step": 9955
+    },
+    {
+      "epoch": 2.3971119133574006,
+      "grad_norm": 1.59375,
+      "learning_rate": 3.2426332173972076e-05,
+      "loss": 0.4637,
+      "step": 9960
+    },
+    {
+      "epoch": 2.3983152827918173,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3.239817560145335e-05,
+      "loss": 0.4696,
+      "step": 9965
+    },
+    {
+      "epoch": 2.3995186522262335,
+      "grad_norm": 1.5390625,
+      "learning_rate": 3.2370067572853505e-05,
+      "loss": 0.4652,
+      "step": 9970
+    },
+    {
+      "epoch": 2.4007220216606497,
+      "grad_norm": 1.53125,
+      "learning_rate": 3.234200813306026e-05,
+      "loss": 0.5054,
+      "step": 9975
+    },
+    {
+      "epoch": 2.401925391095066,
+      "grad_norm": 1.734375,
+      "learning_rate": 3.2313997326883785e-05,
+      "loss": 0.4773,
+      "step": 9980
+    },
+    {
+      "epoch": 2.4031287605294827,
+      "grad_norm": 1.640625,
+      "learning_rate": 3.228603519905654e-05,
+      "loss": 0.5104,
+      "step": 9985
+    },
+    {
+      "epoch": 2.404332129963899,
+      "grad_norm": 1.53125,
+      "learning_rate": 3.225812179423331e-05,
+      "loss": 0.4805,
+      "step": 9990
+    },
+    {
+      "epoch": 2.405535499398315,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.2230257156991e-05,
+      "loss": 0.468,
+      "step": 9995
+    },
+    {
+      "epoch": 2.406738868832732,
+      "grad_norm": 1.828125,
+      "learning_rate": 3.22024413318287e-05,
+      "loss": 0.4634,
+      "step": 10000
+    },
+    {
+      "epoch": 2.406738868832732,
+      "eval_loss": 0.42871612310409546,
+      "eval_runtime": 2.6519,
+      "eval_samples_per_second": 75.418,
+      "eval_steps_per_second": 75.418,
+      "step": 10000
+    },
+    {
+      "epoch": 2.407942238267148,
+      "grad_norm": 1.5546875,
+      "learning_rate": 3.2174674363167515e-05,
+      "loss": 0.4666,
+      "step": 10005
+    },
+    {
+      "epoch": 2.4091456077015643,
+      "grad_norm": 1.640625,
+      "learning_rate": 3.2146956295350504e-05,
+      "loss": 0.5036,
+      "step": 10010
+    },
+    {
+      "epoch": 2.410348977135981,
+      "grad_norm": 1.6484375,
+      "learning_rate": 3.2119287172642696e-05,
+      "loss": 0.4966,
+      "step": 10015
+    },
+    {
+      "epoch": 2.411552346570397,
+      "grad_norm": 1.578125,
+      "learning_rate": 3.2091667039230886e-05,
+      "loss": 0.4953,
+      "step": 10020
+    },
+    {
+      "epoch": 2.4127557160048134,
+      "grad_norm": 1.8828125,
+      "learning_rate": 3.206409593922369e-05,
+      "loss": 0.48,
+      "step": 10025
+    },
+    {
+      "epoch": 2.4139590854392297,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.203657391665138e-05,
+      "loss": 0.4826,
+      "step": 10030
+    },
+    {
+      "epoch": 2.4151624548736463,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.200910101546588e-05,
+      "loss": 0.4572,
+      "step": 10035
+    },
+    {
+      "epoch": 2.4163658243080626,
+      "grad_norm": 1.640625,
+      "learning_rate": 3.1981677279540655e-05,
+      "loss": 0.48,
+      "step": 10040
+    },
+    {
+      "epoch": 2.417569193742479,
+      "grad_norm": 1.734375,
+      "learning_rate": 3.195430275267066e-05,
+      "loss": 0.4708,
+      "step": 10045
+    },
+    {
+      "epoch": 2.4187725631768955,
+      "grad_norm": 1.6484375,
+      "learning_rate": 3.1926977478572265e-05,
+      "loss": 0.4647,
+      "step": 10050
+    },
+    {
+      "epoch": 2.4199759326113117,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.189970150088317e-05,
+      "loss": 0.4696,
+      "step": 10055
+    },
+    {
+      "epoch": 2.421179302045728,
+      "grad_norm": 1.65625,
+      "learning_rate": 3.1872474863162365e-05,
+      "loss": 0.4835,
+      "step": 10060
+    },
+    {
+      "epoch": 2.422382671480144,
+      "grad_norm": 1.6015625,
+      "learning_rate": 3.1845297608890044e-05,
+      "loss": 0.4729,
+      "step": 10065
+    },
+    {
+      "epoch": 2.423586040914561,
+      "grad_norm": 1.6171875,
+      "learning_rate": 3.1818169781467544e-05,
+      "loss": 0.474,
+      "step": 10070
+    },
+    {
+      "epoch": 2.424789410348977,
+      "grad_norm": 1.640625,
+      "learning_rate": 3.1791091424217244e-05,
+      "loss": 0.4806,
+      "step": 10075
+    },
+    {
+      "epoch": 2.4259927797833933,
+      "grad_norm": 1.6875,
+      "learning_rate": 3.176406258038254e-05,
+      "loss": 0.4609,
+      "step": 10080
+    },
+    {
+      "epoch": 2.42719614921781,
+      "grad_norm": 1.6328125,
+      "learning_rate": 3.1737083293127764e-05,
+      "loss": 0.482,
+      "step": 10085
+    },
+    {
+      "epoch": 2.4283995186522263,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.171015360553807e-05,
+      "loss": 0.4992,
+      "step": 10090
+    },
+    {
+      "epoch": 2.4296028880866425,
+      "grad_norm": 1.734375,
+      "learning_rate": 3.1683273560619464e-05,
+      "loss": 0.4687,
+      "step": 10095
+    },
+    {
+      "epoch": 2.430806257521059,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.165644320129861e-05,
+      "loss": 0.4969,
+      "step": 10100
+    },
+    {
+      "epoch": 2.4320096269554754,
+      "grad_norm": 1.6328125,
+      "learning_rate": 3.162966257042287e-05,
+      "loss": 0.4818,
+      "step": 10105
+    },
+    {
+      "epoch": 2.4332129963898916,
+      "grad_norm": 1.5546875,
+      "learning_rate": 3.160293171076016e-05,
+      "loss": 0.4966,
+      "step": 10110
+    },
+    {
+      "epoch": 2.434416365824308,
+      "grad_norm": 1.78125,
+      "learning_rate": 3.157625066499894e-05,
+      "loss": 0.4543,
+      "step": 10115
+    },
+    {
+      "epoch": 2.4356197352587245,
+      "grad_norm": 1.703125,
+      "learning_rate": 3.1549619475748115e-05,
+      "loss": 0.4877,
+      "step": 10120
+    },
+    {
+      "epoch": 2.436823104693141,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.1523038185536936e-05,
+      "loss": 0.4581,
+      "step": 10125
+    },
+    {
+      "epoch": 2.438026474127557,
+      "grad_norm": 1.65625,
+      "learning_rate": 3.1496506836815015e-05,
+      "loss": 0.4877,
+      "step": 10130
+    },
+    {
+      "epoch": 2.4392298435619737,
+      "grad_norm": 1.703125,
+      "learning_rate": 3.147002547195218e-05,
+      "loss": 0.5074,
+      "step": 10135
+    },
+    {
+      "epoch": 2.44043321299639,
+      "grad_norm": 1.71875,
+      "learning_rate": 3.144359413323845e-05,
+      "loss": 0.5168,
+      "step": 10140
+    },
+    {
+      "epoch": 2.441636582430806,
+      "grad_norm": 1.625,
+      "learning_rate": 3.1417212862883916e-05,
+      "loss": 0.4843,
+      "step": 10145
+    },
+    {
+      "epoch": 2.4428399518652224,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.139088170301878e-05,
+      "loss": 0.4809,
+      "step": 10150
+    },
+    {
+      "epoch": 2.444043321299639,
+      "grad_norm": 1.6796875,
+      "learning_rate": 3.136460069569316e-05,
+      "loss": 0.4872,
+      "step": 10155
+    },
+    {
+      "epoch": 2.4452466907340553,
+      "grad_norm": 1.59375,
+      "learning_rate": 3.133836988287711e-05,
+      "loss": 0.4707,
+      "step": 10160
+    },
+    {
+      "epoch": 2.4464500601684716,
+      "grad_norm": 1.8828125,
+      "learning_rate": 3.1312189306460496e-05,
+      "loss": 0.4945,
+      "step": 10165
+    },
+    {
+      "epoch": 2.4476534296028882,
+      "grad_norm": 1.515625,
+      "learning_rate": 3.128605900825301e-05,
+      "loss": 0.4746,
+      "step": 10170
+    },
+    {
+      "epoch": 2.4488567990373045,
+      "grad_norm": 1.734375,
+      "learning_rate": 3.1259979029984e-05,
+      "loss": 0.4841,
+      "step": 10175
+    },
+    {
+      "epoch": 2.4500601684717207,
+      "grad_norm": 1.75,
+      "learning_rate": 3.1233949413302476e-05,
+      "loss": 0.4884,
+      "step": 10180
+    },
+    {
+      "epoch": 2.4512635379061374,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.120797019977703e-05,
+      "loss": 0.5056,
+      "step": 10185
+    },
+    {
+      "epoch": 2.4524669073405536,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.118204143089574e-05,
+      "loss": 0.484,
+      "step": 10190
+    },
+    {
+      "epoch": 2.45367027677497,
+      "grad_norm": 1.6015625,
+      "learning_rate": 3.1156163148066154e-05,
+      "loss": 0.4626,
+      "step": 10195
+    },
+    {
+      "epoch": 2.4548736462093865,
+      "grad_norm": 1.828125,
+      "learning_rate": 3.1130335392615145e-05,
+      "loss": 0.5227,
+      "step": 10200
+    },
+    {
+      "epoch": 2.4560770156438028,
+      "grad_norm": 1.671875,
+      "learning_rate": 3.110455820578897e-05,
+      "loss": 0.529,
+      "step": 10205
+    },
+    {
+      "epoch": 2.457280385078219,
+      "grad_norm": 1.6640625,
+      "learning_rate": 3.107883162875305e-05,
+      "loss": 0.486,
+      "step": 10210
+    },
+    {
+      "epoch": 2.4584837545126352,
+      "grad_norm": 1.6171875,
+      "learning_rate": 3.1053155702592044e-05,
+      "loss": 0.4785,
+      "step": 10215
+    },
+    {
+      "epoch": 2.459687123947052,
+      "grad_norm": 1.765625,
+      "learning_rate": 3.1027530468309706e-05,
+      "loss": 0.4844,
+      "step": 10220
+    },
+    {
+      "epoch": 2.460890493381468,
+      "grad_norm": 1.625,
+      "learning_rate": 3.1001955966828806e-05,
+      "loss": 0.4732,
+      "step": 10225
+    },
+    {
+      "epoch": 2.4620938628158844,
+      "grad_norm": 1.515625,
+      "learning_rate": 3.097643223899114e-05,
+      "loss": 0.474,
+      "step": 10230
+    },
+    {
+      "epoch": 2.4632972322503006,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.095095932555738e-05,
+      "loss": 0.4966,
+      "step": 10235
+    },
+    {
+      "epoch": 2.4645006016847173,
+      "grad_norm": 1.5625,
+      "learning_rate": 3.0925537267207096e-05,
+      "loss": 0.4779,
+      "step": 10240
+    },
+    {
+      "epoch": 2.4657039711191335,
+      "grad_norm": 1.6484375,
+      "learning_rate": 3.090016610453859e-05,
+      "loss": 0.4876,
+      "step": 10245
+    },
+    {
+      "epoch": 2.4669073405535498,
+      "grad_norm": 1.75,
+      "learning_rate": 3.0874845878068933e-05,
+      "loss": 0.4808,
+      "step": 10250
+    },
+    {
+      "epoch": 2.4681107099879664,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.084957662823381e-05,
+      "loss": 0.4721,
+      "step": 10255
+    },
+    {
+      "epoch": 2.4693140794223827,
+      "grad_norm": 1.7109375,
+      "learning_rate": 3.082435839538753e-05,
+      "loss": 0.5265,
+      "step": 10260
+    },
+    {
+      "epoch": 2.470517448856799,
+      "grad_norm": 1.6015625,
+      "learning_rate": 3.079919121980291e-05,
+      "loss": 0.46,
+      "step": 10265
+    },
+    {
+      "epoch": 2.4717208182912156,
+      "grad_norm": 1.671875,
