diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,10856 @@
+{
+  "best_metric": 0.812837658080022,
+  "best_model_checkpoint": "/data/hungnm/unisentiment/modernBERT-large-sentiment/checkpoint-4611",
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 7685,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0032530904359141183,
+      "grad_norm": 0.0,
+      "learning_rate": 0.0,
+      "loss": 2.3733,
+      "step": 5
+    },
+    {
+      "epoch": 0.006506180871828237,
+      "grad_norm": 0.0,
+      "learning_rate": 0.0,
+      "loss": 2.3771,
+      "step": 10
+    },
+    {
+      "epoch": 0.009759271307742356,
+      "grad_norm": 0.0,
+      "learning_rate": 0.0,
+      "loss": 2.3846,
+      "step": 15
+    },
+    {
+      "epoch": 0.013012361743656473,
+      "grad_norm": 6.415463447570801,
+      "learning_rate": 1.2987012987012988e-06,
+      "loss": 2.3719,
+      "step": 20
+    },
+    {
+      "epoch": 0.01626545217957059,
+      "grad_norm": 6.333994388580322,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 2.3437,
+      "step": 25
+    },
+    {
+      "epoch": 0.01951854261548471,
+      "grad_norm": 8.240402221679688,
+      "learning_rate": 7.792207792207792e-06,
+      "loss": 2.2279,
+      "step": 30
+    },
+    {
+      "epoch": 0.02277163305139883,
+      "grad_norm": 3.9838759899139404,
+      "learning_rate": 1.103896103896104e-05,
+      "loss": 2.079,
+      "step": 35
+    },
+    {
+      "epoch": 0.026024723487312947,
+      "grad_norm": 4.472020626068115,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 1.997,
+      "step": 40
+    },
+    {
+      "epoch": 0.029277813923227064,
+      "grad_norm": 3.7005715370178223,
+      "learning_rate": 1.7532467532467535e-05,
+      "loss": 1.9341,
+      "step": 45
+    },
+    {
+      "epoch": 0.03253090435914118,
+      "grad_norm": 13.776549339294434,
+      "learning_rate": 2.012987012987013e-05,
+      "loss": 1.8824,
+      "step": 50
+    },
+    {
+      "epoch": 0.035783994795055306,
+      "grad_norm": 21.78818130493164,
+      "learning_rate": 2.272727272727273e-05,
+      "loss": 1.8142,
+      "step": 55
+    },
+    {
+      "epoch": 0.03903708523096942,
+      "grad_norm": 5.216381549835205,
+      "learning_rate": 2.5324675324675325e-05,
+      "loss": 1.7531,
+      "step": 60
+    },
+    {
+      "epoch": 0.04229017566688354,
+      "grad_norm": 1.8160972595214844,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 1.7342,
+      "step": 65
+    },
+    {
+      "epoch": 0.04554326610279766,
+      "grad_norm": 6.6145339012146,
+      "learning_rate": 3.181818181818182e-05,
+      "loss": 1.719,
+      "step": 70
+    },
+    {
+      "epoch": 0.048796356538711776,
+      "grad_norm": 2.8823325634002686,
+      "learning_rate": 3.506493506493507e-05,
+      "loss": 1.6266,
+      "step": 75
+    },
+    {
+      "epoch": 0.05204944697462589,
+      "grad_norm": 7.7064433097839355,
+      "learning_rate": 3.831168831168831e-05,
+      "loss": 1.5322,
+      "step": 80
+    },
+    {
+      "epoch": 0.05530253741054001,
+      "grad_norm": 2.749300241470337,
+      "learning_rate": 4.155844155844156e-05,
+      "loss": 1.5303,
+      "step": 85
+    },
+    {
+      "epoch": 0.05855562784645413,
+      "grad_norm": 2.898153305053711,
+      "learning_rate": 4.4805194805194805e-05,
+      "loss": 1.4395,
+      "step": 90
+    },
+    {
+      "epoch": 0.06180871828236825,
+      "grad_norm": 2.754175901412964,
+      "learning_rate": 4.8051948051948054e-05,
+      "loss": 1.3824,
+      "step": 95
+    },
+    {
+      "epoch": 0.06506180871828236,
+      "grad_norm": 6.261497974395752,
+      "learning_rate": 4.99999914743261e-05,
+      "loss": 1.3976,
+      "step": 100
+    },
+    {
+      "epoch": 0.06831489915419649,
+      "grad_norm": 1.882625699043274,
+      "learning_rate": 4.9999895560561514e-05,
+      "loss": 1.3925,
+      "step": 105
+    },
+    {
+      "epoch": 0.07156798959011061,
+      "grad_norm": 1.972509503364563,
+      "learning_rate": 4.9999693076350204e-05,
+      "loss": 1.3423,
+      "step": 110
+    },
+    {
+      "epoch": 0.07482108002602472,
+      "grad_norm": 3.2671749591827393,
+      "learning_rate": 4.999938402255531e-05,
+      "loss": 1.2724,
+      "step": 115
+    },
+    {
+      "epoch": 0.07807417046193885,
+      "grad_norm": 2.316319465637207,
+      "learning_rate": 4.9998968400494294e-05,
+      "loss": 1.2621,
+      "step": 120
+    },
+    {
+      "epoch": 0.08132726089785296,
+      "grad_norm": 2.291067361831665,
+      "learning_rate": 4.9998446211938876e-05,
+      "loss": 1.2517,
+      "step": 125
+    },
+    {
+      "epoch": 0.08458035133376708,
+      "grad_norm": 2.1083593368530273,
+      "learning_rate": 4.999781745911506e-05,
+      "loss": 1.2589,
+      "step": 130
+    },
+    {
+      "epoch": 0.08783344176968119,
+      "grad_norm": 2.5555965900421143,
+      "learning_rate": 4.9997082144703124e-05,
+      "loss": 1.2316,
+      "step": 135
+    },
+    {
+      "epoch": 0.09108653220559532,
+      "grad_norm": 1.8004491329193115,
+      "learning_rate": 4.999624027183758e-05,
+      "loss": 1.2197,
+      "step": 140
+    },
+    {
+      "epoch": 0.09433962264150944,
+      "grad_norm": 1.7682017087936401,
+      "learning_rate": 4.999529184410721e-05,
+      "loss": 1.1816,
+      "step": 145
+    },
+    {
+      "epoch": 0.09759271307742355,
+      "grad_norm": 2.857111692428589,
+      "learning_rate": 4.999423686555498e-05,
+      "loss": 1.1797,
+      "step": 150
+    },
+    {
+      "epoch": 0.10084580351333768,
+      "grad_norm": 3.4952378273010254,
+      "learning_rate": 4.999307534067812e-05,
+      "loss": 1.1883,
+      "step": 155
+    },
+    {
+      "epoch": 0.10409889394925179,
+      "grad_norm": 2.0805113315582275,
+      "learning_rate": 4.999180727442799e-05,
+      "loss": 1.2026,
+      "step": 160
+    },
+    {
+      "epoch": 0.10735198438516591,
+      "grad_norm": 2.541172504425049,
+      "learning_rate": 4.9990432672210174e-05,
+      "loss": 1.1911,
+      "step": 165
+    },
+    {
+      "epoch": 0.11060507482108002,
+      "grad_norm": 2.9252259731292725,
+      "learning_rate": 4.9988951539884365e-05,
+      "loss": 1.2091,
+      "step": 170
+    },
+    {
+      "epoch": 0.11385816525699415,
+      "grad_norm": 1.487844705581665,
+      "learning_rate": 4.9987363883764396e-05,
+      "loss": 1.1773,
+      "step": 175
+    },
+    {
+      "epoch": 0.11711125569290826,
+      "grad_norm": 1.4860082864761353,
+      "learning_rate": 4.9985669710618156e-05,
+      "loss": 1.1507,
+      "step": 180
+    },
+    {
+      "epoch": 0.12036434612882238,
+      "grad_norm": 1.6352834701538086,
+      "learning_rate": 4.9983869027667656e-05,
+      "loss": 1.111,
+      "step": 185
+    },
+    {
+      "epoch": 0.1236174365647365,
+      "grad_norm": 1.2237991094589233,
+      "learning_rate": 4.99819618425889e-05,
+      "loss": 1.1241,
+      "step": 190
+    },
+    {
+      "epoch": 0.12687052700065063,
+      "grad_norm": 1.7817842960357666,
+      "learning_rate": 4.997994816351191e-05,
+      "loss": 1.1474,
+      "step": 195
+    },
+    {
+      "epoch": 0.13012361743656473,
+      "grad_norm": 1.440572738647461,
+      "learning_rate": 4.997782799902065e-05,
+      "loss": 1.1401,
+      "step": 200
+    },
+    {
+      "epoch": 0.13337670787247885,
+      "grad_norm": 1.1736341714859009,
+      "learning_rate": 4.997560135815307e-05,
+      "loss": 1.1317,
+      "step": 205
+    },
+    {
+      "epoch": 0.13662979830839297,
+      "grad_norm": 1.5103353261947632,
+      "learning_rate": 4.997326825040094e-05,
+      "loss": 1.1289,
+      "step": 210
+    },
+    {
+      "epoch": 0.1398828887443071,
+      "grad_norm": 2.190202236175537,
+      "learning_rate": 4.997082868570993e-05,
+      "loss": 1.1173,
+      "step": 215
+    },
+    {
+      "epoch": 0.14313597918022122,
+      "grad_norm": 1.1619236469268799,
+      "learning_rate": 4.9968282674479486e-05,
+      "loss": 1.1033,
+      "step": 220
+    },
+    {
+      "epoch": 0.14638906961613532,
+      "grad_norm": 1.4644910097122192,
+      "learning_rate": 4.9965630227562866e-05,
+      "loss": 1.0935,
+      "step": 225
+    },
+    {
+      "epoch": 0.14964216005204944,
+      "grad_norm": 1.8269380331039429,
+      "learning_rate": 4.9962871356266994e-05,
+      "loss": 1.0968,
+      "step": 230
+    },
+    {
+      "epoch": 0.15289525048796357,
+      "grad_norm": 1.0320109128952026,
+      "learning_rate": 4.996000607235248e-05,
+      "loss": 1.0718,
+      "step": 235
+    },
+    {
+      "epoch": 0.1561483409238777,
+      "grad_norm": 2.302537202835083,
+      "learning_rate": 4.995703438803359e-05,
+      "loss": 1.1044,
+      "step": 240
+    },
+    {
+      "epoch": 0.1594014313597918,
+      "grad_norm": 2.2435052394866943,
+      "learning_rate": 4.995395631597809e-05,
+      "loss": 1.0936,
+      "step": 245
+    },
+    {
+      "epoch": 0.16265452179570591,
+      "grad_norm": 2.0816986560821533,
+      "learning_rate": 4.995077186930731e-05,
+      "loss": 1.1036,
+      "step": 250
+    },
+    {
+      "epoch": 0.16590761223162004,
+      "grad_norm": 1.0913221836090088,
+      "learning_rate": 4.994748106159602e-05,
+      "loss": 1.0906,
+      "step": 255
+    },
+    {
+      "epoch": 0.16916070266753416,
+      "grad_norm": 1.5220658779144287,
+      "learning_rate": 4.9944083906872405e-05,
+      "loss": 1.0789,
+      "step": 260
+    },
+    {
+      "epoch": 0.1724137931034483,
+      "grad_norm": 2.121184825897217,
+      "learning_rate": 4.994058041961796e-05,
+      "loss": 1.0718,
+      "step": 265
+    },
+    {
+      "epoch": 0.17566688353936238,
+      "grad_norm": 1.4716627597808838,
+      "learning_rate": 4.9936970614767485e-05,
+      "loss": 1.0378,
+      "step": 270
+    },
+    {
+      "epoch": 0.1789199739752765,
+      "grad_norm": 1.700437068939209,
+      "learning_rate": 4.993325450770898e-05,
+      "loss": 1.0586,
+      "step": 275
+    },
+    {
+      "epoch": 0.18217306441119063,
+      "grad_norm": 1.87433660030365,
+      "learning_rate": 4.9929432114283614e-05,
+      "loss": 1.0486,
+      "step": 280
+    },
+    {
+      "epoch": 0.18542615484710476,
+      "grad_norm": 1.4089707136154175,
+      "learning_rate": 4.992550345078559e-05,
+      "loss": 1.0547,
+      "step": 285
+    },
+    {
+      "epoch": 0.18867924528301888,
+      "grad_norm": 1.6576772928237915,
+      "learning_rate": 4.992146853396219e-05,
+      "loss": 1.033,
+      "step": 290
+    },
+    {
+      "epoch": 0.19193233571893298,
+      "grad_norm": 1.2037906646728516,
+      "learning_rate": 4.9917327381013585e-05,
+      "loss": 1.0335,
+      "step": 295
+    },
+    {
+      "epoch": 0.1951854261548471,
+      "grad_norm": 1.2297475337982178,
+      "learning_rate": 4.9913080009592824e-05,
+      "loss": 1.0508,
+      "step": 300
+    },
+    {
+      "epoch": 0.19843851659076123,
+      "grad_norm": 1.6239577531814575,
+      "learning_rate": 4.990872643780577e-05,
+      "loss": 1.0528,
+      "step": 305
+    },
+    {
+      "epoch": 0.20169160702667535,
+      "grad_norm": 1.4611440896987915,
+      "learning_rate": 4.9904266684210964e-05,
+      "loss": 1.0437,
+      "step": 310
+    },
+    {
+      "epoch": 0.20494469746258945,
+      "grad_norm": 1.5891550779342651,
+      "learning_rate": 4.989970076781961e-05,
+      "loss": 1.0262,
+      "step": 315
+    },
+    {
+      "epoch": 0.20819778789850357,
+      "grad_norm": 1.6164276599884033,
+      "learning_rate": 4.9895028708095474e-05,
+      "loss": 1.0388,
+      "step": 320
+    },
+    {
+      "epoch": 0.2114508783344177,
+      "grad_norm": 1.509124517440796,
+      "learning_rate": 4.989025052495476e-05,
+      "loss": 1.0332,
+      "step": 325
+    },
+    {
+      "epoch": 0.21470396877033182,
+      "grad_norm": 1.2852742671966553,
+      "learning_rate": 4.988536623876609e-05,
+      "loss": 1.0306,
+      "step": 330
+    },
+    {
+      "epoch": 0.21795705920624595,
+      "grad_norm": 1.3699005842208862,
+      "learning_rate": 4.988037587035036e-05,
+      "loss": 1.0266,
+      "step": 335
+    },
+    {
+      "epoch": 0.22121014964216004,
+      "grad_norm": 1.5352535247802734,
+      "learning_rate": 4.98752794409807e-05,
+      "loss": 1.0281,
+      "step": 340
+    },
+    {
+      "epoch": 0.22446324007807417,
+      "grad_norm": 1.1084707975387573,
+      "learning_rate": 4.9870076972382354e-05,
+      "loss": 0.9941,
+      "step": 345
+    },
+    {
+      "epoch": 0.2277163305139883,
+      "grad_norm": 2.580686330795288,
+      "learning_rate": 4.9864768486732585e-05,
+      "loss": 1.022,
+      "step": 350
+    },
+    {
+      "epoch": 0.23096942094990242,
+      "grad_norm": 22.110929489135742,
+      "learning_rate": 4.98593540066606e-05,
+      "loss": 1.0305,
+      "step": 355
+    },
+    {
+      "epoch": 0.2342225113858165,
+      "grad_norm": 1.32615065574646,
+      "learning_rate": 4.985383355524743e-05,
+      "loss": 1.0218,
+      "step": 360
+    },
+    {
+      "epoch": 0.23747560182173064,
+      "grad_norm": 1.7118041515350342,
+      "learning_rate": 4.984820715602585e-05,
+      "loss": 1.0216,
+      "step": 365
+    },
+    {
+      "epoch": 0.24072869225764476,
+      "grad_norm": 0.9846453070640564,
+      "learning_rate": 4.984247483298029e-05,
+      "loss": 1.0244,
+      "step": 370
+    },
+    {
+      "epoch": 0.24398178269355889,
+      "grad_norm": 1.1930924654006958,
+      "learning_rate": 4.9836636610546697e-05,
+      "loss": 1.0333,
+      "step": 375
+    },
+    {
+      "epoch": 0.247234873129473,
+      "grad_norm": 1.5189563035964966,
+      "learning_rate": 4.983069251361244e-05,
+      "loss": 1.0261,
+      "step": 380
+    },
+    {
+      "epoch": 0.2504879635653871,
+      "grad_norm": 1.648707389831543,
+      "learning_rate": 4.982464256751624e-05,
+      "loss": 1.032,
+      "step": 385
+    },
+    {
+      "epoch": 0.25374105400130126,
+      "grad_norm": 1.6429810523986816,
+      "learning_rate": 4.981848679804803e-05,
+      "loss": 1.0325,
+      "step": 390
+    },
+    {
+      "epoch": 0.25699414443721535,
+      "grad_norm": 1.6516709327697754,
+      "learning_rate": 4.981222523144882e-05,
+      "loss": 1.0036,
+      "step": 395
+    },
+    {
+      "epoch": 0.26024723487312945,
+      "grad_norm": 1.458806037902832,
+      "learning_rate": 4.980585789441066e-05,
+      "loss": 1.0104,
+      "step": 400
+    },
+    {
+      "epoch": 0.2635003253090436,
+      "grad_norm": 1.695860505104065,
+      "learning_rate": 4.979938481407645e-05,
+      "loss": 1.0072,
+      "step": 405
+    },
+    {
+      "epoch": 0.2667534157449577,
+      "grad_norm": 1.897962212562561,
+      "learning_rate": 4.9792806018039876e-05,
+      "loss": 1.0102,
+      "step": 410
+    },
+    {
+      "epoch": 0.27000650618087185,
+      "grad_norm": 1.5890872478485107,
+      "learning_rate": 4.9786121534345265e-05,
+      "loss": 0.9945,
+      "step": 415
+    },
+    {
+      "epoch": 0.27325959661678595,
+      "grad_norm": 1.090980887413025,
+      "learning_rate": 4.977933139148746e-05,
+      "loss": 0.9897,
+      "step": 420
+    },
+    {
+      "epoch": 0.27651268705270005,
+      "grad_norm": 1.3130533695220947,
+      "learning_rate": 4.977243561841174e-05,
+      "loss": 1.0161,
+      "step": 425
+    },
+    {
+      "epoch": 0.2797657774886142,
+      "grad_norm": 1.4305720329284668,
+      "learning_rate": 4.976543424451365e-05,
+      "loss": 0.9802,
+      "step": 430
+    },
+    {
+      "epoch": 0.2830188679245283,
+      "grad_norm": 1.3603191375732422,
+      "learning_rate": 4.975832729963888e-05,
+      "loss": 0.9904,
+      "step": 435
+    },
+    {
+      "epoch": 0.28627195836044245,
+      "grad_norm": 4.940343856811523,
+      "learning_rate": 4.9751114814083186e-05,
+      "loss": 0.9873,
+      "step": 440
+    },
+    {
+      "epoch": 0.28952504879635654,
+      "grad_norm": 2.8304941654205322,
+      "learning_rate": 4.974379681859221e-05,
+      "loss": 0.9855,
+      "step": 445
+    },
+    {
+      "epoch": 0.29277813923227064,
+      "grad_norm": 1.6684479713439941,
+      "learning_rate": 4.973637334436135e-05,
+      "loss": 1.0027,
+      "step": 450
+    },
+    {
+      "epoch": 0.2960312296681848,
+      "grad_norm": 1.036837100982666,
+      "learning_rate": 4.972884442303566e-05,
+      "loss": 0.999,
+      "step": 455
+    },
+    {
+      "epoch": 0.2992843201040989,
+      "grad_norm": 1.512636423110962,
+      "learning_rate": 4.972121008670971e-05,
+      "loss": 0.9988,
+      "step": 460
+    },
+    {
+      "epoch": 0.302537410540013,
+      "grad_norm": 1.7532439231872559,
+      "learning_rate": 4.97134703679274e-05,
+      "loss": 0.984,
+      "step": 465
+    },
+    {
+      "epoch": 0.30579050097592714,
+      "grad_norm": 1.0935814380645752,
+      "learning_rate": 4.970562529968189e-05,
+      "loss": 0.9725,
+      "step": 470
+    },
+    {
+      "epoch": 0.30904359141184123,
+      "grad_norm": 0.9008370041847229,
+      "learning_rate": 4.969767491541543e-05,
+      "loss": 0.9788,
+      "step": 475
+    },
+    {
+      "epoch": 0.3122966818477554,
+      "grad_norm": 0.9959591627120972,
+      "learning_rate": 4.9689619249019174e-05,
+      "loss": 0.9821,
+      "step": 480
+    },
+    {
+      "epoch": 0.3155497722836695,
+      "grad_norm": 1.1844329833984375,
+      "learning_rate": 4.9681458334833114e-05,
+      "loss": 0.9588,
+      "step": 485
+    },
+    {
+      "epoch": 0.3188028627195836,
+      "grad_norm": 1.136703372001648,
+      "learning_rate": 4.9673192207645894e-05,
+      "loss": 0.9948,
+      "step": 490
+    },
+    {
+      "epoch": 0.32205595315549773,
+      "grad_norm": 1.3169392347335815,
+      "learning_rate": 4.9664820902694654e-05,
+      "loss": 0.9924,
+      "step": 495
+    },
+    {
+      "epoch": 0.32530904359141183,
+      "grad_norm": 0.863590657711029,
+      "learning_rate": 4.9656344455664885e-05,
+      "loss": 0.9875,
+      "step": 500
+    },
+    {
+      "epoch": 0.328562134027326,
+      "grad_norm": 1.396112084388733,
+      "learning_rate": 4.9647762902690295e-05,
+      "loss": 0.9765,
+      "step": 505
+    },
+    {
+      "epoch": 0.3318152244632401,
+      "grad_norm": 2.4344260692596436,
+      "learning_rate": 4.963907628035264e-05,
+      "loss": 0.9848,
+      "step": 510
+    },
+    {
+      "epoch": 0.3350683148991542,
+      "grad_norm": 2.1281023025512695,
+      "learning_rate": 4.963028462568154e-05,
+      "loss": 0.9926,
+      "step": 515
+    },
+    {
+      "epoch": 0.3383214053350683,
+      "grad_norm": 1.2993323802947998,
+      "learning_rate": 4.9621387976154396e-05,
+      "loss": 0.9782,
+      "step": 520
+    },
+    {
+      "epoch": 0.3415744957709824,
+      "grad_norm": 1.3118032217025757,
+      "learning_rate": 4.961238636969616e-05,
+      "loss": 0.9885,
+      "step": 525
+    },
+    {
+      "epoch": 0.3448275862068966,
+      "grad_norm": 3.1309878826141357,
+      "learning_rate": 4.960327984467919e-05,
+      "loss": 0.9786,
+      "step": 530
+    },
+    {
+      "epoch": 0.34808067664281067,
+      "grad_norm": 0.9523271918296814,
+      "learning_rate": 4.9594068439923115e-05,
+      "loss": 0.9742,
+      "step": 535
+    },
+    {
+      "epoch": 0.35133376707872477,
+      "grad_norm": 1.010175347328186,
+      "learning_rate": 4.958475219469464e-05,
+      "loss": 0.9569,
+      "step": 540
+    },
+    {
+      "epoch": 0.3545868575146389,
+      "grad_norm": 0.8311709761619568,
+      "learning_rate": 4.9575331148707385e-05,
+      "loss": 0.9638,
+      "step": 545
+    },
+    {
+      "epoch": 0.357839947950553,
+      "grad_norm": 2.4854001998901367,
+      "learning_rate": 4.9565805342121716e-05,
+      "loss": 0.9537,
+      "step": 550
+    },
+    {
+      "epoch": 0.36109303838646717,
+      "grad_norm": 11.345256805419922,
+      "learning_rate": 4.955617481554459e-05,
+      "loss": 0.9738,
+      "step": 555
+    },
+    {
+      "epoch": 0.36434612882238127,
+      "grad_norm": 1.5204081535339355,
+      "learning_rate": 4.954643961002936e-05,
+      "loss": 0.9907,
+      "step": 560
+    },
+    {
+      "epoch": 0.36759921925829536,
+      "grad_norm": 1.0884438753128052,
+      "learning_rate": 4.95365997670756e-05,
+      "loss": 0.9653,
+      "step": 565
+    },
+    {
+      "epoch": 0.3708523096942095,
+      "grad_norm": 0.7718302011489868,
+      "learning_rate": 4.952665532862895e-05,
+      "loss": 0.9573,
+      "step": 570
+    },
+    {
+      "epoch": 0.3741054001301236,
+      "grad_norm": 1.0595511198043823,
+      "learning_rate": 4.9516606337080904e-05,
+      "loss": 0.9535,
+      "step": 575
+    },
+    {
+      "epoch": 0.37735849056603776,
+      "grad_norm": 0.9665298461914062,
+      "learning_rate": 4.950645283526868e-05,
+      "loss": 0.9614,
+      "step": 580
+    },
+    {
+      "epoch": 0.38061158100195186,
+      "grad_norm": 2.5156960487365723,
+      "learning_rate": 4.949619486647497e-05,
+      "loss": 0.9699,
+      "step": 585
+    },
+    {
+      "epoch": 0.38386467143786596,
+      "grad_norm": 0.983572244644165,
+      "learning_rate": 4.948583247442783e-05,
+      "loss": 0.9419,
+      "step": 590
+    },
+    {
+      "epoch": 0.3871177618737801,
+      "grad_norm": 1.5895793437957764,
+      "learning_rate": 4.9475365703300416e-05,
+      "loss": 0.9431,
+      "step": 595
+    },
+    {
+      "epoch": 0.3903708523096942,
+      "grad_norm": 0.7313794493675232,
+      "learning_rate": 4.9464794597710864e-05,
+      "loss": 0.9517,
+      "step": 600
+    },
+    {
+      "epoch": 0.3936239427456083,
+      "grad_norm": 0.8579323887825012,
+      "learning_rate": 4.945411920272205e-05,
+      "loss": 0.9623,
+      "step": 605
+    },
+    {
+      "epoch": 0.39687703318152245,
+      "grad_norm": 1.7319488525390625,
+      "learning_rate": 4.944333956384144e-05,
+      "loss": 0.9652,
+      "step": 610
+    },
+    {
+      "epoch": 0.40013012361743655,
+      "grad_norm": 0.8903234601020813,
+      "learning_rate": 4.943245572702086e-05,
+      "loss": 0.9601,
+      "step": 615
+    },
+    {
+      "epoch": 0.4033832140533507,
+      "grad_norm": 0.8647845387458801,
+      "learning_rate": 4.942146773865631e-05,
+      "loss": 0.9503,
+      "step": 620
+    },
+    {
+      "epoch": 0.4066363044892648,
+      "grad_norm": 1.0848610401153564,
+      "learning_rate": 4.941037564558779e-05,
+      "loss": 0.9482,
+      "step": 625
+    },
+    {
+      "epoch": 0.4098893949251789,
+      "grad_norm": 1.0989733934402466,
+      "learning_rate": 4.939917949509907e-05,
+      "loss": 0.9418,
+      "step": 630
+    },
+    {
+      "epoch": 0.41314248536109305,
+      "grad_norm": 0.8608380556106567,
+      "learning_rate": 4.938787933491749e-05,
+      "loss": 0.943,
+      "step": 635
+    },
+    {
+      "epoch": 0.41639557579700714,
+      "grad_norm": 2.864649772644043,
+      "learning_rate": 4.937647521321378e-05,
+      "loss": 0.948,
+      "step": 640
+    },
+    {
+      "epoch": 0.4196486662329213,
+      "grad_norm": 0.8896744847297668,
+      "learning_rate": 4.936496717860184e-05,
+      "loss": 0.9573,
+      "step": 645
+    },
+    {
+      "epoch": 0.4229017566688354,
+      "grad_norm": 1.1625382900238037,
+      "learning_rate": 4.935335528013853e-05,
+      "loss": 0.9497,
+      "step": 650
+    },
+    {
+      "epoch": 0.4261548471047495,
+      "grad_norm": 0.8717337846755981,
+      "learning_rate": 4.934163956732345e-05,
+      "loss": 0.9442,
+      "step": 655
+    },
+    {
+      "epoch": 0.42940793754066364,
+      "grad_norm": 0.9472817778587341,
+      "learning_rate": 4.932982009009879e-05,
+      "loss": 0.941,
+      "step": 660
+    },
+    {
+      "epoch": 0.43266102797657774,
+      "grad_norm": 1.1837284564971924,
+      "learning_rate": 4.931789689884901e-05,
+      "loss": 0.9521,
+      "step": 665
+    },
+    {
+      "epoch": 0.4359141184124919,
+      "grad_norm": 0.7896617650985718,
+      "learning_rate": 4.9305870044400725e-05,
+      "loss": 0.9407,
+      "step": 670
+    },
+    {
+      "epoch": 0.439167208848406,
+      "grad_norm": 1.0529261827468872,
+      "learning_rate": 4.9293739578022444e-05,
+      "loss": 0.9483,
+      "step": 675
+    },
+    {
+      "epoch": 0.4424202992843201,
+      "grad_norm": 1.018526315689087,
+      "learning_rate": 4.928150555142436e-05,
+      "loss": 0.9474,
+      "step": 680
+    },
+    {
+      "epoch": 0.44567338972023424,
+      "grad_norm": 1.2944765090942383,
+      "learning_rate": 4.926916801675809e-05,
+      "loss": 0.9548,
+      "step": 685
+    },
+    {
+      "epoch": 0.44892648015614833,
+      "grad_norm": 1.1999421119689941,
+      "learning_rate": 4.925672702661653e-05,
+      "loss": 0.9313,
+      "step": 690
+    },
+    {
+      "epoch": 0.4521795705920625,
+      "grad_norm": 1.117092490196228,
+      "learning_rate": 4.92441826340336e-05,
+      "loss": 0.9212,
+      "step": 695
+    },
+    {
+      "epoch": 0.4554326610279766,
+      "grad_norm": 0.976388692855835,
+      "learning_rate": 4.923153489248395e-05,
+      "loss": 0.9258,
+      "step": 700
+    },
+    {
+      "epoch": 0.4586857514638907,
+      "grad_norm": 1.206737756729126,
+      "learning_rate": 4.921878385588284e-05,
+      "loss": 0.897,
+      "step": 705
+    },
+    {
+      "epoch": 0.46193884189980483,
+      "grad_norm": 0.9215301871299744,
+      "learning_rate": 4.920592957858584e-05,
+      "loss": 0.9646,
+      "step": 710
+    },
+    {
+      "epoch": 0.4651919323357189,
+      "grad_norm": 2.4111385345458984,
+      "learning_rate": 4.9192972115388634e-05,
+      "loss": 0.9249,
+      "step": 715
+    },
+    {
+      "epoch": 0.468445022771633,
+      "grad_norm": 1.1773241758346558,
+      "learning_rate": 4.9179911521526734e-05,
+      "loss": 0.9289,
+      "step": 720
+    },
+    {
+      "epoch": 0.4716981132075472,
+      "grad_norm": 1.2238733768463135,
+      "learning_rate": 4.9166747852675325e-05,
+      "loss": 0.9672,
+      "step": 725
+    },
+    {
+      "epoch": 0.4749512036434613,
+      "grad_norm": 0.8060822486877441,
+      "learning_rate": 4.9153481164948964e-05,
+      "loss": 0.9571,
+      "step": 730
+    },
+    {
+      "epoch": 0.4782042940793754,
+      "grad_norm": 1.0834110975265503,
+      "learning_rate": 4.914011151490135e-05,
+      "loss": 0.925,
+      "step": 735
+    },
+    {
+      "epoch": 0.4814573845152895,
+      "grad_norm": 2.0932958126068115,
+      "learning_rate": 4.912663895952511e-05,
+      "loss": 0.9022,
+      "step": 740
+    },
+    {
+      "epoch": 0.4847104749512036,
+      "grad_norm": 0.9362598061561584,
+      "learning_rate": 4.911306355625154e-05,
+      "loss": 0.9569,
+      "step": 745
+    },
+    {
+      "epoch": 0.48796356538711777,
+      "grad_norm": 0.8350614905357361,
+      "learning_rate": 4.909938536295034e-05,
+      "loss": 0.9283,
+      "step": 750
+    },
+    {
+      "epoch": 0.49121665582303187,
+      "grad_norm": 0.7186315059661865,
+      "learning_rate": 4.908560443792941e-05,
+      "loss": 0.916,
+      "step": 755
+    },
+    {
+      "epoch": 0.494469746258946,
+      "grad_norm": 0.8941408395767212,
+      "learning_rate": 4.907172083993457e-05,
+      "loss": 0.9132,
+      "step": 760
+    },
+    {
+      "epoch": 0.4977228366948601,
+      "grad_norm": 0.6881601214408875,
+      "learning_rate": 4.9057734628149296e-05,
+      "loss": 0.9401,
+      "step": 765
+    },
+    {
+      "epoch": 0.5009759271307742,
+      "grad_norm": 0.7907692790031433,
+      "learning_rate": 4.904364586219454e-05,
+      "loss": 0.9453,
+      "step": 770
+    },
+    {
+      "epoch": 0.5042290175666884,
+      "grad_norm": 0.6884419322013855,
+      "learning_rate": 4.902945460212839e-05,
+      "loss": 0.9213,
+      "step": 775
+    },
+    {
+      "epoch": 0.5074821080026025,
+      "grad_norm": 0.9783698916435242,
+      "learning_rate": 4.9015160908445846e-05,
+      "loss": 0.9324,
+      "step": 780
+    },
+    {
+      "epoch": 0.5107351984385166,
+      "grad_norm": 0.670250415802002,
+      "learning_rate": 4.900076484207857e-05,
+      "loss": 0.9186,
+      "step": 785
+    },
+    {
+      "epoch": 0.5139882888744307,
+      "grad_norm": 0.6589803695678711,
+      "learning_rate": 4.8986266464394645e-05,
+      "loss": 0.9245,
+      "step": 790
+    },
+    {
+      "epoch": 0.5172413793103449,
+      "grad_norm": 0.8181180953979492,
+      "learning_rate": 4.8971665837198266e-05,
+      "loss": 0.9117,
+      "step": 795
+    },
+    {
+      "epoch": 0.5204944697462589,
+      "grad_norm": 0.9122630953788757,
+      "learning_rate": 4.89569630227295e-05,
+      "loss": 0.916,
+      "step": 800
+    },
+    {
+      "epoch": 0.523747560182173,
+      "grad_norm": 1.7749128341674805,
+      "learning_rate": 4.894215808366404e-05,
+      "loss": 0.927,
+      "step": 805
+    },
+    {
+      "epoch": 0.5270006506180872,
+      "grad_norm": 0.708722710609436,
+      "learning_rate": 4.892725108311289e-05,
+      "loss": 0.9378,
+      "step": 810
+    },
+    {
+      "epoch": 0.5302537410540012,
+      "grad_norm": 0.7337002754211426,
+      "learning_rate": 4.891224208462217e-05,
+      "loss": 0.9042,
+      "step": 815
+    },
+    {
+      "epoch": 0.5335068314899154,
+      "grad_norm": 0.6495442986488342,
+      "learning_rate": 4.889713115217276e-05,
+      "loss": 0.9271,
+      "step": 820
+    },
+    {
+      "epoch": 0.5367599219258296,
+      "grad_norm": 0.6206863522529602,
+      "learning_rate": 4.8881918350180076e-05,
+      "loss": 0.9292,
+      "step": 825
+    },
+    {
+      "epoch": 0.5400130123617437,
+      "grad_norm": 1.1107088327407837,
+      "learning_rate": 4.886660374349381e-05,
+      "loss": 0.9002,
+      "step": 830
+    },
+    {
+      "epoch": 0.5432661027976577,
+      "grad_norm": 0.7245462536811829,
+      "learning_rate": 4.885118739739759e-05,
+      "loss": 0.898,
+      "step": 835
+    },
+    {
+      "epoch": 0.5465191932335719,
+      "grad_norm": 0.7149853110313416,
+      "learning_rate": 4.883566937760879e-05,
+      "loss": 0.9105,
+      "step": 840
+    },
+    {
+      "epoch": 0.549772283669486,
+      "grad_norm": 0.8408038020133972,
+      "learning_rate": 4.882004975027816e-05,
+      "loss": 0.9059,
+      "step": 845
+    },
+    {
+      "epoch": 0.5530253741054001,
+      "grad_norm": 0.8386579155921936,
+      "learning_rate": 4.880432858198962e-05,
+      "loss": 0.9148,
+      "step": 850
+    },
+    {
+      "epoch": 0.5562784645413142,
+      "grad_norm": 0.6639278531074524,
+      "learning_rate": 4.878850593975992e-05,
+      "loss": 0.8995,
+      "step": 855
+    },
+    {
+      "epoch": 0.5595315549772284,
+      "grad_norm": 0.8974155187606812,
+      "learning_rate": 4.8772581891038385e-05,
+      "loss": 0.9406,
+      "step": 860
+    },
+    {
+      "epoch": 0.5627846454131424,
+      "grad_norm": 1.0255565643310547,
+      "learning_rate": 4.875655650370662e-05,
+      "loss": 0.9067,
+      "step": 865
+    },
+    {
+      "epoch": 0.5660377358490566,
+      "grad_norm": 1.0870089530944824,
+      "learning_rate": 4.87404298460782e-05,
+      "loss": 0.9239,
+      "step": 870
+    },
+    {
+      "epoch": 0.5692908262849707,
+      "grad_norm": 0.7773581743240356,
+      "learning_rate": 4.872420198689845e-05,
+      "loss": 0.9284,
+      "step": 875
+    },
+    {
+      "epoch": 0.5725439167208849,
+      "grad_norm": 0.6731524467468262,
+      "learning_rate": 4.870787299534404e-05,
+      "loss": 0.9122,
+      "step": 880
+    },
+    {
+      "epoch": 0.5757970071567989,
+      "grad_norm": 0.8352004885673523,
+      "learning_rate": 4.869144294102279e-05,
+      "loss": 0.906,
+      "step": 885
+    },
+    {
+      "epoch": 0.5790500975927131,
+      "grad_norm": 0.7254254817962646,
+      "learning_rate": 4.8674911893973305e-05,
+      "loss": 0.899,
+      "step": 890
+    },
+    {
+      "epoch": 0.5823031880286272,
+      "grad_norm": 0.6976432204246521,
+      "learning_rate": 4.8658279924664754e-05,
+      "loss": 0.9104,
+      "step": 895
+    },
+    {
+      "epoch": 0.5855562784645413,
+      "grad_norm": 0.7411409616470337,
+      "learning_rate": 4.8641547103996456e-05,
+      "loss": 0.9216,
+      "step": 900
+    },
+    {
+      "epoch": 0.5888093689004554,
+      "grad_norm": 0.8754427433013916,
+      "learning_rate": 4.862471350329769e-05,
+      "loss": 0.9156,
+      "step": 905
+    },
+    {
+      "epoch": 0.5920624593363696,
+      "grad_norm": 0.7311625480651855,
+      "learning_rate": 4.8607779194327344e-05,
+      "loss": 0.9125,
+      "step": 910
+    },
+    {
+      "epoch": 0.5953155497722836,
+      "grad_norm": 0.7393909096717834,
+      "learning_rate": 4.8590744249273566e-05,
+      "loss": 0.9011,
+      "step": 915
+    },
+    {
+      "epoch": 0.5985686402081978,
+      "grad_norm": 0.7025457620620728,
+      "learning_rate": 4.857360874075355e-05,
+      "loss": 0.9198,
+      "step": 920
+    },
+    {
+      "epoch": 0.6018217306441119,
+      "grad_norm": 0.7867306470870972,
+      "learning_rate": 4.855637274181314e-05,
+      "loss": 0.8803,
+      "step": 925
+    },
+    {
+      "epoch": 0.605074821080026,
+      "grad_norm": 0.6452615857124329,
+      "learning_rate": 4.853903632592657e-05,
+      "loss": 0.9104,
+      "step": 930
+    },
+    {
+      "epoch": 0.6083279115159401,
+      "grad_norm": 0.8884091377258301,
+      "learning_rate": 4.852159956699614e-05,
+      "loss": 0.9119,
+      "step": 935
+    },
+    {
+      "epoch": 0.6115810019518543,
+      "grad_norm": 0.5305378437042236,
+      "learning_rate": 4.850406253935188e-05,
+      "loss": 0.9296,
+      "step": 940
+    },
+    {
+      "epoch": 0.6148340923877684,
+      "grad_norm": 3.5500142574310303,
+      "learning_rate": 4.848642531775126e-05,
+      "loss": 0.8996,
+      "step": 945
+    },
+    {
+      "epoch": 0.6180871828236825,
+      "grad_norm": 0.7085503339767456,
+      "learning_rate": 4.846868797737886e-05,
+      "loss": 0.9045,
+      "step": 950
+    },
+    {
+      "epoch": 0.6213402732595966,
+      "grad_norm": 0.9540202021598816,
+      "learning_rate": 4.8450850593846035e-05,
+      "loss": 0.9041,
+      "step": 955
+    },
+    {
+      "epoch": 0.6245933636955108,
+      "grad_norm": 0.7558302879333496,
+      "learning_rate": 4.843291324319064e-05,
+      "loss": 0.8998,
+      "step": 960
+    },
+    {
+      "epoch": 0.6278464541314248,
+      "grad_norm": 1.2235993146896362,
+      "learning_rate": 4.8414876001876636e-05,
+      "loss": 0.8977,
+      "step": 965
+    },
+    {
+      "epoch": 0.631099544567339,
+      "grad_norm": 0.9432989954948425,
+      "learning_rate": 4.839673894679383e-05,
+      "loss": 0.9013,
+      "step": 970
+    },
+    {
+      "epoch": 0.6343526350032531,
+      "grad_norm": 0.8516787886619568,
+      "learning_rate": 4.83785021552575e-05,
+      "loss": 0.9195,
+      "step": 975
+    },
+    {
+      "epoch": 0.6376057254391672,
+      "grad_norm": 0.9901612401008606,
+      "learning_rate": 4.836016570500809e-05,
+      "loss": 0.8917,
+      "step": 980
+    },
+    {
+      "epoch": 0.6408588158750813,
+      "grad_norm": 0.7319976687431335,
+      "learning_rate": 4.834172967421088e-05,
+      "loss": 0.8961,
+      "step": 985
+    },
+    {
+      "epoch": 0.6441119063109955,
+      "grad_norm": 0.8301442265510559,
+      "learning_rate": 4.832319414145565e-05,
+      "loss": 0.9025,
+      "step": 990
+    },
+    {
+      "epoch": 0.6473649967469096,
+      "grad_norm": 0.7469923496246338,
+      "learning_rate": 4.8304559185756303e-05,
+      "loss": 0.8908,
+      "step": 995
+    },
+    {
+      "epoch": 0.6506180871828237,
+      "grad_norm": 0.7159551382064819,
+      "learning_rate": 4.828582488655062e-05,
+      "loss": 0.8948,
+      "step": 1000
+    },
+    {
+      "epoch": 0.6538711776187378,
+      "grad_norm": 1.0059654712677002,
+      "learning_rate": 4.826699132369983e-05,
+      "loss": 0.9101,
+      "step": 1005
+    },
+    {
+      "epoch": 0.657124268054652,
+      "grad_norm": 0.9471872448921204,
+      "learning_rate": 4.824805857748831e-05,
+      "loss": 0.9172,
+      "step": 1010
+    },
+    {
+      "epoch": 0.660377358490566,
+      "grad_norm": 0.7876858711242676,
+      "learning_rate": 4.822902672862325e-05,
+      "loss": 0.8933,
+      "step": 1015
+    },
+    {
+      "epoch": 0.6636304489264802,
+      "grad_norm": 0.7322363257408142,
+      "learning_rate": 4.82098958582343e-05,
+      "loss": 0.897,
+      "step": 1020
+    },
+    {
+      "epoch": 0.6668835393623943,
+      "grad_norm": 0.7154924273490906,
+      "learning_rate": 4.819066604787321e-05,
+      "loss": 0.8931,
+      "step": 1025
+    },
+    {
+      "epoch": 0.6701366297983083,
+      "grad_norm": 0.759896457195282,
+      "learning_rate": 4.817133737951352e-05,
+      "loss": 0.894,
+      "step": 1030
+    },
+    {
+      "epoch": 0.6733897202342225,
+      "grad_norm": 0.9226410984992981,
+      "learning_rate": 4.815190993555013e-05,
+      "loss": 0.884,
+      "step": 1035
+    },
+    {
+      "epoch": 0.6766428106701367,
+      "grad_norm": 0.7603817582130432,
+      "learning_rate": 4.8132383798799077e-05,
+      "loss": 0.901,
+      "step": 1040
+    },
+    {
+      "epoch": 0.6798959011060507,
+      "grad_norm": 0.5785139203071594,
+      "learning_rate": 4.811275905249705e-05,
+      "loss": 0.9105,
+      "step": 1045
+    },
+    {
+      "epoch": 0.6831489915419648,
+      "grad_norm": 0.9068583846092224,
+      "learning_rate": 4.8093035780301135e-05,
+      "loss": 0.8941,
+      "step": 1050
+    },
+    {
+      "epoch": 0.686402081977879,
+      "grad_norm": 0.6756225228309631,
+      "learning_rate": 4.807321406628838e-05,
+      "loss": 0.9318,
+      "step": 1055
+    },
+    {
+      "epoch": 0.6896551724137931,
+      "grad_norm": 0.5247710943222046,
+      "learning_rate": 4.805329399495552e-05,
+      "loss": 0.8878,
+      "step": 1060
+    },
+    {
+      "epoch": 0.6929082628497072,
+      "grad_norm": 0.7933164238929749,
+      "learning_rate": 4.8033275651218525e-05,
+      "loss": 0.8926,
+      "step": 1065
+    },
+    {
+      "epoch": 0.6961613532856213,
+      "grad_norm": 0.7385444641113281,
+      "learning_rate": 4.8013159120412324e-05,
+      "loss": 0.9179,
+      "step": 1070
+    },
+    {
+      "epoch": 0.6994144437215355,
+      "grad_norm": 0.5647626519203186,
+      "learning_rate": 4.7992944488290357e-05,
+      "loss": 0.8982,
+      "step": 1075
+    },
+    {
+      "epoch": 0.7026675341574495,
+      "grad_norm": 0.7482002973556519,
+      "learning_rate": 4.79726318410243e-05,
+      "loss": 0.8903,
+      "step": 1080
+    },
+    {
+      "epoch": 0.7059206245933637,
+      "grad_norm": 0.7001275420188904,
+      "learning_rate": 4.7952221265203626e-05,
+      "loss": 0.9044,
+      "step": 1085
+    },
+    {
+      "epoch": 0.7091737150292778,
+      "grad_norm": 0.7466359734535217,
+      "learning_rate": 4.793171284783525e-05,
+      "loss": 0.8829,
+      "step": 1090
+    },
+    {
+      "epoch": 0.7124268054651919,
+      "grad_norm": 0.938048779964447,
+      "learning_rate": 4.791110667634321e-05,
+      "loss": 0.9097,
+      "step": 1095
+    },
+    {
+      "epoch": 0.715679895901106,
+      "grad_norm": 1.033058762550354,
+      "learning_rate": 4.789040283856822e-05,
+      "loss": 0.8829,
+      "step": 1100
+    },
+    {
+      "epoch": 0.7189329863370202,
+      "grad_norm": 0.7184875011444092,
+      "learning_rate": 4.7869601422767326e-05,
+      "loss": 0.9007,
+      "step": 1105
+    },
+    {
+      "epoch": 0.7221860767729343,
+      "grad_norm": 0.6718536615371704,
+      "learning_rate": 4.784870251761357e-05,
+      "loss": 0.8909,
+      "step": 1110
+    },
+    {
+      "epoch": 0.7254391672088484,
+      "grad_norm": 0.854129433631897,
+      "learning_rate": 4.782770621219552e-05,
+      "loss": 0.9017,
+      "step": 1115
+    },
+    {
+      "epoch": 0.7286922576447625,
+      "grad_norm": 0.6812130212783813,
+      "learning_rate": 4.7806612596017e-05,
+      "loss": 0.8995,
+      "step": 1120
+    },
+    {
+      "epoch": 0.7319453480806767,
+      "grad_norm": 0.6341739892959595,
+      "learning_rate": 4.778542175899662e-05,
+      "loss": 0.865,
+      "step": 1125
+    },
+    {
+      "epoch": 0.7351984385165907,
+      "grad_norm": 0.7255128026008606,
+      "learning_rate": 4.776413379146743e-05,
+      "loss": 0.8991,
+      "step": 1130
+    },
+    {
+      "epoch": 0.7384515289525049,
+      "grad_norm": 0.6501480340957642,
+      "learning_rate": 4.7742748784176554e-05,
+      "loss": 0.896,
+      "step": 1135
+    },
+    {
+      "epoch": 0.741704619388419,
+      "grad_norm": 1.5047345161437988,
+      "learning_rate": 4.7721266828284754e-05,
+      "loss": 0.9003,
+      "step": 1140
+    },
+    {
+      "epoch": 0.7449577098243331,
+      "grad_norm": 0.9915900230407715,
+      "learning_rate": 4.769968801536608e-05,
+      "loss": 0.887,
+      "step": 1145
+    },
+    {
+      "epoch": 0.7482108002602472,
+      "grad_norm": 1.0044691562652588,
+      "learning_rate": 4.767801243740746e-05,
+      "loss": 0.8908,
+      "step": 1150
+    },
+    {
+      "epoch": 0.7514638906961614,
+      "grad_norm": 0.5450899004936218,
+      "learning_rate": 4.765624018680833e-05,
+      "loss": 0.9114,
+      "step": 1155
+    },
+    {
+      "epoch": 0.7547169811320755,
+      "grad_norm": 0.6787106394767761,
+      "learning_rate": 4.763437135638021e-05,
+      "loss": 0.9027,
+      "step": 1160
+    },
+    {
+      "epoch": 0.7579700715679896,
+      "grad_norm": 0.8962056636810303,
+      "learning_rate": 4.761240603934633e-05,
+      "loss": 0.9099,
+      "step": 1165
+    },
+    {
+      "epoch": 0.7612231620039037,
+      "grad_norm": 0.7145211100578308,
+      "learning_rate": 4.759034432934123e-05,
+      "loss": 0.909,
+      "step": 1170
+    },
+    {
+      "epoch": 0.7644762524398179,
+      "grad_norm": 0.5293973684310913,
+      "learning_rate": 4.7568186320410356e-05,
+      "loss": 0.8734,
+      "step": 1175
+    },
+    {
+      "epoch": 0.7677293428757319,
+      "grad_norm": 0.741165041923523,
+      "learning_rate": 4.754593210700966e-05,
+      "loss": 0.866,
+      "step": 1180
+    },
+    {
+      "epoch": 0.7709824333116461,
+      "grad_norm": 0.7917879819869995,
+      "learning_rate": 4.7523581784005187e-05,
+      "loss": 0.8871,
+      "step": 1185
+    },
+    {
+      "epoch": 0.7742355237475602,
+      "grad_norm": 0.7449229955673218,
+      "learning_rate": 4.750113544667271e-05,
+      "loss": 0.8966,
+      "step": 1190
+    },
+    {
+      "epoch": 0.7774886141834743,
+      "grad_norm": 0.6702096462249756,
+      "learning_rate": 4.7478593190697254e-05,
+      "loss": 0.8784,
+      "step": 1195
+    },
+    {
+      "epoch": 0.7807417046193884,
+      "grad_norm": 0.6181638240814209,
+      "learning_rate": 4.745595511217277e-05,
+      "loss": 0.9148,
+      "step": 1200
+    },
+    {
+      "epoch": 0.7839947950553026,
+      "grad_norm": 0.6590509414672852,
+      "learning_rate": 4.743322130760166e-05,
+      "loss": 0.8738,
+      "step": 1205
+    },
+    {
+      "epoch": 0.7872478854912166,
+      "grad_norm": 0.6860212087631226,
+      "learning_rate": 4.7410391873894386e-05,
+      "loss": 0.8986,
+      "step": 1210
+    },
+    {
+      "epoch": 0.7905009759271308,
+      "grad_norm": 0.8018529415130615,
+      "learning_rate": 4.73874669083691e-05,
+      "loss": 0.8894,
+      "step": 1215
+    },
+    {
+      "epoch": 0.7937540663630449,
+      "grad_norm": 0.49539148807525635,
+      "learning_rate": 4.736444650875114e-05,
+      "loss": 0.8812,
+      "step": 1220
+    },
+    {
+      "epoch": 0.7970071567989591,
+      "grad_norm": 0.5640125870704651,
+      "learning_rate": 4.7341330773172686e-05,
+      "loss": 0.8681,
+      "step": 1225
+    },
+    {
+      "epoch": 0.8002602472348731,
+      "grad_norm": 0.7945250272750854,
+      "learning_rate": 4.731811980017234e-05,
+      "loss": 0.8608,
+      "step": 1230
+    },
+    {
+      "epoch": 0.8035133376707873,
+      "grad_norm": 1.0091153383255005,
+      "learning_rate": 4.729481368869465e-05,
+      "loss": 0.8853,
+      "step": 1235
+    },
+    {
+      "epoch": 0.8067664281067014,
+      "grad_norm": 0.6991880536079407,
+      "learning_rate": 4.727141253808974e-05,
+      "loss": 0.8783,
+      "step": 1240
+    },
+    {
+      "epoch": 0.8100195185426154,
+      "grad_norm": 1.1925946474075317,
+      "learning_rate": 4.724791644811287e-05,
+      "loss": 0.8685,
+      "step": 1245
+    },
+    {
+      "epoch": 0.8132726089785296,
+      "grad_norm": 0.5790075659751892,
+      "learning_rate": 4.722432551892402e-05,
+      "loss": 0.8715,
+      "step": 1250
+    },
+    {
+      "epoch": 0.8165256994144438,
+      "grad_norm": 0.6202079653739929,
+      "learning_rate": 4.720063985108743e-05,
+      "loss": 0.8673,
+      "step": 1255
+    },
+    {
+      "epoch": 0.8197787898503578,
+      "grad_norm": 0.6250977516174316,
+      "learning_rate": 4.717685954557123e-05,
+      "loss": 0.8636,
+      "step": 1260
+    },
+    {
+      "epoch": 0.8230318802862719,
+      "grad_norm": 0.5988998413085938,
+      "learning_rate": 4.715298470374694e-05,
+      "loss": 0.8929,
+      "step": 1265
+    },
+    {
+      "epoch": 0.8262849707221861,
+      "grad_norm": 0.6314589977264404,
+      "learning_rate": 4.712901542738908e-05,
+      "loss": 0.8759,
+      "step": 1270
+    },
+    {
+      "epoch": 0.8295380611581002,
+      "grad_norm": 2.416761875152588,
+      "learning_rate": 4.7104951818674755e-05,
+      "loss": 0.8854,
+      "step": 1275
+    },
+    {
+      "epoch": 0.8327911515940143,
+      "grad_norm": 0.6815042495727539,
+      "learning_rate": 4.7080793980183165e-05,
+      "loss": 0.8801,
+      "step": 1280
+    },
+    {
+      "epoch": 0.8360442420299284,
+      "grad_norm": 0.763461709022522,
+      "learning_rate": 4.7056542014895204e-05,
+      "loss": 0.8805,
+      "step": 1285
+    },
+    {
+      "epoch": 0.8392973324658426,
+      "grad_norm": 0.6846510767936707,
+      "learning_rate": 4.703219602619302e-05,
+      "loss": 0.8847,
+      "step": 1290
+    },
+    {
+      "epoch": 0.8425504229017566,
+      "grad_norm": 0.7311460375785828,
+      "learning_rate": 4.7007756117859566e-05,
+      "loss": 0.8802,
+      "step": 1295
+    },
+    {
+      "epoch": 0.8458035133376708,
+      "grad_norm": 0.6256663203239441,
+      "learning_rate": 4.698322239407814e-05,
+      "loss": 0.872,
+      "step": 1300
+    },
+    {
+      "epoch": 0.8490566037735849,
+      "grad_norm": 0.7773202061653137,
+      "learning_rate": 4.695859495943199e-05,
+      "loss": 0.8902,
+      "step": 1305
+    },
+    {
+      "epoch": 0.852309694209499,
+      "grad_norm": 0.6700133085250854,
+      "learning_rate": 4.6933873918903816e-05,
+      "loss": 0.8713,
+      "step": 1310
+    },
+    {
+      "epoch": 0.8555627846454131,
+      "grad_norm": 1.7503775358200073,
+      "learning_rate": 4.690905937787536e-05,
+      "loss": 0.8763,
+      "step": 1315
+    },
+    {
+      "epoch": 0.8588158750813273,
+      "grad_norm": 0.7265576720237732,
+      "learning_rate": 4.688415144212692e-05,
+      "loss": 0.8808,
+      "step": 1320
+    },
+    {
+      "epoch": 0.8620689655172413,
+      "grad_norm": 0.9674301147460938,
+      "learning_rate": 4.685915021783694e-05,
+      "loss": 0.8953,
+      "step": 1325
+    },
+    {
+      "epoch": 0.8653220559531555,
+      "grad_norm": 0.7183430194854736,
+      "learning_rate": 4.683405581158153e-05,
+      "loss": 0.8679,
+      "step": 1330
+    },
+    {
+      "epoch": 0.8685751463890696,
+      "grad_norm": 0.6251899003982544,
+      "learning_rate": 4.6808868330334024e-05,
+      "loss": 0.9087,
+      "step": 1335
+    },
+    {
+      "epoch": 0.8718282368249838,
+      "grad_norm": 0.6841866970062256,
+      "learning_rate": 4.67835878814645e-05,
+      "loss": 0.8935,
+      "step": 1340
+    },
+    {
+      "epoch": 0.8750813272608978,
+      "grad_norm": 0.7957316637039185,
+      "learning_rate": 4.675821457273938e-05,
+      "loss": 0.8677,
+      "step": 1345
+    },
+    {
+      "epoch": 0.878334417696812,
+      "grad_norm": 0.6032708287239075,
+      "learning_rate": 4.67327485123209e-05,
+      "loss": 0.8975,
+      "step": 1350
+    },
+    {
+      "epoch": 0.8815875081327261,
+      "grad_norm": 0.7431797981262207,
+      "learning_rate": 4.6707189808766684e-05,
+      "loss": 0.8919,
+      "step": 1355
+    },
+    {
+      "epoch": 0.8848405985686402,
+      "grad_norm": 0.6148476004600525,
+      "learning_rate": 4.6681538571029295e-05,
+      "loss": 0.8548,
+      "step": 1360
+    },
+    {
+      "epoch": 0.8880936890045543,
+      "grad_norm": 0.6982467770576477,
+      "learning_rate": 4.665579490845574e-05,
+      "loss": 0.8858,
+      "step": 1365
+    },
+    {
+      "epoch": 0.8913467794404685,
+      "grad_norm": 0.6519801616668701,
+      "learning_rate": 4.662995893078702e-05,
+      "loss": 0.8957,
+      "step": 1370
+    },
+    {
+      "epoch": 0.8945998698763825,
+      "grad_norm": 0.6641463041305542,
+      "learning_rate": 4.660403074815767e-05,
+      "loss": 0.86,
+      "step": 1375
+    },
+    {
+      "epoch": 0.8978529603122967,
+      "grad_norm": 0.7551462650299072,
+      "learning_rate": 4.657801047109527e-05,
+      "loss": 0.8709,
+      "step": 1380
+    },
+    {
+      "epoch": 0.9011060507482108,
+      "grad_norm": 0.7857052087783813,
+      "learning_rate": 4.655189821051998e-05,
+      "loss": 0.8539,
+      "step": 1385
+    },
+    {
+      "epoch": 0.904359141184125,
+      "grad_norm": 1.1861300468444824,
+      "learning_rate": 4.6525694077744076e-05,
+      "loss": 0.8855,
+      "step": 1390
+    },
+    {
+      "epoch": 0.907612231620039,
+      "grad_norm": 0.8817470073699951,
+      "learning_rate": 4.6499398184471476e-05,
+      "loss": 0.8734,
+      "step": 1395
+    },
+    {
+      "epoch": 0.9108653220559532,
+      "grad_norm": 0.6167863011360168,
+      "learning_rate": 4.647301064279725e-05,
+      "loss": 0.8765,
+      "step": 1400
+    },
+    {
+      "epoch": 0.9141184124918673,
+      "grad_norm": 0.8413434624671936,
+      "learning_rate": 4.644653156520715e-05,
+      "loss": 0.8889,
+      "step": 1405
+    },
+    {
+      "epoch": 0.9173715029277814,
+      "grad_norm": 0.5738908052444458,
+      "learning_rate": 4.6419961064577134e-05,
+      "loss": 0.8479,
+      "step": 1410
+    },
+    {
+      "epoch": 0.9206245933636955,
+      "grad_norm": 0.9078507423400879,
+      "learning_rate": 4.6393299254172875e-05,
+      "loss": 0.881,
+      "step": 1415
+    },
+    {
+      "epoch": 0.9238776837996097,
+      "grad_norm": 1.190238356590271,
+      "learning_rate": 4.63665462476493e-05,
+      "loss": 0.8692,
+      "step": 1420
+    },
+    {
+      "epoch": 0.9271307742355237,
+      "grad_norm": 0.5501294136047363,
+      "learning_rate": 4.633970215905007e-05,
+      "loss": 0.8792,
+      "step": 1425
+    },
+    {
+      "epoch": 0.9303838646714379,
+      "grad_norm": 0.6713528633117676,
+      "learning_rate": 4.631276710280713e-05,
+      "loss": 0.861,
+      "step": 1430
+    },
+    {
+      "epoch": 0.933636955107352,
+      "grad_norm": 0.600857675075531,
+      "learning_rate": 4.6285741193740194e-05,
+      "loss": 0.8657,
+      "step": 1435
+    },
+    {
+      "epoch": 0.936890045543266,
+      "grad_norm": 1.3047159910202026,
+      "learning_rate": 4.625862454705629e-05,
+      "loss": 0.8716,
+      "step": 1440
+    },
+    {
+      "epoch": 0.9401431359791802,
+      "grad_norm": 0.7485547065734863,
+      "learning_rate": 4.623141727834919e-05,
+      "loss": 0.8742,
+      "step": 1445
+    },
+    {
+      "epoch": 0.9433962264150944,
+      "grad_norm": 0.7072353959083557,
+      "learning_rate": 4.620411950359903e-05,
+      "loss": 0.8659,
+      "step": 1450
+    },
+    {
+      "epoch": 0.9466493168510085,
+      "grad_norm": 0.5867493748664856,
+      "learning_rate": 4.617673133917175e-05,
+      "loss": 0.8864,
+      "step": 1455
+    },
+    {
+      "epoch": 0.9499024072869225,
+      "grad_norm": 0.6515786647796631,
+      "learning_rate": 4.614925290181858e-05,
+      "loss": 0.8841,
+      "step": 1460
+    },
+    {
+      "epoch": 0.9531554977228367,
+      "grad_norm": 0.7220116853713989,
+      "learning_rate": 4.612168430867559e-05,
+      "loss": 0.88,
+      "step": 1465
+    },
+    {
+      "epoch": 0.9564085881587508,
+      "grad_norm": 0.5353178381919861,
+      "learning_rate": 4.6094025677263155e-05,
+      "loss": 0.8578,
+      "step": 1470
+    },
+    {
+      "epoch": 0.9596616785946649,
+      "grad_norm": 1.0675499439239502,
+      "learning_rate": 4.606627712548548e-05,
+      "loss": 0.8705,
+      "step": 1475
+    },
+    {
+      "epoch": 0.962914769030579,
+      "grad_norm": 0.6946088671684265,
+      "learning_rate": 4.6038438771630074e-05,
+      "loss": 0.8707,
+      "step": 1480
+    },
+    {
+      "epoch": 0.9661678594664932,
+      "grad_norm": 0.6132957339286804,
+      "learning_rate": 4.601051073436728e-05,
+      "loss": 0.872,
+      "step": 1485
+    },
+    {
+      "epoch": 0.9694209499024072,
+      "grad_norm": 2.741361379623413,
+      "learning_rate": 4.5982493132749724e-05,
+      "loss": 0.8711,
+      "step": 1490
+    },
+    {
+      "epoch": 0.9726740403383214,
+      "grad_norm": 0.6481953859329224,