+      "learning_rate": 3.0774075141671253e-05,
+      "loss": 0.4614,
+      "step": 10270
+    },
+    {
+      "epoch": 2.472924187725632,
+      "grad_norm": 1.6796875,
+      "learning_rate": 3.074901020110223e-05,
+      "loss": 0.4906,
+      "step": 10275
+    },
+    {
+      "epoch": 2.474127557160048,
+      "grad_norm": 1.6796875,
+      "learning_rate": 3.0723996438123863e-05,
+      "loss": 0.4863,
+      "step": 10280
+    },
+    {
+      "epoch": 2.4753309265944647,
+      "grad_norm": 1.5703125,
+      "learning_rate": 3.0699033892682455e-05,
+      "loss": 0.4468,
+      "step": 10285
+    },
+    {
+      "epoch": 2.476534296028881,
+      "grad_norm": 1.65625,
+      "learning_rate": 3.0674122604642494e-05,
+      "loss": 0.4698,
+      "step": 10290
+    },
+    {
+      "epoch": 2.477737665463297,
+      "grad_norm": 1.5625,
+      "learning_rate": 3.064926261378663e-05,
+      "loss": 0.4576,
+      "step": 10295
+    },
+    {
+      "epoch": 2.4789410348977134,
+      "grad_norm": 1.8984375,
+      "learning_rate": 3.062445395981557e-05,
+      "loss": 0.472,
+      "step": 10300
+    },
+    {
+      "epoch": 2.48014440433213,
+      "grad_norm": 1.8515625,
+      "learning_rate": 3.0599696682348065e-05,
+      "loss": 0.4512,
+      "step": 10305
+    },
+    {
+      "epoch": 2.4813477737665464,
+      "grad_norm": 1.78125,
+      "learning_rate": 3.057499082092081e-05,
+      "loss": 0.4875,
+      "step": 10310
+    },
+    {
+      "epoch": 2.4825511432009626,
+      "grad_norm": 1.59375,
+      "learning_rate": 3.055033641498837e-05,
+      "loss": 0.4799,
+      "step": 10315
+    },
+    {
+      "epoch": 2.483754512635379,
+      "grad_norm": 1.75,
+      "learning_rate": 3.052573350392318e-05,
+      "loss": 0.5623,
+      "step": 10320
+    },
+    {
+      "epoch": 2.4849578820697955,
+      "grad_norm": 1.7578125,
+      "learning_rate": 3.050118212701538e-05,
+      "loss": 0.5132,
+      "step": 10325
+    },
+    {
+      "epoch": 2.4861612515042117,
+      "grad_norm": 1.546875,
+      "learning_rate": 3.0476682323472873e-05,
+      "loss": 0.4654,
+      "step": 10330
+    },
+    {
+      "epoch": 2.487364620938628,
+      "grad_norm": 1.6796875,
+      "learning_rate": 3.0452234132421153e-05,
+      "loss": 0.4787,
+      "step": 10335
+    },
+    {
+      "epoch": 2.4885679903730447,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.0427837592903322e-05,
+      "loss": 0.4643,
+      "step": 10340
+    },
+    {
+      "epoch": 2.489771359807461,
+      "grad_norm": 1.78125,
+      "learning_rate": 3.0403492743879963e-05,
+      "loss": 0.4728,
+      "step": 10345
+    },
+    {
+      "epoch": 2.490974729241877,
+      "grad_norm": 1.65625,
+      "learning_rate": 3.037919962422914e-05,
+      "loss": 0.4616,
+      "step": 10350
+    },
+    {
+      "epoch": 2.492178098676294,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.0354958272746315e-05,
+      "loss": 0.4936,
+      "step": 10355
+    },
+    {
+      "epoch": 2.49338146811071,
+      "grad_norm": 1.796875,
+      "learning_rate": 3.0330768728144233e-05,
+      "loss": 0.4876,
+      "step": 10360
+    },
+    {
+      "epoch": 2.4945848375451263,
+      "grad_norm": 1.5703125,
+      "learning_rate": 3.0306631029052953e-05,
+      "loss": 0.4492,
+      "step": 10365
+    },
+    {
+      "epoch": 2.495788206979543,
+      "grad_norm": 1.7578125,
+      "learning_rate": 3.0282545214019703e-05,
+      "loss": 0.4701,
+      "step": 10370
+    },
+    {
+      "epoch": 2.496991576413959,
+      "grad_norm": 1.640625,
+      "learning_rate": 3.0258511321508874e-05,
+      "loss": 0.4946,
+      "step": 10375
+    },
+    {
+      "epoch": 2.4981949458483754,
+      "grad_norm": 1.640625,
+      "learning_rate": 3.0234529389901923e-05,
+      "loss": 0.4841,
+      "step": 10380
+    },
+    {
+      "epoch": 2.4993983152827917,
+      "grad_norm": 1.75,
+      "learning_rate": 3.0210599457497347e-05,
+      "loss": 0.4744,
+      "step": 10385
+    },
+    {
+      "epoch": 2.5006016847172083,
+      "grad_norm": 1.625,
+      "learning_rate": 3.0186721562510583e-05,
+      "loss": 0.4639,
+      "step": 10390
+    },
+    {
+      "epoch": 2.5018050541516246,
+      "grad_norm": 1.6171875,
+      "learning_rate": 3.016289574307397e-05,
+      "loss": 0.482,
+      "step": 10395
+    },
+    {
+      "epoch": 2.503008423586041,
+      "grad_norm": 1.6640625,
+      "learning_rate": 3.01391220372367e-05,
+      "loss": 0.4678,
+      "step": 10400
+    },
+    {
+      "epoch": 2.504211793020457,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.011540048296471e-05,
+      "loss": 0.4825,
+      "step": 10405
+    },
+    {
+      "epoch": 2.5054151624548737,
+      "grad_norm": 1.6875,
+      "learning_rate": 3.009173111814069e-05,
+      "loss": 0.4693,
+      "step": 10410
+    },
+    {
+      "epoch": 2.50661853188929,
+      "grad_norm": 1.6015625,
+      "learning_rate": 3.006811398056394e-05,
+      "loss": 0.4785,
+      "step": 10415
+    },
+    {
+      "epoch": 2.507821901323706,
+      "grad_norm": 1.59375,
+      "learning_rate": 3.0044549107950403e-05,
+      "loss": 0.4637,
+      "step": 10420
+    },
+    {
+      "epoch": 2.509025270758123,
+      "grad_norm": 1.6875,
+      "learning_rate": 3.0021036537932523e-05,
+      "loss": 0.5127,
+      "step": 10425
+    },
+    {
+      "epoch": 2.510228640192539,
+      "grad_norm": 1.7734375,
+      "learning_rate": 2.9997576308059222e-05,
+      "loss": 0.4805,
+      "step": 10430
+    },
+    {
+      "epoch": 2.5114320096269553,
+      "grad_norm": 1.6953125,
+      "learning_rate": 2.997416845579585e-05,
+      "loss": 0.4855,
+      "step": 10435
+    },
+    {
+      "epoch": 2.512635379061372,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.9950813018524096e-05,
+      "loss": 0.4735,
+      "step": 10440
+    },
+    {
+      "epoch": 2.5138387484957883,
+      "grad_norm": 1.65625,
+      "learning_rate": 2.992751003354196e-05,
+      "loss": 0.4839,
+      "step": 10445
+    },
+    {
+      "epoch": 2.5150421179302045,
+      "grad_norm": 1.7890625,
+      "learning_rate": 2.9904259538063662e-05,
+      "loss": 0.4965,
+      "step": 10450
+    },
+    {
+      "epoch": 2.516245487364621,
+      "grad_norm": 1.5859375,
+      "learning_rate": 2.9881061569219613e-05,
+      "loss": 0.499,
+      "step": 10455
+    },
+    {
+      "epoch": 2.5174488567990374,
+      "grad_norm": 1.7109375,
+      "learning_rate": 2.9857916164056317e-05,
+      "loss": 0.4927,
+      "step": 10460
+    },
+    {
+      "epoch": 2.5186522262334536,
+      "grad_norm": 1.609375,
+      "learning_rate": 2.9834823359536362e-05,
+      "loss": 0.4586,
+      "step": 10465
+    },
+    {
+      "epoch": 2.51985559566787,
+      "grad_norm": 1.796875,
+      "learning_rate": 2.981178319253831e-05,
+      "loss": 0.4548,
+      "step": 10470
+    },
+    {
+      "epoch": 2.5210589651022866,
+      "grad_norm": 1.8671875,
+      "learning_rate": 2.9788795699856688e-05,
+      "loss": 0.4943,
+      "step": 10475
+    },
+    {
+      "epoch": 2.522262334536703,
+      "grad_norm": 1.625,
+      "learning_rate": 2.976586091820188e-05,
+      "loss": 0.4588,
+      "step": 10480
+    },
+    {
+      "epoch": 2.523465703971119,
+      "grad_norm": 1.7109375,
+      "learning_rate": 2.9742978884200102e-05,
+      "loss": 0.5151,
+      "step": 10485
+    },
+    {
+      "epoch": 2.5246690734055353,
+      "grad_norm": 1.75,
+      "learning_rate": 2.9720149634393347e-05,
+      "loss": 0.5076,
+      "step": 10490
+    },
+    {
+      "epoch": 2.525872442839952,
+      "grad_norm": 1.6328125,
+      "learning_rate": 2.969737320523928e-05,
+      "loss": 0.4618,
+      "step": 10495
+    },
+    {
+      "epoch": 2.527075812274368,
+      "grad_norm": 1.6875,
+      "learning_rate": 2.967464963311125e-05,
+      "loss": 0.466,
+      "step": 10500
+    },
+    {
+      "epoch": 2.527075812274368,
+      "eval_loss": 0.42753535509109497,
+      "eval_runtime": 2.6456,
+      "eval_samples_per_second": 75.598,
+      "eval_steps_per_second": 75.598,
+      "step": 10500
+    },
+    {
+      "epoch": 2.5282791817087844,
+      "grad_norm": 1.625,
+      "learning_rate": 2.9651978954298154e-05,
+      "loss": 0.4825,
+      "step": 10505
+    },
+    {
+      "epoch": 2.529482551143201,
+      "grad_norm": 1.7109375,
+      "learning_rate": 2.9629361205004465e-05,
+      "loss": 0.4836,
+      "step": 10510
+    },
+    {
+      "epoch": 2.5306859205776173,
+      "grad_norm": 1.734375,
+      "learning_rate": 2.960679642135009e-05,
+      "loss": 0.5091,
+      "step": 10515
+    },
+    {
+      "epoch": 2.5318892900120336,
+      "grad_norm": 1.6875,
+      "learning_rate": 2.9584284639370386e-05,
+      "loss": 0.4936,
+      "step": 10520
+    },
+    {
+      "epoch": 2.5330926594464502,
+      "grad_norm": 1.546875,
+      "learning_rate": 2.956182589501603e-05,
+      "loss": 0.471,
+      "step": 10525
+    },
+    {
+      "epoch": 2.5342960288808665,
+      "grad_norm": 1.6953125,
+      "learning_rate": 2.953942022415303e-05,
+      "loss": 0.4968,
+      "step": 10530
+    },
+    {
+      "epoch": 2.5354993983152827,
+      "grad_norm": 1.7734375,
+      "learning_rate": 2.9517067662562634e-05,
+      "loss": 0.4462,
+      "step": 10535
+    },
+    {
+      "epoch": 2.5367027677496994,
+      "grad_norm": 1.6171875,
+      "learning_rate": 2.9494768245941254e-05,
+      "loss": 0.4922,
+      "step": 10540
+    },
+    {
+      "epoch": 2.5379061371841156,
+      "grad_norm": 1.734375,
+      "learning_rate": 2.9472522009900453e-05,
+      "loss": 0.5037,
+      "step": 10545
+    },
+    {
+      "epoch": 2.539109506618532,
+      "grad_norm": 1.65625,
+      "learning_rate": 2.945032898996685e-05,
+      "loss": 0.4743,
+      "step": 10550
+    },
+    {
+      "epoch": 2.5403128760529485,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.942818922158211e-05,
+      "loss": 0.4727,
+      "step": 10555
+    },
+    {
+      "epoch": 2.5415162454873648,
+      "grad_norm": 1.625,
+      "learning_rate": 2.9406102740102805e-05,
+      "loss": 0.489,
+      "step": 10560
+    },
+    {
+      "epoch": 2.542719614921781,
+      "grad_norm": 1.578125,
+      "learning_rate": 2.938406958080045e-05,
+      "loss": 0.4562,
+      "step": 10565
+    },
+    {
+      "epoch": 2.5439229843561972,
+      "grad_norm": 1.75,
+      "learning_rate": 2.93620897788614e-05,
+      "loss": 0.4893,
+      "step": 10570
+    },
+    {
+      "epoch": 2.5451263537906135,
+      "grad_norm": 1.6796875,
+      "learning_rate": 2.934016336938679e-05,
+      "loss": 0.4842,
+      "step": 10575
+    },
+    {
+      "epoch": 2.54632972322503,
+      "grad_norm": 1.6171875,
+      "learning_rate": 2.9318290387392497e-05,
+      "loss": 0.4988,
+      "step": 10580
+    },
+    {