+      "learning_rate": 4.595438608621183e-05,
+      "loss": 0.8804,
+      "step": 1495
+    },
+    {
+      "epoch": 0.9759271307742355,
+      "grad_norm": 0.8871548771858215,
+      "learning_rate": 4.592618971456933e-05,
+      "loss": 0.863,
+      "step": 1500
+    },
+    {
+      "epoch": 0.9791802212101497,
+      "grad_norm": 1.2673571109771729,
+      "learning_rate": 4.5897904138018724e-05,
+      "loss": 0.8781,
+      "step": 1505
+    },
+    {
+      "epoch": 0.9824333116460637,
+      "grad_norm": 0.5219647288322449,
+      "learning_rate": 4.586952947713677e-05,
+      "loss": 0.8738,
+      "step": 1510
+    },
+    {
+      "epoch": 0.9856864020819779,
+      "grad_norm": 0.7620292901992798,
+      "learning_rate": 4.584106585287998e-05,
+      "loss": 0.8602,
+      "step": 1515
+    },
+    {
+      "epoch": 0.988939492517892,
+      "grad_norm": 5.0167717933654785,
+      "learning_rate": 4.581251338658412e-05,
+      "loss": 0.879,
+      "step": 1520
+    },
+    {
+      "epoch": 0.9921925829538061,
+      "grad_norm": 0.6157656311988831,
+      "learning_rate": 4.578387219996366e-05,
+      "loss": 0.8645,
+      "step": 1525
+    },
+    {
+      "epoch": 0.9954456733897202,
+      "grad_norm": 0.6330501437187195,
+      "learning_rate": 4.5755142415111264e-05,
+      "loss": 0.8549,
+      "step": 1530
+    },
+    {
+      "epoch": 0.9986987638256344,
+      "grad_norm": 0.7185651063919067,
+      "learning_rate": 4.572632415449729e-05,
+      "loss": 0.8799,
+      "step": 1535
+    },
+    {
+      "epoch": 1.0,
+      "eval_f1": 0.8050111210499576,
+      "eval_loss": 0.432861328125,
+      "eval_precision": 0.8078559249569004,
+      "eval_recall": 0.8036506251146921,
+      "eval_runtime": 475.6379,
+      "eval_samples_per_second": 827.173,
+      "eval_steps_per_second": 0.809,
+      "step": 1537
+    },
+    {
+      "epoch": 1.0019518542615484,
+      "grad_norm": 1.2476260662078857,
+      "learning_rate": 4.5697417540969234e-05,
+      "loss": 0.8628,
+      "step": 1540
+    },
+    {
+      "epoch": 1.0052049446974627,
+      "grad_norm": 1.6661646366119385,
+      "learning_rate": 4.566842269775126e-05,
+      "loss": 0.8106,
+      "step": 1545
+    },
+    {
+      "epoch": 1.0084580351333767,
+      "grad_norm": 7.054599285125732,
+      "learning_rate": 4.563933974844361e-05,
+      "loss": 0.7696,
+      "step": 1550
+    },
+    {
+      "epoch": 1.0117111255692908,
+      "grad_norm": 0.7829424142837524,
+      "learning_rate": 4.561016881702212e-05,
+      "loss": 0.8057,
+      "step": 1555
+    },
+    {
+      "epoch": 1.014964216005205,
+      "grad_norm": 0.5904113054275513,
+      "learning_rate": 4.5580910027837673e-05,
+      "loss": 0.8178,
+      "step": 1560
+    },
+    {
+      "epoch": 1.018217306441119,
+      "grad_norm": 1.8633893728256226,
+      "learning_rate": 4.555156350561569e-05,
+      "loss": 0.8021,
+      "step": 1565
+    },
+    {
+      "epoch": 1.0214703968770331,
+      "grad_norm": 0.935964047908783,
+      "learning_rate": 4.5522129375455555e-05,
+      "loss": 0.7791,
+      "step": 1570
+    },
+    {
+      "epoch": 1.0247234873129474,
+      "grad_norm": 1.3689883947372437,
+      "learning_rate": 4.5492607762830145e-05,
+      "loss": 0.814,
+      "step": 1575
+    },
+    {
+      "epoch": 1.0279765777488614,
+      "grad_norm": 0.8765047788619995,
+      "learning_rate": 4.546299879358523e-05,
+      "loss": 0.8149,
+      "step": 1580
+    },
+    {
+      "epoch": 1.0312296681847755,
+      "grad_norm": 1.1618647575378418,
+      "learning_rate": 4.5433302593939e-05,
+      "loss": 0.7935,
+      "step": 1585
+    },
+    {
+      "epoch": 1.0344827586206897,
+      "grad_norm": 0.7140945196151733,
+      "learning_rate": 4.540351929048146e-05,
+      "loss": 0.7859,
+      "step": 1590
+    },
+    {
+      "epoch": 1.0377358490566038,
+      "grad_norm": 0.9278448820114136,
+      "learning_rate": 4.537364901017393e-05,
+      "loss": 0.8269,
+      "step": 1595
+    },
+    {
+      "epoch": 1.0409889394925178,
+      "grad_norm": 0.7428409457206726,
+      "learning_rate": 4.534369188034853e-05,
+      "loss": 0.806,
+      "step": 1600
+    },
+    {
+      "epoch": 1.044242029928432,
+      "grad_norm": 0.7308477759361267,
+      "learning_rate": 4.5313648028707557e-05,
+      "loss": 0.7991,
+      "step": 1605
+    },
+    {
+      "epoch": 1.047495120364346,
+      "grad_norm": 0.7885825037956238,
+      "learning_rate": 4.528351758332303e-05,
+      "loss": 0.7896,
+      "step": 1610
+    },
+    {
+      "epoch": 1.0507482108002602,
+      "grad_norm": 0.8900930285453796,
+      "learning_rate": 4.525330067263608e-05,
+      "loss": 0.791,
+      "step": 1615
+    },
+    {
+      "epoch": 1.0540013012361744,
+      "grad_norm": 0.7243936061859131,
+      "learning_rate": 4.5222997425456446e-05,
+      "loss": 0.8118,
+      "step": 1620
+    },
+    {
+      "epoch": 1.0572543916720885,
+      "grad_norm": 0.7627750039100647,
+      "learning_rate": 4.519260797096187e-05,
+      "loss": 0.7967,
+      "step": 1625
+    },
+    {
+      "epoch": 1.0605074821080025,
+      "grad_norm": 0.7201557755470276,
+      "learning_rate": 4.5162132438697615e-05,
+      "loss": 0.8087,
+      "step": 1630
+    },
+    {
+      "epoch": 1.0637605725439168,
+      "grad_norm": 1.3476982116699219,
+      "learning_rate": 4.513157095857586e-05,
+      "loss": 0.8152,
+      "step": 1635
+    },
+    {
+      "epoch": 1.0670136629798308,
+      "grad_norm": 2.0568456649780273,
+      "learning_rate": 4.510092366087518e-05,
+      "loss": 0.7879,
+      "step": 1640
+    },
+    {
+      "epoch": 1.070266753415745,
+      "grad_norm": 0.805178701877594,
+      "learning_rate": 4.507019067623997e-05,
+      "loss": 0.8083,
+      "step": 1645
+    },
+    {
+      "epoch": 1.073519843851659,
+      "grad_norm": 0.8525136709213257,
+      "learning_rate": 4.5039372135679883e-05,
+      "loss": 0.8044,
+      "step": 1650
+    },
+    {
+      "epoch": 1.0767729342875731,
+      "grad_norm": 0.7201101183891296,
+      "learning_rate": 4.5008468170569295e-05,
+      "loss": 0.798,
+      "step": 1655
+    },
+    {
+      "epoch": 1.0800260247234874,
+      "grad_norm": 0.8228124976158142,
+      "learning_rate": 4.497747891264675e-05,
+      "loss": 0.7921,
+      "step": 1660
+    },
+    {
+      "epoch": 1.0832791151594015,
+      "grad_norm": 0.8848757147789001,
+      "learning_rate": 4.494640449401434e-05,
+      "loss": 0.789,
+      "step": 1665
+    },
+    {
+      "epoch": 1.0865322055953155,
+      "grad_norm": 0.7168120741844177,
+      "learning_rate": 4.491524504713722e-05,
+      "loss": 0.8081,
+      "step": 1670
+    },
+    {
+      "epoch": 1.0897852960312298,
+      "grad_norm": 0.7164594531059265,
+      "learning_rate": 4.4884000704842976e-05,
+      "loss": 0.8004,
+      "step": 1675
+    },
+    {
+      "epoch": 1.0930383864671438,
+      "grad_norm": 0.7822607159614563,
+      "learning_rate": 4.485267160032112e-05,
+      "loss": 0.811,
+      "step": 1680
+    },
+    {
+      "epoch": 1.0962914769030578,
+      "grad_norm": 1.0780830383300781,
+      "learning_rate": 4.4821257867122475e-05,
+      "loss": 0.8068,
+      "step": 1685
+    },
+    {
+      "epoch": 1.099544567338972,
+      "grad_norm": 0.9002332091331482,
+      "learning_rate": 4.478975963915861e-05,
+      "loss": 0.7883,
+      "step": 1690
+    },
+    {
+      "epoch": 1.1027976577748861,
+      "grad_norm": 0.8772884011268616,
+      "learning_rate": 4.475817705070132e-05,
+      "loss": 0.8103,
+      "step": 1695
+    },
+    {
+      "epoch": 1.1060507482108002,
+      "grad_norm": 0.7703087329864502,
+      "learning_rate": 4.472651023638196e-05,
+      "loss": 0.7852,
+      "step": 1700
+    },
+    {
+      "epoch": 1.1093038386467144,
+      "grad_norm": 0.6608516573905945,
+      "learning_rate": 4.469475933119098e-05,
+      "loss": 0.8177,
+      "step": 1705
+    },
+    {
+      "epoch": 1.1125569290826285,
+      "grad_norm": 0.6942402720451355,
+      "learning_rate": 4.4662924470477255e-05,
+      "loss": 0.7958,
+      "step": 1710
+    },
+    {
+      "epoch": 1.1158100195185425,
+      "grad_norm": 1.1164368391036987,
+      "learning_rate": 4.4631005789947576e-05,
+      "loss": 0.8265,
+      "step": 1715
+    },
+    {
+      "epoch": 1.1190631099544568,
+      "grad_norm": 0.9671571850776672,
+      "learning_rate": 4.4599003425666026e-05,
+      "loss": 0.828,
+      "step": 1720
+    },
+    {
+      "epoch": 1.1223162003903708,
+      "grad_norm": 0.7158306837081909,
+      "learning_rate": 4.456691751405343e-05,
+      "loss": 0.8068,
+      "step": 1725
+    },
+    {
+      "epoch": 1.1255692908262849,
+      "grad_norm": 0.7458908557891846,
+      "learning_rate": 4.453474819188675e-05,
+      "loss": 0.8044,
+      "step": 1730
+    },
+    {
+      "epoch": 1.1288223812621991,
+      "grad_norm": 0.832358181476593,
+      "learning_rate": 4.450249559629853e-05,
+      "loss": 0.8041,
+      "step": 1735
+    },
+    {
+      "epoch": 1.1320754716981132,
+      "grad_norm": 1.0899025201797485,
+      "learning_rate": 4.447015986477628e-05,
+      "loss": 0.8171,
+      "step": 1740
+    },
+    {
+      "epoch": 1.1353285621340272,
+      "grad_norm": 0.9264464378356934,
+      "learning_rate": 4.443774113516192e-05,
+      "loss": 0.7795,
+      "step": 1745
+    },
+    {
+      "epoch": 1.1385816525699415,
+      "grad_norm": 1.837517261505127,
+      "learning_rate": 4.440523954565114e-05,
+      "loss": 0.8011,
+      "step": 1750
+    },
+    {
+      "epoch": 1.1418347430058555,
+      "grad_norm": 0.7183268070220947,
+      "learning_rate": 4.437265523479291e-05,
+      "loss": 0.8071,
+      "step": 1755
+    },
+    {
+      "epoch": 1.1450878334417696,
+      "grad_norm": 1.3461638689041138,
+      "learning_rate": 4.433998834148877e-05,
+      "loss": 0.8142,
+      "step": 1760
+    },
+    {
+      "epoch": 1.1483409238776838,
+      "grad_norm": 1.060806155204773,
+      "learning_rate": 4.430723900499232e-05,
+      "loss": 0.7857,
+      "step": 1765
+    },
+    {
+      "epoch": 1.1515940143135979,
+      "grad_norm": 0.9301265478134155,
+      "learning_rate": 4.427440736490861e-05,
+      "loss": 0.8009,
+      "step": 1770
+    },
+    {
+      "epoch": 1.1548471047495121,
+      "grad_norm": 1.4558310508728027,
+      "learning_rate": 4.4241493561193515e-05,
+      "loss": 0.8203,
+      "step": 1775
+    },
+    {
+      "epoch": 1.1581001951854262,
+      "grad_norm": 0.7694815993309021,
+      "learning_rate": 4.4208497734153177e-05,
+      "loss": 0.7825,
+      "step": 1780
+    },
+    {
+      "epoch": 1.1613532856213402,
+      "grad_norm": 0.7246577739715576,
+      "learning_rate": 4.417542002444339e-05,
+      "loss": 0.8157,
+      "step": 1785
+    },
+    {
+      "epoch": 1.1646063760572545,
+      "grad_norm": 1.0033169984817505,
+      "learning_rate": 4.4142260573068993e-05,
+      "loss": 0.8013,
+      "step": 1790
+    },
+    {
+      "epoch": 1.1678594664931685,
+      "grad_norm": 1.7446528673171997,
+      "learning_rate": 4.410901952138326e-05,
+      "loss": 0.8004,
+      "step": 1795
+    },
+    {
+      "epoch": 1.1711125569290826,
+      "grad_norm": 1.0588308572769165,
+      "learning_rate": 4.407569701108737e-05,
+      "loss": 0.8055,
+      "step": 1800
+    },
+    {
+      "epoch": 1.1743656473649968,
+      "grad_norm": 0.7162896990776062,
+      "learning_rate": 4.404229318422968e-05,
+      "loss": 0.8091,
+      "step": 1805
+    },
+    {
+      "epoch": 1.1776187378009109,
+      "grad_norm": 1.3299797773361206,
+      "learning_rate": 4.400880818320521e-05,
+      "loss": 0.8068,
+      "step": 1810
+    },
+    {
+      "epoch": 1.180871828236825,
+      "grad_norm": 0.7723076939582825,
+      "learning_rate": 4.397524215075504e-05,
+      "loss": 0.8065,
+      "step": 1815
+    },
+    {
+      "epoch": 1.1841249186727392,
+      "grad_norm": 0.700183629989624,
+      "learning_rate": 4.3941595229965636e-05,
+      "loss": 0.8006,
+      "step": 1820
+    },
+    {
+      "epoch": 1.1873780091086532,
+      "grad_norm": 0.7281947135925293,
+      "learning_rate": 4.390786756426829e-05,
+      "loss": 0.8026,
+      "step": 1825
+    },
+    {
+      "epoch": 1.1906310995445673,
+      "grad_norm": 0.8081494569778442,
+      "learning_rate": 4.3874059297438515e-05,
+      "loss": 0.7887,
+      "step": 1830
+    },
+    {
+      "epoch": 1.1938841899804815,
+      "grad_norm": 0.749001681804657,
+      "learning_rate": 4.384017057359538e-05,
+      "loss": 0.8007,
+      "step": 1835
+    },
+    {
+      "epoch": 1.1971372804163956,
+      "grad_norm": 0.8392378091812134,
+      "learning_rate": 4.380620153720095e-05,
+      "loss": 0.8228,
+      "step": 1840
+    },
+    {
+      "epoch": 1.2003903708523098,
+      "grad_norm": 0.9195884466171265,
+      "learning_rate": 4.377215233305966e-05,
+      "loss": 0.8009,
+      "step": 1845
+    },
+    {
+      "epoch": 1.2036434612882239,
+      "grad_norm": 0.9343377947807312,
+      "learning_rate": 4.373802310631765e-05,
+      "loss": 0.7785,
+      "step": 1850
+    },
+    {
+      "epoch": 1.206896551724138,
+      "grad_norm": 0.9939282536506653,
+      "learning_rate": 4.370381400246221e-05,
+      "loss": 0.8228,
+      "step": 1855
+    },
+    {
+      "epoch": 1.2101496421600522,
+      "grad_norm": 4.1950907707214355,
+      "learning_rate": 4.366952516732114e-05,
+      "loss": 0.8051,
+      "step": 1860
+    },
+    {
+      "epoch": 1.2134027325959662,
+      "grad_norm": 1.1269813776016235,
+      "learning_rate": 4.3635156747062105e-05,
+      "loss": 0.8059,
+      "step": 1865
+    },
+    {
+      "epoch": 1.2166558230318802,
+      "grad_norm": 1.8155053853988647,
+      "learning_rate": 4.360070888819203e-05,
+      "loss": 0.8157,
+      "step": 1870
+    },
+    {
+      "epoch": 1.2199089134677945,
+      "grad_norm": 0.7301884889602661,
+      "learning_rate": 4.356618173755648e-05,
+      "loss": 0.786,
+      "step": 1875
+    },
+    {
+      "epoch": 1.2231620039037086,
+      "grad_norm": 0.6721681952476501,
+      "learning_rate": 4.353157544233902e-05,
+      "loss": 0.818,
+      "step": 1880
+    },
+    {
+      "epoch": 1.2264150943396226,
+      "grad_norm": 0.9088836312294006,
+      "learning_rate": 4.349689015006061e-05,
+      "loss": 0.7883,
+      "step": 1885
+    },
+    {
+      "epoch": 1.2296681847755369,
+      "grad_norm": 0.7416070103645325,
+      "learning_rate": 4.3462126008578936e-05,
+      "loss": 0.8033,
+      "step": 1890
+    },
+    {
+      "epoch": 1.232921275211451,
+      "grad_norm": 0.6465050578117371,
+      "learning_rate": 4.342728316608783e-05,
+      "loss": 0.8111,
+      "step": 1895
+    },
+    {
+      "epoch": 1.236174365647365,
+      "grad_norm": 0.870087206363678,
+      "learning_rate": 4.3392361771116604e-05,
+      "loss": 0.8307,
+      "step": 1900
+    },
+    {
+      "epoch": 1.2394274560832792,
+      "grad_norm": 0.7588717937469482,
+      "learning_rate": 4.335736197252942e-05,
+      "loss": 0.8024,
+      "step": 1905
+    },
+    {
+      "epoch": 1.2426805465191932,
+      "grad_norm": 24.7558650970459,
+      "learning_rate": 4.332228391952469e-05,
+      "loss": 0.8089,
+      "step": 1910
+    },
+    {
+      "epoch": 1.2459336369551073,
+      "grad_norm": 0.7845138907432556,
+      "learning_rate": 4.328712776163436e-05,
+      "loss": 0.8092,
+      "step": 1915
+    },
+    {
+      "epoch": 1.2491867273910215,
+      "grad_norm": 0.5970214605331421,
+      "learning_rate": 4.325189364872337e-05,
+      "loss": 0.768,
+      "step": 1920
+    },
+    {
+      "epoch": 1.2524398178269356,
+      "grad_norm": 0.6692010164260864,
+      "learning_rate": 4.321658173098895e-05,
+      "loss": 0.7974,
+      "step": 1925
+    },
+    {
+      "epoch": 1.2556929082628496,
+      "grad_norm": 0.6391417384147644,
+      "learning_rate": 4.318119215896001e-05,
+      "loss": 0.8151,
+      "step": 1930
+    },
+    {
+      "epoch": 1.258945998698764,
+      "grad_norm": 1.4557222127914429,
+      "learning_rate": 4.314572508349646e-05,
+      "loss": 0.817,
+      "step": 1935
+    },
+    {
+      "epoch": 1.262199089134678,
+      "grad_norm": 0.6106812357902527,
+      "learning_rate": 4.311018065578864e-05,
+      "loss": 0.8154,
+      "step": 1940
+    },
+    {
+      "epoch": 1.265452179570592,
+      "grad_norm": 0.7062475681304932,
+      "learning_rate": 4.307455902735659e-05,
+      "loss": 0.8106,
+      "step": 1945
+    },
+    {
+      "epoch": 1.2687052700065062,
+      "grad_norm": 0.6937923431396484,
+      "learning_rate": 4.303886035004947e-05,
+      "loss": 0.8193,
+      "step": 1950
+    },
+    {
+      "epoch": 1.2719583604424203,
+      "grad_norm": 0.5820494294166565,
+      "learning_rate": 4.3003084776044855e-05,
+      "loss": 0.8166,
+      "step": 1955
+    },
+    {
+      "epoch": 1.2752114508783343,
+      "grad_norm": 0.9809524416923523,
+      "learning_rate": 4.2967232457848154e-05,
+      "loss": 0.7983,
+      "step": 1960
+    },
+    {
+      "epoch": 1.2784645413142486,
+      "grad_norm": 0.6147664785385132,
+      "learning_rate": 4.293130354829191e-05,
+      "loss": 0.8195,
+      "step": 1965
+    },
+    {
+      "epoch": 1.2817176317501626,
+      "grad_norm": 0.632622241973877,
+      "learning_rate": 4.289529820053515e-05,
+      "loss": 0.7964,
+      "step": 1970
+    },
+    {
+      "epoch": 1.2849707221860767,
+      "grad_norm": 0.5672839283943176,
+      "learning_rate": 4.285921656806276e-05,
+      "loss": 0.7854,
+      "step": 1975
+    },
+    {
+      "epoch": 1.288223812621991,
+      "grad_norm": 0.7861791849136353,
+      "learning_rate": 4.2823058804684815e-05,
+      "loss": 0.7772,
+      "step": 1980
+    },
+    {
+      "epoch": 1.291476903057905,
+      "grad_norm": 0.627170741558075,
+      "learning_rate": 4.2786825064535905e-05,
+      "loss": 0.8033,
+      "step": 1985
+    },
+    {
+      "epoch": 1.294729993493819,
+      "grad_norm": 0.7018983960151672,
+      "learning_rate": 4.275051550207453e-05,
+      "loss": 0.8067,
+      "step": 1990
+    },
+    {
+      "epoch": 1.2979830839297333,
+      "grad_norm": 0.6708381772041321,
+      "learning_rate": 4.2714130272082365e-05,
+      "loss": 0.8019,
+      "step": 1995
+    },
+    {
+      "epoch": 1.3012361743656473,
+      "grad_norm": 1.4676740169525146,
+      "learning_rate": 4.267766952966369e-05,
+      "loss": 0.8366,
+      "step": 2000
+    },
+    {
+      "epoch": 1.3044892648015614,
+      "grad_norm": 0.6410109996795654,
+      "learning_rate": 4.2641133430244644e-05,
+      "loss": 0.7767,
+      "step": 2005
+    },
+    {
+      "epoch": 1.3077423552374756,
+      "grad_norm": 0.6521077752113342,
+      "learning_rate": 4.2604522129572624e-05,
+      "loss": 0.8134,
+      "step": 2010
+    },
+    {
+      "epoch": 1.3109954456733897,
+      "grad_norm": 0.605689525604248,
+      "learning_rate": 4.256783578371557e-05,
+      "loss": 0.7803,
+      "step": 2015
+    },
+    {
+      "epoch": 1.3142485361093037,
+      "grad_norm": 1.1350992918014526,
+      "learning_rate": 4.253107454906137e-05,
+      "loss": 0.787,
+      "step": 2020
+    },
+    {
+      "epoch": 1.317501626545218,
+      "grad_norm": 0.7411300539970398,
+      "learning_rate": 4.2494238582317114e-05,
+      "loss": 0.8047,
+      "step": 2025
+    },
+    {
+      "epoch": 1.320754716981132,
+      "grad_norm": 0.8492873311042786,
+      "learning_rate": 4.2457328040508484e-05,
+      "loss": 0.788,
+      "step": 2030
+    },
+    {
+      "epoch": 1.3240078074170463,
+      "grad_norm": 0.7676826119422913,
+      "learning_rate": 4.2420343080979035e-05,
+      "loss": 0.808,
+      "step": 2035
+    },
+    {
+      "epoch": 1.3272608978529603,
+      "grad_norm": 0.6066752076148987,
+      "learning_rate": 4.238328386138959e-05,
+      "loss": 0.7894,
+      "step": 2040
+    },
+    {
+      "epoch": 1.3305139882888743,
+      "grad_norm": 0.9500682353973389,
+      "learning_rate": 4.234615053971751e-05,
+      "loss": 0.7933,
+      "step": 2045
+    },
+    {
+      "epoch": 1.3337670787247886,
+      "grad_norm": 0.6935995221138,
+      "learning_rate": 4.230894327425604e-05,
+      "loss": 0.7949,
+      "step": 2050
+    },
+    {
+      "epoch": 1.3370201691607027,
+      "grad_norm": 0.701280415058136,
+      "learning_rate": 4.227166222361364e-05,
+      "loss": 0.7879,
+      "step": 2055
+    },
+    {
+      "epoch": 1.340273259596617,
+      "grad_norm": 0.6840693354606628,
+      "learning_rate": 4.2234307546713305e-05,
+      "loss": 0.8095,
+      "step": 2060
+    },
+    {
+      "epoch": 1.343526350032531,
+      "grad_norm": 0.6785464286804199,
+      "learning_rate": 4.219687940279188e-05,
+      "loss": 0.7931,
+      "step": 2065
+    },
+    {
+      "epoch": 1.346779440468445,
+      "grad_norm": 1.4129618406295776,
+      "learning_rate": 4.2159377951399385e-05,
+      "loss": 0.8222,
+      "step": 2070
+    },
+    {
+      "epoch": 1.3500325309043593,
+      "grad_norm": 0.9987890124320984,
+      "learning_rate": 4.212180335239836e-05,
+      "loss": 0.8086,
+      "step": 2075
+    },
+    {
+      "epoch": 1.3532856213402733,
+      "grad_norm": 0.7902525067329407,
+      "learning_rate": 4.208415576596315e-05,
+      "loss": 0.808,
+      "step": 2080
+    },
+    {
+      "epoch": 1.3565387117761873,
+      "grad_norm": 1.0341746807098389,
+      "learning_rate": 4.2046435352579206e-05,
+      "loss": 0.8218,
+      "step": 2085
+    },
+    {
+      "epoch": 1.3597918022121016,
+      "grad_norm": 0.6914475560188293,
+      "learning_rate": 4.200864227304247e-05,
+      "loss": 0.8033,
+      "step": 2090
+    },
+    {
+      "epoch": 1.3630448926480156,
+      "grad_norm": 0.8766419291496277,
+      "learning_rate": 4.1970776688458624e-05,
+      "loss": 0.7885,
+      "step": 2095
+    },
+    {
+      "epoch": 1.3662979830839297,
+      "grad_norm": 0.7936528325080872,
+      "learning_rate": 4.1932838760242445e-05,
+      "loss": 0.8163,
+      "step": 2100
+    },
+    {
+      "epoch": 1.369551073519844,
+      "grad_norm": 0.6358763575553894,
+      "learning_rate": 4.189482865011706e-05,
+      "loss": 0.8042,
+      "step": 2105
+    },
+    {
+      "epoch": 1.372804163955758,
+      "grad_norm": 0.6236315965652466,
+      "learning_rate": 4.1856746520113345e-05,
+      "loss": 0.7926,
+      "step": 2110
+    },
+    {
+      "epoch": 1.376057254391672,
+      "grad_norm": 0.6494086980819702,
+      "learning_rate": 4.181859253256916e-05,
+      "loss": 0.8001,
+      "step": 2115
+    },
+    {
+      "epoch": 1.3793103448275863,
+      "grad_norm": 0.5840818881988525,
+      "learning_rate": 4.178036685012868e-05,
+      "loss": 0.7936,
+      "step": 2120
+    },
+    {
+      "epoch": 1.3825634352635003,
+      "grad_norm": 0.6631473898887634,
+      "learning_rate": 4.174206963574171e-05,
+      "loss": 0.8022,
+      "step": 2125
+    },
+    {
+      "epoch": 1.3858165256994144,
+      "grad_norm": 0.6331456303596497,
+      "learning_rate": 4.1703701052662974e-05,
+      "loss": 0.8023,
+      "step": 2130
+    },
+    {
+      "epoch": 1.3890696161353286,
+      "grad_norm": 0.9553548693656921,
+      "learning_rate": 4.166526126445145e-05,
+      "loss": 0.7927,
+      "step": 2135
+    },
+    {
+      "epoch": 1.3923227065712427,
+      "grad_norm": 0.5648055076599121,
+      "learning_rate": 4.162675043496963e-05,
+      "loss": 0.8094,
+      "step": 2140
+    },
+    {
+      "epoch": 1.3955757970071567,
+      "grad_norm": 0.7545254826545715,
+      "learning_rate": 4.158816872838285e-05,
+      "loss": 0.7898,
+      "step": 2145
+    },
+    {
+      "epoch": 1.398828887443071,
+      "grad_norm": 0.6327040791511536,
+      "learning_rate": 4.1549516309158586e-05,
+      "loss": 0.7933,
+      "step": 2150
+    },
+    {
+      "epoch": 1.402081977878985,
+      "grad_norm": 0.897237241268158,
+      "learning_rate": 4.151079334206577e-05,
+      "loss": 0.8026,
+      "step": 2155
+    },
+    {
+      "epoch": 1.405335068314899,
+      "grad_norm": 0.6637735962867737,
+      "learning_rate": 4.147199999217402e-05,
+      "loss": 0.8089,
+      "step": 2160
+    },
+    {
+      "epoch": 1.4085881587508133,
+      "grad_norm": 0.7697905898094177,
+      "learning_rate": 4.143313642485302e-05,
+      "loss": 0.7986,
+      "step": 2165
+    },
+    {
+      "epoch": 1.4118412491867274,
+      "grad_norm": 0.6807245016098022,
+      "learning_rate": 4.139420280577177e-05,
+      "loss": 0.8095,
+      "step": 2170
+    },
+    {
+      "epoch": 1.4150943396226414,
+      "grad_norm": 0.6346644163131714,
+      "learning_rate": 4.1355199300897894e-05,
+      "loss": 0.7874,
+      "step": 2175
+    },
+    {
+      "epoch": 1.4183474300585557,
+      "grad_norm": 1.3532865047454834,
+      "learning_rate": 4.1316126076496935e-05,
+      "loss": 0.8077,
+      "step": 2180
+    },
+    {
+      "epoch": 1.4216005204944697,
+      "grad_norm": 0.7018231749534607,
+      "learning_rate": 4.127698329913161e-05,
+      "loss": 0.7826,
+      "step": 2185
+    },
+    {
+      "epoch": 1.4248536109303838,
+      "grad_norm": 0.6970395445823669,
+      "learning_rate": 4.1237771135661164e-05,
+      "loss": 0.7959,
+      "step": 2190
+    },
+    {
+      "epoch": 1.428106701366298,
+      "grad_norm": 0.802807092666626,
+      "learning_rate": 4.119848975324059e-05,
+      "loss": 0.8093,
+      "step": 2195
+    },
+    {
+      "epoch": 1.431359791802212,
+      "grad_norm": 0.6809616684913635,
+      "learning_rate": 4.115913931931997e-05,
+      "loss": 0.7949,
+      "step": 2200
+    },
+    {
+      "epoch": 1.434612882238126,
+      "grad_norm": 0.6927191615104675,
+      "learning_rate": 4.1119720001643745e-05,
+      "loss": 0.8072,
+      "step": 2205
+    },
+    {
+      "epoch": 1.4378659726740404,
+      "grad_norm": 0.8336883187294006,
+      "learning_rate": 4.108023196824998e-05,
+      "loss": 0.8058,
+      "step": 2210
+    },
+    {
+      "epoch": 1.4411190631099544,
+      "grad_norm": 0.9360595345497131,
+      "learning_rate": 4.1040675387469685e-05,
+      "loss": 0.796,
+      "step": 2215
+    },
+    {
+      "epoch": 1.4443721535458685,
+      "grad_norm": 0.7317126989364624,
+      "learning_rate": 4.1001050427926045e-05,
+      "loss": 0.789,
+      "step": 2220
+    },
+    {
+      "epoch": 1.4476252439817827,
+      "grad_norm": 0.7633559703826904,
+      "learning_rate": 4.0961357258533774e-05,
+      "loss": 0.7885,
+      "step": 2225
+    },
+    {
+      "epoch": 1.4508783344176968,
+      "grad_norm": 0.6734256744384766,
+      "learning_rate": 4.0921596048498315e-05,
+      "loss": 0.7852,
+      "step": 2230
+    },
+    {