+      "epoch": 2.5475330926594464,
+      "grad_norm": 1.71875,
+      "learning_rate": 2.929647086780906e-05,
+      "loss": 0.4788,
+      "step": 10585
+    },
+    {
+      "epoch": 2.5487364620938626,
+      "grad_norm": 1.6953125,
+      "learning_rate": 2.9274704845481668e-05,
+      "loss": 0.5066,
+      "step": 10590
+    },
+    {
+      "epoch": 2.5499398315282793,
+      "grad_norm": 1.7421875,
+      "learning_rate": 2.925299235517005e-05,
+      "loss": 0.478,
+      "step": 10595
+    },
+    {
+      "epoch": 2.5511432009626955,
+      "grad_norm": 1.6796875,
+      "learning_rate": 2.923133343154847e-05,
+      "loss": 0.4756,
+      "step": 10600
+    },
+    {
+      "epoch": 2.5523465703971118,
+      "grad_norm": 1.7578125,
+      "learning_rate": 2.9209728109205602e-05,
+      "loss": 0.4817,
+      "step": 10605
+    },
+    {
+      "epoch": 2.5535499398315284,
+      "grad_norm": 1.7265625,
+      "learning_rate": 2.9188176422644586e-05,
+      "loss": 0.4795,
+      "step": 10610
+    },
+    {
+      "epoch": 2.5547533092659447,
+      "grad_norm": 1.65625,
+      "learning_rate": 2.916667840628286e-05,
+      "loss": 0.4618,
+      "step": 10615
+    },
+    {
+      "epoch": 2.555956678700361,
+      "grad_norm": 1.7109375,
+      "learning_rate": 2.9145234094452152e-05,
+      "loss": 0.4792,
+      "step": 10620
+    },
+    {
+      "epoch": 2.5571600481347776,
+      "grad_norm": 1.8359375,
+      "learning_rate": 2.912384352139846e-05,
+      "loss": 0.5195,
+      "step": 10625
+    },
+    {
+      "epoch": 2.558363417569194,
+      "grad_norm": 1.6796875,
+      "learning_rate": 2.9102506721281932e-05,
+      "loss": 0.4711,
+      "step": 10630
+    },
+    {
+      "epoch": 2.55956678700361,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.908122372817685e-05,
+      "loss": 0.5014,
+      "step": 10635
+    },
+    {
+      "epoch": 2.5607701564380267,
+      "grad_norm": 1.5625,
+      "learning_rate": 2.9059994576071566e-05,
+      "loss": 0.4875,
+      "step": 10640
+    },
+    {
+      "epoch": 2.561973525872443,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.9038819298868456e-05,
+      "loss": 0.4798,
+      "step": 10645
+    },
+    {
+      "epoch": 2.563176895306859,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.9017697930383852e-05,
+      "loss": 0.5266,
+      "step": 10650
+    },
+    {
+      "epoch": 2.5643802647412755,
+      "grad_norm": 1.6171875,
+      "learning_rate": 2.8996630504348013e-05,
+      "loss": 0.4718,
+      "step": 10655
+    },
+    {
+      "epoch": 2.5655836341756917,
+      "grad_norm": 1.6875,
+      "learning_rate": 2.897561705440501e-05,
+      "loss": 0.47,
+      "step": 10660
+    },
+    {
+      "epoch": 2.5667870036101084,
+      "grad_norm": 1.65625,
+      "learning_rate": 2.8954657614112776e-05,
+      "loss": 0.4771,
+      "step": 10665
+    },
+    {
+      "epoch": 2.5679903730445246,
+      "grad_norm": 1.6796875,
+      "learning_rate": 2.893375221694295e-05,
+      "loss": 0.4754,
+      "step": 10670
+    },
+    {
+      "epoch": 2.569193742478941,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.891290089628086e-05,
+      "loss": 0.4988,
+      "step": 10675
+    },
+    {
+      "epoch": 2.5703971119133575,
+      "grad_norm": 1.734375,
+      "learning_rate": 2.889210368542552e-05,
+      "loss": 0.4776,
+      "step": 10680
+    },
+    {
+      "epoch": 2.5716004813477737,
+      "grad_norm": 1.6171875,
+      "learning_rate": 2.8871360617589482e-05,
+      "loss": 0.4859,
+      "step": 10685
+    },
+    {
+      "epoch": 2.57280385078219,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.8850671725898866e-05,
+      "loss": 0.4934,
+      "step": 10690
+    },
+    {
+      "epoch": 2.5740072202166067,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.883003704339326e-05,
+      "loss": 0.5091,
+      "step": 10695
+    },
+    {
+      "epoch": 2.575210589651023,
+      "grad_norm": 1.8203125,
+      "learning_rate": 2.880945660302568e-05,
+      "loss": 0.5044,
+      "step": 10700
+    },
+    {
+      "epoch": 2.576413959085439,
+      "grad_norm": 1.734375,
+      "learning_rate": 2.8788930437662534e-05,
+      "loss": 0.4895,
+      "step": 10705
+    },
+    {
+      "epoch": 2.577617328519856,
+      "grad_norm": 1.6796875,
+      "learning_rate": 2.8768458580083534e-05,
+      "loss": 0.4935,
+      "step": 10710
+    },
+    {
+      "epoch": 2.578820697954272,
+      "grad_norm": 1.8125,
+      "learning_rate": 2.874804106298168e-05,
+      "loss": 0.4641,
+      "step": 10715
+    },
+    {
+      "epoch": 2.5800240673886883,
+      "grad_norm": 1.6875,
+      "learning_rate": 2.8727677918963175e-05,
+      "loss": 0.493,
+      "step": 10720
+    },
+    {
+      "epoch": 2.581227436823105,
+      "grad_norm": 1.71875,
+      "learning_rate": 2.8707369180547415e-05,
+      "loss": 0.4825,
+      "step": 10725
+    },
+    {
+      "epoch": 2.582430806257521,
+      "grad_norm": 1.6796875,
+      "learning_rate": 2.868711488016689e-05,
+      "loss": 0.4867,
+      "step": 10730
+    },
+    {
+      "epoch": 2.5836341756919374,
+      "grad_norm": 1.6953125,
+      "learning_rate": 2.8666915050167162e-05,
+      "loss": 0.4928,
+      "step": 10735
+    },
+    {
+      "epoch": 2.5848375451263537,
+      "grad_norm": 1.6328125,
+      "learning_rate": 2.8646769722806797e-05,
+      "loss": 0.4806,
+      "step": 10740
+    },
+    {
+      "epoch": 2.58604091456077,
+      "grad_norm": 1.890625,
+      "learning_rate": 2.8626678930257337e-05,
+      "loss": 0.5266,
+      "step": 10745
+    },
+    {
+      "epoch": 2.5872442839951866,
+      "grad_norm": 1.484375,
+      "learning_rate": 2.8606642704603222e-05,
+      "loss": 0.4748,
+      "step": 10750
+    },
+    {
+      "epoch": 2.588447653429603,
+      "grad_norm": 1.625,
+      "learning_rate": 2.8586661077841757e-05,
+      "loss": 0.4731,
+      "step": 10755
+    },
+    {
+      "epoch": 2.589651022864019,
+      "grad_norm": 1.609375,
+      "learning_rate": 2.8566734081883053e-05,
+      "loss": 0.4626,
+      "step": 10760
+    },
+    {
+      "epoch": 2.5908543922984357,
+      "grad_norm": 1.484375,
+      "learning_rate": 2.8546861748549954e-05,
+      "loss": 0.4794,
+      "step": 10765
+    },
+    {
+      "epoch": 2.592057761732852,
+      "grad_norm": 1.5703125,
+      "learning_rate": 2.852704410957806e-05,
+      "loss": 0.5056,
+      "step": 10770
+    },
+    {
+      "epoch": 2.593261131167268,
+      "grad_norm": 1.9375,
+      "learning_rate": 2.8507281196615577e-05,
+      "loss": 0.481,
+      "step": 10775
+    },
+    {
+      "epoch": 2.594464500601685,
+      "grad_norm": 1.625,
+      "learning_rate": 2.8487573041223333e-05,
+      "loss": 0.4982,
+      "step": 10780
+    },
+    {
+      "epoch": 2.595667870036101,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.846791967487472e-05,
+      "loss": 0.4746,
+      "step": 10785
+    },
+    {
+      "epoch": 2.5968712394705173,
+      "grad_norm": 1.7109375,
+      "learning_rate": 2.8448321128955624e-05,
+      "loss": 0.4864,
+      "step": 10790
+    },
+    {
+      "epoch": 2.598074608904934,
+      "grad_norm": 1.90625,
+      "learning_rate": 2.842877743476436e-05,
+      "loss": 0.4606,
+      "step": 10795
+    },
+    {
+      "epoch": 2.5992779783393503,
+      "grad_norm": 1.5078125,
+      "learning_rate": 2.84092886235117e-05,
+      "loss": 0.4661,
+      "step": 10800
+    },
+    {
+      "epoch": 2.6004813477737665,
+      "grad_norm": 1.65625,
+      "learning_rate": 2.8389854726320735e-05,
+      "loss": 0.4523,
+      "step": 10805
+    },
+    {
+      "epoch": 2.601684717208183,
+      "grad_norm": 1.625,
+      "learning_rate": 2.8370475774226854e-05,
+      "loss": 0.4783,
+      "step": 10810
+    },
+    {
+      "epoch": 2.6028880866425994,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.8351151798177713e-05,
+      "loss": 0.4668,
+      "step": 10815
+    },
+    {
+      "epoch": 2.6040914560770156,
+      "grad_norm": 1.7109375,
+      "learning_rate": 2.8331882829033165e-05,
+      "loss": 0.4396,
+      "step": 10820
+    },
+    {
+      "epoch": 2.605294825511432,
+      "grad_norm": 1.6953125,
+      "learning_rate": 2.8312668897565252e-05,
+      "loss": 0.4702,
+      "step": 10825
+    },
+    {
+      "epoch": 2.606498194945848,
+      "grad_norm": 1.578125,
+      "learning_rate": 2.8293510034458053e-05,
+      "loss": 0.4539,
+      "step": 10830
+    },
+    {
+      "epoch": 2.607701564380265,
+      "grad_norm": 1.7421875,
+      "learning_rate": 2.8274406270307774e-05,
+      "loss": 0.4816,
+      "step": 10835
+    },
+    {
+      "epoch": 2.608904933814681,
+      "grad_norm": 1.6328125,
+      "learning_rate": 2.825535763562259e-05,
+      "loss": 0.5128,
+      "step": 10840
+    },
+    {
+      "epoch": 2.6101083032490973,
+      "grad_norm": 1.640625,
+      "learning_rate": 2.823636416082266e-05,
+      "loss": 0.4873,
+      "step": 10845
+    },
+    {
+      "epoch": 2.611311672683514,
+      "grad_norm": 1.6875,
+      "learning_rate": 2.8217425876240016e-05,
+      "loss": 0.4858,
+      "step": 10850
+    },
+    {
+      "epoch": 2.61251504211793,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.81985428121186e-05,
+      "loss": 0.5068,
+      "step": 10855
+    },
+    {
+      "epoch": 2.6137184115523464,
+      "grad_norm": 1.7109375,
+      "learning_rate": 2.8179714998614137e-05,
+      "loss": 0.4981,
+      "step": 10860
+    },
+    {
+      "epoch": 2.614921780986763,
+      "grad_norm": 1.625,
+      "learning_rate": 2.816094246579413e-05,
+      "loss": 0.4599,
+      "step": 10865
+    },
+    {
+      "epoch": 2.6161251504211793,
+      "grad_norm": 1.75,
+      "learning_rate": 2.81422252436378e-05,
+      "loss": 0.5052,
+      "step": 10870
+    },
+    {
+      "epoch": 2.6173285198555956,
+      "grad_norm": 1.6328125,
+      "learning_rate": 2.8123563362036032e-05,
+      "loss": 0.4965,
+      "step": 10875
+    },
+    {
+      "epoch": 2.6185318892900122,
+      "grad_norm": 1.625,
+      "learning_rate": 2.8104956850791343e-05,
+      "loss": 0.4629,
+      "step": 10880
+    },
+    {
+      "epoch": 2.6197352587244285,
+      "grad_norm": 1.625,
+      "learning_rate": 2.808640573961781e-05,
+      "loss": 0.4899,
+      "step": 10885
+    },
+    {
+      "epoch": 2.6209386281588447,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.8067910058141053e-05,
+      "loss": 0.4821,
+      "step": 10890
+    },
+    {
+      "epoch": 2.6221419975932614,
+      "grad_norm": 1.5390625,
+      "learning_rate": 2.804946983589817e-05,
+      "loss": 0.4627,
+      "step": 10895
+    },
+    {
+      "epoch": 2.6233453670276776,
+      "grad_norm": 1.6328125,
+      "learning_rate": 2.8031085102337683e-05,
+      "loss": 0.4992,
+      "step": 10900
+    },
+    {
+      "epoch": 2.624548736462094,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.8012755886819522e-05,
+      "loss": 0.4943,
+      "step": 10905
+    },
+    {
+      "epoch": 2.62575210589651,
+      "grad_norm": 1.7109375,
+      "learning_rate": 2.799448221861491e-05,
+      "loss": 0.4827,
+      "step": 10910
+    },
+    {
+      "epoch": 2.6269554753309263,
+      "grad_norm": 1.6953125,
+      "learning_rate": 2.797626412690642e-05,
+      "loss": 0.5237,
+      "step": 10915
+    },
+    {
+      "epoch": 2.628158844765343,
+      "grad_norm": 1.8359375,
+      "learning_rate": 2.795810164078783e-05,
+      "loss": 0.4704,
+      "step": 10920
+    },
+    {
+      "epoch": 2.6293622141997592,
+      "grad_norm": 1.609375,
+      "learning_rate": 2.793999478926414e-05,
+      "loss": 0.4923,
+      "step": 10925
+    },
+    {
+      "epoch": 2.6305655836341755,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.7921943601251483e-05,
+      "loss": 0.4792,
+      "step": 10930
+    },
+    {
+      "epoch": 2.631768953068592,
+      "grad_norm": 1.75,
+      "learning_rate": 2.790394810557712e-05,
+      "loss": 0.4865,
+      "step": 10935
+    },
+    {
+      "epoch": 2.6329723225030084,
+      "grad_norm": 1.65625,
+      "learning_rate": 2.7886008330979353e-05,
+      "loss": 0.4697,
+      "step": 10940
+    },
+    {
+      "epoch": 2.6341756919374246,
+      "grad_norm": 1.796875,
+      "learning_rate": 2.7868124306107516e-05,
+      "loss": 0.4771,
+      "step": 10945
+    },
+    {
+      "epoch": 2.6353790613718413,
+      "grad_norm": 1.5078125,
+      "learning_rate": 2.785029605952191e-05,
+      "loss": 0.4955,
+      "step": 10950
+    },
+    {
+      "epoch": 2.6365824308062575,
+      "grad_norm": 1.8125,
+      "learning_rate": 2.783252361969374e-05,
+      "loss": 0.4593,
+      "step": 10955
+    },
+    {
+      "epoch": 2.6377858002406738,
+      "grad_norm": 1.7109375,
+      "learning_rate": 2.781480701500512e-05,
+      "loss": 0.5014,
+      "step": 10960
+    },
+    {
+      "epoch": 2.6389891696750905,
+      "grad_norm": 1.796875,
+      "learning_rate": 2.7797146273748968e-05,
+      "loss": 0.4649,
+      "step": 10965
+    },
+    {
+      "epoch": 2.6401925391095067,
+      "grad_norm": 1.546875,
+      "learning_rate": 2.7779541424128996e-05,
+      "loss": 0.4714,
+      "step": 10970
+    },
+    {
+      "epoch": 2.641395908543923,
+      "grad_norm": 1.7265625,
+      "learning_rate": 2.7761992494259687e-05,
+      "loss": 0.507,
+      "step": 10975
+    },
+    {
+      "epoch": 2.6425992779783396,
+      "grad_norm": 1.6171875,
+      "learning_rate": 2.7744499512166185e-05,
+      "loss": 0.4885,
+      "step": 10980
+    },
+    {
+      "epoch": 2.643802647412756,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.77270625057843e-05,
+      "loss": 0.4656,
+      "step": 10985
+    },
+    {
+      "epoch": 2.645006016847172,
+      "grad_norm": 1.7578125,
+      "learning_rate": 2.7709681502960464e-05,
+      "loss": 0.4905,
+      "step": 10990
+    },
+    {
+      "epoch": 2.6462093862815883,
+      "grad_norm": 1.6796875,
+      "learning_rate": 2.769235653145165e-05,
+      "loss": 0.4678,
+      "step": 10995
+    },
+    {
+      "epoch": 2.6474127557160045,
+      "grad_norm": 1.6953125,
+      "learning_rate": 2.7675087618925355e-05,
+      "loss": 0.4929,
+      "step": 11000
+    },
+    {
+      "epoch": 2.6474127557160045,
+      "eval_loss": 0.42500078678131104,
+      "eval_runtime": 2.6494,
+      "eval_samples_per_second": 75.49,
+      "eval_steps_per_second": 75.49,
+      "step": 11000
+    },
+    {
+      "epoch": 2.648616125150421,
+      "grad_norm": 1.5390625,
+      "learning_rate": 2.765787479295958e-05,
+      "loss": 0.4801,
+      "step": 11005
+    },
+    {
+      "epoch": 2.6498194945848375,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.7640718081042713e-05,
+      "loss": 0.4399,
+      "step": 11010
+    },
+    {
+      "epoch": 2.6510228640192537,
+      "grad_norm": 1.6171875,
+      "learning_rate": 2.7623617510573552e-05,
+      "loss": 0.4657,
+      "step": 11015
+    },
+    {
+      "epoch": 2.6522262334536704,
+      "grad_norm": 1.796875,
+      "learning_rate": 2.7606573108861258e-05,
+      "loss": 0.4789,
+      "step": 11020
+    },
+    {
+      "epoch": 2.6534296028880866,
+      "grad_norm": 1.6328125,
+      "learning_rate": 2.7589584903125248e-05,
+      "loss": 0.4456,
+      "step": 11025
+    },
+    {
+      "epoch": 2.654632972322503,
+      "grad_norm": 1.609375,
+      "learning_rate": 2.757265292049524e-05,
+      "loss": 0.4729,
+      "step": 11030
+    },
+    {
+      "epoch": 2.6558363417569195,
+      "grad_norm": 1.734375,
+      "learning_rate": 2.755577718801113e-05,
+      "loss": 0.4684,
+      "step": 11035
+    },
+    {
+      "epoch": 2.6570397111913358,
+      "grad_norm": 1.5859375,
+      "learning_rate": 2.7538957732623012e-05,
+      "loss": 0.4481,
+      "step": 11040
+    },
+    {
+      "epoch": 2.658243080625752,
+      "grad_norm": 1.8984375,
+      "learning_rate": 2.7522194581191066e-05,
+      "loss": 0.4906,
+      "step": 11045
+    },
+    {
+      "epoch": 2.6594464500601687,
+      "grad_norm": 1.875,
+      "learning_rate": 2.7505487760485622e-05,
+      "loss": 0.513,
+      "step": 11050
+    },
+    {
+      "epoch": 2.660649819494585,
+      "grad_norm": 1.6796875,
+      "learning_rate": 2.7488837297186986e-05,
+      "loss": 0.4924,
+      "step": 11055
+    },
+    {
+      "epoch": 2.661853188929001,
+      "grad_norm": 1.640625,
+      "learning_rate": 2.7472243217885502e-05,
+      "loss": 0.4782,
+      "step": 11060
+    },
+    {
+      "epoch": 2.663056558363418,
+      "grad_norm": 1.6953125,
+      "learning_rate": 2.745570554908146e-05,
+      "loss": 0.5242,
+      "step": 11065
+    },
+    {
+      "epoch": 2.664259927797834,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.7439224317185042e-05,
+      "loss": 0.4725,
+      "step": 11070
+    },
+    {
+      "epoch": 2.6654632972322503,
+      "grad_norm": 1.625,
+      "learning_rate": 2.7422799548516356e-05,
+      "loss": 0.5028,
+      "step": 11075
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 1.5234375,
+      "learning_rate": 2.7406431269305273e-05,
+      "loss": 0.4748,
+      "step": 11080
+    },
+    {
+      "epoch": 2.667870036101083,
+      "grad_norm": 1.5234375,
+      "learning_rate": 2.7390119505691513e-05,
+      "loss": 0.464,
+      "step": 11085
+    },
+    {
+      "epoch": 2.6690734055354994,
+      "grad_norm": 1.6171875,
+      "learning_rate": 2.7373864283724494e-05,
+      "loss": 0.4731,
+      "step": 11090
+    },
+    {
+      "epoch": 2.6702767749699157,
+      "grad_norm": 1.5703125,
+      "learning_rate": 2.735766562936337e-05,
+      "loss": 0.4776,
+      "step": 11095
+    },
+    {
+      "epoch": 2.671480144404332,
+      "grad_norm": 1.5234375,
+      "learning_rate": 2.7341523568476932e-05,
+      "loss": 0.4903,
+      "step": 11100
+    },
+    {
+      "epoch": 2.6726835138387486,
+      "grad_norm": 1.609375,
+      "learning_rate": 2.7325438126843623e-05,
+      "loss": 0.4393,
+      "step": 11105
+    },
+    {
+      "epoch": 2.673886883273165,
+      "grad_norm": 1.65625,
+      "learning_rate": 2.7309409330151448e-05,
+      "loss": 0.4801,
+      "step": 11110
+    },
+    {
+      "epoch": 2.675090252707581,
+      "grad_norm": 1.609375,
+      "learning_rate": 2.7293437203997948e-05,
+      "loss": 0.477,
+      "step": 11115
+    },
+    {
+      "epoch": 2.6762936221419977,
+      "grad_norm": 1.6953125,
+      "learning_rate": 2.7277521773890184e-05,
+      "loss": 0.4975,
+      "step": 11120
+    },
+    {
+      "epoch": 2.677496991576414,
+      "grad_norm": 1.7265625,
+      "learning_rate": 2.7261663065244633e-05,
+      "loss": 0.4837,
+      "step": 11125
+    },
+    {
+      "epoch": 2.67870036101083,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.7245861103387245e-05,
+      "loss": 0.4679,
+      "step": 11130
+    },
+    {
+      "epoch": 2.679903730445247,
+      "grad_norm": 1.8828125,
+      "learning_rate": 2.72301159135533e-05,
+      "loss": 0.5291,
+      "step": 11135
+    },
+    {
+      "epoch": 2.681107099879663,
+      "grad_norm": 1.546875,
+      "learning_rate": 2.7214427520887443e-05,
+      "loss": 0.4773,
+      "step": 11140
+    },
+    {
+      "epoch": 2.6823104693140793,
+      "grad_norm": 1.625,
+      "learning_rate": 2.7198795950443602e-05,
+      "loss": 0.4797,
+      "step": 11145
+    },
+    {
+      "epoch": 2.683513838748496,
+      "grad_norm": 1.71875,
+      "learning_rate": 2.7183221227184965e-05,
+      "loss": 0.4914,
+      "step": 11150
+    },
+    {
+      "epoch": 2.6847172081829123,
+      "grad_norm": 1.6953125,
+      "learning_rate": 2.716770337598394e-05,
+      "loss": 0.4819,
+      "step": 11155
+    },
+    {
+      "epoch": 2.6859205776173285,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.715224242162211e-05,
+      "loss": 0.4945,
+      "step": 11160
+    },
+    {
+      "epoch": 2.6871239470517447,
+      "grad_norm": 1.7109375,
+      "learning_rate": 2.713683838879019e-05,
+      "loss": 0.5188,
+      "step": 11165
+    },
+    {
+      "epoch": 2.6883273164861614,
+      "grad_norm": 1.7265625,
+      "learning_rate": 2.712149130208799e-05,
+      "loss": 0.4976,
+      "step": 11170
+    },
+    {
+      "epoch": 2.6895306859205776,
+      "grad_norm": 1.6328125,
+      "learning_rate": 2.7106201186024398e-05,
+      "loss": 0.4624,
+      "step": 11175
+    },
+    {
+      "epoch": 2.690734055354994,
+      "grad_norm": 1.7734375,
+      "learning_rate": 2.7090968065017305e-05,
+      "loss": 0.515,
+      "step": 11180
+    },
+    {
+      "epoch": 2.69193742478941,
+      "grad_norm": 1.6953125,
+      "learning_rate": 2.7075791963393585e-05,
+      "loss": 0.4496,
+      "step": 11185
+    },
+    {
+      "epoch": 2.693140794223827,
+      "grad_norm": 1.6328125,
+      "learning_rate": 2.7060672905389052e-05,
+      "loss": 0.4344,
+      "step": 11190
+    },
+    {
+      "epoch": 2.694344163658243,
+      "grad_norm": 1.546875,
+      "learning_rate": 2.704561091514843e-05,
+      "loss": 0.4846,
+      "step": 11195
+    },
+    {
+      "epoch": 2.6955475330926593,
+      "grad_norm": 1.671875,
+      "learning_rate": 2.7030606016725304e-05,
+      "loss": 0.4739,
+      "step": 11200
+    },
+    {
+      "epoch": 2.696750902527076,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.7015658234082083e-05,
+      "loss": 0.455,
+      "step": 11205
+    },
+    {
+      "epoch": 2.697954271961492,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.7000767591089962e-05,
+      "loss": 0.5107,
+      "step": 11210
+    },
+    {
+      "epoch": 2.6991576413959084,