+      "epoch": 1.4541314248536108,
+      "grad_norm": 1.5330891609191895,
+      "learning_rate": 4.088176696731517e-05,
+      "loss": 0.8067,
+      "step": 2235
+    },
+    {
+      "epoch": 1.457384515289525,
+      "grad_norm": 1.4686920642852783,
+      "learning_rate": 4.084187018476918e-05,
+      "loss": 0.8,
+      "step": 2240
+    },
+    {
+      "epoch": 1.460637605725439,
+      "grad_norm": 0.8620485067367554,
+      "learning_rate": 4.0801905870933764e-05,
+      "loss": 0.7865,
+      "step": 2245
+    },
+    {
+      "epoch": 1.4638906961613534,
+      "grad_norm": 0.8238881826400757,
+      "learning_rate": 4.076187419617024e-05,
+      "loss": 0.8486,
+      "step": 2250
+    },
+    {
+      "epoch": 1.4671437865972674,
+      "grad_norm": 0.5753189921379089,
+      "learning_rate": 4.072177533112703e-05,
+      "loss": 0.7975,
+      "step": 2255
+    },
+    {
+      "epoch": 1.4703968770331814,
+      "grad_norm": 0.707665741443634,
+      "learning_rate": 4.068160944673903e-05,
+      "loss": 0.8067,
+      "step": 2260
+    },
+    {
+      "epoch": 1.4736499674690957,
+      "grad_norm": 0.702339768409729,
+      "learning_rate": 4.0641376714226795e-05,
+      "loss": 0.7823,
+      "step": 2265
+    },
+    {
+      "epoch": 1.4769030579050098,
+      "grad_norm": 1.500267505645752,
+      "learning_rate": 4.060107730509587e-05,
+      "loss": 0.8159,
+      "step": 2270
+    },
+    {
+      "epoch": 1.480156148340924,
+      "grad_norm": 0.8150926232337952,
+      "learning_rate": 4.0560711391135986e-05,
+      "loss": 0.8227,
+      "step": 2275
+    },
+    {
+      "epoch": 1.483409238776838,
+      "grad_norm": 0.8220822215080261,
+      "learning_rate": 4.052027914442043e-05,
+      "loss": 0.8067,
+      "step": 2280
+    },
+    {
+      "epoch": 1.486662329212752,
+      "grad_norm": 0.6167823672294617,
+      "learning_rate": 4.047978073730519e-05,
+      "loss": 0.8158,
+      "step": 2285
+    },
+    {
+      "epoch": 1.4899154196486664,
+      "grad_norm": 36.838592529296875,
+      "learning_rate": 4.043921634242836e-05,
+      "loss": 0.8178,
+      "step": 2290
+    },
+    {
+      "epoch": 1.4931685100845804,
+      "grad_norm": 0.6391186714172363,
+      "learning_rate": 4.039858613270927e-05,
+      "loss": 0.7954,
+      "step": 2295
+    },
+    {
+      "epoch": 1.4964216005204944,
+      "grad_norm": 0.7283660769462585,
+      "learning_rate": 4.035789028134782e-05,
+      "loss": 0.8045,
+      "step": 2300
+    },
+    {
+      "epoch": 1.4996746909564087,
+      "grad_norm": 1.2124541997909546,
+      "learning_rate": 4.031712896182376e-05,
+      "loss": 0.7941,
+      "step": 2305
+    },
+    {
+      "epoch": 1.5029277813923227,
+      "grad_norm": 0.6527173519134521,
+      "learning_rate": 4.0276302347895864e-05,
+      "loss": 0.808,
+      "step": 2310
+    },
+    {
+      "epoch": 1.5061808718282368,
+      "grad_norm": 0.7583323121070862,
+      "learning_rate": 4.023541061360131e-05,
+      "loss": 0.8228,
+      "step": 2315
+    },
+    {
+      "epoch": 1.509433962264151,
+      "grad_norm": 2.344557762145996,
+      "learning_rate": 4.019445393325483e-05,
+      "loss": 0.798,
+      "step": 2320
+    },
+    {
+      "epoch": 1.512687052700065,
+      "grad_norm": 0.8352357149124146,
+      "learning_rate": 4.0153432481448027e-05,
+      "loss": 0.7912,
+      "step": 2325
+    },
+    {
+      "epoch": 1.5159401431359791,
+      "grad_norm": 0.8066681623458862,
+      "learning_rate": 4.01123464330486e-05,
+      "loss": 0.8125,
+      "step": 2330
+    },
+    {
+      "epoch": 1.5191932335718934,
+      "grad_norm": 0.7398434281349182,
+      "learning_rate": 4.007119596319962e-05,
+      "loss": 0.7997,
+      "step": 2335
+    },
+    {
+      "epoch": 1.5224463240078074,
+      "grad_norm": 0.9586179852485657,
+      "learning_rate": 4.002998124731879e-05,
+      "loss": 0.7994,
+      "step": 2340
+    },
+    {
+      "epoch": 1.5256994144437215,
+      "grad_norm": 0.6949347853660583,
+      "learning_rate": 3.998870246109767e-05,
+      "loss": 0.8192,
+      "step": 2345
+    },
+    {
+      "epoch": 1.5289525048796357,
+      "grad_norm": 1.5228863954544067,
+      "learning_rate": 3.994735978050094e-05,
+      "loss": 0.7902,
+      "step": 2350
+    },
+    {
+      "epoch": 1.5322055953155498,
+      "grad_norm": 0.6021274328231812,
+      "learning_rate": 3.990595338176564e-05,
+      "loss": 0.7995,
+      "step": 2355
+    },
+    {
+      "epoch": 1.5354586857514638,
+      "grad_norm": 0.6251325607299805,
+      "learning_rate": 3.986448344140047e-05,
+      "loss": 0.7943,
+      "step": 2360
+    },
+    {
+      "epoch": 1.538711776187378,
+      "grad_norm": 0.6837208867073059,
+      "learning_rate": 3.9822950136184946e-05,
+      "loss": 0.8093,
+      "step": 2365
+    },
+    {
+      "epoch": 1.5419648666232921,
+      "grad_norm": 1.162960171699524,
+      "learning_rate": 3.978135364316874e-05,
+      "loss": 0.8309,
+      "step": 2370
+    },
+    {
+      "epoch": 1.5452179570592062,
+      "grad_norm": 0.6852900385856628,
+      "learning_rate": 3.973969413967086e-05,
+      "loss": 0.803,
+      "step": 2375
+    },
+    {
+      "epoch": 1.5484710474951204,
+      "grad_norm": 0.8108793497085571,
+      "learning_rate": 3.9697971803278924e-05,
+      "loss": 0.8035,
+      "step": 2380
+    },
+    {
+      "epoch": 1.5517241379310345,
+      "grad_norm": 0.7817934155464172,
+      "learning_rate": 3.9656186811848395e-05,
+      "loss": 0.7872,
+      "step": 2385
+    },
+    {
+      "epoch": 1.5549772283669485,
+      "grad_norm": 0.6819909811019897,
+      "learning_rate": 3.9614339343501836e-05,
+      "loss": 0.7978,
+      "step": 2390
+    },
+    {
+      "epoch": 1.5582303188028628,
+      "grad_norm": 1.547499179840088,
+      "learning_rate": 3.9572429576628114e-05,
+      "loss": 0.8042,
+      "step": 2395
+    },
+    {
+      "epoch": 1.5614834092387768,
+      "grad_norm": 0.6589548587799072,
+      "learning_rate": 3.9530457689881684e-05,
+      "loss": 0.8112,
+      "step": 2400
+    },
+    {
+      "epoch": 1.5647364996746909,
+      "grad_norm": 1.0360349416732788,
+      "learning_rate": 3.94884238621818e-05,
+      "loss": 0.7949,
+      "step": 2405
+    },
+    {
+      "epoch": 1.5679895901106051,
+      "grad_norm": 0.8563544154167175,
+      "learning_rate": 3.944632827271176e-05,
+      "loss": 0.8193,
+      "step": 2410
+    },
+    {
+      "epoch": 1.5712426805465192,
+      "grad_norm": 0.8043304681777954,
+      "learning_rate": 3.940417110091816e-05,
+      "loss": 0.7986,
+      "step": 2415
+    },
+    {
+      "epoch": 1.5744957709824332,
+      "grad_norm": 0.7991480827331543,
+      "learning_rate": 3.9361952526510085e-05,
+      "loss": 0.791,
+      "step": 2420
+    },
+    {
+      "epoch": 1.5777488614183475,
+      "grad_norm": 0.8276852369308472,
+      "learning_rate": 3.9319672729458376e-05,
+      "loss": 0.8065,
+      "step": 2425
+    },
+    {
+      "epoch": 1.5810019518542615,
+      "grad_norm": 0.601204514503479,
+      "learning_rate": 3.927733188999486e-05,
+      "loss": 0.803,
+      "step": 2430
+    },
+    {
+      "epoch": 1.5842550422901756,
+      "grad_norm": 0.77410888671875,
+      "learning_rate": 3.92349301886116e-05,
+      "loss": 0.8056,
+      "step": 2435
+    },
+    {
+      "epoch": 1.5875081327260898,
+      "grad_norm": 1.0822545289993286,
+      "learning_rate": 3.9192467806060044e-05,
+      "loss": 0.7781,
+      "step": 2440
+    },
+    {
+      "epoch": 1.5907612231620039,
+      "grad_norm": 0.602203905582428,
+      "learning_rate": 3.914994492335038e-05,
+      "loss": 0.7805,
+      "step": 2445
+    },
+    {
+      "epoch": 1.594014313597918,
+      "grad_norm": 0.6420490741729736,
+      "learning_rate": 3.910736172175066e-05,
+      "loss": 0.8081,
+      "step": 2450
+    },
+    {
+      "epoch": 1.5972674040338322,
+      "grad_norm": 1.321158766746521,
+      "learning_rate": 3.9064718382786076e-05,
+      "loss": 0.8251,
+      "step": 2455
+    },
+    {
+      "epoch": 1.6005204944697464,
+      "grad_norm": 1.2666637897491455,
+      "learning_rate": 3.9022015088238174e-05,
+      "loss": 0.8017,
+      "step": 2460
+    },
+    {
+      "epoch": 1.6037735849056602,
+      "grad_norm": 0.7312076687812805,
+      "learning_rate": 3.897925202014409e-05,
+      "loss": 0.8193,
+      "step": 2465
+    },
+    {
+      "epoch": 1.6070266753415745,
+      "grad_norm": 0.6371551156044006,
+      "learning_rate": 3.8936429360795745e-05,
+      "loss": 0.8028,
+      "step": 2470
+    },
+    {
+      "epoch": 1.6102797657774888,
+      "grad_norm": 0.7501451373100281,
+      "learning_rate": 3.88935472927391e-05,
+      "loss": 0.809,
+      "step": 2475
+    },
+    {
+      "epoch": 1.6135328562134026,
+      "grad_norm": 0.6462128162384033,
+      "learning_rate": 3.885060599877337e-05,
+      "loss": 0.7898,
+      "step": 2480
+    },
+    {
+      "epoch": 1.6167859466493169,
+      "grad_norm": 0.6399087905883789,
+      "learning_rate": 3.880760566195023e-05,
+      "loss": 0.7848,
+      "step": 2485
+    },
+    {
+      "epoch": 1.6200390370852311,
+      "grad_norm": 0.6313064694404602,
+      "learning_rate": 3.876454646557305e-05,
+      "loss": 0.7907,
+      "step": 2490
+    },
+    {
+      "epoch": 1.623292127521145,
+      "grad_norm": 0.5535691976547241,
+      "learning_rate": 3.872142859319612e-05,
+      "loss": 0.8071,
+      "step": 2495
+    },
+    {
+      "epoch": 1.6265452179570592,
+      "grad_norm": 0.7361935377120972,
+      "learning_rate": 3.867825222862383e-05,
+      "loss": 0.8034,
+      "step": 2500
+    },
+    {
+      "epoch": 1.6297983083929735,
+      "grad_norm": 0.7017170190811157,
+      "learning_rate": 3.863501755590994e-05,
+      "loss": 0.8202,
+      "step": 2505
+    },
+    {
+      "epoch": 1.6330513988288873,
+      "grad_norm": 2.15995717048645,
+      "learning_rate": 3.8591724759356734e-05,
+      "loss": 0.8064,
+      "step": 2510
+    },
+    {
+      "epoch": 1.6363044892648015,
+      "grad_norm": 0.6328789591789246,
+      "learning_rate": 3.854837402351431e-05,
+      "loss": 0.7959,
+      "step": 2515
+    },
+    {
+      "epoch": 1.6395575797007158,
+      "grad_norm": 0.7377115488052368,
+      "learning_rate": 3.8504965533179724e-05,
+      "loss": 0.7826,
+      "step": 2520
+    },
+    {
+      "epoch": 1.6428106701366298,
+      "grad_norm": 0.7871087193489075,
+      "learning_rate": 3.8461499473396246e-05,
+      "loss": 0.7836,
+      "step": 2525
+    },
+    {
+      "epoch": 1.6460637605725439,
+      "grad_norm": 0.7310390472412109,
+      "learning_rate": 3.841797602945254e-05,
+      "loss": 0.82,
+      "step": 2530
+    },
+    {
+      "epoch": 1.6493168510084582,
+      "grad_norm": 0.6992024779319763,
+      "learning_rate": 3.837439538688189e-05,
+      "loss": 0.7865,
+      "step": 2535
+    },
+    {
+      "epoch": 1.6525699414443722,
+      "grad_norm": 0.6299036741256714,
+      "learning_rate": 3.833075773146142e-05,
+      "loss": 0.8071,
+      "step": 2540
+    },
+    {
+      "epoch": 1.6558230318802862,
+      "grad_norm": 0.7134023904800415,
+      "learning_rate": 3.828706324921128e-05,
+      "loss": 0.7931,
+      "step": 2545
+    },
+    {
+      "epoch": 1.6590761223162005,
+      "grad_norm": 0.9033751487731934,
+      "learning_rate": 3.824331212639388e-05,
+      "loss": 0.8144,
+      "step": 2550
+    },
+    {
+      "epoch": 1.6623292127521145,
+      "grad_norm": 0.6907141208648682,
+      "learning_rate": 3.8199504549513055e-05,
+      "loss": 0.7866,
+      "step": 2555
+    },
+    {
+      "epoch": 1.6655823031880286,
+      "grad_norm": 0.9282029271125793,
+      "learning_rate": 3.81556407053133e-05,
+      "loss": 0.7935,
+      "step": 2560
+    },
+    {
+      "epoch": 1.6688353936239428,
+      "grad_norm": 0.8306324481964111,
+      "learning_rate": 3.811172078077899e-05,
+      "loss": 0.7896,
+      "step": 2565
+    },
+    {
+      "epoch": 1.6720884840598569,
+      "grad_norm": 0.8237498998641968,
+      "learning_rate": 3.806774496313355e-05,
+      "loss": 0.7894,
+      "step": 2570
+    },
+    {
+      "epoch": 1.675341574495771,
+      "grad_norm": 1.046177625656128,
+      "learning_rate": 3.802371343983865e-05,
+      "loss": 0.7965,
+      "step": 2575
+    },
+    {
+      "epoch": 1.6785946649316852,
+      "grad_norm": 0.8086977601051331,
+      "learning_rate": 3.797962639859344e-05,
+      "loss": 0.7962,
+      "step": 2580
+    },
+    {
+      "epoch": 1.6818477553675992,
+      "grad_norm": 0.8723990321159363,
+      "learning_rate": 3.7935484027333746e-05,
+      "loss": 0.8,
+      "step": 2585
+    },
+    {
+      "epoch": 1.6851008458035133,
+      "grad_norm": 0.6789833903312683,
+      "learning_rate": 3.7891286514231225e-05,
+      "loss": 0.801,
+      "step": 2590
+    },
+    {
+      "epoch": 1.6883539362394275,
+      "grad_norm": 0.7277116179466248,
+      "learning_rate": 3.784703404769263e-05,
+      "loss": 0.7984,
+      "step": 2595
+    },
+    {
+      "epoch": 1.6916070266753416,
+      "grad_norm": 1.03284752368927,
+      "learning_rate": 3.780272681635894e-05,
+      "loss": 0.8021,
+      "step": 2600
+    },
+    {
+      "epoch": 1.6948601171112556,
+      "grad_norm": 0.9302812218666077,
+      "learning_rate": 3.77583650091046e-05,
+      "loss": 0.8046,
+      "step": 2605
+    },
+    {
+      "epoch": 1.6981132075471699,
+      "grad_norm": 1.2493805885314941,
+      "learning_rate": 3.771394881503673e-05,
+      "loss": 0.805,
+      "step": 2610
+    },
+    {
+      "epoch": 1.701366297983084,
+      "grad_norm": 1.7447010278701782,
+      "learning_rate": 3.766947842349423e-05,
+      "loss": 0.7968,
+      "step": 2615
+    },
+    {
+      "epoch": 1.704619388418998,
+      "grad_norm": 0.5957579016685486,
+      "learning_rate": 3.76249540240471e-05,
+      "loss": 0.7958,
+      "step": 2620
+    },
+    {
+      "epoch": 1.7078724788549122,
+      "grad_norm": 0.8952409625053406,
+      "learning_rate": 3.7580375806495524e-05,
+      "loss": 0.7974,
+      "step": 2625
+    },
+    {
+      "epoch": 1.7111255692908263,
+      "grad_norm": 0.5696317553520203,
+      "learning_rate": 3.753574396086913e-05,
+      "loss": 0.8037,
+      "step": 2630
+    },
+    {
+      "epoch": 1.7143786597267403,
+      "grad_norm": 1.1165753602981567,
+      "learning_rate": 3.7491058677426135e-05,
+      "loss": 0.8038,
+      "step": 2635
+    },
+    {
+      "epoch": 1.7176317501626546,
+      "grad_norm": 0.7049902081489563,
+      "learning_rate": 3.7446320146652556e-05,
+      "loss": 0.7678,
+      "step": 2640
+    },
+    {
+      "epoch": 1.7208848405985686,
+      "grad_norm": 0.6802716851234436,
+      "learning_rate": 3.740152855926139e-05,
+      "loss": 0.8163,
+      "step": 2645
+    },
+    {
+      "epoch": 1.7241379310344827,
+      "grad_norm": 0.808645486831665,
+      "learning_rate": 3.735668410619183e-05,
+      "loss": 0.8045,
+      "step": 2650
+    },
+    {
+      "epoch": 1.727391021470397,
+      "grad_norm": 0.6994810104370117,
+      "learning_rate": 3.7311786978608415e-05,
+      "loss": 0.794,
+      "step": 2655
+    },
+    {
+      "epoch": 1.730644111906311,
+      "grad_norm": 0.9750849604606628,
+      "learning_rate": 3.726683736790022e-05,
+      "loss": 0.8192,
+      "step": 2660
+    },
+    {
+      "epoch": 1.733897202342225,
+      "grad_norm": 0.8701574206352234,
+      "learning_rate": 3.7221835465680024e-05,
+      "loss": 0.8072,
+      "step": 2665
+    },
+    {
+      "epoch": 1.7371502927781393,
+      "grad_norm": 0.7430034875869751,
+      "learning_rate": 3.717678146378357e-05,
+      "loss": 0.8147,
+      "step": 2670
+    },
+    {
+      "epoch": 1.7404033832140533,
+      "grad_norm": 0.7101485729217529,
+      "learning_rate": 3.7131675554268654e-05,
+      "loss": 0.8174,
+      "step": 2675
+    },
+    {
+      "epoch": 1.7436564736499673,
+      "grad_norm": 0.6719719767570496,
+      "learning_rate": 3.7086517929414346e-05,
+      "loss": 0.7912,
+      "step": 2680
+    },
+    {
+      "epoch": 1.7469095640858816,
+      "grad_norm": 0.8749985098838806,
+      "learning_rate": 3.70413087817202e-05,
+      "loss": 0.7914,
+      "step": 2685
+    },
+    {
+      "epoch": 1.7501626545217959,
+      "grad_norm": 0.6824137568473816,
+      "learning_rate": 3.699604830390537e-05,
+      "loss": 0.7974,
+      "step": 2690
+    },
+    {
+      "epoch": 1.7534157449577097,
+      "grad_norm": 0.6715524196624756,
+      "learning_rate": 3.695073668890785e-05,
+      "loss": 0.8134,
+      "step": 2695
+    },
+    {
+      "epoch": 1.756668835393624,
+      "grad_norm": 1.2192811965942383,
+      "learning_rate": 3.690537412988359e-05,
+      "loss": 0.8042,
+      "step": 2700
+    },
+    {
+      "epoch": 1.7599219258295382,
+      "grad_norm": 0.5974992513656616,
+      "learning_rate": 3.685996082020574e-05,
+      "loss": 0.7871,
+      "step": 2705
+    },
+    {
+      "epoch": 1.763175016265452,
+      "grad_norm": 0.7278191447257996,
+      "learning_rate": 3.681449695346376e-05,
+      "loss": 0.8104,
+      "step": 2710
+    },
+    {
+      "epoch": 1.7664281067013663,
+      "grad_norm": 0.662208080291748,
+      "learning_rate": 3.676898272346266e-05,
+      "loss": 0.7831,
+      "step": 2715
+    },
+    {
+      "epoch": 1.7696811971372806,
+      "grad_norm": 0.891823410987854,
+      "learning_rate": 3.6723418324222126e-05,
+      "loss": 0.8045,
+      "step": 2720
+    },
+    {
+      "epoch": 1.7729342875731944,
+      "grad_norm": 0.6170246005058289,
+      "learning_rate": 3.667780394997569e-05,
+      "loss": 0.7881,
+      "step": 2725
+    },
+    {
+      "epoch": 1.7761873780091086,
+      "grad_norm": 0.6385141015052795,
+      "learning_rate": 3.663213979516994e-05,
+      "loss": 0.7926,
+      "step": 2730
+    },
+    {
+      "epoch": 1.779440468445023,
+      "grad_norm": 0.827928900718689,
+      "learning_rate": 3.658642605446367e-05,
+      "loss": 0.7865,
+      "step": 2735
+    },
+    {
+      "epoch": 1.7826935588809367,
+      "grad_norm": 0.8710260987281799,
+      "learning_rate": 3.6540662922727034e-05,
+      "loss": 0.8045,
+      "step": 2740
+    },
+    {
+      "epoch": 1.785946649316851,
+      "grad_norm": 0.7606410980224609,
+      "learning_rate": 3.6494850595040745e-05,
+      "loss": 0.8041,
+      "step": 2745
+    },
+    {
+      "epoch": 1.7891997397527653,
+      "grad_norm": 0.598753809928894,
+      "learning_rate": 3.644898926669524e-05,
+      "loss": 0.8023,
+      "step": 2750
+    },
+    {
+      "epoch": 1.7924528301886793,
+      "grad_norm": 0.6432173848152161,
+      "learning_rate": 3.640307913318982e-05,
+      "loss": 0.8165,
+      "step": 2755
+    },
+    {
+      "epoch": 1.7957059206245933,
+      "grad_norm": 0.6266681551933289,
+      "learning_rate": 3.6357120390231825e-05,
+      "loss": 0.8133,
+      "step": 2760
+    },
+    {
+      "epoch": 1.7989590110605076,
+      "grad_norm": 1.0969923734664917,
+      "learning_rate": 3.6311113233735836e-05,
+      "loss": 0.8016,
+      "step": 2765
+    },
+    {
+      "epoch": 1.8022121014964216,
+      "grad_norm": 1.2453726530075073,
+      "learning_rate": 3.626505785982281e-05,
+      "loss": 0.7391,
+      "step": 2770
+    },
+    {
+      "epoch": 1.8054651919323357,
+      "grad_norm": 0.8763654828071594,
+      "learning_rate": 3.6218954464819224e-05,
+      "loss": 0.7857,
+      "step": 2775
+    },
+    {
+      "epoch": 1.80871828236825,
+      "grad_norm": 0.7538381218910217,
+      "learning_rate": 3.6172803245256284e-05,
+      "loss": 0.7857,
+      "step": 2780
+    },
+    {
+      "epoch": 1.811971372804164,
+      "grad_norm": 0.6540011763572693,
+      "learning_rate": 3.612660439786904e-05,
+      "loss": 0.7982,
+      "step": 2785
+    },
+    {
+      "epoch": 1.815224463240078,
+      "grad_norm": 0.7887677550315857,
+      "learning_rate": 3.608035811959561e-05,
+      "loss": 0.8012,
+      "step": 2790
+    },
+    {
+      "epoch": 1.8184775536759923,
+      "grad_norm": 0.6847140192985535,
+      "learning_rate": 3.603406460757627e-05,
+      "loss": 0.8011,
+      "step": 2795
+    },
+    {
+      "epoch": 1.8217306441119063,
+      "grad_norm": 0.5890961289405823,
+      "learning_rate": 3.598772405915264e-05,
+      "loss": 0.7934,
+      "step": 2800
+    },
+    {
+      "epoch": 1.8249837345478204,
+      "grad_norm": 0.5724996328353882,
+      "learning_rate": 3.594133667186688e-05,
+      "loss": 0.7992,
+      "step": 2805
+    },
+    {
+      "epoch": 1.8282368249837346,
+      "grad_norm": 0.6108303070068359,
+      "learning_rate": 3.58949026434608e-05,
+      "loss": 0.7798,
+      "step": 2810
+    },
+    {
+      "epoch": 1.8314899154196487,
+      "grad_norm": 1.2907392978668213,
+      "learning_rate": 3.584842217187503e-05,
+      "loss": 0.7767,
+      "step": 2815
+    },
+    {
+      "epoch": 1.8347430058555627,
+      "grad_norm": 0.6735864877700806,
+      "learning_rate": 3.580189545524818e-05,
+      "loss": 0.7884,
+      "step": 2820
+    },
+    {
+      "epoch": 1.837996096291477,
+      "grad_norm": 1.0391428470611572,
+      "learning_rate": 3.575532269191599e-05,
+      "loss": 0.7944,
+      "step": 2825
+    },
+    {
+      "epoch": 1.841249186727391,
+      "grad_norm": 0.7969436645507812,
+      "learning_rate": 3.57087040804105e-05,
+      "loss": 0.7756,
+      "step": 2830
+    },
+    {
+      "epoch": 1.844502277163305,
+      "grad_norm": 0.6109620332717896,
+      "learning_rate": 3.566203981945921e-05,
+      "loss": 0.802,
+      "step": 2835
+    },
+    {
+      "epoch": 1.8477553675992193,
+      "grad_norm": 0.6684165000915527,
+      "learning_rate": 3.561533010798418e-05,
+      "loss": 0.7942,
+      "step": 2840
+    },
+    {
+      "epoch": 1.8510084580351334,
+      "grad_norm": 0.6041696071624756,
+      "learning_rate": 3.556857514510123e-05,
+      "loss": 0.8121,
+      "step": 2845
+    },
+    {
+      "epoch": 1.8542615484710474,
+      "grad_norm": 0.844451904296875,
+      "learning_rate": 3.5521775130119095e-05,
+      "loss": 0.7941,
+      "step": 2850
+    },
+    {
+      "epoch": 1.8575146389069617,
+      "grad_norm": 0.8286609649658203,
+      "learning_rate": 3.547493026253854e-05,
+      "loss": 0.7955,
+      "step": 2855
+    },
+    {
+      "epoch": 1.8607677293428757,
+      "grad_norm": 0.5692376494407654,
+      "learning_rate": 3.542804074205155e-05,
+      "loss": 0.7736,
+      "step": 2860
+    },
+    {
+      "epoch": 1.8640208197787898,
+      "grad_norm": 0.80513995885849,
+      "learning_rate": 3.5381106768540426e-05,
+      "loss": 0.819,
+      "step": 2865
+    },
+    {
+      "epoch": 1.867273910214704,
+      "grad_norm": 0.748439371585846,
+      "learning_rate": 3.5334128542077004e-05,
+      "loss": 0.8042,
+      "step": 2870
+    },
+    {
+      "epoch": 1.870527000650618,
+      "grad_norm": 0.5762147903442383,
+      "learning_rate": 3.528710626292174e-05,
+      "loss": 0.7935,
+      "step": 2875
+    },
+    {
+      "epoch": 1.873780091086532,
+      "grad_norm": 0.8298909664154053,
+      "learning_rate": 3.5240040131522876e-05,
+      "loss": 0.796,
+      "step": 2880
+    },
+    {
+      "epoch": 1.8770331815224464,
+      "grad_norm": 0.805218517780304,
+      "learning_rate": 3.519293034851559e-05,
+      "loss": 0.777,
+      "step": 2885
+    },
+    {
+      "epoch": 1.8802862719583604,
+      "grad_norm": 0.617969810962677,
+      "learning_rate": 3.514577711472117e-05,
+      "loss": 0.7925,
+      "step": 2890
+    },
+    {
+      "epoch": 1.8835393623942744,
+      "grad_norm": 0.6062343120574951,
+      "learning_rate": 3.509858063114608e-05,
+      "loss": 0.7853,
+      "step": 2895
+    },
+    {
+      "epoch": 1.8867924528301887,
+      "grad_norm": 0.8101537823677063,
+      "learning_rate": 3.505134109898118e-05,
+      "loss": 0.7601,
+      "step": 2900
+    },
+    {
+      "epoch": 1.8900455432661027,
+      "grad_norm": 0.8260220885276794,
+      "learning_rate": 3.500405871960085e-05,
+      "loss": 0.7946,
+      "step": 2905
+    },
+    {
+      "epoch": 1.8932986337020168,
+      "grad_norm": 0.8925743699073792,
+      "learning_rate": 3.495673369456207e-05,
+      "loss": 0.799,
+      "step": 2910
+    },
+    {
+      "epoch": 1.896551724137931,
+      "grad_norm": 0.5829617977142334,
+      "learning_rate": 3.490936622560368e-05,
+      "loss": 0.8058,
+      "step": 2915
+    },
+    {
+      "epoch": 1.8998048145738453,
+      "grad_norm": 0.6162145137786865,
+      "learning_rate": 3.4861956514645386e-05,
+      "loss": 0.7715,
+      "step": 2920
+    },
+    {
+      "epoch": 1.9030579050097591,
+      "grad_norm": 0.641473114490509,
+      "learning_rate": 3.481450476378703e-05,
+      "loss": 0.7854,
+      "step": 2925
+    },
+    {
+      "epoch": 1.9063109954456734,
+      "grad_norm": 0.6128933429718018,
+      "learning_rate": 3.4767011175307595e-05,
+      "loss": 0.8211,
+      "step": 2930
+    },
+    {
+      "epoch": 1.9095640858815877,
+      "grad_norm": 0.6579800248146057,
+      "learning_rate": 3.4719475951664464e-05,
+      "loss": 0.793,
+      "step": 2935
+    },
+    {
+      "epoch": 1.9128171763175015,
+      "grad_norm": 0.7673735022544861,
+      "learning_rate": 3.4671899295492485e-05,
+      "loss": 0.7965,
+      "step": 2940
+    },
+    {
+      "epoch": 1.9160702667534157,
+      "grad_norm": 0.6426681876182556,
+      "learning_rate": 3.462428140960311e-05,
+      "loss": 0.7817,
+      "step": 2945
+    },
+    {
+      "epoch": 1.91932335718933,
+      "grad_norm": 1.0133545398712158,
+      "learning_rate": 3.4576622496983575e-05,
+      "loss": 0.806,
+      "step": 2950
+    },
+    {
+      "epoch": 1.9225764476252438,
+      "grad_norm": 0.5959199666976929,
+      "learning_rate": 3.452892276079599e-05,
+      "loss": 0.7939,
+      "step": 2955
+    },
+    {
+      "epoch": 1.925829538061158,
+      "grad_norm": 0.6069105267524719,
+      "learning_rate": 3.4481182404376485e-05,
+      "loss": 0.8011,
+      "step": 2960
+    },
+    {
+      "epoch": 1.9290826284970723,
+      "grad_norm": 0.6653848886489868,
+      "learning_rate": 3.443340163123437e-05,
+      "loss": 0.7864,
+      "step": 2965
+    },
+    {
+      "epoch": 1.9323357189329864,
+      "grad_norm": 0.8323044180870056,
+      "learning_rate": 3.4385580645051216e-05,
+      "loss": 0.8096,
+      "step": 2970
+    },
+    {
+      "epoch": 1.9355888093689004,
+      "grad_norm": 0.7751704454421997,
+      "learning_rate": 3.433771964968004e-05,
+      "loss": 0.8101,
+      "step": 2975
+    },
+    {
+      "epoch": 1.9388418998048147,
+      "grad_norm": 0.7757251262664795,
+      "learning_rate": 3.4289818849144384e-05,