+      "grad_norm": 1.6953125,
+      "learning_rate": 2.6985934111528894e-05,
+      "loss": 0.4824,
+      "step": 11215
+    },
+    {
+      "epoch": 2.700361010830325,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.6971157819087537e-05,
+      "loss": 0.4571,
+      "step": 11220
+    },
+    {
+      "epoch": 2.7015643802647413,
+      "grad_norm": 1.640625,
+      "learning_rate": 2.6956438737363216e-05,
+      "loss": 0.4845,
+      "step": 11225
+    },
+    {
+      "epoch": 2.7027677496991576,
+      "grad_norm": 1.828125,
+      "learning_rate": 2.6941776889861912e-05,
+      "loss": 0.4821,
+      "step": 11230
+    },
+    {
+      "epoch": 2.7039711191335742,
+      "grad_norm": 1.671875,
+      "learning_rate": 2.6927172299998184e-05,
+      "loss": 0.4681,
+      "step": 11235
+    },
+    {
+      "epoch": 2.7051744885679905,
+      "grad_norm": 1.6171875,
+      "learning_rate": 2.6912624991095162e-05,
+      "loss": 0.4716,
+      "step": 11240
+    },
+    {
+      "epoch": 2.7063778580024067,
+      "grad_norm": 1.8203125,
+      "learning_rate": 2.6898134986384503e-05,
+      "loss": 0.4784,
+      "step": 11245
+    },
+    {
+      "epoch": 2.707581227436823,
+      "grad_norm": 1.53125,
+      "learning_rate": 2.688370230900633e-05,
+      "loss": 0.4818,
+      "step": 11250
+    },
+    {
+      "epoch": 2.7087845968712396,
+      "grad_norm": 1.6328125,
+      "learning_rate": 2.6869326982009258e-05,
+      "loss": 0.454,
+      "step": 11255
+    },
+    {
+      "epoch": 2.709987966305656,
+      "grad_norm": 1.7109375,
+      "learning_rate": 2.6855009028350265e-05,
+      "loss": 0.4757,
+      "step": 11260
+    },
+    {
+      "epoch": 2.711191335740072,
+      "grad_norm": 1.7265625,
+      "learning_rate": 2.6840748470894747e-05,
+      "loss": 0.4624,
+      "step": 11265
+    },
+    {
+      "epoch": 2.7123947051744883,
+      "grad_norm": 1.7109375,
+      "learning_rate": 2.6826545332416404e-05,
+      "loss": 0.476,
+      "step": 11270
+    },
+    {
+      "epoch": 2.713598074608905,
+      "grad_norm": 1.640625,
+      "learning_rate": 2.681239963559728e-05,
+      "loss": 0.4831,
+      "step": 11275
+    },
+    {
+      "epoch": 2.7148014440433212,
+      "grad_norm": 1.734375,
+      "learning_rate": 2.679831140302764e-05,
+      "loss": 0.4741,
+      "step": 11280
+    },
+    {
+      "epoch": 2.7160048134777375,
+      "grad_norm": 1.7109375,
+      "learning_rate": 2.678428065720602e-05,
+      "loss": 0.4468,
+      "step": 11285
+    },
+    {
+      "epoch": 2.717208182912154,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.6770307420539135e-05,
+      "loss": 0.4755,
+      "step": 11290
+    },
+    {
+      "epoch": 2.7184115523465704,
+      "grad_norm": 1.4765625,
+      "learning_rate": 2.675639171534185e-05,
+      "loss": 0.4792,
+      "step": 11295
+    },
+    {
+      "epoch": 2.7196149217809866,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.6742533563837175e-05,
+      "loss": 0.4592,
+      "step": 11300
+    },
+    {
+      "epoch": 2.7208182912154033,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.6728732988156193e-05,
+      "loss": 0.4599,
+      "step": 11305
+    },
+    {
+      "epoch": 2.7220216606498195,
+      "grad_norm": 1.609375,
+      "learning_rate": 2.6714990010338047e-05,
+      "loss": 0.4659,
+      "step": 11310
+    },
+    {
+      "epoch": 2.7232250300842358,
+      "grad_norm": 1.671875,
+      "learning_rate": 2.6701304652329877e-05,
+      "loss": 0.498,
+      "step": 11315
+    },
+    {
+      "epoch": 2.7244283995186525,
+      "grad_norm": 1.578125,
+      "learning_rate": 2.6687676935986846e-05,
+      "loss": 0.4826,
+      "step": 11320
+    },
+    {
+      "epoch": 2.7256317689530687,
+      "grad_norm": 1.765625,
+      "learning_rate": 2.6674106883072023e-05,
+      "loss": 0.4886,
+      "step": 11325
+    },
+    {
+      "epoch": 2.726835138387485,
+      "grad_norm": 1.6953125,
+      "learning_rate": 2.666059451525643e-05,
+      "loss": 0.4858,
+      "step": 11330
+    },
+    {
+      "epoch": 2.728038507821901,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.664713985411893e-05,
+      "loss": 0.4971,
+      "step": 11335
+    },
+    {
+      "epoch": 2.729241877256318,
+      "grad_norm": 1.6875,
+      "learning_rate": 2.6633742921146254e-05,
+      "loss": 0.4797,
+      "step": 11340
+    },
+    {
+      "epoch": 2.730445246690734,
+      "grad_norm": 1.65625,
+      "learning_rate": 2.6620403737732942e-05,
+      "loss": 0.4846,
+      "step": 11345
+    },
+    {
+      "epoch": 2.7316486161251503,
+      "grad_norm": 1.671875,
+      "learning_rate": 2.660712232518129e-05,
+      "loss": 0.4783,
+      "step": 11350
+    },
+    {
+      "epoch": 2.7328519855595665,
+      "grad_norm": 1.7578125,
+      "learning_rate": 2.659389870470135e-05,
+      "loss": 0.466,
+      "step": 11355
+    },
+    {
+      "epoch": 2.7340553549939832,
+      "grad_norm": 1.8515625,
+      "learning_rate": 2.6580732897410888e-05,
+      "loss": 0.4862,
+      "step": 11360
+    },
+    {
+      "epoch": 2.7352587244283995,
+      "grad_norm": 1.5859375,
+      "learning_rate": 2.6567624924335334e-05,
+      "loss": 0.4569,
+      "step": 11365
+    },
+    {
+      "epoch": 2.7364620938628157,
+      "grad_norm": 1.625,
+      "learning_rate": 2.655457480640776e-05,
+      "loss": 0.4681,
+      "step": 11370
+    },
+    {
+      "epoch": 2.7376654632972324,
+      "grad_norm": 1.5703125,
+      "learning_rate": 2.654158256446884e-05,
+      "loss": 0.4885,
+      "step": 11375
+    },
+    {
+      "epoch": 2.7388688327316486,
+      "grad_norm": 1.59375,
+      "learning_rate": 2.652864821926684e-05,
+      "loss": 0.4908,
+      "step": 11380
+    },
+    {
+      "epoch": 2.740072202166065,
+      "grad_norm": 1.7109375,
+      "learning_rate": 2.651577179145755e-05,
+      "loss": 0.445,
+      "step": 11385
+    },
+    {
+      "epoch": 2.7412755716004815,
+      "grad_norm": 1.71875,
+      "learning_rate": 2.6502953301604253e-05,
+      "loss": 0.4859,
+      "step": 11390
+    },
+    {
+      "epoch": 2.7424789410348978,
+      "grad_norm": 1.515625,
+      "learning_rate": 2.6490192770177752e-05,
+      "loss": 0.4743,
+      "step": 11395
+    },
+    {
+      "epoch": 2.743682310469314,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.6477490217556253e-05,
+      "loss": 0.4595,
+      "step": 11400
+    },
+    {
+      "epoch": 2.7448856799037307,
+      "grad_norm": 1.75,
+      "learning_rate": 2.646484566402537e-05,
+      "loss": 0.4935,
+      "step": 11405
+    },
+    {
+      "epoch": 2.746089049338147,
+      "grad_norm": 1.65625,
+      "learning_rate": 2.6452259129778125e-05,
+      "loss": 0.4911,
+      "step": 11410
+    },
+    {
+      "epoch": 2.747292418772563,
+      "grad_norm": 1.6171875,
+      "learning_rate": 2.643973063491486e-05,
+      "loss": 0.4661,
+      "step": 11415
+    },
+    {
+      "epoch": 2.74849578820698,
+      "grad_norm": 1.578125,
+      "learning_rate": 2.6427260199443243e-05,
+      "loss": 0.4701,
+      "step": 11420
+    },
+    {
+      "epoch": 2.749699157641396,
+      "grad_norm": 1.75,
+      "learning_rate": 2.6414847843278208e-05,
+      "loss": 0.463,
+      "step": 11425
+    },
+    {
+      "epoch": 2.7509025270758123,
+      "grad_norm": 1.71875,
+      "learning_rate": 2.640249358624195e-05,
+      "loss": 0.4793,
+      "step": 11430
+    },
+    {
+      "epoch": 2.7521058965102285,
+      "grad_norm": 1.6796875,
+      "learning_rate": 2.6390197448063887e-05,
+      "loss": 0.4497,
+      "step": 11435
+    },
+    {
+      "epoch": 2.7533092659446448,
+      "grad_norm": 1.765625,
+      "learning_rate": 2.6377959448380596e-05,
+      "loss": 0.4766,
+      "step": 11440
+    },
+    {
+      "epoch": 2.7545126353790614,
+      "grad_norm": 1.640625,
+      "learning_rate": 2.636577960673584e-05,
+      "loss": 0.474,
+      "step": 11445
+    },
+    {
+      "epoch": 2.7557160048134777,
+      "grad_norm": 1.625,
+      "learning_rate": 2.6353657942580473e-05,
+      "loss": 0.4788,
+      "step": 11450
+    },
+    {
+      "epoch": 2.756919374247894,
+      "grad_norm": 1.609375,
+      "learning_rate": 2.634159447527249e-05,
+      "loss": 0.4455,
+      "step": 11455
+    },
+    {
+      "epoch": 2.7581227436823106,
+      "grad_norm": 1.7421875,
+      "learning_rate": 2.6329589224076887e-05,
+      "loss": 0.4846,
+      "step": 11460
+    },
+    {
+      "epoch": 2.759326113116727,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.6317642208165745e-05,
+      "loss": 0.4729,
+      "step": 11465
+    },
+    {
+      "epoch": 2.760529482551143,
+      "grad_norm": 1.6171875,
+      "learning_rate": 2.6305753446618094e-05,
+      "loss": 0.4455,
+      "step": 11470
+    },
+    {
+      "epoch": 2.7617328519855597,
+      "grad_norm": 1.6953125,
+      "learning_rate": 2.6293922958419978e-05,
+      "loss": 0.4674,
+      "step": 11475
+    },
+    {
+      "epoch": 2.762936221419976,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.628215076246436e-05,
+      "loss": 0.4855,
+      "step": 11480
+    },
+    {
+      "epoch": 2.764139590854392,
+      "grad_norm": 1.671875,
+      "learning_rate": 2.62704368775511e-05,
+      "loss": 0.4713,
+      "step": 11485
+    },
+    {
+      "epoch": 2.765342960288809,
+      "grad_norm": 1.6796875,
+      "learning_rate": 2.6258781322386965e-05,
+      "loss": 0.4653,
+      "step": 11490
+    },
+    {
+      "epoch": 2.766546329723225,
+      "grad_norm": 1.640625,
+      "learning_rate": 2.6247184115585548e-05,
+      "loss": 0.4888,
+      "step": 11495
+    },
+    {
+      "epoch": 2.7677496991576414,
+      "grad_norm": 1.828125,
+      "learning_rate": 2.6235645275667272e-05,
+      "loss": 0.5032,
+      "step": 11500
+    },
+    {
+      "epoch": 2.7677496991576414,
+      "eval_loss": 0.4223162531852722,
+      "eval_runtime": 2.7761,
+      "eval_samples_per_second": 72.042,
+      "eval_steps_per_second": 72.042,
+      "step": 11500
+    },
+    {
+      "epoch": 2.768953068592058,
+      "grad_norm": 1.5859375,
+      "learning_rate": 2.6224164821059353e-05,
+      "loss": 0.5094,
+      "step": 11505
+    },
+    {
+      "epoch": 2.7701564380264743,
+      "grad_norm": 1.6796875,
+      "learning_rate": 2.621274277009575e-05,
+      "loss": 0.4785,
+      "step": 11510
+    },
+    {
+      "epoch": 2.7713598074608905,
+      "grad_norm": 1.5703125,
+      "learning_rate": 2.6201379141017168e-05,
+      "loss": 0.4683,
+      "step": 11515
+    },
+    {
+      "epoch": 2.7725631768953067,
+      "grad_norm": 1.625,
+      "learning_rate": 2.6190073951971017e-05,
+      "loss": 0.4826,
+      "step": 11520
+    },
+    {
+      "epoch": 2.773766546329723,
+      "grad_norm": 1.75,
+      "learning_rate": 2.6178827221011365e-05,