+      "loss": 0.7871,
+      "step": 2980
+    },
+    {
+      "epoch": 1.9420949902407287,
+      "grad_norm": 0.6133014559745789,
+      "learning_rate": 3.424187844763751e-05,
+      "loss": 0.7759,
+      "step": 2985
+    },
+    {
+      "epoch": 1.9453480806766428,
+      "grad_norm": 0.6651121973991394,
+      "learning_rate": 3.419389864952145e-05,
+      "loss": 0.8079,
+      "step": 2990
+    },
+    {
+      "epoch": 1.948601171112557,
+      "grad_norm": 0.6148653626441956,
+      "learning_rate": 3.414587965932622e-05,
+      "loss": 0.8075,
+      "step": 2995
+    },
+    {
+      "epoch": 1.951854261548471,
+      "grad_norm": 0.7674732208251953,
+      "learning_rate": 3.409782168174887e-05,
+      "loss": 0.7836,
+      "step": 3000
+    },
+    {
+      "epoch": 1.9551073519843851,
+      "grad_norm": 0.7626878619194031,
+      "learning_rate": 3.404972492165267e-05,
+      "loss": 0.787,
+      "step": 3005
+    },
+    {
+      "epoch": 1.9583604424202994,
+      "grad_norm": 0.6008310914039612,
+      "learning_rate": 3.40015895840662e-05,
+      "loss": 0.8043,
+      "step": 3010
+    },
+    {
+      "epoch": 1.9616135328562134,
+      "grad_norm": 0.6407262682914734,
+      "learning_rate": 3.3953415874182495e-05,
+      "loss": 0.7847,
+      "step": 3015
+    },
+    {
+      "epoch": 1.9648666232921275,
+      "grad_norm": 0.8387401103973389,
+      "learning_rate": 3.390520399735818e-05,
+      "loss": 0.7885,
+      "step": 3020
+    },
+    {
+      "epoch": 1.9681197137280417,
+      "grad_norm": 0.9059749245643616,
+      "learning_rate": 3.385695415911253e-05,
+      "loss": 0.7885,
+      "step": 3025
+    },
+    {
+      "epoch": 1.9713728041639558,
+      "grad_norm": 0.7701427340507507,
+      "learning_rate": 3.38086665651267e-05,
+      "loss": 0.7951,
+      "step": 3030
+    },
+    {
+      "epoch": 1.9746258945998698,
+      "grad_norm": 0.5787619352340698,
+      "learning_rate": 3.376034142124277e-05,
+      "loss": 0.7692,
+      "step": 3035
+    },
+    {
+      "epoch": 1.977878985035784,
+      "grad_norm": 0.5660680532455444,
+      "learning_rate": 3.371197893346288e-05,
+      "loss": 0.7935,
+      "step": 3040
+    },
+    {
+      "epoch": 1.9811320754716981,
+      "grad_norm": 0.6145839095115662,
+      "learning_rate": 3.3663579307948365e-05,
+      "loss": 0.774,
+      "step": 3045
+    },
+    {
+      "epoch": 1.9843851659076122,
+      "grad_norm": 0.9014440178871155,
+      "learning_rate": 3.3615142751018894e-05,
+      "loss": 0.795,
+      "step": 3050
+    },
+    {
+      "epoch": 1.9876382563435264,
+      "grad_norm": 1.5400235652923584,
+      "learning_rate": 3.356666946915152e-05,
+      "loss": 0.8015,
+      "step": 3055
+    },
+    {
+      "epoch": 1.9908913467794405,
+      "grad_norm": 0.7585124373435974,
+      "learning_rate": 3.35181596689799e-05,
+      "loss": 0.7751,
+      "step": 3060
+    },
+    {
+      "epoch": 1.9941444372153545,
+      "grad_norm": 0.675399124622345,
+      "learning_rate": 3.3469613557293345e-05,
+      "loss": 0.7552,
+      "step": 3065
+    },
+    {
+      "epoch": 1.9973975276512688,
+      "grad_norm": 0.7362255454063416,
+      "learning_rate": 3.342103134103593e-05,
+      "loss": 0.7674,
+      "step": 3070
+    },
+    {
+      "epoch": 2.0,
+      "eval_f1": 0.8125089732634697,
+      "eval_loss": 0.419677734375,
+      "eval_precision": 0.812527609369082,
+      "eval_recall": 0.8124974017274719,
+      "eval_runtime": 298.2094,
+      "eval_samples_per_second": 1319.325,
+      "eval_steps_per_second": 1.291,
+      "step": 3074
+    },
+    {
+      "epoch": 2.000650618087183,
+      "grad_norm": 0.6836552023887634,
+      "learning_rate": 3.3372413227305684e-05,
+      "loss": 0.7623,
+      "step": 3075
+    },
+    {
+      "epoch": 2.003903708523097,
+      "grad_norm": 0.7960084676742554,
+      "learning_rate": 3.3323759423353615e-05,
+      "loss": 0.6671,
+      "step": 3080
+    },
+    {
+      "epoch": 2.007156798959011,
+      "grad_norm": 0.7370645999908447,
+      "learning_rate": 3.327507013658291e-05,
+      "loss": 0.6733,
+      "step": 3085
+    },
+    {
+      "epoch": 2.0104098893949254,
+      "grad_norm": 1.088987112045288,
+      "learning_rate": 3.3226345574548e-05,
+      "loss": 0.6668,
+      "step": 3090
+    },
+    {
+      "epoch": 2.013662979830839,
+      "grad_norm": 0.9499191045761108,
+      "learning_rate": 3.317758594495367e-05,
+      "loss": 0.6749,
+      "step": 3095
+    },
+    {
+      "epoch": 2.0169160702667535,
+      "grad_norm": 0.8411104083061218,
+      "learning_rate": 3.312879145565422e-05,
+      "loss": 0.6643,
+      "step": 3100
+    },
+    {
+      "epoch": 2.0201691607026677,
+      "grad_norm": 0.8027604818344116,
+      "learning_rate": 3.307996231465254e-05,
+      "loss": 0.6604,
+      "step": 3105
+    },
+    {
+      "epoch": 2.0234222511385815,
+      "grad_norm": 0.8328303694725037,
+      "learning_rate": 3.303109873009922e-05,
+      "loss": 0.6921,
+      "step": 3110
+    },
+    {
+      "epoch": 2.026675341574496,
+      "grad_norm": 0.8239777088165283,
+      "learning_rate": 3.298220091029171e-05,
+      "loss": 0.6619,
+      "step": 3115
+    },
+    {
+      "epoch": 2.02992843201041,
+      "grad_norm": 0.8349932432174683,
+      "learning_rate": 3.293326906367338e-05,
+      "loss": 0.6289,
+      "step": 3120
+    },
+    {
+      "epoch": 2.033181522446324,
+      "grad_norm": 1.271503210067749,
+      "learning_rate": 3.2884303398832634e-05,
+      "loss": 0.6643,
+      "step": 3125
+    },
+    {
+      "epoch": 2.036434612882238,
+      "grad_norm": 0.8596307039260864,
+      "learning_rate": 3.283530412450207e-05,
+      "loss": 0.6434,
+      "step": 3130
+    },
+    {
+      "epoch": 2.0396877033181524,
+      "grad_norm": 0.8263525366783142,
+      "learning_rate": 3.278627144955754e-05,
+      "loss": 0.6485,
+      "step": 3135
+    },
+    {
+      "epoch": 2.0429407937540662,
+      "grad_norm": 0.8425394892692566,
+      "learning_rate": 3.2737205583017286e-05,
+      "loss": 0.6428,
+      "step": 3140
+    },
+    {
+      "epoch": 2.0461938841899805,
+      "grad_norm": 1.253609299659729,
+      "learning_rate": 3.268810673404102e-05,
+      "loss": 0.6427,
+      "step": 3145
+    },
+    {
+      "epoch": 2.0494469746258948,
+      "grad_norm": 0.8926024436950684,
+      "learning_rate": 3.2638975111929084e-05,
+      "loss": 0.6748,
+      "step": 3150
+    },
+    {
+      "epoch": 2.0527000650618086,
+      "grad_norm": 0.959561288356781,
+      "learning_rate": 3.25898109261215e-05,
+      "loss": 0.6542,
+      "step": 3155
+    },
+    {
+      "epoch": 2.055953155497723,
+      "grad_norm": 1.0102553367614746,
+      "learning_rate": 3.254061438619711e-05,
+      "loss": 0.6661,
+      "step": 3160
+    },
+    {
+      "epoch": 2.059206245933637,
+      "grad_norm": 1.3455661535263062,
+      "learning_rate": 3.249138570187268e-05,
+      "loss": 0.6656,
+      "step": 3165
+    },
+    {
+      "epoch": 2.062459336369551,
+      "grad_norm": 1.2887942790985107,
+      "learning_rate": 3.244212508300201e-05,
+      "loss": 0.6641,
+      "step": 3170
+    },
+    {
+      "epoch": 2.065712426805465,
+      "grad_norm": 0.8388302326202393,
+      "learning_rate": 3.239283273957502e-05,
+      "loss": 0.6493,
+      "step": 3175
+    },
+    {
+      "epoch": 2.0689655172413794,
+      "grad_norm": 1.096605658531189,
+      "learning_rate": 3.2343508881716874e-05,
+      "loss": 0.6664,
+      "step": 3180
+    },
+    {
+      "epoch": 2.0722186076772933,
+      "grad_norm": 0.8951081037521362,
+      "learning_rate": 3.229415371968706e-05,
+      "loss": 0.6772,
+      "step": 3185
+    },
+    {
+      "epoch": 2.0754716981132075,
+      "grad_norm": 0.8592993021011353,
+      "learning_rate": 3.2244767463878525e-05,
+      "loss": 0.6688,
+      "step": 3190
+    },
+    {
+      "epoch": 2.078724788549122,
+      "grad_norm": 0.8971413969993591,
+      "learning_rate": 3.219535032481679e-05,
+      "loss": 0.6739,
+      "step": 3195
+    },
+    {
+      "epoch": 2.0819778789850356,
+      "grad_norm": 2.3976173400878906,
+      "learning_rate": 3.214590251315896e-05,
+      "loss": 0.663,
+      "step": 3200
+    },
+    {
+      "epoch": 2.08523096942095,
+      "grad_norm": 1.024121642112732,
+      "learning_rate": 3.209642423969296e-05,
+      "loss": 0.6618,
+      "step": 3205
+    },
+    {
+      "epoch": 2.088484059856864,
+      "grad_norm": 0.8724255561828613,
+      "learning_rate": 3.204691571533652e-05,
+      "loss": 0.6459,
+      "step": 3210
+    },
+    {
+      "epoch": 2.091737150292778,
+      "grad_norm": 0.9145408272743225,
+      "learning_rate": 3.1997377151136356e-05,
+      "loss": 0.6759,
+      "step": 3215
+    },
+    {
+      "epoch": 2.094990240728692,
+      "grad_norm": 1.3194828033447266,
+      "learning_rate": 3.194780875826723e-05,
+      "loss": 0.6687,
+      "step": 3220
+    },
+    {
+      "epoch": 2.0982433311646065,
+      "grad_norm": 1.0560972690582275,
+      "learning_rate": 3.189821074803103e-05,
+      "loss": 0.6554,
+      "step": 3225
+    },
+    {
+      "epoch": 2.1014964216005203,
+      "grad_norm": 0.8973615765571594,
+      "learning_rate": 3.1848583331855954e-05,
+      "loss": 0.6554,
+      "step": 3230
+    },
+    {
+      "epoch": 2.1047495120364346,
+      "grad_norm": 1.0747287273406982,
+      "learning_rate": 3.1808860368799674e-05,
+      "loss": 0.6729,
+      "step": 3235
+    },
+    {
+      "epoch": 2.108002602472349,
+      "grad_norm": 1.3772774934768677,
+      "learning_rate": 3.1759180555133126e-05,
+      "loss": 0.6449,
+      "step": 3240
+    },
+    {
+      "epoch": 2.1112556929082626,
+      "grad_norm": 0.8960999250411987,
+      "learning_rate": 3.170947192819057e-05,
+      "loss": 0.6678,
+      "step": 3245
+    },
+    {
+      "epoch": 2.114508783344177,
+      "grad_norm": 0.891995906829834,
+      "learning_rate": 3.165973469987168e-05,
+      "loss": 0.6542,
+      "step": 3250
+    },
+    {
+      "epoch": 2.117761873780091,
+      "grad_norm": 0.8801698684692383,
+      "learning_rate": 3.160996908219812e-05,
+      "loss": 0.6588,
+      "step": 3255
+    },
+    {
+      "epoch": 2.121014964216005,
+      "grad_norm": 1.0145418643951416,
+      "learning_rate": 3.1560175287312534e-05,
+      "loss": 0.6561,
+      "step": 3260
+    },
+    {
+      "epoch": 2.1242680546519193,
+      "grad_norm": 1.2028182744979858,
+      "learning_rate": 3.151035352747767e-05,
+      "loss": 0.6689,
+      "step": 3265
+    },
+    {
+      "epoch": 2.1275211450878335,
+      "grad_norm": 0.9466059803962708,
+      "learning_rate": 3.1460504015075525e-05,
+      "loss": 0.6518,
+      "step": 3270
+    },
+    {
+      "epoch": 2.130774235523748,
+      "grad_norm": 0.938401460647583,
+      "learning_rate": 3.141062696260636e-05,
+      "loss": 0.6649,
+      "step": 3275
+    },
+    {
+      "epoch": 2.1340273259596616,
+      "grad_norm": 2.2599849700927734,
+      "learning_rate": 3.1360722582687876e-05,
+      "loss": 0.6607,
+      "step": 3280
+    },
+    {
+      "epoch": 2.137280416395576,
+      "grad_norm": 0.8594357371330261,
+      "learning_rate": 3.1310791088054225e-05,
+      "loss": 0.6572,
+      "step": 3285
+    },
+    {
+      "epoch": 2.14053350683149,
+      "grad_norm": 1.1824400424957275,
+      "learning_rate": 3.126083269155517e-05,
+      "loss": 0.6812,
+      "step": 3290
+    },
+    {
+      "epoch": 2.143786597267404,
+      "grad_norm": 1.1186939477920532,
+      "learning_rate": 3.121084760615515e-05,
+      "loss": 0.6653,
+      "step": 3295
+    },
+    {
+      "epoch": 2.147039687703318,
+      "grad_norm": 0.9989317059516907,
+      "learning_rate": 3.116083604493236e-05,
+      "loss": 0.6934,
+      "step": 3300
+    },
+    {
+      "epoch": 2.1502927781392325,
+      "grad_norm": 0.8350909948348999,
+      "learning_rate": 3.111079822107788e-05,
+      "loss": 0.6447,
+      "step": 3305
+    },
+    {
+      "epoch": 2.1535458685751463,
+      "grad_norm": 0.9244917035102844,
+      "learning_rate": 3.106073434789472e-05,
+      "loss": 0.6492,
+      "step": 3310
+    },
+    {
+      "epoch": 2.1567989590110606,
+      "grad_norm": 0.9779621362686157,
+      "learning_rate": 3.1010644638796956e-05,
+      "loss": 0.6679,
+      "step": 3315
+    },
+    {
+      "epoch": 2.160052049446975,
+      "grad_norm": 1.2001359462738037,
+      "learning_rate": 3.096052930730877e-05,
+      "loss": 0.6752,
+      "step": 3320
+    },
+    {
+      "epoch": 2.1633051398828886,
+      "grad_norm": 0.8512766361236572,
+      "learning_rate": 3.091038856706361e-05,
+      "loss": 0.6601,
+      "step": 3325
+    },
+    {
+      "epoch": 2.166558230318803,
+      "grad_norm": 3.723388433456421,
+      "learning_rate": 3.086022263180318e-05,
+      "loss": 0.6976,
+      "step": 3330
+    },
+    {
+      "epoch": 2.169811320754717,
+      "grad_norm": 0.9627018570899963,
+      "learning_rate": 3.081003171537665e-05,
+      "loss": 0.666,
+      "step": 3335
+    },
+    {
+      "epoch": 2.173064411190631,
+      "grad_norm": 0.9980655312538147,
+      "learning_rate": 3.075981603173963e-05,
+      "loss": 0.6729,
+      "step": 3340
+    },
+    {
+      "epoch": 2.1763175016265452,
+      "grad_norm": 1.0389659404754639,
+      "learning_rate": 3.070957579495333e-05,
+      "loss": 0.6772,
+      "step": 3345
+    },
+    {
+      "epoch": 2.1795705920624595,
+      "grad_norm": 1.0569038391113281,
+      "learning_rate": 3.065931121918364e-05,
+      "loss": 0.6711,
+      "step": 3350
+    },
+    {
+      "epoch": 2.1828236824983733,
+      "grad_norm": 1.2096091508865356,
+      "learning_rate": 3.060902251870017e-05,
+      "loss": 0.6699,
+      "step": 3355
+    },
+    {
+      "epoch": 2.1860767729342876,
+      "grad_norm": 1.1020735502243042,
+      "learning_rate": 3.0558709907875385e-05,
+      "loss": 0.6426,
+      "step": 3360
+    },
+    {
+      "epoch": 2.189329863370202,
+      "grad_norm": 0.9492762088775635,
+      "learning_rate": 3.0508373601183695e-05,
+      "loss": 0.6712,
+      "step": 3365
+    },
+    {
+      "epoch": 2.1925829538061157,
+      "grad_norm": 0.9105232954025269,
+      "learning_rate": 3.045801381320048e-05,
+      "loss": 0.6722,
+      "step": 3370
+    },
+    {
+      "epoch": 2.19583604424203,
+      "grad_norm": 0.8517335653305054,
+      "learning_rate": 3.0407630758601256e-05,
+      "loss": 0.6727,
+      "step": 3375
+    },
+    {
+      "epoch": 2.199089134677944,
+      "grad_norm": 0.9834687113761902,
+      "learning_rate": 3.035722465216071e-05,
+      "loss": 0.6581,
+      "step": 3380
+    },
+    {
+      "epoch": 2.202342225113858,
+      "grad_norm": 1.1220489740371704,
+      "learning_rate": 3.030679570875177e-05,
+      "loss": 0.6648,
+      "step": 3385
+    },
+    {
+      "epoch": 2.2055953155497723,
+      "grad_norm": 1.169081211090088,
+      "learning_rate": 3.0256344143344765e-05,
+      "loss": 0.6567,
+      "step": 3390
+    },
+    {
+      "epoch": 2.2088484059856865,
+      "grad_norm": 1.303977131843567,
+      "learning_rate": 3.02058701710064e-05,
+      "loss": 0.6429,
+      "step": 3395
+    },
+    {
+      "epoch": 2.2121014964216004,
+      "grad_norm": 1.1148353815078735,
+      "learning_rate": 3.0155374006898946e-05,
+      "loss": 0.686,
+      "step": 3400
+    },
+    {
+      "epoch": 2.2153545868575146,
+      "grad_norm": 1.0681456327438354,
+      "learning_rate": 3.010485586627924e-05,
+      "loss": 0.6585,
+      "step": 3405
+    },
+    {
+      "epoch": 2.218607677293429,
+      "grad_norm": 1.259251356124878,
+      "learning_rate": 3.005431596449782e-05,
+      "loss": 0.6525,
+      "step": 3410
+    },
+    {
+      "epoch": 2.2218607677293427,
+      "grad_norm": 1.041143774986267,
+      "learning_rate": 3.0003754516997984e-05,
+      "loss": 0.6983,
+      "step": 3415
+    },
+    {
+      "epoch": 2.225113858165257,
+      "grad_norm": 0.8610508441925049,
+      "learning_rate": 2.9953171739314867e-05,
+      "loss": 0.6491,
+      "step": 3420
+    },
+    {
+      "epoch": 2.2283669486011712,
+      "grad_norm": 1.259366750717163,
+      "learning_rate": 2.9902567847074537e-05,
+      "loss": 0.6771,
+      "step": 3425
+    },
+    {
+      "epoch": 2.231620039037085,
+      "grad_norm": 0.8394737243652344,
+      "learning_rate": 2.9851943055993088e-05,
+      "loss": 0.6659,
+      "step": 3430
+    },
+    {
+      "epoch": 2.2348731294729993,
+      "grad_norm": 0.8857366442680359,
+      "learning_rate": 2.980129758187567e-05,
+      "loss": 0.6525,
+      "step": 3435
+    },
+    {
+      "epoch": 2.2381262199089136,
+      "grad_norm": 1.0085824728012085,
+      "learning_rate": 2.9750631640615617e-05,
+      "loss": 0.6437,
+      "step": 3440
+    },
+    {
+      "epoch": 2.2413793103448274,
+      "grad_norm": 0.8579273819923401,
+      "learning_rate": 2.969994544819352e-05,
+      "loss": 0.6583,
+      "step": 3445
+    },
+    {
+      "epoch": 2.2446324007807417,
+      "grad_norm": 1.7075871229171753,
+      "learning_rate": 2.9649239220676285e-05,
+      "loss": 0.6572,
+      "step": 3450
+    },
+    {
+      "epoch": 2.247885491216656,
+      "grad_norm": 1.0663272142410278,
+      "learning_rate": 2.959851317421622e-05,
+      "loss": 0.656,
+      "step": 3455
+    },
+    {
+      "epoch": 2.2511385816525697,
+      "grad_norm": 1.072970986366272,
+      "learning_rate": 2.9547767525050142e-05,
+      "loss": 0.6809,
+      "step": 3460
+    },
+    {
+      "epoch": 2.254391672088484,
+      "grad_norm": 0.9220450520515442,
+      "learning_rate": 2.9497002489498393e-05,
+      "loss": 0.6804,
+      "step": 3465
+    },
+    {
+      "epoch": 2.2576447625243983,
+      "grad_norm": 1.0572975873947144,
+      "learning_rate": 2.9446218283964e-05,
+      "loss": 0.6561,
+      "step": 3470
+    },
+    {
+      "epoch": 2.260897852960312,
+      "grad_norm": 1.0907764434814453,
+      "learning_rate": 2.939541512493167e-05,
+      "loss": 0.6555,
+      "step": 3475
+    },
+    {
+      "epoch": 2.2641509433962264,
+      "grad_norm": 0.9391184449195862,
+      "learning_rate": 2.9344593228966925e-05,
+      "loss": 0.6512,
+      "step": 3480
+    },
+    {
+      "epoch": 2.2674040338321406,
+      "grad_norm": 0.9960388541221619,
+      "learning_rate": 2.929375281271517e-05,
+      "loss": 0.6694,
+      "step": 3485
+    },
+    {
+      "epoch": 2.2706571242680544,
+      "grad_norm": 0.9363919496536255,
+      "learning_rate": 2.9242894092900725e-05,
+      "loss": 0.6748,
+      "step": 3490
+    },
+    {
+      "epoch": 2.2739102147039687,
+      "grad_norm": 0.9546090364456177,
+      "learning_rate": 2.9192017286325973e-05,
+      "loss": 0.6509,
+      "step": 3495
+    },
+    {
+      "epoch": 2.277163305139883,
+      "grad_norm": 1.040223240852356,
+      "learning_rate": 2.9141122609870364e-05,
+      "loss": 0.6519,
+      "step": 3500
+    },
+    {
+      "epoch": 2.280416395575797,
+      "grad_norm": 0.921008288860321,
+      "learning_rate": 2.909021028048955e-05,
+      "loss": 0.6645,
+      "step": 3505
+    },
+    {
+      "epoch": 2.283669486011711,
+      "grad_norm": 1.0411720275878906,
+      "learning_rate": 2.9039280515214428e-05,
+      "loss": 0.6607,
+      "step": 3510
+    },
+    {
+      "epoch": 2.2869225764476253,
+      "grad_norm": 0.9046562910079956,
+      "learning_rate": 2.898833353115021e-05,
+      "loss": 0.6588,
+      "step": 3515
+    },
+    {
+      "epoch": 2.290175666883539,
+      "grad_norm": 0.932158350944519,
+      "learning_rate": 2.8937369545475517e-05,
+      "loss": 0.6718,
+      "step": 3520
+    },
+    {
+      "epoch": 2.2934287573194534,
+      "grad_norm": 1.0239461660385132,
+      "learning_rate": 2.8886388775441457e-05,
+      "loss": 0.6896,
+      "step": 3525
+    },
+    {
+      "epoch": 2.2966818477553677,
+      "grad_norm": 2.0023818016052246,
+      "learning_rate": 2.8835391438370664e-05,
+      "loss": 0.6653,
+      "step": 3530
+    },
+    {
+      "epoch": 2.2999349381912815,
+      "grad_norm": 1.012035608291626,
+      "learning_rate": 2.8784377751656416e-05,
+      "loss": 0.657,
+      "step": 3535
+    },
+    {
+      "epoch": 2.3031880286271957,
+      "grad_norm": 0.922524631023407,
+      "learning_rate": 2.873334793276166e-05,
+      "loss": 0.6805,
+      "step": 3540
+    },
+    {
+      "epoch": 2.30644111906311,
+      "grad_norm": 1.0384365320205688,
+      "learning_rate": 2.8682302199218148e-05,
+      "loss": 0.6643,
+      "step": 3545
+    },
+    {
+      "epoch": 2.3096942094990243,
+      "grad_norm": 2.3190078735351562,
+      "learning_rate": 2.8631240768625446e-05,
+      "loss": 0.6761,
+      "step": 3550
+    },
+    {
+      "epoch": 2.312947299934938,
+      "grad_norm": 1.3237860202789307,
+      "learning_rate": 2.8580163858650038e-05,
+      "loss": 0.6532,
+      "step": 3555
+    },
+    {
+      "epoch": 2.3162003903708523,
+      "grad_norm": 1.0602989196777344,
+      "learning_rate": 2.85290716870244e-05,
+      "loss": 0.6546,
+      "step": 3560
+    },
+    {
+      "epoch": 2.3194534808067666,
+      "grad_norm": 1.1202350854873657,
+      "learning_rate": 2.8477964471546077e-05,
+      "loss": 0.6703,
+      "step": 3565
+    },
+    {
+      "epoch": 2.3227065712426804,
+      "grad_norm": 0.7759214043617249,
+      "learning_rate": 2.8426842430076712e-05,
+      "loss": 0.6569,
+      "step": 3570
+    },
+    {
+      "epoch": 2.3259596616785947,
+      "grad_norm": 1.013677716255188,
+      "learning_rate": 2.8375705780541173e-05,
+      "loss": 0.6719,
+      "step": 3575
+    },
+    {
+      "epoch": 2.329212752114509,
+      "grad_norm": 0.9612709283828735,
+      "learning_rate": 2.8324554740926594e-05,
+      "loss": 0.6685,
+      "step": 3580
+    },
+    {
+      "epoch": 2.3324658425504228,
+      "grad_norm": 0.9653801321983337,
+      "learning_rate": 2.827338952928146e-05,
+      "loss": 0.6578,
+      "step": 3585
+    },
+    {
+      "epoch": 2.335718932986337,
+      "grad_norm": 0.9308706521987915,
+      "learning_rate": 2.8222210363714653e-05,
+      "loss": 0.6446,
+      "step": 3590
+    },
+    {
+      "epoch": 2.3389720234222513,
+      "grad_norm": 0.9803590774536133,
+      "learning_rate": 2.8171017462394546e-05,
+      "loss": 0.6395,
+      "step": 3595
+    },
+    {
+      "epoch": 2.342225113858165,
+      "grad_norm": 1.5392948389053345,
+      "learning_rate": 2.8119811043548063e-05,
+      "loss": 0.6452,
+      "step": 3600
+    },
+    {
+      "epoch": 2.3454782042940794,
+      "grad_norm": 1.1688051223754883,
+      "learning_rate": 2.806859132545975e-05,
+      "loss": 0.6619,
+      "step": 3605
+    },
+    {
+      "epoch": 2.3487312947299936,
+      "grad_norm": 2.6195671558380127,
+      "learning_rate": 2.801735852647086e-05,
+      "loss": 0.6603,
+      "step": 3610
+    },
+    {
+      "epoch": 2.3519843851659075,
+      "grad_norm": 0.9085677266120911,
+      "learning_rate": 2.79661128649784e-05,
+      "loss": 0.6562,
+      "step": 3615
+    },
+    {
+      "epoch": 2.3552374756018217,
+      "grad_norm": 1.1864854097366333,
+      "learning_rate": 2.791485455943419e-05,
+      "loss": 0.6566,
+      "step": 3620
+    },
+    {
+      "epoch": 2.358490566037736,
+      "grad_norm": 1.0113550424575806,
+      "learning_rate": 2.7863583828343964e-05,
+      "loss": 0.6555,
+      "step": 3625
+    },
+    {
+      "epoch": 2.36174365647365,
+      "grad_norm": 0.8967930674552917,
+      "learning_rate": 2.7812300890266442e-05,
+      "loss": 0.6351,
+      "step": 3630
+    },
+    {
+      "epoch": 2.364996746909564,
+      "grad_norm": 0.9505155682563782,
+      "learning_rate": 2.7761005963812337e-05,
+      "loss": 0.6717,
+      "step": 3635
+    },
+    {
+      "epoch": 2.3682498373454783,
+      "grad_norm": 17.054515838623047,
+      "learning_rate": 2.7709699267643503e-05,
+      "loss": 0.6866,
+      "step": 3640
+    },
+    {
+      "epoch": 2.371502927781392,
+      "grad_norm": 1.044188141822815,
+      "learning_rate": 2.7658381020471964e-05,
+      "loss": 0.6564,
+      "step": 3645
+    },
+    {
+      "epoch": 2.3747560182173064,
+      "grad_norm": 1.0670133829116821,
+      "learning_rate": 2.7607051441058958e-05,
+      "loss": 0.677,
+      "step": 3650
+    },
+    {
+      "epoch": 2.3780091086532207,
+      "grad_norm": 0.9037075638771057,
+      "learning_rate": 2.7555710748214064e-05,
+      "loss": 0.6675,
+      "step": 3655
+    },
+    {
+      "epoch": 2.3812621990891345,
+      "grad_norm": 1.2479665279388428,
+      "learning_rate": 2.75043591607942e-05,
+      "loss": 0.6359,
+      "step": 3660
+    },
+    {
+      "epoch": 2.3845152895250488,
+      "grad_norm": 1.097076654434204,
+      "learning_rate": 2.7452996897702765e-05,
+      "loss": 0.6477,
+      "step": 3665
+    },
+    {
+      "epoch": 2.387768379960963,
+      "grad_norm": 0.9353396892547607,
+      "learning_rate": 2.7401624177888636e-05,
+      "loss": 0.6452,
+      "step": 3670
+    },
+    {
+      "epoch": 2.391021470396877,
+      "grad_norm": 1.2035560607910156,
+      "learning_rate": 2.7350241220345274e-05,
+      "loss": 0.6522,
+      "step": 3675
+    },
+    {
+      "epoch": 2.394274560832791,
+      "grad_norm": 1.0560545921325684,
+      "learning_rate": 2.729884824410979e-05,
+      "loss": 0.6655,
+      "step": 3680
+    },
+    {
+      "epoch": 2.3975276512687054,
+      "grad_norm": 1.2608094215393066,
+      "learning_rate": 2.724744546826199e-05,
+      "loss": 0.6566,
+      "step": 3685
+    },
+    {
+      "epoch": 2.4007807417046196,
+      "grad_norm": 1.0126628875732422,
+      "learning_rate": 2.719603311192347e-05,
+      "loss": 0.6596,
+      "step": 3690
+    },
+    {
+      "epoch": 2.4040338321405335,
+      "grad_norm": 0.8379278182983398,
+      "learning_rate": 2.7144611394256653e-05,
+      "loss": 0.6581,
+      "step": 3695
+    },
+    {
+      "epoch": 2.4072869225764477,
+      "grad_norm": 1.0668292045593262,
+      "learning_rate": 2.7093180534463863e-05,
+      "loss": 0.6623,
+      "step": 3700
+    },
+    {
+      "epoch": 2.410540013012362,
+      "grad_norm": 2.791311264038086,
+      "learning_rate": 2.7041740751786408e-05,
+      "loss": 0.6364,
+      "step": 3705
+    },
+    {
+      "epoch": 2.413793103448276,
+      "grad_norm": 1.5181926488876343,
+      "learning_rate": 2.6990292265503646e-05,
+      "loss": 0.6522,
+      "step": 3710
+    },
+    {
+      "epoch": 2.41704619388419,
+      "grad_norm": 1.028696894645691,
+      "learning_rate": 2.6938835294931996e-05,
+      "loss": 0.6755,
+      "step": 3715
+    },
+    {
+      "epoch": 2.4202992843201043,
+      "grad_norm": 2.595792531967163,
+      "learning_rate": 2.6887370059424078e-05,
+      "loss": 0.6704,
+      "step": 3720
+    },
+    {
+      "epoch": 2.423552374756018,
+      "grad_norm": 1.041309118270874,
+      "learning_rate": 2.6835896778367738e-05,
+      "loss": 0.6489,
+      "step": 3725
+    },
+    {
+      "epoch": 2.4268054651919324,
+      "grad_norm": 1.087664246559143,
+      "learning_rate": 2.6784415671185104e-05,
+      "loss": 0.6521,
+      "step": 3730
+    },
+    {
+      "epoch": 2.4300585556278467,
+      "grad_norm": 1.149911880493164,
+      "learning_rate": 2.6732926957331688e-05,
+      "loss": 0.6461,
+      "step": 3735
+    },
+    {
+      "epoch": 2.4333116460637605,
+      "grad_norm": 0.9765056371688843,
+      "learning_rate": 2.668143085629541e-05,
+      "loss": 0.6408,
+      "step": 3740
+    },
+    {
+      "epoch": 2.4365647364996748,
+      "grad_norm": 1.053946614265442,
+      "learning_rate": 2.6629927587595688e-05,
+      "loss": 0.658,
+      "step": 3745
+    },
+    {
+      "epoch": 2.439817826935589,
+      "grad_norm": 1.1712669134140015,
+      "learning_rate": 2.65784173707825e-05,
+      "loss": 0.6504,
+      "step": 3750
+    },
+    {
+      "epoch": 2.443070917371503,
+      "grad_norm": 0.9881893396377563,
+      "learning_rate": 2.6526900425435425e-05,
+      "loss": 0.6709,
+      "step": 3755
+    },
+    {
+      "epoch": 2.446324007807417,
+      "grad_norm": 0.8471539616584778,
+      "learning_rate": 2.6475376971162734e-05,
+      "loss": 0.6754,
+      "step": 3760
+    },
+    {
+      "epoch": 2.4495770982433314,
+      "grad_norm": 0.9622436165809631,
+      "learning_rate": 2.642384722760046e-05,
+      "loss": 0.6597,
+      "step": 3765
+    },
+    {
+      "epoch": 2.452830188679245,
+      "grad_norm": 0.7975105047225952,
+      "learning_rate": 2.6372311414411427e-05,
+      "loss": 0.662,
+      "step": 3770
+    },
+    {
+      "epoch": 2.4560832791151594,
+      "grad_norm": 1.1026084423065186,
+      "learning_rate": 2.6320769751284335e-05,
+      "loss": 0.6651,
+      "step": 3775
+    },
+    {
+      "epoch": 2.4593363695510737,
+      "grad_norm": 1.0752489566802979,
+      "learning_rate": 2.6269222457932824e-05,
+      "loss": 0.6366,
+      "step": 3780
+    },
+    {
+      "epoch": 2.4625894599869875,
+      "grad_norm": 0.8716150522232056,
+      "learning_rate": 2.621766975409453e-05,
+      "loss": 0.6653,
+      "step": 3785
+    },
+    {
+      "epoch": 2.465842550422902,
+      "grad_norm": 1.0649757385253906,
+      "learning_rate": 2.616611185953018e-05,
+      "loss": 0.6869,
+      "step": 3790
+    },
+    {
+      "epoch": 2.469095640858816,
+      "grad_norm": 1.258510947227478,
+      "learning_rate": 2.6114548994022576e-05,
+      "loss": 0.6564,
+      "step": 3795
+    },
+    {
+      "epoch": 2.47234873129473,
+      "grad_norm": 0.9622224569320679,
+      "learning_rate": 2.6062981377375762e-05,
+      "loss": 0.6615,
+      "step": 3800
+    },
+    {
+      "epoch": 2.475601821730644,
+      "grad_norm": 2.6835622787475586,
+      "learning_rate": 2.6011409229414003e-05,
+      "loss": 0.6793,
+      "step": 3805
+    },
+    {
+      "epoch": 2.4788549121665584,
+      "grad_norm": 1.3989510536193848,
+      "learning_rate": 2.59598327699809e-05,
+      "loss": 0.6462,
+      "step": 3810
+    },
+    {
+      "epoch": 2.482108002602472,
+      "grad_norm": 0.9427128434181213,
+      "learning_rate": 2.5908252218938423e-05,
+      "loss": 0.6489,
+      "step": 3815
+    },
+    {
+      "epoch": 2.4853610930383865,
+      "grad_norm": 1.0668915510177612,
+      "learning_rate": 2.585666779616598e-05,
+      "loss": 0.6633,
+      "step": 3820
+    },
+    {
+      "epoch": 2.4886141834743007,
+      "grad_norm": 0.9310431480407715,
+      "learning_rate": 2.5805079721559494e-05,
+      "loss": 0.6798,
+      "step": 3825
+    },
+    {
+      "epoch": 2.4918672739102146,
+      "grad_norm": 1.0935829877853394,
+      "learning_rate": 2.5753488215030448e-05,
+      "loss": 0.669,
+      "step": 3830
+    },
+    {
+      "epoch": 2.495120364346129,
+      "grad_norm": 0.8651193380355835,
+      "learning_rate": 2.5701893496504953e-05,
+      "loss": 0.6789,
+      "step": 3835
+    },
+    {
+      "epoch": 2.498373454782043,
+      "grad_norm": 1.0802959203720093,
+      "learning_rate": 2.5650295785922817e-05,
+      "loss": 0.6656,
+      "step": 3840
+    },
+    {
+      "epoch": 2.501626545217957,
+      "grad_norm": 1.0248281955718994,
+      "learning_rate": 2.5598695303236615e-05,
+      "loss": 0.6506,
+      "step": 3845
+    },
+    {
+      "epoch": 2.504879635653871,
+      "grad_norm": 1.0692211389541626,
+      "learning_rate": 2.5547092268410703e-05,
+      "loss": 0.6667,
+      "step": 3850
+    },
+    {
+      "epoch": 2.5081327260897854,
+      "grad_norm": 1.2791413068771362,
+      "learning_rate": 2.5495486901420362e-05,
+      "loss": 0.6506,
+      "step": 3855
+    },
+    {
+      "epoch": 2.5113858165256993,
+      "grad_norm": 1.1521552801132202,
+      "learning_rate": 2.5443879422250767e-05,
+      "loss": 0.6698,
+      "step": 3860
+    },
+    {
+      "epoch": 2.5146389069616135,
+      "grad_norm": 0.9791963696479797,
+      "learning_rate": 2.539227005089614e-05,
+      "loss": 0.6732,
+      "step": 3865
+    },
+    {
+      "epoch": 2.517891997397528,
+      "grad_norm": 0.9954161643981934,
+      "learning_rate": 2.5340659007358742e-05,
+      "loss": 0.6599,
+      "step": 3870
+    },
+    {
+      "epoch": 2.5211450878334416,
+      "grad_norm": 1.1552543640136719,
+      "learning_rate": 2.5289046511647972e-05,
+      "loss": 0.6849,
+      "step": 3875
+    },
+    {
+      "epoch": 2.524398178269356,
+      "grad_norm": 0.9102405905723572,
+      "learning_rate": 2.523743278377943e-05,
+      "loss": 0.6375,
+      "step": 3880
+    },
+    {
+      "epoch": 2.52765126870527,
+      "grad_norm": 1.0647556781768799,
+      "learning_rate": 2.518581804377394e-05,
+      "loss": 0.6585,
+      "step": 3885
+    },
+    {
+      "epoch": 2.530904359141184,
+      "grad_norm": 1.022453784942627,
+      "learning_rate": 2.5134202511656658e-05,
+      "loss": 0.6667,
+      "step": 3890
+    },
+    {
+      "epoch": 2.534157449577098,
+      "grad_norm": 0.9109323024749756,
+      "learning_rate": 2.5082586407456134e-05,
+      "loss": 0.6754,
+      "step": 3895
+    },
+    {
+      "epoch": 2.5374105400130125,
+      "grad_norm": 0.8672389388084412,
+      "learning_rate": 2.5030969951203316e-05,
+      "loss": 0.6432,
+      "step": 3900
+    },
+    {
+      "epoch": 2.5406636304489263,
+      "grad_norm": 1.022900938987732,
+      "learning_rate": 2.4979353362930685e-05,
+      "loss": 0.6512,
+      "step": 3905
+    },
+    {
+      "epoch": 2.5439167208848406,
+      "grad_norm": 0.9560525417327881,
+      "learning_rate": 2.492773686267128e-05,
+      "loss": 0.6528,
+      "step": 3910
+    },
+    {
+      "epoch": 2.547169811320755,
+      "grad_norm": 1.1724059581756592,
+      "learning_rate": 2.4876120670457754e-05,
+      "loss": 0.7026,
+      "step": 3915
+    },
+    {
+      "epoch": 2.5504229017566686,
+      "grad_norm": 0.9830509424209595,
+      "learning_rate": 2.482450500632145e-05,
+      "loss": 0.6411,
+      "step": 3920
+    },
+    {
+      "epoch": 2.553675992192583,
+      "grad_norm": 0.8978729844093323,
+      "learning_rate": 2.477289009029147e-05,
+      "loss": 0.6656,
+      "step": 3925
+    },
+    {
+      "epoch": 2.556929082628497,
+      "grad_norm": 2.0030722618103027,
+      "learning_rate": 2.4721276142393714e-05,
+      "loss": 0.6554,
+      "step": 3930
+    },
+    {
+      "epoch": 2.560182173064411,
+      "grad_norm": 14.189367294311523,
+      "learning_rate": 2.4669663382649967e-05,
+      "loss": 0.6196,
+      "step": 3935
+    },
+    {
+      "epoch": 2.5634352635003252,
+      "grad_norm": 1.2099864482879639,
+      "learning_rate": 2.4618052031076933e-05,
+      "loss": 0.651,
+      "step": 3940
+    },
+    {
+      "epoch": 2.5666883539362395,
+      "grad_norm": 0.9091722965240479,
+      "learning_rate": 2.4566442307685325e-05,
+      "loss": 0.6533,
+      "step": 3945
+    },
+    {
+      "epoch": 2.5699414443721533,
+      "grad_norm": 1.1715190410614014,
+      "learning_rate": 2.4514834432478927e-05,
+      "loss": 0.6578,
+      "step": 3950
+    },
+    {
+      "epoch": 2.5731945348080676,
+      "grad_norm": 0.9423794150352478,
+      "learning_rate": 2.4463228625453607e-05,
+      "loss": 0.665,
+      "step": 3955
+    },
+    {
+      "epoch": 2.576447625243982,
+      "grad_norm": 0.8811748027801514,
+      "learning_rate": 2.4411625106596457e-05,
+      "loss": 0.6589,
+      "step": 3960
+    },
+    {
+      "epoch": 2.5797007156798957,
+      "grad_norm": 0.9299075603485107,
+      "learning_rate": 2.43600240958848e-05,
+      "loss": 0.6568,
+      "step": 3965
+    },
+    {
+      "epoch": 2.58295380611581,
+      "grad_norm": 0.9680789709091187,
+      "learning_rate": 2.4308425813285255e-05,
+      "loss": 0.654,
+      "step": 3970
+    },
+    {
+      "epoch": 2.586206896551724,
+      "grad_norm": 1.6087170839309692,
+      "learning_rate": 2.425683047875282e-05,
+      "loss": 0.6812,
+      "step": 3975
+    },
+    {
+      "epoch": 2.589459986987638,
+      "grad_norm": 0.9408166408538818,
+      "learning_rate": 2.420523831222994e-05,
+      "loss": 0.6292,
+      "step": 3980
+    },
+    {
+      "epoch": 2.5927130774235523,
+      "grad_norm": 0.9801003336906433,
+      "learning_rate": 2.4153649533645545e-05,
+      "loss": 0.6536,
+      "step": 3985
+    },
+    {
+      "epoch": 2.5959661678594665,
+      "grad_norm": 0.908278226852417,
+      "learning_rate": 2.4102064362914108e-05,
+      "loss": 0.6494,
+      "step": 3990
+    },
+    {
+      "epoch": 2.5992192582953804,
+      "grad_norm": 2.5500690937042236,
+      "learning_rate": 2.4050483019934737e-05,
+      "loss": 0.6338,
+      "step": 3995
+    },
+    {
+      "epoch": 2.6024723487312946,
+      "grad_norm": 1.1047178506851196,
+      "learning_rate": 2.3998905724590237e-05,
+      "loss": 0.656,
+      "step": 4000
+    },
+    {
+      "epoch": 2.605725439167209,
+      "grad_norm": 1.0765790939331055,
+      "learning_rate": 2.3947332696746122e-05,
+      "loss": 0.6445,
+      "step": 4005
+    },
+    {
+      "epoch": 2.6089785296031227,
+      "grad_norm": 0.9311931133270264,
+      "learning_rate": 2.3895764156249746e-05,
+      "loss": 0.6472,
+      "step": 4010
+    },
+    {
+      "epoch": 2.612231620039037,
+      "grad_norm": 1.5003751516342163,
+      "learning_rate": 2.3844200322929323e-05,
+      "loss": 0.6713,
+      "step": 4015
+    },
+    {
+      "epoch": 2.6154847104749512,
+      "grad_norm": 0.8282221555709839,
+      "learning_rate": 2.3792641416592994e-05,
+      "loss": 0.6709,
+      "step": 4020
+    },
+    {
+      "epoch": 2.618737800910865,
+      "grad_norm": 1.0039994716644287,
+      "learning_rate": 2.3741087657027912e-05,
+      "loss": 0.6723,
+      "step": 4025
+    },
+    {
+      "epoch": 2.6219908913467793,
+      "grad_norm": 1.1094609498977661,
+      "learning_rate": 2.3689539263999286e-05,
+      "loss": 0.6519,
+      "step": 4030
+    },
+    {
+      "epoch": 2.6252439817826936,
+      "grad_norm": 0.9982606768608093,
+      "learning_rate": 2.3637996457249434e-05,
+      "loss": 0.6444,
+      "step": 4035
+    },
+    {
+      "epoch": 2.6284970722186074,
+      "grad_norm": 0.9692511558532715,
+      "learning_rate": 2.3586459456496877e-05,
+      "loss": 0.6525,
+      "step": 4040
+    },
+    {
+      "epoch": 2.6317501626545217,
+      "grad_norm": 1.5806814432144165,
+      "learning_rate": 2.3534928481435388e-05,
+      "loss": 0.6756,
+      "step": 4045
+    },
+    {
+      "epoch": 2.635003253090436,
+      "grad_norm": 0.9672693014144897,
+      "learning_rate": 2.348340375173303e-05,
+      "loss": 0.635,
+      "step": 4050
+    },
+    {
+      "epoch": 2.63825634352635,
+      "grad_norm": 0.9741607904434204,
+      "learning_rate": 2.3442188612229703e-05,
+      "loss": 0.6786,
+      "step": 4055
+    },
+    {
+      "epoch": 2.641509433962264,
+      "grad_norm": 1.4344401359558105,
+      "learning_rate": 2.3390675677651777e-05,
+      "loss": 0.6695,
+      "step": 4060
+    },
+    {
+      "epoch": 2.6447625243981783,
+      "grad_norm": 0.9737870693206787,
+      "learning_rate": 2.3339169603358997e-05,
+      "loss": 0.6562,
+      "step": 4065
+    },
+    {
+      "epoch": 2.6480156148340925,
+      "grad_norm": 1.122745156288147,
+      "learning_rate": 2.328767060891328e-05,
+      "loss": 0.6428,
+      "step": 4070
+    },
+    {
+      "epoch": 2.6512687052700064,
+      "grad_norm": 1.1388943195343018,
+      "learning_rate": 2.323617891384638e-05,
+      "loss": 0.6592,
+      "step": 4075
+    },
+    {
+      "epoch": 2.6545217957059206,
+      "grad_norm": 1.3511914014816284,
+      "learning_rate": 2.3184694737658942e-05,
+      "loss": 0.6811,
+      "step": 4080
+    },
+    {
+      "epoch": 2.657774886141835,
+      "grad_norm": 0.9977470636367798,
+      "learning_rate": 2.3133218299819536e-05,
+      "loss": 0.6489,
+      "step": 4085
+    },
+    {
+      "epoch": 2.6610279765777487,
+      "grad_norm": 6.271300792694092,
+      "learning_rate": 2.308174981976377e-05,
+      "loss": 0.6539,
+      "step": 4090
+    },
+    {
+      "epoch": 2.664281067013663,
+      "grad_norm": 1.12664794921875,
+      "learning_rate": 2.3030289516893306e-05,
+      "loss": 0.6874,
+      "step": 4095
+    },
+    {
+      "epoch": 2.6675341574495772,
+      "grad_norm": 1.1613657474517822,
+      "learning_rate": 2.2978837610574964e-05,
+      "loss": 0.649,
+      "step": 4100
+    },
+    {
+      "epoch": 2.6707872478854915,
+      "grad_norm": 0.9490109086036682,
+      "learning_rate": 2.2927394320139765e-05,
+      "loss": 0.6316,
+      "step": 4105
+    },
+    {
+      "epoch": 2.6740403383214053,
+      "grad_norm": 1.0231504440307617,
+      "learning_rate": 2.2875959864882002e-05,
+      "loss": 0.6496,
+      "step": 4110
+    },
+    {
+      "epoch": 2.6772934287573196,
+      "grad_norm": 1.2208635807037354,
+      "learning_rate": 2.2824534464058314e-05,
+      "loss": 0.6763,
+      "step": 4115
+    },
+    {
+      "epoch": 2.680546519193234,
+      "grad_norm": 1.1819181442260742,
+      "learning_rate": 2.2773118336886724e-05,
+      "loss": 0.6631,
+      "step": 4120
+    },
+    {
+      "epoch": 2.6837996096291477,
+      "grad_norm": 0.9770584106445312,
+      "learning_rate": 2.2721711702545735e-05,
+      "loss": 0.6727,
+      "step": 4125
+    },
+    {
+      "epoch": 2.687052700065062,
+      "grad_norm": 0.9494980573654175,
+      "learning_rate": 2.267031478017339e-05,
+      "loss": 0.6555,
+      "step": 4130
+    },
+    {
+      "epoch": 2.690305790500976,
+      "grad_norm": 1.0622323751449585,
+      "learning_rate": 2.2618927788866316e-05,
+      "loss": 0.6688,
+      "step": 4135
+    },
+    {
+      "epoch": 2.69355888093689,
+      "grad_norm": 1.1738066673278809,
+      "learning_rate": 2.2567550947678812e-05,
+      "loss": 0.6665,
+      "step": 4140
+    },
+    {
+      "epoch": 2.6968119713728043,
+      "grad_norm": 0.9498983025550842,
+      "learning_rate": 2.2516184475621915e-05,
+      "loss": 0.6593,
+      "step": 4145
+    },
+    {
+      "epoch": 2.7000650618087185,
+      "grad_norm": 0.9943968057632446,
+      "learning_rate": 2.246482859166245e-05,
+      "loss": 0.6394,
+      "step": 4150
+    },
+    {
+      "epoch": 2.7033181522446323,
+      "grad_norm": 0.9715344905853271,
+      "learning_rate": 2.2413483514722117e-05,
+      "loss": 0.6451,
+      "step": 4155
+    },
+    {
+      "epoch": 2.7065712426805466,
+      "grad_norm": 0.9957199692726135,
+      "learning_rate": 2.2362149463676536e-05,
+      "loss": 0.654,
+      "step": 4160
+    },
+    {
+      "epoch": 2.709824333116461,
+      "grad_norm": 1.3196548223495483,
+      "learning_rate": 2.231082665735433e-05,
+      "loss": 0.6377,
+      "step": 4165
+    },
+    {
+      "epoch": 2.7130774235523747,
+      "grad_norm": 1.0311577320098877,
+      "learning_rate": 2.22595153145362e-05,
+      "loss": 0.6532,
+      "step": 4170
+    },
+    {
+      "epoch": 2.716330513988289,
+      "grad_norm": 0.967720627784729,
+      "learning_rate": 2.220821565395395e-05,
+      "loss": 0.6723,
+      "step": 4175
+    },
+    {
+      "epoch": 2.719583604424203,
+      "grad_norm": 0.9087415933609009,
+      "learning_rate": 2.215692789428962e-05,
+      "loss": 0.6738,
+      "step": 4180
+    },
+    {
+      "epoch": 2.722836694860117,
+      "grad_norm": 0.9315296411514282,
+      "learning_rate": 2.21056522541745e-05,
+      "loss": 0.6727,
+      "step": 4185
+    },
+    {
+      "epoch": 2.7260897852960313,
+      "grad_norm": 1.8225336074829102,
+      "learning_rate": 2.2054388952188205e-05,
+      "loss": 0.6787,
+      "step": 4190
+    },
+    {
+      "epoch": 2.7293428757319456,
+      "grad_norm": 1.4423229694366455,
+      "learning_rate": 2.2003138206857782e-05,
+      "loss": 0.6549,
+      "step": 4195
+    },
+    {
+      "epoch": 2.7325959661678594,
+      "grad_norm": 0.9800527691841125,
+      "learning_rate": 2.1951900236656732e-05,
+      "loss": 0.6622,
+      "step": 4200
+    },
+    {
+      "epoch": 2.7358490566037736,
+      "grad_norm": 1.0587884187698364,
+      "learning_rate": 2.1900675260004102e-05,
+      "loss": 0.6575,
+      "step": 4205
+    },
+    {
+      "epoch": 2.739102147039688,
+      "grad_norm": 0.9304030537605286,
+      "learning_rate": 2.1849463495263546e-05,
+      "loss": 0.6374,
+      "step": 4210
+    },
+    {
+      "epoch": 2.7423552374756017,
+      "grad_norm": 1.906974196434021,
+      "learning_rate": 2.1798265160742413e-05,
+      "loss": 0.6516,
+      "step": 4215
+    },
+    {
+      "epoch": 2.745608327911516,
+      "grad_norm": 1.2081124782562256,
+      "learning_rate": 2.1747080474690778e-05,
+      "loss": 0.6872,
+      "step": 4220
+    },
+    {
+      "epoch": 2.7488614183474303,
+      "grad_norm": 0.8841726779937744,
+      "learning_rate": 2.169590965530056e-05,
+      "loss": 0.6585,
+      "step": 4225
+    },
+    {
+      "epoch": 2.752114508783344,
+      "grad_norm": 1.3173036575317383,
+      "learning_rate": 2.1644752920704534e-05,
+      "loss": 0.6584,
+      "step": 4230
+    },
+    {
+      "epoch": 2.7553675992192583,
+      "grad_norm": 1.285091519355774,
+      "learning_rate": 2.1593610488975468e-05,
+      "loss": 0.6578,
+      "step": 4235
+    },
+    {
+      "epoch": 2.7586206896551726,
+      "grad_norm": 1.0926772356033325,
+      "learning_rate": 2.1542482578125143e-05,
+      "loss": 0.6489,
+      "step": 4240
+    },
+    {
+      "epoch": 2.7618737800910864,
+      "grad_norm": 1.1347663402557373,
+      "learning_rate": 2.149136940610343e-05,
+      "loss": 0.6455,
+      "step": 4245
+    },
+    {
+      "epoch": 2.7651268705270007,
+      "grad_norm": 1.2697584629058838,
+      "learning_rate": 2.1440271190797403e-05,
+      "loss": 0.6708,
+      "step": 4250
+    },
+    {
+      "epoch": 2.768379960962915,
+      "grad_norm": 0.9204789996147156,
+      "learning_rate": 2.1389188150030344e-05,
+      "loss": 0.6889,
+      "step": 4255
+    },
+    {
+      "epoch": 2.7716330513988288,
+      "grad_norm": 1.154973030090332,
+      "learning_rate": 2.1338120501560862e-05,
+      "loss": 0.6653,
+      "step": 4260
+    },
+    {
+      "epoch": 2.774886141834743,
+      "grad_norm": 0.9574093222618103,
+      "learning_rate": 2.128706846308196e-05,
+      "loss": 0.6548,
+      "step": 4265
+    },
+    {
+      "epoch": 2.7781392322706573,
+      "grad_norm": 0.9033051133155823,
+      "learning_rate": 2.123603225222007e-05,
+      "loss": 0.6521,
+      "step": 4270
+    },
+    {
+      "epoch": 2.781392322706571,
+      "grad_norm": 0.9509397745132446,
+      "learning_rate": 2.11850120865342e-05,
+      "loss": 0.6542,
+      "step": 4275
+    },
+    {
+      "epoch": 2.7846454131424854,
+      "grad_norm": 0.9007295370101929,
+      "learning_rate": 2.1134008183514906e-05,
+      "loss": 0.6354,
+      "step": 4280
+    },
+    {
+      "epoch": 2.7878985035783996,
+      "grad_norm": 1.7208142280578613,
+      "learning_rate": 2.108302076058346e-05,
+      "loss": 0.6691,
+      "step": 4285
+    },
+    {
+      "epoch": 2.7911515940143135,
+      "grad_norm": 0.8821661472320557,
+      "learning_rate": 2.1032050035090865e-05,
+      "loss": 0.6784,
+      "step": 4290
+    },
+    {
+      "epoch": 2.7944046844502277,
+      "grad_norm": 0.9296684861183167,
+      "learning_rate": 2.0981096224316944e-05,
+      "loss": 0.6635,
+      "step": 4295
+    },
+    {
+      "epoch": 2.797657774886142,
+      "grad_norm": 1.5931028127670288,
+      "learning_rate": 2.093015954546942e-05,
+      "loss": 0.6763,
+      "step": 4300
+    },
+    {
+      "epoch": 2.800910865322056,
+      "grad_norm": 3.0784621238708496,
+      "learning_rate": 2.0879240215683e-05,
+      "loss": 0.6472,
+      "step": 4305
+    },
+    {
+      "epoch": 2.80416395575797,
+      "grad_norm": 1.1016740798950195,
+      "learning_rate": 2.0828338452018396e-05,
+      "loss": 0.6822,
+      "step": 4310
+    },
+    {
+      "epoch": 2.8074170461938843,
+      "grad_norm": 1.020324945449829,
+      "learning_rate": 2.0777454471461476e-05,
+      "loss": 0.6374,
+      "step": 4315
+    },
+    {
+      "epoch": 2.810670136629798,
+      "grad_norm": 1.3198879957199097,
+      "learning_rate": 2.0726588490922288e-05,
+      "loss": 0.6525,
+      "step": 4320
+    },
+    {
+      "epoch": 2.8139232270657124,
+      "grad_norm": 1.068981647491455,
+      "learning_rate": 2.0675740727234142e-05,
+      "loss": 0.6243,
+      "step": 4325
+    },
+    {
+      "epoch": 2.8171763175016267,
+      "grad_norm": 1.2015876770019531,
+      "learning_rate": 2.062491139715271e-05,
+      "loss": 0.6779,
+      "step": 4330
+    },
+    {
+      "epoch": 2.8204294079375405,
+      "grad_norm": 0.9587366580963135,
+      "learning_rate": 2.057410071735506e-05,
+      "loss": 0.6594,
+      "step": 4335
+    },
+    {
+      "epoch": 2.8236824983734548,
+      "grad_norm": 0.9465182423591614,
+      "learning_rate": 2.0523308904438775e-05,
+      "loss": 0.6545,
+      "step": 4340
+    },
+    {
+      "epoch": 2.826935588809369,
+      "grad_norm": 0.9994252324104309,
+      "learning_rate": 2.0472536174921017e-05,
+      "loss": 0.651,
+      "step": 4345
+    },
+    {
+      "epoch": 2.830188679245283,
+      "grad_norm": 0.9640632271766663,
+      "learning_rate": 2.0421782745237574e-05,
+      "loss": 0.6617,
+      "step": 4350
+    },
+    {
+      "epoch": 2.833441769681197,
+      "grad_norm": 0.8937436938285828,
+      "learning_rate": 2.0371048831741987e-05,
+      "loss": 0.6459,
+      "step": 4355
+    },
+    {
+      "epoch": 2.8366948601171114,
+      "grad_norm": 1.2173950672149658,
+      "learning_rate": 2.0320334650704594e-05,
+      "loss": 0.658,
+      "step": 4360
+    },
+    {
+      "epoch": 2.839947950553025,
+      "grad_norm": 1.0267751216888428,
+      "learning_rate": 2.0269640418311608e-05,
+      "loss": 0.6494,
+      "step": 4365
+    },
+    {
+      "epoch": 2.8432010409889394,
+      "grad_norm": 3.8074207305908203,
+      "learning_rate": 2.021896635066421e-05,
+      "loss": 0.6454,
+      "step": 4370
+    },
+    {
+      "epoch": 2.8464541314248537,
+      "grad_norm": 1.9503079652786255,
+      "learning_rate": 2.0168312663777638e-05,
+      "loss": 0.6446,
+      "step": 4375
+    },
+    {
+      "epoch": 2.8497072218607675,
+      "grad_norm": 0.942142903804779,
+      "learning_rate": 2.011767957358021e-05,
+      "loss": 0.6374,
+      "step": 4380
+    },
+    {
+      "epoch": 2.852960312296682,
+      "grad_norm": 1.3679735660552979,
+      "learning_rate": 2.0067067295912494e-05,
+      "loss": 0.6554,
+      "step": 4385
+    },
+    {
+      "epoch": 2.856213402732596,
+      "grad_norm": 1.217850923538208,
+      "learning_rate": 2.0016476046526305e-05,
+      "loss": 0.6332,
+      "step": 4390
+    },
+    {
+      "epoch": 2.85946649316851,
+      "grad_norm": 0.9636296033859253,
+      "learning_rate": 1.996590604108383e-05,
+      "loss": 0.6601,
+      "step": 4395
+    },
+    {
+      "epoch": 2.862719583604424,
+      "grad_norm": 1.0870169401168823,
+      "learning_rate": 1.991535749515668e-05,
+      "loss": 0.6488,
+      "step": 4400
+    },
+    {
+      "epoch": 2.8659726740403384,
+      "grad_norm": 0.9022223353385925,
+      "learning_rate": 1.9864830624225005e-05,
+      "loss": 0.6379,
+      "step": 4405
+    },
+    {
+      "epoch": 2.869225764476252,
+      "grad_norm": 0.9370593428611755,
+      "learning_rate": 1.981432564367657e-05,
+      "loss": 0.6288,
+      "step": 4410
+    },
+    {
+      "epoch": 2.8724788549121665,
+      "grad_norm": 1.368965744972229,
+      "learning_rate": 1.976384276880578e-05,
+      "loss": 0.6723,
+      "step": 4415
+    },
+    {
+      "epoch": 2.8757319453480807,
+      "grad_norm": 1.0498238801956177,
+      "learning_rate": 1.971338221481285e-05,
+      "loss": 0.6348,
+      "step": 4420
+    },
+    {
+      "epoch": 2.8789850357839946,
+      "grad_norm": 11.295182228088379,
+      "learning_rate": 1.966294419680283e-05,
+      "loss": 0.6694,
+      "step": 4425
+    },
+    {
+      "epoch": 2.882238126219909,
+      "grad_norm": 1.0129636526107788,
+      "learning_rate": 1.96125289297847e-05,
+      "loss": 0.6533,
+      "step": 4430
+    },
+    {
+      "epoch": 2.885491216655823,
+      "grad_norm": 3.2188525199890137,
+      "learning_rate": 1.9562136628670464e-05,
+      "loss": 0.6661,
+      "step": 4435
+    },
+    {
+      "epoch": 2.888744307091737,
+      "grad_norm": 1.130278468132019,
+      "learning_rate": 1.9511767508274214e-05,
+      "loss": 0.6489,
+      "step": 4440
+    },
+    {
+      "epoch": 2.891997397527651,
+      "grad_norm": 1.2190179824829102,
+      "learning_rate": 1.946142178331124e-05,
+      "loss": 0.6323,
+      "step": 4445
+    },
+    {
+      "epoch": 2.8952504879635654,
+      "grad_norm": 0.9841941595077515,
+      "learning_rate": 1.9411099668397085e-05,
+      "loss": 0.6623,
+      "step": 4450
+    },
+    {
+      "epoch": 2.8985035783994793,
+      "grad_norm": 0.9542006850242615,
+      "learning_rate": 1.9360801378046666e-05,
+      "loss": 0.6389,
+      "step": 4455
+    },
+    {
+      "epoch": 2.9017566688353935,
+      "grad_norm": 1.037227988243103,
+      "learning_rate": 1.931052712667332e-05,
+      "loss": 0.6504,
+      "step": 4460
+    },
+    {
+      "epoch": 2.905009759271308,
+      "grad_norm": 1.146600365638733,
+      "learning_rate": 1.9260277128587936e-05,