+      "loss": 0.4548,
+      "step": 11525
+    },
+    {
+      "epoch": 2.7749699157641396,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.6167638966098933e-05,
+      "loss": 0.4904,
+      "step": 11530
+    },
+    {
+      "epoch": 2.776173285198556,
+      "grad_norm": 1.7265625,
+      "learning_rate": 2.6156509205101063e-05,
+      "loss": 0.5256,
+      "step": 11535
+    },
+    {
+      "epoch": 2.777376654632972,
+      "grad_norm": 1.609375,
+      "learning_rate": 2.6145437955791663e-05,
+      "loss": 0.5097,
+      "step": 11540
+    },
+    {
+      "epoch": 2.778580024067389,
+      "grad_norm": 1.578125,
+      "learning_rate": 2.6134425235851232e-05,
+      "loss": 0.4746,
+      "step": 11545
+    },
+    {
+      "epoch": 2.779783393501805,
+      "grad_norm": 1.609375,
+      "learning_rate": 2.612347106286677e-05,
+      "loss": 0.4973,
+      "step": 11550
+    },
+    {
+      "epoch": 2.7809867629362213,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.6112575454331793e-05,
+      "loss": 0.4872,
+      "step": 11555
+    },
+    {
+      "epoch": 2.782190132370638,
+      "grad_norm": 1.5703125,
+      "learning_rate": 2.6101738427646286e-05,
+      "loss": 0.451,
+      "step": 11560
+    },
+    {
+      "epoch": 2.783393501805054,
+      "grad_norm": 1.7890625,
+      "learning_rate": 2.6090960000116686e-05,
+      "loss": 0.5001,
+      "step": 11565
+    },
+    {
+      "epoch": 2.7845968712394704,
+      "grad_norm": 1.6796875,
+      "learning_rate": 2.6080240188955846e-05,
+      "loss": 0.4847,
+      "step": 11570
+    },
+    {
+      "epoch": 2.785800240673887,
+      "grad_norm": 1.6171875,
+      "learning_rate": 2.6069579011283002e-05,
+      "loss": 0.4543,
+      "step": 11575
+    },
+    {
+      "epoch": 2.7870036101083033,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.605897648412377e-05,
+      "loss": 0.4936,
+      "step": 11580
+    },
+    {
+      "epoch": 2.7882069795427196,
+      "grad_norm": 1.65625,
+      "learning_rate": 2.604843262441009e-05,
+      "loss": 0.4913,
+      "step": 11585
+    },
+    {
+      "epoch": 2.7894103489771362,
+      "grad_norm": 1.5546875,
+      "learning_rate": 2.603794744898022e-05,
+      "loss": 0.501,
+      "step": 11590
+    },
+    {
+      "epoch": 2.7906137184115525,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.6027520974578697e-05,
+      "loss": 0.5053,
+      "step": 11595
+    },
+    {
+      "epoch": 2.7918170878459687,
+      "grad_norm": 1.578125,
+      "learning_rate": 2.601715321785631e-05,
+      "loss": 0.47,
+      "step": 11600
+    },
+    {
+      "epoch": 2.793020457280385,
+      "grad_norm": 1.640625,
+      "learning_rate": 2.6006844195370077e-05,
+      "loss": 0.4937,
+      "step": 11605
+    },
+    {
+      "epoch": 2.794223826714801,
+      "grad_norm": 1.6953125,
+      "learning_rate": 2.599659392358323e-05,
+      "loss": 0.4883,
+      "step": 11610
+    },
+    {
+      "epoch": 2.795427196149218,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.5986402418865176e-05,
+      "loss": 0.4875,
+      "step": 11615
+    },
+    {
+      "epoch": 2.796630565583634,
+      "grad_norm": 1.6953125,
+      "learning_rate": 2.597626969749146e-05,
+      "loss": 0.4724,
+      "step": 11620
+    },
+    {
+      "epoch": 2.7978339350180503,
+      "grad_norm": 1.5859375,
+      "learning_rate": 2.5966195775643762e-05,
+      "loss": 0.4903,
+      "step": 11625
+    },
+    {
+      "epoch": 2.799037304452467,
+      "grad_norm": 1.6953125,
+      "learning_rate": 2.5956180669409856e-05,
+      "loss": 0.4942,
+      "step": 11630
+    },
+    {
+      "epoch": 2.8002406738868832,
+      "grad_norm": 1.7578125,
+      "learning_rate": 2.594622439478359e-05,
+      "loss": 0.4572,
+      "step": 11635
+    },
+    {
+      "epoch": 2.8014440433212995,
+      "grad_norm": 1.7421875,
+      "learning_rate": 2.5936326967664868e-05,
+      "loss": 0.4629,
+      "step": 11640
+    },
+    {
+      "epoch": 2.802647412755716,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.592648840385961e-05,
+      "loss": 0.4768,
+      "step": 11645
+    },
+    {
+      "epoch": 2.8038507821901324,
+      "grad_norm": 1.6875,
+      "learning_rate": 2.591670871907972e-05,
+      "loss": 0.491,
+      "step": 11650
+    },
+    {
+      "epoch": 2.8050541516245486,
+      "grad_norm": 1.6796875,
+      "learning_rate": 2.59069879289431e-05,
+      "loss": 0.4437,
+      "step": 11655
+    },
+    {
+      "epoch": 2.8062575210589653,
+      "grad_norm": 1.640625,
+      "learning_rate": 2.5897326048973566e-05,
+      "loss": 0.4843,
+      "step": 11660
+    },
+    {
+      "epoch": 2.8074608904933815,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.588772309460089e-05,
+      "loss": 0.4542,
+      "step": 11665
+    },
+    {
+      "epoch": 2.808664259927798,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.587817908116073e-05,
+      "loss": 0.4739,
+      "step": 11670
+    },
+    {
+      "epoch": 2.8098676293622145,
+      "grad_norm": 1.5859375,
+      "learning_rate": 2.586869402389459e-05,
+      "loss": 0.4724,
+      "step": 11675
+    },
+    {
+      "epoch": 2.8110709987966307,
+      "grad_norm": 1.6171875,
+      "learning_rate": 2.5859267937949874e-05,
+      "loss": 0.4864,
+      "step": 11680
+    },
+    {
+      "epoch": 2.812274368231047,
+      "grad_norm": 1.6171875,
+      "learning_rate": 2.584990083837976e-05,
+      "loss": 0.4378,
+      "step": 11685
+    },
+    {
+      "epoch": 2.813477737665463,
+      "grad_norm": 1.578125,
+      "learning_rate": 2.5840592740143267e-05,
+      "loss": 0.4654,
+      "step": 11690
+    },
+    {
+      "epoch": 2.8146811070998794,
+      "grad_norm": 1.5859375,
+      "learning_rate": 2.5831343658105154e-05,
+      "loss": 0.4958,
+      "step": 11695
+    },
+    {
+      "epoch": 2.815884476534296,
+      "grad_norm": 1.6875,
+      "learning_rate": 2.5822153607035953e-05,
+      "loss": 0.4828,
+      "step": 11700
+    },
+    {
+      "epoch": 2.8170878459687123,
+      "grad_norm": 1.7265625,
+      "learning_rate": 2.5813022601611946e-05,
+      "loss": 0.4764,
+      "step": 11705
+    },
+    {
+      "epoch": 2.8182912154031285,
+      "grad_norm": 1.7734375,
+      "learning_rate": 2.5803950656415066e-05,
+      "loss": 0.4836,
+      "step": 11710
+    },
+    {
+      "epoch": 2.8194945848375452,
+      "grad_norm": 1.5234375,
+      "learning_rate": 2.5794937785932978e-05,
+      "loss": 0.4733,
+      "step": 11715
+    },
+    {
+      "epoch": 2.8206979542719615,
+      "grad_norm": 1.875,
+      "learning_rate": 2.5785984004558992e-05,
+      "loss": 0.4774,
+      "step": 11720
+    },
+    {
+      "epoch": 2.8219013237063777,
+      "grad_norm": 1.796875,
+      "learning_rate": 2.5777089326592036e-05,
+      "loss": 0.4972,
+      "step": 11725
+    },
+    {
+      "epoch": 2.8231046931407944,
+      "grad_norm": 1.859375,
+      "learning_rate": 2.5768253766236677e-05,
+      "loss": 0.4992,
+      "step": 11730
+    },
+    {
+      "epoch": 2.8243080625752106,
+      "grad_norm": 1.6875,
+      "learning_rate": 2.5759477337603057e-05,
+      "loss": 0.4865,
+      "step": 11735
+    },
+    {
+      "epoch": 2.825511432009627,
+      "grad_norm": 1.609375,
+      "learning_rate": 2.5750760054706888e-05,
+      "loss": 0.4935,
+      "step": 11740
+    },
+    {
+      "epoch": 2.8267148014440435,
+      "grad_norm": 1.7109375,
+      "learning_rate": 2.5742101931469435e-05,
+      "loss": 0.488,
+      "step": 11745
+    },
+    {
+      "epoch": 2.8279181708784598,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.5733502981717494e-05,
+      "loss": 0.4821,
+      "step": 11750
+    },
+    {
+      "epoch": 2.829121540312876,
+      "grad_norm": 1.6328125,
+      "learning_rate": 2.572496321918333e-05,
+      "loss": 0.4761,
+      "step": 11755
+    },
+    {
+      "epoch": 2.8303249097472927,
+      "grad_norm": 1.625,
+      "learning_rate": 2.571648265750473e-05,
+      "loss": 0.4941,
+      "step": 11760
+    },
+    {
+      "epoch": 2.831528279181709,
+      "grad_norm": 1.546875,
+      "learning_rate": 2.5708061310224904e-05,
+      "loss": 0.4951,
+      "step": 11765
+    },
+    {
+      "epoch": 2.832731648616125,
+      "grad_norm": 1.5390625,
+      "learning_rate": 2.5699699190792516e-05,
+      "loss": 0.4537,
+      "step": 11770
+    },
+    {
+      "epoch": 2.8339350180505414,
+      "grad_norm": 1.625,
+      "learning_rate": 2.5691396312561637e-05,
+      "loss": 0.4732,
+      "step": 11775
+    },
+    {
+      "epoch": 2.8351383874849576,
+      "grad_norm": 1.8046875,
+      "learning_rate": 2.5683152688791736e-05,
+      "loss": 0.4944,
+      "step": 11780
+    },
+    {
+      "epoch": 2.8363417569193743,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.5674968332647663e-05,
+      "loss": 0.467,
+      "step": 11785
+    },
+    {
+      "epoch": 2.8375451263537905,
+      "grad_norm": 1.6953125,
+      "learning_rate": 2.56668432571996e-05,
+      "loss": 0.4785,
+      "step": 11790
+    },
+    {
+      "epoch": 2.8387484957882068,
+      "grad_norm": 1.75,
+      "learning_rate": 2.5658777475423076e-05,
+      "loss": 0.4583,
+      "step": 11795
+    },
+    {
+      "epoch": 2.8399518652226234,
+      "grad_norm": 1.796875,
+      "learning_rate": 2.565077100019892e-05,
+      "loss": 0.488,
+      "step": 11800
+    },
+    {
+      "epoch": 2.8411552346570397,
+      "grad_norm": 1.609375,
+      "learning_rate": 2.5642823844313247e-05,
+      "loss": 0.5022,
+      "step": 11805
+    },
+    {
+      "epoch": 2.842358604091456,
+      "grad_norm": 1.640625,
+      "learning_rate": 2.563493602045745e-05,
+      "loss": 0.4925,
+      "step": 11810
+    },
+    {
+      "epoch": 2.8435619735258726,
+      "grad_norm": 1.7265625,
+      "learning_rate": 2.5627107541228168e-05,
+      "loss": 0.4912,
+      "step": 11815
+    },
+    {
+      "epoch": 2.844765342960289,
+      "grad_norm": 1.7734375,
+      "learning_rate": 2.5619338419127275e-05,
+      "loss": 0.4603,
+      "step": 11820
+    },
+    {
+      "epoch": 2.845968712394705,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.561162866656183e-05,
+      "loss": 0.4702,
+      "step": 11825
+    },
+    {
+      "epoch": 2.8471720818291217,
+      "grad_norm": 1.5625,
+      "learning_rate": 2.5603978295844106e-05,
+      "loss": 0.4548,
+      "step": 11830
+    },
+    {
+      "epoch": 2.848375451263538,
+      "grad_norm": 1.78125,
+      "learning_rate": 2.5596387319191524e-05,
+      "loss": 0.5103,
+      "step": 11835
+    },
+    {
+      "epoch": 2.849578820697954,
+      "grad_norm": 1.46875,
+      "learning_rate": 2.5588855748726685e-05,
+      "loss": 0.4699,
+      "step": 11840
+    },
+    {
+      "epoch": 2.850782190132371,