+      "loss": 0.6558,
+      "step": 4465
+    },
+    {
+      "epoch": 2.9082628497072216,
+      "grad_norm": 0.8806731104850769,
+      "learning_rate": 1.921005159799798e-05,
+      "loss": 0.6534,
+      "step": 4470
+    },
+    {
+      "epoch": 2.911515940143136,
+      "grad_norm": 0.926228404045105,
+      "learning_rate": 1.915985074900664e-05,
+      "loss": 0.6416,
+      "step": 4475
+    },
+    {
+      "epoch": 2.91476903057905,
+      "grad_norm": 3.157443046569824,
+      "learning_rate": 1.9109674795611898e-05,
+      "loss": 0.6614,
+      "step": 4480
+    },
+    {
+      "epoch": 2.918022121014964,
+      "grad_norm": 0.9215693473815918,
+      "learning_rate": 1.9059523951705585e-05,
+      "loss": 0.6738,
+      "step": 4485
+    },
+    {
+      "epoch": 2.921275211450878,
+      "grad_norm": 1.0931910276412964,
+      "learning_rate": 1.900939843107251e-05,
+      "loss": 0.6587,
+      "step": 4490
+    },
+    {
+      "epoch": 2.9245283018867925,
+      "grad_norm": 0.8158690333366394,
+      "learning_rate": 1.895929844738954e-05,
+      "loss": 0.6562,
+      "step": 4495
+    },
+    {
+      "epoch": 2.9277813923227067,
+      "grad_norm": 0.8992384672164917,
+      "learning_rate": 1.8909224214224662e-05,
+      "loss": 0.6552,
+      "step": 4500
+    },
+    {
+      "epoch": 2.9310344827586206,
+      "grad_norm": 0.9259112477302551,
+      "learning_rate": 1.885917594503611e-05,
+      "loss": 0.6391,
+      "step": 4505
+    },
+    {
+      "epoch": 2.934287573194535,
+      "grad_norm": 1.9078396558761597,
+      "learning_rate": 1.8809153853171426e-05,
+      "loss": 0.6419,
+      "step": 4510
+    },
+    {
+      "epoch": 2.937540663630449,
+      "grad_norm": 1.015743374824524,
+      "learning_rate": 1.875915815186657e-05,
+      "loss": 0.6437,
+      "step": 4515
+    },
+    {
+      "epoch": 2.940793754066363,
+      "grad_norm": 0.8620851039886475,
+      "learning_rate": 1.8709189054244996e-05,
+      "loss": 0.6496,
+      "step": 4520
+    },
+    {
+      "epoch": 2.944046844502277,
+      "grad_norm": 0.9002664089202881,
+      "learning_rate": 1.865924677331677e-05,
+      "loss": 0.6468,
+      "step": 4525
+    },
+    {
+      "epoch": 2.9472999349381914,
+      "grad_norm": 1.9784495830535889,
+      "learning_rate": 1.8609331521977623e-05,
+      "loss": 0.6707,
+      "step": 4530
+    },
+    {
+      "epoch": 2.9505530253741052,
+      "grad_norm": 0.902123749256134,
+      "learning_rate": 1.8559443513008067e-05,
+      "loss": 0.6545,
+      "step": 4535
+    },
+    {
+      "epoch": 2.9538061158100195,
+      "grad_norm": 1.239797592163086,
+      "learning_rate": 1.8509582959072486e-05,
+      "loss": 0.641,
+      "step": 4540
+    },
+    {
+      "epoch": 2.9570592062459338,
+      "grad_norm": 1.3703653812408447,
+      "learning_rate": 1.8459750072718235e-05,
+      "loss": 0.6373,
+      "step": 4545
+    },
+    {
+      "epoch": 2.960312296681848,
+      "grad_norm": 3.020883083343506,
+      "learning_rate": 1.8409945066374706e-05,
+      "loss": 0.6318,
+      "step": 4550
+    },
+    {
+      "epoch": 2.963565387117762,
+      "grad_norm": 1.026063084602356,
+      "learning_rate": 1.8360168152352472e-05,
+      "loss": 0.6294,
+      "step": 4555
+    },
+    {
+      "epoch": 2.966818477553676,
+      "grad_norm": 0.9441201090812683,
+      "learning_rate": 1.8310419542842327e-05,
+      "loss": 0.6543,
+      "step": 4560
+    },
+    {
+      "epoch": 2.9700715679895904,
+      "grad_norm": 1.139441728591919,
+      "learning_rate": 1.826069944991442e-05,
+      "loss": 0.6433,
+      "step": 4565
+    },
+    {
+      "epoch": 2.973324658425504,
+      "grad_norm": 1.0269153118133545,
+      "learning_rate": 1.821100808551735e-05,
+      "loss": 0.6669,
+      "step": 4570
+    },
+    {
+      "epoch": 2.9765777488614185,
+      "grad_norm": 0.9721339344978333,
+      "learning_rate": 1.8161345661477215e-05,
+      "loss": 0.6481,
+      "step": 4575
+    },
+    {
+      "epoch": 2.9798308392973327,
+      "grad_norm": 1.0770370960235596,
+      "learning_rate": 1.811171238949679e-05,
+      "loss": 0.6532,
+      "step": 4580
+    },
+    {
+      "epoch": 2.9830839297332465,
+      "grad_norm": 0.8994714021682739,
+      "learning_rate": 1.8062108481154545e-05,
+      "loss": 0.6503,
+      "step": 4585
+    },
+    {
+      "epoch": 2.986337020169161,
+      "grad_norm": 1.0329276323318481,
+      "learning_rate": 1.801253414790379e-05,
+      "loss": 0.6295,
+      "step": 4590
+    },
+    {
+      "epoch": 2.989590110605075,
+      "grad_norm": 5.5768232345581055,
+      "learning_rate": 1.796298960107177e-05,
+      "loss": 0.6711,
+      "step": 4595
+    },
+    {
+      "epoch": 2.992843201040989,
+      "grad_norm": 1.2496637105941772,
+      "learning_rate": 1.7913475051858744e-05,
+      "loss": 0.6501,
+      "step": 4600
+    },
+    {
+      "epoch": 2.996096291476903,
+      "grad_norm": 1.1241545677185059,
+      "learning_rate": 1.7863990711337093e-05,
+      "loss": 0.6334,
+      "step": 4605
+    },
+    {
+      "epoch": 2.9993493819128174,
+      "grad_norm": 1.352886438369751,
+      "learning_rate": 1.7814536790450437e-05,
+      "loss": 0.6427,
+      "step": 4610
+    },
+    {
+      "epoch": 3.0,
+      "eval_f1": 0.812837658080022,
+      "eval_loss": 0.43994140625,
+      "eval_precision": 0.8124680334875379,
+      "eval_recall": 0.8133431620386601,
+      "eval_runtime": 406.4496,
+      "eval_samples_per_second": 967.98,
+      "eval_steps_per_second": 0.947,
+      "step": 4611
+    },
+    {
+      "epoch": 3.0026024723487312,
+      "grad_norm": 1.0044734477996826,
+      "learning_rate": 1.7765113500012706e-05,
+      "loss": 0.5134,
+      "step": 4615
+    },
+    {
+      "epoch": 3.0058555627846455,
+      "grad_norm": 1.0504697561264038,
+      "learning_rate": 1.771572105070727e-05,
+      "loss": 0.4684,
+      "step": 4620
+    },
+    {
+      "epoch": 3.0091086532205593,
+      "grad_norm": 1.143985390663147,
+      "learning_rate": 1.766635965308603e-05,
+      "loss": 0.4485,
+      "step": 4625
+    },
+    {
+      "epoch": 3.0123617436564736,
+      "grad_norm": 1.933708667755127,
+      "learning_rate": 1.7617029517568502e-05,
+      "loss": 0.469,
+      "step": 4630
+    },
+    {
+      "epoch": 3.015614834092388,
+      "grad_norm": 1.3680862188339233,
+      "learning_rate": 1.756773085444095e-05,
+      "loss": 0.4556,
+      "step": 4635
+    },
+    {
+      "epoch": 3.018867924528302,
+      "grad_norm": 1.3127816915512085,
+      "learning_rate": 1.7518463873855486e-05,
+      "loss": 0.4536,
+      "step": 4640
+    },
+    {
+      "epoch": 3.022121014964216,
+      "grad_norm": 1.4003212451934814,
+      "learning_rate": 1.746922878582914e-05,
+      "loss": 0.4553,
+      "step": 4645
+    },
+    {
+      "epoch": 3.02537410540013,
+      "grad_norm": 1.3466031551361084,
+      "learning_rate": 1.7420025800243e-05,
+      "loss": 0.4634,
+      "step": 4650
+    },
+    {
+      "epoch": 3.0286271958360445,
+      "grad_norm": 1.3510533571243286,
+      "learning_rate": 1.7370855126841314e-05,
+      "loss": 0.4385,
+      "step": 4655
+    },
+    {
+      "epoch": 3.0318802862719583,
+      "grad_norm": 1.4002180099487305,
+      "learning_rate": 1.732171697523059e-05,
+      "loss": 0.4455,
+      "step": 4660
+    },
+    {
+      "epoch": 3.0351333767078725,
+      "grad_norm": 1.432400107383728,
+      "learning_rate": 1.7272611554878678e-05,
+      "loss": 0.443,
+      "step": 4665
+    },
+    {
+      "epoch": 3.038386467143787,
+      "grad_norm": 3.641195774078369,
+      "learning_rate": 1.722353907511393e-05,
+      "loss": 0.4351,
+      "step": 4670
+    },
+    {
+      "epoch": 3.0416395575797006,
+      "grad_norm": 1.3562887907028198,
+      "learning_rate": 1.717449974512426e-05,
+      "loss": 0.4452,
+      "step": 4675
+    },
+    {
+      "epoch": 3.044892648015615,
+      "grad_norm": 1.4275323152542114,
+      "learning_rate": 1.7125493773956265e-05,
+      "loss": 0.4424,
+      "step": 4680
+    },
+    {
+      "epoch": 3.048145738451529,
+      "grad_norm": 1.6079832315444946,
+      "learning_rate": 1.7076521370514355e-05,
+      "loss": 0.4532,
+      "step": 4685
+    },
+    {
+      "epoch": 3.051398828887443,
+      "grad_norm": 1.3562263250350952,
+      "learning_rate": 1.7027582743559843e-05,
+      "loss": 0.4417,
+      "step": 4690
+    },
+    {
+      "epoch": 3.0546519193233572,
+      "grad_norm": 1.7932298183441162,
+      "learning_rate": 1.6978678101710043e-05,
+      "loss": 0.4375,
+      "step": 4695
+    },
+    {
+      "epoch": 3.0579050097592715,
+      "grad_norm": 1.6265188455581665,
+      "learning_rate": 1.6929807653437412e-05,
+      "loss": 0.4307,
+      "step": 4700
+    },
+    {
+      "epoch": 3.0611581001951853,
+      "grad_norm": 1.4614055156707764,
+      "learning_rate": 1.6880971607068646e-05,
+      "loss": 0.4275,
+      "step": 4705
+    },
+    {
+      "epoch": 3.0644111906310996,
+      "grad_norm": 2.0057411193847656,
+      "learning_rate": 1.6832170170783776e-05,
+      "loss": 0.4359,
+      "step": 4710
+    },
+    {
+      "epoch": 3.067664281067014,
+      "grad_norm": 1.3620643615722656,
+      "learning_rate": 1.6783403552615314e-05,
+      "loss": 0.4267,
+      "step": 4715
+    },
+    {
+      "epoch": 3.0709173715029277,
+      "grad_norm": 1.4945169687271118,
+      "learning_rate": 1.6734671960447333e-05,
+      "loss": 0.4401,
+      "step": 4720
+    },
+    {
+      "epoch": 3.074170461938842,
+      "grad_norm": 1.7050446271896362,
+      "learning_rate": 1.6685975602014604e-05,
+      "loss": 0.4553,
+      "step": 4725
+    },
+    {
+      "epoch": 3.077423552374756,
+      "grad_norm": 2.7887024879455566,
+      "learning_rate": 1.6637314684901713e-05,
+      "loss": 0.4312,
+      "step": 4730
+    },
+    {
+      "epoch": 3.08067664281067,
+      "grad_norm": 1.5501539707183838,
+      "learning_rate": 1.658868941654213e-05,
+      "loss": 0.4495,
+      "step": 4735
+    },
+    {
+      "epoch": 3.0839297332465843,
+      "grad_norm": 2.2847344875335693,
+      "learning_rate": 1.6540100004217402e-05,
+      "loss": 0.4535,
+      "step": 4740
+    },
+    {
+      "epoch": 3.0871828236824985,
+      "grad_norm": 1.3970599174499512,
+      "learning_rate": 1.6491546655056208e-05,
+      "loss": 0.4329,
+      "step": 4745
+    },
+    {
+      "epoch": 3.0904359141184123,
+      "grad_norm": 1.7147552967071533,
+      "learning_rate": 1.644302957603349e-05,
+      "loss": 0.4465,
+      "step": 4750
+    },
+    {
+      "epoch": 3.0936890045543266,
+      "grad_norm": 1.4633418321609497,
+      "learning_rate": 1.6394548973969588e-05,
+      "loss": 0.4163,
+      "step": 4755
+    },
+    {
+      "epoch": 3.096942094990241,
+      "grad_norm": 2.099158763885498,
+      "learning_rate": 1.634610505552934e-05,
+      "loss": 0.4554,
+      "step": 4760
+    },
+    {
+      "epoch": 3.1001951854261547,
+      "grad_norm": 1.6046409606933594,
+      "learning_rate": 1.6297698027221216e-05,
+      "loss": 0.4331,
+      "step": 4765
+    },
+    {
+      "epoch": 3.103448275862069,
+      "grad_norm": 1.6139500141143799,
+      "learning_rate": 1.6249328095396415e-05,
+      "loss": 0.451,
+      "step": 4770
+    },
+    {
+      "epoch": 3.106701366297983,
+      "grad_norm": 2.9759113788604736,
+      "learning_rate": 1.6200995466248014e-05,
+      "loss": 0.4275,
+      "step": 4775
+    },
+    {
+      "epoch": 3.109954456733897,
+      "grad_norm": 1.3965107202529907,
+      "learning_rate": 1.6152700345810063e-05,
+      "loss": 0.4394,
+      "step": 4780
+    },
+    {
+      "epoch": 3.1132075471698113,
+      "grad_norm": 1.3388948440551758,
+      "learning_rate": 1.6104442939956733e-05,
+      "loss": 0.4496,
+      "step": 4785
+    },
+    {
+      "epoch": 3.1164606376057256,
+      "grad_norm": 1.3315989971160889,
+      "learning_rate": 1.6056223454401396e-05,
+      "loss": 0.4359,
+      "step": 4790
+    },
+    {
+      "epoch": 3.1197137280416394,
+      "grad_norm": 1.7022862434387207,
+      "learning_rate": 1.6008042094695825e-05,
+      "loss": 0.4239,
+      "step": 4795
+    },
+    {
+      "epoch": 3.1229668184775536,
+      "grad_norm": 4.944558620452881,
+      "learning_rate": 1.5959899066229218e-05,
+      "loss": 0.4322,
+      "step": 4800
+    },
+    {
+      "epoch": 3.126219908913468,
+      "grad_norm": 2.622772693634033,
+      "learning_rate": 1.5911794574227402e-05,
+      "loss": 0.4512,
+      "step": 4805
+    },
+    {
+      "epoch": 3.1294729993493817,
+      "grad_norm": 1.4150971174240112,
+      "learning_rate": 1.5863728823751923e-05,
+      "loss": 0.4152,
+      "step": 4810
+    },
+    {
+      "epoch": 3.132726089785296,
+      "grad_norm": 1.6928874254226685,
+      "learning_rate": 1.5815702019699168e-05,
+      "loss": 0.4388,
+      "step": 4815
+    },
+    {
+      "epoch": 3.1359791802212102,
+      "grad_norm": 1.9819401502609253,
+      "learning_rate": 1.576771436679952e-05,
+      "loss": 0.4358,
+      "step": 4820
+    },
+    {
+      "epoch": 3.139232270657124,
+      "grad_norm": 4.905421733856201,
+      "learning_rate": 1.5719766069616457e-05,
+      "loss": 0.4296,
+      "step": 4825
+    },
+    {
+      "epoch": 3.1424853610930383,
+      "grad_norm": 1.4010788202285767,
+      "learning_rate": 1.5671857332545685e-05,
+      "loss": 0.4252,
+      "step": 4830
+    },
+    {
+      "epoch": 3.1457384515289526,
+      "grad_norm": 1.8857386112213135,
+      "learning_rate": 1.5623988359814285e-05,
+      "loss": 0.4344,
+      "step": 4835
+    },
+    {
+      "epoch": 3.1489915419648664,
+      "grad_norm": 1.8753392696380615,
+      "learning_rate": 1.5576159355479812e-05,
+      "loss": 0.4331,
+      "step": 4840
+    },
+    {
+      "epoch": 3.1522446324007807,
+      "grad_norm": 1.5203431844711304,
+      "learning_rate": 1.5528370523429465e-05,
+      "loss": 0.4366,
+      "step": 4845
+    },
+    {
+      "epoch": 3.155497722836695,
+      "grad_norm": 1.7210590839385986,
+      "learning_rate": 1.5480622067379176e-05,
+      "loss": 0.4161,
+      "step": 4850
+    },
+    {
+      "epoch": 3.1587508132726088,
+      "grad_norm": 1.7494614124298096,
+      "learning_rate": 1.5432914190872757e-05,
+      "loss": 0.428,
+      "step": 4855
+    },
+    {
+      "epoch": 3.162003903708523,
+      "grad_norm": 1.7109274864196777,
+      "learning_rate": 1.538524709728106e-05,
+      "loss": 0.423,
+      "step": 4860
+    },
+    {
+      "epoch": 3.1652569941444373,
+      "grad_norm": 2.6436519622802734,
+      "learning_rate": 1.533762098980107e-05,
+      "loss": 0.4308,
+      "step": 4865
+    },
+    {
+      "epoch": 3.168510084580351,
+      "grad_norm": 6.52752161026001,
+      "learning_rate": 1.5290036071455055e-05,
+      "loss": 0.4425,
+      "step": 4870
+    },
+    {
+      "epoch": 3.1717631750162654,
+      "grad_norm": 1.7809419631958008,
+      "learning_rate": 1.5242492545089698e-05,
+      "loss": 0.4444,
+      "step": 4875
+    },
+    {
+      "epoch": 3.1750162654521796,
+      "grad_norm": 4.83860969543457,
+      "learning_rate": 1.5194990613375253e-05,
+      "loss": 0.4315,
+      "step": 4880
+    },
+    {
+      "epoch": 3.178269355888094,
+      "grad_norm": 1.6642948389053345,
+      "learning_rate": 1.5147530478804634e-05,
+      "loss": 0.4348,
+      "step": 4885
+    },
+    {
+      "epoch": 3.1815224463240077,
+      "grad_norm": 1.948080062866211,
+      "learning_rate": 1.5100112343692604e-05,
+      "loss": 0.4334,
+      "step": 4890
+    },
+    {
+      "epoch": 3.184775536759922,
+      "grad_norm": 1.5025594234466553,
+      "learning_rate": 1.5052736410174877e-05,
+      "loss": 0.4436,
+      "step": 4895
+    },
+    {
+      "epoch": 3.1880286271958362,
+      "grad_norm": 1.6708216667175293,
+      "learning_rate": 1.5005402880207273e-05,
+      "loss": 0.446,
+      "step": 4900
+    },
+    {
+      "epoch": 3.19128171763175,
+      "grad_norm": 3.007356643676758,
+      "learning_rate": 1.495811195556486e-05,
+      "loss": 0.42,
+      "step": 4905
+    },
+    {
+      "epoch": 3.1945348080676643,
+      "grad_norm": 1.6949506998062134,
+      "learning_rate": 1.4910863837841068e-05,
+      "loss": 0.4226,
+      "step": 4910
+    },
+    {
+      "epoch": 3.1977878985035786,
+      "grad_norm": 2.3278913497924805,
+      "learning_rate": 1.4863658728446864e-05,
+      "loss": 0.4101,
+      "step": 4915
+    },
+    {
+      "epoch": 3.2010409889394924,
+      "grad_norm": 1.6102315187454224,
+      "learning_rate": 1.4816496828609878e-05,
+      "loss": 0.4139,
+      "step": 4920
+    },
+    {
+      "epoch": 3.2042940793754067,
+      "grad_norm": 1.5101537704467773,
+      "learning_rate": 1.476937833937352e-05,
+      "loss": 0.4361,
+      "step": 4925
+    },
+    {
+      "epoch": 3.207547169811321,
+      "grad_norm": 1.7118359804153442,
+      "learning_rate": 1.472230346159619e-05,
+      "loss": 0.4197,
+      "step": 4930
+    },
+    {
+      "epoch": 3.2108002602472347,
+      "grad_norm": 2.325700044631958,
+      "learning_rate": 1.4675272395950345e-05,
+      "loss": 0.4322,
+      "step": 4935
+    },
+    {
+      "epoch": 3.214053350683149,
+      "grad_norm": 2.856618642807007,
+      "learning_rate": 1.46282853429217e-05,
+      "loss": 0.439,
+      "step": 4940
+    },
+    {
+      "epoch": 3.2173064411190633,
+      "grad_norm": 1.6420326232910156,
+      "learning_rate": 1.4581342502808321e-05,
+      "loss": 0.4078,
+      "step": 4945
+    },
+    {
+      "epoch": 3.220559531554977,
+      "grad_norm": 1.6206941604614258,
+      "learning_rate": 1.4534444075719839e-05,
+      "loss": 0.4031,
+      "step": 4950
+    },
+    {
+      "epoch": 3.2238126219908914,
+      "grad_norm": 2.0076100826263428,
+      "learning_rate": 1.4487590261576542e-05,
+      "loss": 0.428,
+      "step": 4955
+    },
+    {
+      "epoch": 3.2270657124268056,
+      "grad_norm": 1.8135625123977661,
+      "learning_rate": 1.4440781260108521e-05,
+      "loss": 0.4177,
+      "step": 4960
+    },
+    {
+      "epoch": 3.2303188028627194,
+      "grad_norm": 1.494215488433838,
+      "learning_rate": 1.4394017270854887e-05,
+      "loss": 0.4233,
+      "step": 4965
+    },
+    {
+      "epoch": 3.2335718932986337,
+      "grad_norm": 1.5636515617370605,
+      "learning_rate": 1.4347298493162823e-05,
+      "loss": 0.4354,
+      "step": 4970
+    },
+    {
+      "epoch": 3.236824983734548,
+      "grad_norm": 1.34303879737854,
+      "learning_rate": 1.4300625126186806e-05,
+      "loss": 0.4313,
+      "step": 4975
+    },
+    {
+      "epoch": 3.240078074170462,
+      "grad_norm": 1.4629467725753784,
+      "learning_rate": 1.4253997368887717e-05,
+      "loss": 0.4272,
+      "step": 4980
+    },
+    {
+      "epoch": 3.243331164606376,
+      "grad_norm": 1.7214086055755615,
+      "learning_rate": 1.4207415420032044e-05,
+      "loss": 0.435,
+      "step": 4985
+    },
+    {
+      "epoch": 3.2465842550422903,
+      "grad_norm": 1.5725369453430176,
+      "learning_rate": 1.4160879478190974e-05,
+      "loss": 0.4239,
+      "step": 4990
+    },
+    {
+      "epoch": 3.249837345478204,
+      "grad_norm": 2.499541997909546,
+      "learning_rate": 1.411438974173957e-05,
+      "loss": 0.438,
+      "step": 4995
+    },
+    {
+      "epoch": 3.2530904359141184,
+      "grad_norm": 1.4437288045883179,
+      "learning_rate": 1.4067946408855953e-05,
+      "loss": 0.427,
+      "step": 5000
+    },
+    {
+      "epoch": 3.2563435263500327,
+      "grad_norm": 1.851952075958252,
+      "learning_rate": 1.4021549677520415e-05,
+      "loss": 0.4419,
+      "step": 5005
+    },
+    {
+      "epoch": 3.2595966167859465,
+      "grad_norm": 1.586517572402954,
+      "learning_rate": 1.3975199745514587e-05,
+      "loss": 0.4193,
+      "step": 5010
+    },
+    {
+      "epoch": 3.2628497072218607,
+      "grad_norm": 1.516689419746399,
+      "learning_rate": 1.392889681042063e-05,
+      "loss": 0.4273,
+      "step": 5015
+    },
+    {
+      "epoch": 3.266102797657775,
+      "grad_norm": 1.5535813570022583,
+      "learning_rate": 1.3882641069620339e-05,
+      "loss": 0.4326,
+      "step": 5020
+    },
+    {
+      "epoch": 3.269355888093689,
+      "grad_norm": 2.197828531265259,
+      "learning_rate": 1.3836432720294329e-05,
+      "loss": 0.4381,
+      "step": 5025
+    },
+    {
+      "epoch": 3.272608978529603,
+      "grad_norm": 1.452176570892334,
+      "learning_rate": 1.3790271959421219e-05,
+      "loss": 0.426,
+      "step": 5030
+    },
+    {
+      "epoch": 3.2758620689655173,
+      "grad_norm": 1.550866723060608,
+      "learning_rate": 1.3744158983776733e-05,
+      "loss": 0.4284,
+      "step": 5035
+    },
+    {
+      "epoch": 3.279115159401431,
+      "grad_norm": 1.6225146055221558,
+      "learning_rate": 1.3698093989932904e-05,
+      "loss": 0.4319,
+      "step": 5040
+    },
+    {
+      "epoch": 3.2823682498373454,
+      "grad_norm": 1.6624497175216675,
+      "learning_rate": 1.3652077174257249e-05,
+      "loss": 0.4291,
+      "step": 5045
+    },
+    {
+      "epoch": 3.2856213402732597,
+      "grad_norm": 2.7444093227386475,
+      "learning_rate": 1.3606108732911882e-05,
+      "loss": 0.4292,
+      "step": 5050
+    },
+    {
+      "epoch": 3.288874430709174,
+      "grad_norm": 1.5146102905273438,
+      "learning_rate": 1.3560188861852702e-05,
+      "loss": 0.4466,
+      "step": 5055
+    },
+    {
+      "epoch": 3.2921275211450878,
+      "grad_norm": 1.6835765838623047,
+      "learning_rate": 1.3514317756828587e-05,
+      "loss": 0.4188,
+      "step": 5060
+    },
+    {
+      "epoch": 3.295380611581002,
+      "grad_norm": 1.7336374521255493,
+      "learning_rate": 1.3468495613380533e-05,
+      "loss": 0.4331,
+      "step": 5065
+    },
+    {
+      "epoch": 3.2986337020169163,
+      "grad_norm": 1.7082960605621338,
+      "learning_rate": 1.3422722626840791e-05,
+      "loss": 0.4161,
+      "step": 5070
+    },
+    {
+      "epoch": 3.30188679245283,
+      "grad_norm": 2.2675247192382812,
+      "learning_rate": 1.3376998992332076e-05,
+      "loss": 0.4111,
+      "step": 5075
+    },
+    {
+      "epoch": 3.3051398828887444,
+      "grad_norm": 1.4138269424438477,
+      "learning_rate": 1.3331324904766745e-05,
+      "loss": 0.4244,
+      "step": 5080
+    },
+    {
+      "epoch": 3.3083929733246586,
+      "grad_norm": 2.2490038871765137,
+      "learning_rate": 1.328570055884592e-05,
+      "loss": 0.4208,
+      "step": 5085
+    },
+    {
+      "epoch": 3.3116460637605725,
+      "grad_norm": 1.6177328824996948,
+      "learning_rate": 1.3240126149058685e-05,
+      "loss": 0.4227,
+      "step": 5090
+    },
+    {
+      "epoch": 3.3148991541964867,
+      "grad_norm": 1.7124465703964233,
+      "learning_rate": 1.3194601869681272e-05,
+      "loss": 0.4202,
+      "step": 5095
+    },
+    {
+      "epoch": 3.318152244632401,
+      "grad_norm": 1.7719941139221191,
+      "learning_rate": 1.3149127914776196e-05,
+      "loss": 0.417,
+      "step": 5100
+    },
+    {
+      "epoch": 3.321405335068315,
+      "grad_norm": 1.6123124361038208,
+      "learning_rate": 1.3103704478191448e-05,
+      "loss": 0.4398,
+      "step": 5105
+    },
+    {
+      "epoch": 3.324658425504229,
+      "grad_norm": 1.724311113357544,
+      "learning_rate": 1.3058331753559688e-05,
+      "loss": 0.4456,
+      "step": 5110
+    },
+    {
+      "epoch": 3.3279115159401433,
+      "grad_norm": 1.976486086845398,
+      "learning_rate": 1.301300993429738e-05,
+      "loss": 0.4207,
+      "step": 5115
+    },
+    {
+      "epoch": 3.331164606376057,
+      "grad_norm": 1.5178265571594238,
+      "learning_rate": 1.296773921360398e-05,
+      "loss": 0.4138,
+      "step": 5120
+    },
+    {
+      "epoch": 3.3344176968119714,
+      "grad_norm": 2.115887403488159,
+      "learning_rate": 1.2922519784461154e-05,
+      "loss": 0.4237,
+      "step": 5125
+    },
+    {
+      "epoch": 3.3376707872478857,
+      "grad_norm": 1.584922194480896,
+      "learning_rate": 1.2877351839631884e-05,
+      "loss": 0.4331,
+      "step": 5130
+    },
+    {
+      "epoch": 3.3409238776837995,
+      "grad_norm": 1.9146180152893066,
+      "learning_rate": 1.283223557165969e-05,
+      "loss": 0.4213,
+      "step": 5135
+    },
+    {
+      "epoch": 3.3441769681197138,
+      "grad_norm": 2.934779644012451,
+      "learning_rate": 1.2787171172867826e-05,
+      "loss": 0.4304,
+      "step": 5140
+    },
+    {
+      "epoch": 3.347430058555628,
+      "grad_norm": 1.7485958337783813,
+      "learning_rate": 1.2742158835358412e-05,
+      "loss": 0.4402,
+      "step": 5145
+    },
+    {
+      "epoch": 3.350683148991542,
+      "grad_norm": 1.497227668762207,
+      "learning_rate": 1.2697198751011641e-05,
+      "loss": 0.4235,
+      "step": 5150
+    },
+    {
+      "epoch": 3.353936239427456,
+      "grad_norm": 1.3970587253570557,
+      "learning_rate": 1.2652291111484962e-05,
+      "loss": 0.41,
+      "step": 5155
+    },
+    {
+      "epoch": 3.3571893298633704,
+      "grad_norm": 2.238862991333008,
+      "learning_rate": 1.2607436108212278e-05,
+      "loss": 0.4398,
+      "step": 5160
+    },
+    {
+      "epoch": 3.360442420299284,
+      "grad_norm": 4.999053478240967,
+      "learning_rate": 1.256263393240309e-05,
+      "loss": 0.4316,
+      "step": 5165
+    },
+    {
+      "epoch": 3.3636955107351985,
+      "grad_norm": 2.9387760162353516,
+      "learning_rate": 1.25178847750417e-05,
+      "loss": 0.4391,
+      "step": 5170
+    },
+    {
+      "epoch": 3.3669486011711127,
+      "grad_norm": 7.236008167266846,
+      "learning_rate": 1.2473188826886428e-05,
+      "loss": 0.4247,
+      "step": 5175
+    },
+    {
+      "epoch": 3.3702016916070265,
+      "grad_norm": 1.8067480325698853,
+      "learning_rate": 1.2428546278468753e-05,
+      "loss": 0.4363,
+      "step": 5180
+    },
+    {
+      "epoch": 3.373454782042941,
+      "grad_norm": 2.2515077590942383,
+      "learning_rate": 1.2383957320092512e-05,
+      "loss": 0.4218,
+      "step": 5185
+    },
+    {
+      "epoch": 3.376707872478855,
+      "grad_norm": 2.5399014949798584,
+      "learning_rate": 1.2339422141833127e-05,
+      "loss": 0.4344,
+      "step": 5190
+    },
+    {
+      "epoch": 3.379960962914769,
+      "grad_norm": 3.233678102493286,
+      "learning_rate": 1.2294940933536725e-05,
+      "loss": 0.4214,
+      "step": 5195
+    },
+    {
+      "epoch": 3.383214053350683,
+      "grad_norm": 1.5409492254257202,
+      "learning_rate": 1.2250513884819403e-05,
+      "loss": 0.4225,
+      "step": 5200
+    },
+    {
+      "epoch": 3.3864671437865974,
+      "grad_norm": 1.2861021757125854,