+      "grad_norm": 1.796875,
+      "learning_rate": 2.5581383596477283e-05,
+      "loss": 0.4721,
+      "step": 11845
+    },
+    {
+      "epoch": 2.851985559566787,
+      "grad_norm": 1.6796875,
+      "learning_rate": 2.5573970874376144e-05,
+      "loss": 0.4692,
+      "step": 11850
+    },
+    {
+      "epoch": 2.8531889290012034,
+      "grad_norm": 1.6875,
+      "learning_rate": 2.5566617594261185e-05,
+      "loss": 0.5246,
+      "step": 11855
+    },
+    {
+      "epoch": 2.8543922984356196,
+      "grad_norm": 1.71875,
+      "learning_rate": 2.5559323767875385e-05,
+      "loss": 0.4567,
+      "step": 11860
+    },
+    {
+      "epoch": 2.855595667870036,
+      "grad_norm": 1.6953125,
+      "learning_rate": 2.555208940686679e-05,
+      "loss": 0.4832,
+      "step": 11865
+    },
+    {
+      "epoch": 2.8567990373044525,
+      "grad_norm": 1.65625,
+      "learning_rate": 2.5544914522788472e-05,
+      "loss": 0.4807,
+      "step": 11870
+    },
+    {
+      "epoch": 2.8580024067388687,
+      "grad_norm": 1.5625,
+      "learning_rate": 2.553779912709853e-05,
+      "loss": 0.4649,
+      "step": 11875
+    },
+    {
+      "epoch": 2.859205776173285,
+      "grad_norm": 1.4375,
+      "learning_rate": 2.5530743231160042e-05,
+      "loss": 0.4521,
+      "step": 11880
+    },
+    {
+      "epoch": 2.8604091456077017,
+      "grad_norm": 1.5625,
+      "learning_rate": 2.55237468462411e-05,
+      "loss": 0.5021,
+      "step": 11885
+    },
+    {
+      "epoch": 2.861612515042118,
+      "grad_norm": 1.7734375,
+      "learning_rate": 2.551680998351472e-05,
+      "loss": 0.4874,
+      "step": 11890
+    },
+    {
+      "epoch": 2.862815884476534,
+      "grad_norm": 1.71875,
+      "learning_rate": 2.550993265405889e-05,
+      "loss": 0.47,
+      "step": 11895
+    },
+    {
+      "epoch": 2.864019253910951,
+      "grad_norm": 1.65625,
+      "learning_rate": 2.5503114868856515e-05,
+      "loss": 0.506,
+      "step": 11900
+    },
+    {
+      "epoch": 2.865222623345367,
+      "grad_norm": 1.609375,
+      "learning_rate": 2.5496356638795408e-05,
+      "loss": 0.4731,
+      "step": 11905
+    },
+    {
+      "epoch": 2.8664259927797833,
+      "grad_norm": 1.65625,
+      "learning_rate": 2.5489657974668276e-05,
+      "loss": 0.4912,
+      "step": 11910
+    },
+    {
+      "epoch": 2.8676293622142,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.5483018887172713e-05,
+      "loss": 0.4489,
+      "step": 11915
+    },
+    {
+      "epoch": 2.868832731648616,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.5476439386911137e-05,
+      "loss": 0.4685,
+      "step": 11920
+    },
+    {
+      "epoch": 2.8700361010830324,
+      "grad_norm": 1.671875,
+      "learning_rate": 2.546991948439085e-05,
+      "loss": 0.4934,
+      "step": 11925
+    },
+    {
+      "epoch": 2.871239470517449,
+      "grad_norm": 1.734375,
+      "learning_rate": 2.546345919002395e-05,
+      "loss": 0.482,
+      "step": 11930
+    },
+    {
+      "epoch": 2.8724428399518653,
+      "grad_norm": 1.5859375,
+      "learning_rate": 2.545705851412734e-05,
+      "loss": 0.4905,
+      "step": 11935
+    },
+    {
+      "epoch": 2.8736462093862816,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.545071746692274e-05,
+      "loss": 0.496,
+      "step": 11940
+    },
+    {
+      "epoch": 2.874849578820698,
+      "grad_norm": 1.671875,
+      "learning_rate": 2.544443605853662e-05,
+      "loss": 0.4808,
+      "step": 11945
+    },
+    {
+      "epoch": 2.8760529482551145,
+      "grad_norm": 1.6171875,
+      "learning_rate": 2.5438214299000206e-05,
+      "loss": 0.5077,
+      "step": 11950
+    },
+    {
+      "epoch": 2.8772563176895307,
+      "grad_norm": 1.75,
+      "learning_rate": 2.54320521982495e-05,
+      "loss": 0.4662,
+      "step": 11955
+    },
+    {
+      "epoch": 2.878459687123947,
+      "grad_norm": 1.7734375,
+      "learning_rate": 2.542594976612519e-05,
+      "loss": 0.4772,
+      "step": 11960
+    },
+    {
+      "epoch": 2.879663056558363,
+      "grad_norm": 1.609375,
+      "learning_rate": 2.5419907012372698e-05,
+      "loss": 0.4876,
+      "step": 11965
+    },
+    {
+      "epoch": 2.88086642599278,
+      "grad_norm": 1.71875,
+      "learning_rate": 2.5413923946642128e-05,
+      "loss": 0.477,
+      "step": 11970
+    },
+    {
+      "epoch": 2.882069795427196,
+      "grad_norm": 1.7265625,
+      "learning_rate": 2.5408000578488285e-05,
+      "loss": 0.4921,
+      "step": 11975
+    },
+    {
+      "epoch": 2.8832731648616123,
+      "grad_norm": 1.765625,
+      "learning_rate": 2.5402136917370614e-05,
+      "loss": 0.4814,
+      "step": 11980
+    },
+    {
+      "epoch": 2.884476534296029,
+      "grad_norm": 1.5625,
+      "learning_rate": 2.539633297265323e-05,
+      "loss": 0.4581,
+      "step": 11985
+    },
+    {
+      "epoch": 2.8856799037304453,
+      "grad_norm": 1.7734375,
+      "learning_rate": 2.5390588753604865e-05,
+      "loss": 0.4747,
+      "step": 11990
+    },
+    {
+      "epoch": 2.8868832731648615,
+      "grad_norm": 1.65625,
+      "learning_rate": 2.538490426939888e-05,
+      "loss": 0.4919,
+      "step": 11995
+    },
+    {
+      "epoch": 2.888086642599278,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.537927952911324e-05,
+      "loss": 0.5019,
+      "step": 12000
+    },
+    {
+      "epoch": 2.888086642599278,
+      "eval_loss": 0.42064329981803894,
+      "eval_runtime": 2.6982,
+      "eval_samples_per_second": 74.123,
+      "eval_steps_per_second": 74.123,
+      "step": 12000
+    },
+    {
+      "epoch": 2.8892900120336944,
+      "grad_norm": 1.640625,
+      "learning_rate": 2.537371454173051e-05,
+      "loss": 0.4696,
+      "step": 12005
+    },
+    {
+      "epoch": 2.8904933814681106,
+      "grad_norm": 1.6171875,
+      "learning_rate": 2.536820931613781e-05,
+      "loss": 0.4803,
+      "step": 12010
+    },
+    {
+      "epoch": 2.8916967509025273,
+      "grad_norm": 1.7890625,
+      "learning_rate": 2.5362763861126836e-05,
+      "loss": 0.4403,
+      "step": 12015
+    },
+    {
+      "epoch": 2.8929001203369435,
+      "grad_norm": 1.640625,
+      "learning_rate": 2.535737818539384e-05,
+      "loss": 0.5083,
+      "step": 12020
+    },
+    {
+      "epoch": 2.89410348977136,
+      "grad_norm": 1.7265625,
+      "learning_rate": 2.5352052297539577e-05,
+      "loss": 0.4773,
+      "step": 12025
+    },
+    {
+      "epoch": 2.895306859205776,
+      "grad_norm": 1.71875,
+      "learning_rate": 2.5346786206069368e-05,
+      "loss": 0.4629,
+      "step": 12030
+    },
+    {
+      "epoch": 2.8965102286401927,
+      "grad_norm": 1.5625,
+      "learning_rate": 2.5341579919392997e-05,
+      "loss": 0.4756,
+      "step": 12035
+    },
+    {
+      "epoch": 2.897713598074609,
+      "grad_norm": 1.5703125,
+      "learning_rate": 2.5336433445824774e-05,
+      "loss": 0.467,
+      "step": 12040
+    },
+    {
+      "epoch": 2.898916967509025,
+      "grad_norm": 1.625,
+      "learning_rate": 2.533134679358347e-05,
+      "loss": 0.4646,
+      "step": 12045
+    },
+    {
+      "epoch": 2.9001203369434414,
+      "grad_norm": 1.5078125,
+      "learning_rate": 2.5326319970792325e-05,
+      "loss": 0.4964,
+      "step": 12050
+    },
+    {
+      "epoch": 2.901323706377858,
+      "grad_norm": 1.5859375,
+      "learning_rate": 2.5321352985479046e-05,
+      "loss": 0.4581,
+      "step": 12055
+    },
+    {
+      "epoch": 2.9025270758122743,
+      "grad_norm": 1.7890625,
+      "learning_rate": 2.531644584557576e-05,
+      "loss": 0.4934,
+      "step": 12060
+    },
+    {
+      "epoch": 2.9037304452466906,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.5311598558919045e-05,
+      "loss": 0.4836,
+      "step": 12065
+    },
+    {
+      "epoch": 2.9049338146811072,
+      "grad_norm": 1.5859375,
+      "learning_rate": 2.5306811133249887e-05,
+      "loss": 0.4881,
+      "step": 12070
+    },
+    {
+      "epoch": 2.9061371841155235,
+      "grad_norm": 1.6796875,
+      "learning_rate": 2.5302083576213657e-05,
+      "loss": 0.4912,
+      "step": 12075
+    },
+    {
+      "epoch": 2.9073405535499397,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.5297415895360148e-05,
+      "loss": 0.4502,
+      "step": 12080
+    },
+    {
+      "epoch": 2.9085439229843564,
+      "grad_norm": 1.71875,
+      "learning_rate": 2.5292808098143517e-05,
+      "loss": 0.4689,
+      "step": 12085
+    },
+    {
+      "epoch": 2.9097472924187726,
+      "grad_norm": 1.7109375,
+      "learning_rate": 2.5288260191922284e-05,
+      "loss": 0.4673,
+      "step": 12090
+    },
+    {
+      "epoch": 2.910950661853189,
+      "grad_norm": 1.765625,
+      "learning_rate": 2.5283772183959324e-05,
+      "loss": 0.4968,
+      "step": 12095
+    },
+    {
+      "epoch": 2.9121540312876055,
+      "grad_norm": 1.5625,
+      "learning_rate": 2.5279344081421883e-05,
+      "loss": 0.4661,
+      "step": 12100
+    },
+    {
+      "epoch": 2.9133574007220218,
+      "grad_norm": 1.640625,
+      "learning_rate": 2.5274975891381498e-05,
+      "loss": 0.4769,
+      "step": 12105
+    },
+    {
+      "epoch": 2.914560770156438,
+      "grad_norm": 1.625,
+      "learning_rate": 2.5270667620814064e-05,
+      "loss": 0.4779,
+      "step": 12110
+    },
+    {
+      "epoch": 2.9157641395908542,
+      "grad_norm": 1.671875,
+      "learning_rate": 2.5266419276599755e-05,
+      "loss": 0.4539,
+      "step": 12115
+    },
+    {
+      "epoch": 2.916967509025271,
+      "grad_norm": 1.6171875,
+      "learning_rate": 2.5262230865523076e-05,
+      "loss": 0.5058,
+      "step": 12120
+    },
+    {
+      "epoch": 2.918170878459687,
+      "grad_norm": 1.671875,
+      "learning_rate": 2.5258102394272793e-05,
+      "loss": 0.491,
+      "step": 12125
+    },
+    {
+      "epoch": 2.9193742478941034,
+      "grad_norm": 1.6328125,
+      "learning_rate": 2.5254033869441966e-05,
+      "loss": 0.4759,
+      "step": 12130
+    },
+    {
+      "epoch": 2.9205776173285196,
+      "grad_norm": 1.734375,
+      "learning_rate": 2.5250025297527914e-05,
+      "loss": 0.4659,
+      "step": 12135
+    },
+    {
+      "epoch": 2.9217809867629363,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.5246076684932223e-05,
+      "loss": 0.4559,
+      "step": 12140
+    },
+    {
+      "epoch": 2.9229843561973525,
+      "grad_norm": 1.6328125,
+      "learning_rate": 2.5242188037960707e-05,
+      "loss": 0.4981,
+      "step": 12145
+    },
+    {
+      "epoch": 2.9241877256317688,
+      "grad_norm": 1.734375,
+      "learning_rate": 2.5238359362823442e-05,
+      "loss": 0.4798,
+      "step": 12150
+    },
+    {
+      "epoch": 2.9253910950661854,
+      "grad_norm": 1.6171875,
+      "learning_rate": 2.5234590665634706e-05,
+      "loss": 0.475,
+      "step": 12155
+    },
+    {
+      "epoch": 2.9265944645006017,
+      "grad_norm": 1.765625,
+      "learning_rate": 2.523088195241301e-05,
+      "loss": 0.505,
+      "step": 12160
+    },
+    {
+      "epoch": 2.927797833935018,
+      "grad_norm": 1.5859375,
+      "learning_rate": 2.5227233229081072e-05,
+      "loss": 0.5083,
+      "step": 12165
+    },
+    {
+      "epoch": 2.9290012033694346,
+      "grad_norm": 1.5390625,
+      "learning_rate": 2.5223644501465792e-05,
+      "loss": 0.4681,
+      "step": 12170
+    },
+    {
+      "epoch": 2.930204572803851,
+      "grad_norm": 1.6875,
+      "learning_rate": 2.5220115775298276e-05,
+      "loss": 0.4649,
+      "step": 12175
+    },
+    {
+      "epoch": 2.931407942238267,
+      "grad_norm": 1.84375,
+      "learning_rate": 2.52166470562138e-05,
+      "loss": 0.4824,
+      "step": 12180
+    },
+    {
+      "epoch": 2.9326113116726837,
+      "grad_norm": 1.6328125,
+      "learning_rate": 2.5213238349751808e-05,
+      "loss": 0.4894,
+      "step": 12185
+    },
+    {
+      "epoch": 2.9338146811071,
+      "grad_norm": 1.640625,
+      "learning_rate": 2.5209889661355926e-05,
+      "loss": 0.4793,
+      "step": 12190
+    },
+    {
+      "epoch": 2.935018050541516,
+      "grad_norm": 1.7421875,
+      "learning_rate": 2.52066009963739e-05,
+      "loss": 0.4906,
+      "step": 12195
+    },
+    {
+      "epoch": 2.9362214199759324,
+      "grad_norm": 1.7109375,
+      "learning_rate": 2.5203372360057653e-05,
+      "loss": 0.4909,
+      "step": 12200
+    },
+    {
+      "epoch": 2.937424789410349,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.5200203757563216e-05,
+      "loss": 0.479,
+      "step": 12205
+    },
+    {
+      "epoch": 2.9386281588447654,
+      "grad_norm": 1.6796875,
+      "learning_rate": 2.5197095193950768e-05,
+      "loss": 0.4705,
+      "step": 12210
+    },
+    {
+      "epoch": 2.9398315282791816,
+      "grad_norm": 1.625,
+      "learning_rate": 2.5194046674184594e-05,
+      "loss": 0.4807,
+      "step": 12215
+    },
+    {
+      "epoch": 2.941034897713598,
+      "grad_norm": 1.625,
+      "learning_rate": 2.51910582031331e-05,
+      "loss": 0.4625,
+      "step": 12220
+    },
+    {
+      "epoch": 2.9422382671480145,
+      "grad_norm": 1.6328125,
+      "learning_rate": 2.5188129785568787e-05,
+      "loss": 0.4968,
+      "step": 12225
+    },
+    {
+      "epoch": 2.9434416365824307,
+      "grad_norm": 1.6015625,
+      "learning_rate": 2.518526142616826e-05,
+      "loss": 0.4911,
+      "step": 12230
+    },
+    {
+      "epoch": 2.944645006016847,
+      "grad_norm": 1.640625,
+      "learning_rate": 2.5182453129512218e-05,
+      "loss": 0.5157,
+      "step": 12235
+    },
+    {
+      "epoch": 2.9458483754512637,
+      "grad_norm": 1.546875,
+      "learning_rate": 2.5179704900085426e-05,
+      "loss": 0.5027,
+      "step": 12240
+    },
+    {
+      "epoch": 2.94705174488568,
+      "grad_norm": 1.59375,
+      "learning_rate": 2.5177016742276727e-05,
+      "loss": 0.4583,
+      "step": 12245
+    },
+    {
+      "epoch": 2.948255114320096,
+      "grad_norm": 1.59375,
+      "learning_rate": 2.5174388660379044e-05,
+      "loss": 0.4926,
+      "step": 12250
+    },
+    {
+      "epoch": 2.949458483754513,
+      "grad_norm": 1.6953125,
+      "learning_rate": 2.517182065858935e-05,
+      "loss": 0.4835,
+      "step": 12255
+    },
+    {
+      "epoch": 2.950661853188929,
+      "grad_norm": 1.578125,
+      "learning_rate": 2.516931274100866e-05,
+      "loss": 0.4722,
+      "step": 12260
+    },
+    {
+      "epoch": 2.9518652226233453,
+      "grad_norm": 2.234375,
+      "learning_rate": 2.516686491164207e-05,
+      "loss": 0.4709,
+      "step": 12265
+    },
+    {
+      "epoch": 2.953068592057762,
+      "grad_norm": 1.7109375,
+      "learning_rate": 2.516447717439868e-05,
+      "loss": 0.4793,
+      "step": 12270
+    },
+    {
+      "epoch": 2.954271961492178,
+      "grad_norm": 1.65625,
+      "learning_rate": 2.5162149533091646e-05,
+      "loss": 0.4718,
+      "step": 12275
+    },
+    {
+      "epoch": 2.9554753309265944,
+      "grad_norm": 1.5546875,
+      "learning_rate": 2.5159881991438154e-05,
+      "loss": 0.4704,
+      "step": 12280
+    },
+    {
+      "epoch": 2.956678700361011,
+      "grad_norm": 1.734375,
+      "learning_rate": 2.5157674553059392e-05,
+      "loss": 0.466,
+      "step": 12285
+    },
+    {
+      "epoch": 2.9578820697954273,
+      "grad_norm": 1.65625,
+      "learning_rate": 2.515552722148059e-05,
+      "loss": 0.4939,
+      "step": 12290
+    },
+    {
+      "epoch": 2.9590854392298436,
+      "grad_norm": 1.6171875,
+      "learning_rate": 2.515344000013098e-05,
+      "loss": 0.4697,
+      "step": 12295
+    },
+    {
+      "epoch": 2.96028880866426,
+      "grad_norm": 1.7421875,
+      "learning_rate": 2.515141289234378e-05,
+      "loss": 0.4388,
+      "step": 12300
+    },
+    {
+      "epoch": 2.961492178098676,
+      "grad_norm": 1.6875,
+      "learning_rate": 2.5149445901356243e-05,
+      "loss": 0.4616,
+      "step": 12305
+    },
+    {
+      "epoch": 2.9626955475330927,
+      "grad_norm": 1.671875,
+      "learning_rate": 2.514753903030959e-05,
+      "loss": 0.469,
+      "step": 12310
+    },
+    {
+      "epoch": 2.963898916967509,
+      "grad_norm": 1.453125,
+      "learning_rate": 2.5145692282249044e-05,
+      "loss": 0.4812,
+      "step": 12315
+    },
+    {
+      "epoch": 2.965102286401925,
+      "grad_norm": 1.7578125,
+      "learning_rate": 2.514390566012381e-05,
+      "loss": 0.4706,
+      "step": 12320
+    },
+    {
+      "epoch": 2.966305655836342,
+      "grad_norm": 1.7421875,
+      "learning_rate": 2.514217916678708e-05,
+      "loss": 0.477,
+      "step": 12325
+    },
+    {
+      "epoch": 2.967509025270758,
+      "grad_norm": 1.75,
+      "learning_rate": 2.514051280499601e-05,
+      "loss": 0.4592,
+      "step": 12330
+    },
+    {
+      "epoch": 2.9687123947051743,
+      "grad_norm": 1.6171875,
+      "learning_rate": 2.5138906577411732e-05,
+      "loss": 0.4594,
+      "step": 12335
+    },
+    {
+      "epoch": 2.969915764139591,
+      "grad_norm": 1.5625,
+      "learning_rate": 2.5137360486599356e-05,
+      "loss": 0.4699,
+      "step": 12340
+    },
+    {
+      "epoch": 2.9711191335740073,
+      "grad_norm": 1.6171875,
+      "learning_rate": 2.513587453502794e-05,
+      "loss": 0.477,
+      "step": 12345
+    },
+    {
+      "epoch": 2.9723225030084235,
+      "grad_norm": 1.625,
+      "learning_rate": 2.513444872507051e-05,
+      "loss": 0.4775,
+      "step": 12350
+    },
+    {
+      "epoch": 2.97352587244284,
+      "grad_norm": 1.9140625,
+      "learning_rate": 2.513308305900404e-05,
+      "loss": 0.4711,
+      "step": 12355
+    },
+    {
+      "epoch": 2.9747292418772564,
+      "grad_norm": 1.6171875,
+      "learning_rate": 2.5131777539009473e-05,
+      "loss": 0.4933,
+      "step": 12360
+    },
+    {
+      "epoch": 2.9759326113116726,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.5130532167171676e-05,
+      "loss": 0.4859,
+      "step": 12365
+    },
+    {
+      "epoch": 2.9771359807460893,
+      "grad_norm": 1.6328125,
+      "learning_rate": 2.5129346945479483e-05,
+      "loss": 0.48,
+      "step": 12370
+    },
+    {
+      "epoch": 2.9783393501805056,
+      "grad_norm": 1.8984375,
+      "learning_rate": 2.5128221875825652e-05,
+      "loss": 0.4538,
+      "step": 12375
+    },
+    {
+      "epoch": 2.979542719614922,
+      "grad_norm": 1.7109375,
+      "learning_rate": 2.5127156960006893e-05,
+      "loss": 0.4736,
+      "step": 12380
+    },
+    {
+      "epoch": 2.980746089049338,
+      "grad_norm": 1.625,
+      "learning_rate": 2.5126152199723852e-05,
+      "loss": 0.4635,
+      "step": 12385
+    },
+    {
+      "epoch": 2.9819494584837543,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.51252075965811e-05,
+      "loss": 0.5018,
+      "step": 12390
+    },
+    {
+      "epoch": 2.983152827918171,
+      "grad_norm": 1.609375,
+      "learning_rate": 2.512432315208714e-05,
+      "loss": 0.4592,
+      "step": 12395
+    },
+    {
+      "epoch": 2.984356197352587,
+      "grad_norm": 1.6875,
+      "learning_rate": 2.5123498867654397e-05,
+      "loss": 0.4874,
+      "step": 12400
+    },
+    {
+      "epoch": 2.9855595667870034,
+      "grad_norm": 1.53125,
+      "learning_rate": 2.5122734744599256e-05,
+      "loss": 0.4315,
+      "step": 12405
+    },
+    {
+      "epoch": 2.98676293622142,
+      "grad_norm": 1.796875,
+      "learning_rate": 2.5122030784141974e-05,
+      "loss": 0.4757,
+      "step": 12410
+    },
+    {
+      "epoch": 2.9879663056558363,
+      "grad_norm": 1.75,
+      "learning_rate": 2.5121386987406766e-05,
+      "loss": 0.4613,
+      "step": 12415
+    },
+    {
+      "epoch": 2.9891696750902526,
+      "grad_norm": 1.6875,
+      "learning_rate": 2.5120803355421764e-05,
+      "loss": 0.4887,
+      "step": 12420
+    },
+    {
+      "epoch": 2.9903730445246692,
+      "grad_norm": 1.59375,
+      "learning_rate": 2.5120279889119002e-05,
+      "loss": 0.4892,
+      "step": 12425
+    },
+    {
+      "epoch": 2.9915764139590855,
+      "grad_norm": 1.578125,
+      "learning_rate": 2.5119816589334445e-05,
+      "loss": 0.4503,
+      "step": 12430
+    },
+    {
+      "epoch": 2.9927797833935017,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.511941345680798e-05,
+      "loss": 0.4954,
+      "step": 12435
+    },
+    {
+      "epoch": 2.9939831528279184,
+      "grad_norm": 1.625,
+      "learning_rate": 2.5119070492183377e-05,
+      "loss": 0.4456,
+      "step": 12440
+    },
+    {
+      "epoch": 2.9951865222623346,
+      "grad_norm": 1.6171875,
+      "learning_rate": 2.5118787696008367e-05,
+      "loss": 0.4645,
+      "step": 12445
+    },
+    {
+      "epoch": 2.996389891696751,
+      "grad_norm": 1.578125,
+      "learning_rate": 2.5118565068734546e-05,
+      "loss": 0.4891,
+      "step": 12450
+    },
+    {
+      "epoch": 2.9975932611311675,
+      "grad_norm": 1.7265625,
+      "learning_rate": 2.5118402610717452e-05,
+      "loss": 0.4804,
+      "step": 12455
+    },
+    {
+      "epoch": 2.9987966305655838,
+      "grad_norm": 1.890625,
+      "learning_rate": 2.5118300322216534e-05,
+      "loss": 0.4957,
+      "step": 12460
+    },
+    {
+      "epoch": 2.99927797833935,
+      "eval_loss": 0.42031019926071167,
+      "eval_runtime": 2.6521,
+      "eval_samples_per_second": 75.411,
+      "eval_steps_per_second": 75.411,
+      "step": 12462
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 12465,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.00021780783616e+17,
+  "train_batch_size": 48,
+  "trial_name": null,
+  "trial_params": null
+}