+      "learning_rate": 1.2206141185066359e-05,
+      "loss": 0.4268,
+      "step": 5205
+    },
+    {
+      "epoch": 3.3897202342225112,
+      "grad_norm": 1.417588472366333,
+      "learning_rate": 1.2161823023431133e-05,
+      "loss": 0.4188,
+      "step": 5210
+    },
+    {
+      "epoch": 3.3929733246584255,
+      "grad_norm": 3.5109989643096924,
+      "learning_rate": 1.2117559588834757e-05,
+      "loss": 0.4243,
+      "step": 5215
+    },
+    {
+      "epoch": 3.3962264150943398,
+      "grad_norm": 6.47197151184082,
+      "learning_rate": 1.207335106996497e-05,
+      "loss": 0.4229,
+      "step": 5220
+    },
+    {
+      "epoch": 3.3994795055302536,
+      "grad_norm": 1.6867964267730713,
+      "learning_rate": 1.2029197655275442e-05,
+      "loss": 0.4287,
+      "step": 5225
+    },
+    {
+      "epoch": 3.402732595966168,
+      "grad_norm": 1.575161099433899,
+      "learning_rate": 1.1985099532984917e-05,
+      "loss": 0.4304,
+      "step": 5230
+    },
+    {
+      "epoch": 3.405985686402082,
+      "grad_norm": 1.444969654083252,
+      "learning_rate": 1.1941056891076432e-05,
+      "loss": 0.4325,
+      "step": 5235
+    },
+    {
+      "epoch": 3.409238776837996,
+      "grad_norm": 1.6448862552642822,
+      "learning_rate": 1.1897069917296555e-05,
+      "loss": 0.4358,
+      "step": 5240
+    },
+    {
+      "epoch": 3.41249186727391,
+      "grad_norm": 1.5217825174331665,
+      "learning_rate": 1.1853138799154514e-05,
+      "loss": 0.4404,
+      "step": 5245
+    },
+    {
+      "epoch": 3.4157449577098244,
+      "grad_norm": 10.740922927856445,
+      "learning_rate": 1.1809263723921438e-05,
+      "loss": 0.4421,
+      "step": 5250
+    },
+    {
+      "epoch": 3.4189980481457383,
+      "grad_norm": 2.3015806674957275,
+      "learning_rate": 1.1765444878629583e-05,
+      "loss": 0.4376,
+      "step": 5255
+    },
+    {
+      "epoch": 3.4222511385816525,
+      "grad_norm": 1.4035224914550781,
+      "learning_rate": 1.1721682450071476e-05,
+      "loss": 0.453,
+      "step": 5260
+    },
+    {
+      "epoch": 3.425504229017567,
+      "grad_norm": 1.4976873397827148,
+      "learning_rate": 1.167797662479915e-05,
+      "loss": 0.4411,
+      "step": 5265
+    },
+    {
+      "epoch": 3.4287573194534806,
+      "grad_norm": 1.4919233322143555,
+      "learning_rate": 1.1634327589123373e-05,
+      "loss": 0.4357,
+      "step": 5270
+    },
+    {
+      "epoch": 3.432010409889395,
+      "grad_norm": 1.5440754890441895,
+      "learning_rate": 1.1590735529112806e-05,
+      "loss": 0.4288,
+      "step": 5275
+    },
+    {
+      "epoch": 3.435263500325309,
+      "grad_norm": 1.637131690979004,
+      "learning_rate": 1.1547200630593224e-05,
+      "loss": 0.4379,
+      "step": 5280
+    },
+    {
+      "epoch": 3.438516590761223,
+      "grad_norm": 1.525527834892273,
+      "learning_rate": 1.1503723079146766e-05,
+      "loss": 0.4158,
+      "step": 5285
+    },
+    {
+      "epoch": 3.441769681197137,
+      "grad_norm": 1.5343090295791626,
+      "learning_rate": 1.1460303060111083e-05,
+      "loss": 0.4128,
+      "step": 5290
+    },
+    {
+      "epoch": 3.4450227716330515,
+      "grad_norm": 1.7940443754196167,
+      "learning_rate": 1.1416940758578567e-05,
+      "loss": 0.4389,
+      "step": 5295
+    },
+    {
+      "epoch": 3.4482758620689653,
+      "grad_norm": 1.8386902809143066,
+      "learning_rate": 1.137363635939561e-05,
+      "loss": 0.4537,
+      "step": 5300
+    },
+    {
+      "epoch": 3.4515289525048796,
+      "grad_norm": 1.498874306678772,
+      "learning_rate": 1.1330390047161729e-05,
+      "loss": 0.4297,
+      "step": 5305
+    },
+    {
+      "epoch": 3.454782042940794,
+      "grad_norm": 1.4000130891799927,
+      "learning_rate": 1.1287202006228858e-05,
+      "loss": 0.4235,
+      "step": 5310
+    },
+    {
+      "epoch": 3.4580351333767076,
+      "grad_norm": 6.055357933044434,
+      "learning_rate": 1.1244072420700502e-05,
+      "loss": 0.4282,
+      "step": 5315
+    },
+    {
+      "epoch": 3.461288223812622,
+      "grad_norm": 1.7397581338882446,
+      "learning_rate": 1.1201001474431022e-05,
+      "loss": 0.4053,
+      "step": 5320
+    },
+    {
+      "epoch": 3.464541314248536,
+      "grad_norm": 1.7397915124893188,
+      "learning_rate": 1.1157989351024767e-05,
+      "loss": 0.4213,
+      "step": 5325
+    },
+    {
+      "epoch": 3.46779440468445,
+      "grad_norm": 1.6159850358963013,
+      "learning_rate": 1.1115036233835349e-05,
+      "loss": 0.4241,
+      "step": 5330
+    },
+    {
+      "epoch": 3.4710474951203643,
+      "grad_norm": 1.4938251972198486,
+      "learning_rate": 1.1072142305964855e-05,
+      "loss": 0.425,
+      "step": 5335
+    },
+    {
+      "epoch": 3.4743005855562785,
+      "grad_norm": 2.087301015853882,
+      "learning_rate": 1.102930775026306e-05,
+      "loss": 0.4224,
+      "step": 5340
+    },
+    {
+      "epoch": 3.4775536759921923,
+      "grad_norm": 1.5332581996917725,
+      "learning_rate": 1.098653274932662e-05,
+      "loss": 0.4281,
+      "step": 5345
+    },
+    {
+      "epoch": 3.4808067664281066,
+      "grad_norm": 1.481587529182434,
+      "learning_rate": 1.094381748549835e-05,
+      "loss": 0.4083,
+      "step": 5350
+    },
+    {
+      "epoch": 3.484059856864021,
+      "grad_norm": 1.8097538948059082,
+      "learning_rate": 1.0901162140866395e-05,
+      "loss": 0.4231,
+      "step": 5355
+    },
+    {
+      "epoch": 3.487312947299935,
+      "grad_norm": 1.6291629076004028,
+      "learning_rate": 1.0858566897263475e-05,
+      "loss": 0.4298,
+      "step": 5360
+    },
+    {
+      "epoch": 3.490566037735849,
+      "grad_norm": 1.3684316873550415,
+      "learning_rate": 1.081603193626611e-05,
+      "loss": 0.4045,
+      "step": 5365
+    },
+    {
+      "epoch": 3.493819128171763,
+      "grad_norm": 1.9993515014648438,
+      "learning_rate": 1.0773557439193865e-05,
+      "loss": 0.4223,
+      "step": 5370
+    },
+    {
+      "epoch": 3.4970722186076775,
+      "grad_norm": 1.364740014076233,
+      "learning_rate": 1.0731143587108533e-05,
+      "loss": 0.4472,
+      "step": 5375
+    },
+    {
+      "epoch": 3.5003253090435913,
+      "grad_norm": 1.4979236125946045,
+      "learning_rate": 1.0688790560813388e-05,
+      "loss": 0.4232,
+      "step": 5380
+    },
+    {
+      "epoch": 3.5035783994795056,
+      "grad_norm": 1.574204921722412,
+      "learning_rate": 1.064649854085244e-05,
+      "loss": 0.4228,
+      "step": 5385
+    },
+    {
+      "epoch": 3.5068314899154194,
+      "grad_norm": 2.251814842224121,
+      "learning_rate": 1.0604267707509608e-05,
+      "loss": 0.4091,
+      "step": 5390
+    },
+    {
+      "epoch": 3.5100845803513336,
+      "grad_norm": 2.445500135421753,
+      "learning_rate": 1.0562098240807989e-05,
+      "loss": 0.4257,
+      "step": 5395
+    },
+    {
+      "epoch": 3.513337670787248,
+      "grad_norm": 1.3687506914138794,
+      "learning_rate": 1.0519990320509104e-05,
+      "loss": 0.4132,
+      "step": 5400
+    },
+    {
+      "epoch": 3.516590761223162,
+      "grad_norm": 1.6088290214538574,
+      "learning_rate": 1.0477944126112097e-05,
+      "loss": 0.4151,
+      "step": 5405
+    },
+    {
+      "epoch": 3.519843851659076,
+      "grad_norm": 1.3225458860397339,
+      "learning_rate": 1.0435959836852967e-05,
+      "loss": 0.4173,
+      "step": 5410
+    },
+    {
+      "epoch": 3.5230969420949902,
+      "grad_norm": 2.5459144115448,
+      "learning_rate": 1.0394037631703867e-05,
+      "loss": 0.4344,
+      "step": 5415
+    },
+    {
+      "epoch": 3.5263500325309045,
+      "grad_norm": 1.7091197967529297,
+      "learning_rate": 1.0352177689372256e-05,
+      "loss": 0.4328,
+      "step": 5420
+    },
+    {
+      "epoch": 3.5296031229668183,
+      "grad_norm": 1.490488052368164,
+      "learning_rate": 1.0310380188300178e-05,
+      "loss": 0.4153,
+      "step": 5425
+    },
+    {
+      "epoch": 3.5328562134027326,
+      "grad_norm": 8.786995887756348,
+      "learning_rate": 1.0268645306663532e-05,
+      "loss": 0.4466,
+      "step": 5430
+    },
+    {
+      "epoch": 3.536109303838647,
+      "grad_norm": 2.5409798622131348,
+      "learning_rate": 1.0226973222371253e-05,
+      "loss": 0.4174,
+      "step": 5435
+    },
+    {
+      "epoch": 3.5393623942745607,
+      "grad_norm": 2.2137346267700195,
+      "learning_rate": 1.0185364113064577e-05,
+      "loss": 0.4296,
+      "step": 5440
+    },
+    {
+      "epoch": 3.542615484710475,
+      "grad_norm": 1.8226664066314697,
+      "learning_rate": 1.0143818156116323e-05,
+      "loss": 0.4076,
+      "step": 5445
+    },
+    {
+      "epoch": 3.545868575146389,
+      "grad_norm": 1.7435983419418335,
+      "learning_rate": 1.0102335528630061e-05,
+      "loss": 0.4241,
+      "step": 5450
+    },
+    {
+      "epoch": 3.5491216655823035,
+      "grad_norm": 2.1171414852142334,
+      "learning_rate": 1.0060916407439413e-05,
+      "loss": 0.4094,
+      "step": 5455
+    },
+    {
+      "epoch": 3.5523747560182173,
+      "grad_norm": 1.3854146003723145,
+      "learning_rate": 1.0019560969107302e-05,
+      "loss": 0.4165,
+      "step": 5460
+    },
+    {
+      "epoch": 3.5556278464541315,
+      "grad_norm": 1.7370872497558594,
+      "learning_rate": 9.978269389925157e-06,
+      "loss": 0.4179,
+      "step": 5465
+    },
+    {
+      "epoch": 3.558880936890046,
+      "grad_norm": 2.7952463626861572,
+      "learning_rate": 9.937041845912188e-06,
+      "loss": 0.4366,
+      "step": 5470
+    },
+    {
+      "epoch": 3.5621340273259596,
+      "grad_norm": 2.453916311264038,
+      "learning_rate": 9.895878512814647e-06,
+      "loss": 0.4105,
+      "step": 5475
+    },
+    {
+      "epoch": 3.565387117761874,
+      "grad_norm": 3.428210973739624,
+      "learning_rate": 9.854779566105068e-06,
+      "loss": 0.4235,
+      "step": 5480
+    },
+    {
+      "epoch": 3.568640208197788,
+      "grad_norm": 1.5725935697555542,
+      "learning_rate": 9.813745180981502e-06,
+      "loss": 0.4165,
+      "step": 5485
+    },
+    {
+      "epoch": 3.571893298633702,
+      "grad_norm": 1.8538109064102173,
+      "learning_rate": 9.772775532366774e-06,
+      "loss": 0.4427,
+      "step": 5490
+    },
+    {
+      "epoch": 3.5751463890696162,
+      "grad_norm": 2.118178606033325,
+      "learning_rate": 9.731870794907789e-06,
+      "loss": 0.4299,
+      "step": 5495
+    },
+    {
+      "epoch": 3.5783994795055305,
+      "grad_norm": 6.738338470458984,
+      "learning_rate": 9.691031142974707e-06,
+      "loss": 0.433,
+      "step": 5500
+    },
+    {
+      "epoch": 3.5816525699414443,
+      "grad_norm": 2.768328905105591,
+      "learning_rate": 9.65025675066025e-06,
+      "loss": 0.4251,
+      "step": 5505
+    },
+    {
+      "epoch": 3.5849056603773586,
+      "grad_norm": 1.6791824102401733,
+      "learning_rate": 9.609547791778964e-06,
+      "loss": 0.4195,
+      "step": 5510
+    },
+    {
+      "epoch": 3.588158750813273,
+      "grad_norm": 1.43679940700531,
+      "learning_rate": 9.568904439866444e-06,
+      "loss": 0.4125,
+      "step": 5515
+    },
+    {
+      "epoch": 3.5914118412491867,
+      "grad_norm": 1.60779869556427,
+      "learning_rate": 9.528326868178616e-06,
+      "loss": 0.4309,
+      "step": 5520
+    },
+    {
+      "epoch": 3.594664931685101,
+      "grad_norm": 2.161658763885498,
+      "learning_rate": 9.487815249691012e-06,
+      "loss": 0.4391,
+      "step": 5525
+    },
+    {
+      "epoch": 3.597918022121015,
+      "grad_norm": 1.7440755367279053,
+      "learning_rate": 9.447369757098002e-06,
+      "loss": 0.4046,
+      "step": 5530
+    },
+    {
+      "epoch": 3.601171112556929,
+      "grad_norm": 1.9223883152008057,
+      "learning_rate": 9.406990562812068e-06,
+      "loss": 0.3904,
+      "step": 5535
+    },
+    {
+      "epoch": 3.6044242029928433,
+      "grad_norm": 1.7541393041610718,
+      "learning_rate": 9.366677838963078e-06,
+      "loss": 0.4238,
+      "step": 5540
+    },
+    {
+      "epoch": 3.6076772934287575,
+      "grad_norm": 1.4891860485076904,
+      "learning_rate": 9.32643175739756e-06,
+      "loss": 0.4173,
+      "step": 5545
+    },
+    {
+      "epoch": 3.6109303838646714,
+      "grad_norm": 2.7674455642700195,
+      "learning_rate": 9.286252489677944e-06,
+      "loss": 0.4033,
+      "step": 5550
+    },
+    {
+      "epoch": 3.6141834743005856,
+      "grad_norm": 2.088984727859497,
+      "learning_rate": 9.246140207081833e-06,
+      "loss": 0.4468,
+      "step": 5555
+    },
+    {
+      "epoch": 3.6174365647365,
+      "grad_norm": 1.516174554824829,
+      "learning_rate": 9.206095080601319e-06,
+      "loss": 0.4194,
+      "step": 5560
+    },
+    {
+      "epoch": 3.6206896551724137,
+      "grad_norm": 2.387362241744995,
+      "learning_rate": 9.16611728094218e-06,
+      "loss": 0.4142,
+      "step": 5565
+    },
+    {
+      "epoch": 3.623942745608328,
+      "grad_norm": 1.8257601261138916,
+      "learning_rate": 9.126206978523202e-06,
+      "loss": 0.4221,
+      "step": 5570
+    },
+    {
+      "epoch": 3.6271958360442422,
+      "grad_norm": 1.815144419670105,
+      "learning_rate": 9.086364343475461e-06,
+      "loss": 0.4356,
+      "step": 5575
+    },
+    {
+      "epoch": 3.630448926480156,
+      "grad_norm": 1.5288091897964478,
+      "learning_rate": 9.04658954564156e-06,
+      "loss": 0.4077,
+      "step": 5580
+    },
+    {
+      "epoch": 3.6337020169160703,
+      "grad_norm": 1.570156455039978,
+      "learning_rate": 9.006882754574914e-06,
+      "loss": 0.4136,
+      "step": 5585
+    },
+    {
+      "epoch": 3.6369551073519846,
+      "grad_norm": 2.161524772644043,
+      "learning_rate": 8.967244139539064e-06,
+      "loss": 0.4204,
+      "step": 5590
+    },
+    {
+      "epoch": 3.6402081977878984,
+      "grad_norm": 1.590711236000061,
+      "learning_rate": 8.927673869506905e-06,
+      "loss": 0.4133,
+      "step": 5595
+    },
+    {
+      "epoch": 3.6434612882238127,
+      "grad_norm": 1.5367971658706665,
+      "learning_rate": 8.888172113159989e-06,
+      "loss": 0.4152,
+      "step": 5600
+    },
+    {
+      "epoch": 3.646714378659727,
+      "grad_norm": 1.5242828130722046,
+      "learning_rate": 8.848739038887822e-06,
+      "loss": 0.4128,
+      "step": 5605
+    },
+    {
+      "epoch": 3.6499674690956407,
+      "grad_norm": 1.4691710472106934,
+      "learning_rate": 8.809374814787124e-06,
+      "loss": 0.4247,
+      "step": 5610
+    },
+    {
+      "epoch": 3.653220559531555,
+      "grad_norm": 3.0192527770996094,
+      "learning_rate": 8.770079608661108e-06,
+      "loss": 0.4279,
+      "step": 5615
+    },
+    {
+      "epoch": 3.6564736499674693,
+      "grad_norm": 1.5646493434906006,
+      "learning_rate": 8.730853588018772e-06,
+      "loss": 0.4055,
+      "step": 5620
+    },
+    {
+      "epoch": 3.659726740403383,
+      "grad_norm": 1.502172589302063,
+      "learning_rate": 8.691696920074214e-06,
+      "loss": 0.4314,
+      "step": 5625
+    },
+    {
+      "epoch": 3.6629798308392973,
+      "grad_norm": 1.4211641550064087,
+      "learning_rate": 8.652609771745862e-06,
+      "loss": 0.4218,
+      "step": 5630
+    },
+    {
+      "epoch": 3.6662329212752116,
+      "grad_norm": 1.905414342880249,
+      "learning_rate": 8.613592309655804e-06,
+      "loss": 0.4178,
+      "step": 5635
+    },
+    {
+      "epoch": 3.6694860117111254,
+      "grad_norm": 2.1291544437408447,
+      "learning_rate": 8.574644700129087e-06,
+      "loss": 0.4224,
+      "step": 5640
+    },
+    {
+      "epoch": 3.6727391021470397,
+      "grad_norm": 1.6421247720718384,
+      "learning_rate": 8.535767109192955e-06,
+      "loss": 0.4378,
+      "step": 5645
+    },
+    {
+      "epoch": 3.675992192582954,
+      "grad_norm": 1.5527359247207642,
+      "learning_rate": 8.496959702576187e-06,
+      "loss": 0.4242,
+      "step": 5650
+    },
+    {
+      "epoch": 3.6792452830188678,
+      "grad_norm": 2.3626434803009033,
+      "learning_rate": 8.458222645708394e-06,
+      "loss": 0.4187,
+      "step": 5655
+    },
+    {
+      "epoch": 3.682498373454782,
+      "grad_norm": 1.3949254751205444,
+      "learning_rate": 8.419556103719279e-06,
+      "loss": 0.4094,
+      "step": 5660
+    },
+    {
+      "epoch": 3.6857514638906963,
+      "grad_norm": 1.7462977170944214,
+      "learning_rate": 8.380960241437947e-06,
+      "loss": 0.4228,
+      "step": 5665
+    },
+    {
+      "epoch": 3.68900455432661,
+      "grad_norm": 1.5994423627853394,
+      "learning_rate": 8.342435223392232e-06,
+      "loss": 0.4216,
+      "step": 5670
+    },
+    {
+      "epoch": 3.6922576447625244,
+      "grad_norm": 2.040391206741333,
+      "learning_rate": 8.303981213807947e-06,
+      "loss": 0.419,
+      "step": 5675
+    },
+    {
+      "epoch": 3.6955107351984386,
+      "grad_norm": 1.5465586185455322,
+      "learning_rate": 8.265598376608211e-06,
+      "loss": 0.4105,
+      "step": 5680
+    },
+    {
+      "epoch": 3.6987638256343525,
+      "grad_norm": 1.965409278869629,
+      "learning_rate": 8.227286875412766e-06,
+      "loss": 0.4056,
+      "step": 5685
+    },
+    {
+      "epoch": 3.7020169160702667,
+      "grad_norm": 1.9103503227233887,
+      "learning_rate": 8.189046873537237e-06,
+      "loss": 0.4142,
+      "step": 5690
+    },
+    {
+      "epoch": 3.705270006506181,
+      "grad_norm": 1.6853419542312622,
+      "learning_rate": 8.150878533992458e-06,
+      "loss": 0.4421,
+      "step": 5695
+    },
+    {
+      "epoch": 3.708523096942095,
+      "grad_norm": 1.7083536386489868,
+      "learning_rate": 8.112782019483813e-06,
+      "loss": 0.4175,
+      "step": 5700
+    },
+    {
+      "epoch": 3.711776187378009,
+      "grad_norm": 1.8322556018829346,
+      "learning_rate": 8.074757492410471e-06,
+      "loss": 0.4202,
+      "step": 5705
+    },
+    {
+      "epoch": 3.7150292778139233,
+      "grad_norm": 1.8045029640197754,
+      "learning_rate": 8.036805114864736e-06,
+      "loss": 0.4232,
+      "step": 5710
+    },
+    {
+      "epoch": 3.718282368249837,
+      "grad_norm": 1.9806746244430542,
+      "learning_rate": 7.998925048631362e-06,
+      "loss": 0.4176,
+      "step": 5715
+    },
+    {
+      "epoch": 3.7215354586857514,
+      "grad_norm": 1.4905263185501099,
+      "learning_rate": 7.96111745518685e-06,
+      "loss": 0.422,
+      "step": 5720
+    },
+    {
+      "epoch": 3.7247885491216657,
+      "grad_norm": 1.4577668905258179,
+      "learning_rate": 7.923382495698758e-06,
+      "loss": 0.406,
+      "step": 5725
+    },
+    {
+      "epoch": 3.7280416395575795,
+      "grad_norm": 2.971972942352295,
+      "learning_rate": 7.88572033102501e-06,
+      "loss": 0.4168,
+      "step": 5730
+    },
+    {
+      "epoch": 3.7312947299934938,
+      "grad_norm": 1.7888929843902588,
+      "learning_rate": 7.848131121713234e-06,
+      "loss": 0.4397,
+      "step": 5735
+    },
+    {
+      "epoch": 3.734547820429408,
+      "grad_norm": 1.5038715600967407,
+      "learning_rate": 7.810615028000045e-06,
+      "loss": 0.4017,
+      "step": 5740
+    },
+    {
+      "epoch": 3.737800910865322,
+      "grad_norm": 1.679883599281311,
+      "learning_rate": 7.773172209810397e-06,
+      "loss": 0.4221,
+      "step": 5745
+    },
+    {
+      "epoch": 3.741054001301236,
+      "grad_norm": 1.669687271118164,
+      "learning_rate": 7.735802826756856e-06,
+      "loss": 0.408,
+      "step": 5750
+    },
+    {
+      "epoch": 3.7443070917371504,
+      "grad_norm": 2.8292086124420166,
+      "learning_rate": 7.698507038138978e-06,
+      "loss": 0.4408,
+      "step": 5755
+    },
+    {
+      "epoch": 3.747560182173064,
+      "grad_norm": 1.6891485452651978,
+      "learning_rate": 7.661285002942572e-06,
+      "loss": 0.4202,
+      "step": 5760
+    },
+    {
+      "epoch": 3.7508132726089785,
+      "grad_norm": 18.631166458129883,
+      "learning_rate": 7.624136879839053e-06,
+      "loss": 0.4368,
+      "step": 5765
+    },
+    {
+      "epoch": 3.7540663630448927,
+      "grad_norm": 2.31217622756958,
+      "learning_rate": 7.5870628271847765e-06,
+      "loss": 0.4099,
+      "step": 5770
+    },
+    {
+      "epoch": 3.7573194534808065,
+      "grad_norm": 1.7663567066192627,
+      "learning_rate": 7.550063003020333e-06,
+      "loss": 0.4242,
+      "step": 5775
+    },
+    {
+      "epoch": 3.760572543916721,
+      "grad_norm": 1.651605486869812,
+      "learning_rate": 7.5131375650698835e-06,
+      "loss": 0.428,
+      "step": 5780
+    },
+    {
+      "epoch": 3.763825634352635,
+      "grad_norm": 1.542211890220642,
+      "learning_rate": 7.476286670740518e-06,
+      "loss": 0.4163,
+      "step": 5785
+    },
+    {
+      "epoch": 3.767078724788549,
+      "grad_norm": 1.7693531513214111,
+      "learning_rate": 7.439510477121536e-06,
+      "loss": 0.4192,
+      "step": 5790
+    },
+    {
+      "epoch": 3.770331815224463,
+      "grad_norm": 2.135603666305542,
+      "learning_rate": 7.402809140983799e-06,
+      "loss": 0.4155,
+      "step": 5795
+    },
+    {
+      "epoch": 3.7735849056603774,
+      "grad_norm": 2.1200711727142334,
+      "learning_rate": 7.366182818779088e-06,
+      "loss": 0.4056,
+      "step": 5800
+    },
+    {
+      "epoch": 3.7768379960962912,
+      "grad_norm": 2.6680068969726562,
+      "learning_rate": 7.329631666639392e-06,
+      "loss": 0.4124,
+      "step": 5805
+    },
+    {
+      "epoch": 3.7800910865322055,
+      "grad_norm": 2.2767679691314697,
+      "learning_rate": 7.2931558403762535e-06,
+      "loss": 0.3993,
+      "step": 5810
+    },
+    {
+      "epoch": 3.7833441769681198,
+      "grad_norm": 1.7897844314575195,
+      "learning_rate": 7.256755495480141e-06,
+      "loss": 0.3827,
+      "step": 5815
+    },
+    {
+      "epoch": 3.7865972674040336,
+      "grad_norm": 1.7870675325393677,
+      "learning_rate": 7.220430787119742e-06,
+      "loss": 0.4199,
+      "step": 5820
+    },
+    {
+      "epoch": 3.789850357839948,
+      "grad_norm": 1.4809794425964355,
+      "learning_rate": 7.184181870141307e-06,
+      "loss": 0.4056,
+      "step": 5825
+    },
+    {
+      "epoch": 3.793103448275862,
+      "grad_norm": 2.3490641117095947,
+      "learning_rate": 7.148008899068029e-06,
+      "loss": 0.4084,
+      "step": 5830
+    },
+    {
+      "epoch": 3.796356538711776,
+      "grad_norm": 2.2586324214935303,
+      "learning_rate": 7.1119120280993295e-06,
+      "loss": 0.4125,
+      "step": 5835
+    },
+    {
+      "epoch": 3.79960962914769,
+      "grad_norm": 3.2746353149414062,
+      "learning_rate": 7.0758914111102335e-06,
+      "loss": 0.3964,
+      "step": 5840
+    },
+    {
+      "epoch": 3.8028627195836044,
+      "grad_norm": 1.9759531021118164,
+      "learning_rate": 7.039947201650726e-06,
+      "loss": 0.4151,
+      "step": 5845
+    },
+    {
+      "epoch": 3.8061158100195187,
+      "grad_norm": 1.395560622215271,
+      "learning_rate": 7.004079552945062e-06,
+      "loss": 0.415,
+      "step": 5850
+    },
+    {
+      "epoch": 3.8093689004554325,
+      "grad_norm": 1.6155577898025513,
+      "learning_rate": 6.968288617891116e-06,
+      "loss": 0.4093,
+      "step": 5855
+    },
+    {
+      "epoch": 3.812621990891347,
+      "grad_norm": 1.647705078125,
+      "learning_rate": 6.932574549059789e-06,
+      "loss": 0.4033,
+      "step": 5860
+    },
+    {
+      "epoch": 3.815875081327261,
+      "grad_norm": 1.5718135833740234,
+      "learning_rate": 6.8969374986942784e-06,
+      "loss": 0.4281,
+      "step": 5865
+    },
+    {
+      "epoch": 3.819128171763175,
+      "grad_norm": 6.746004581451416,
+      "learning_rate": 6.861377618709466e-06,
+      "loss": 0.4199,
+      "step": 5870
+    },
+    {
+      "epoch": 3.822381262199089,
+      "grad_norm": 1.6057871580123901,
+      "learning_rate": 6.825895060691273e-06,
+      "loss": 0.4059,
+      "step": 5875
+    },
+    {
+      "epoch": 3.8256343526350034,
+      "grad_norm": 1.5333271026611328,
+      "learning_rate": 6.790489975896033e-06,
+      "loss": 0.4311,
+      "step": 5880
+    },
+    {
+      "epoch": 3.828887443070917,
+      "grad_norm": 4.261138439178467,
+      "learning_rate": 6.755162515249799e-06,
+      "loss": 0.3987,
+      "step": 5885
+    },
+    {
+      "epoch": 3.8321405335068315,
+      "grad_norm": 1.475321650505066,
+      "learning_rate": 6.719912829347716e-06,
+      "loss": 0.424,
+      "step": 5890
+    },
+    {
+      "epoch": 3.8353936239427457,
+      "grad_norm": 1.6494297981262207,
+      "learning_rate": 6.6847410684534235e-06,
+      "loss": 0.4367,
+      "step": 5895
+    },
+    {
+      "epoch": 3.8386467143786596,
+      "grad_norm": 1.8216992616653442,
+      "learning_rate": 6.649647382498345e-06,
+      "loss": 0.4201,
+      "step": 5900
+    },
+    {
+      "epoch": 3.841899804814574,
+      "grad_norm": 1.7172224521636963,
+      "learning_rate": 6.6146319210810935e-06,
+      "loss": 0.4279,
+      "step": 5905
+    },
+    {
+      "epoch": 3.845152895250488,
+      "grad_norm": 2.8617303371429443,
+      "learning_rate": 6.579694833466843e-06,
+      "loss": 0.4219,
+      "step": 5910
+    },
+    {
+      "epoch": 3.8484059856864024,
+      "grad_norm": 1.705623984336853,
+      "learning_rate": 6.5448362685866485e-06,
+      "loss": 0.4085,
+      "step": 5915
+    },
+    {
+      "epoch": 3.851659076122316,
+      "grad_norm": 1.7146087884902954,
+      "learning_rate": 6.510056375036841e-06,
+      "loss": 0.4089,
+      "step": 5920
+    },
+    {
+      "epoch": 3.8549121665582304,
+      "grad_norm": 1.4763227701187134,
+      "learning_rate": 6.47535530107839e-06,
+      "loss": 0.408,
+      "step": 5925
+    },
+    {
+      "epoch": 3.8581652569941447,
+      "grad_norm": 2.6685636043548584,
+      "learning_rate": 6.440733194636281e-06,
+      "loss": 0.4216,
+      "step": 5930
+    },
+    {
+      "epoch": 3.8614183474300585,
+      "grad_norm": 1.5533685684204102,
+      "learning_rate": 6.406190203298859e-06,
+      "loss": 0.4182,
+      "step": 5935
+    },
+    {
+      "epoch": 3.864671437865973,
+      "grad_norm": 1.7280486822128296,
+      "learning_rate": 6.3717264743172134e-06,
+      "loss": 0.415,
+      "step": 5940
+    },
+    {
+      "epoch": 3.867924528301887,
+      "grad_norm": 1.7348867654800415,
+      "learning_rate": 6.337342154604573e-06,
+      "loss": 0.4046,
+      "step": 5945
+    },
+    {
+      "epoch": 3.871177618737801,
+      "grad_norm": 1.730487585067749,
+      "learning_rate": 6.303037390735634e-06,
+      "loss": 0.4186,
+      "step": 5950
+    },
+    {
+      "epoch": 3.874430709173715,
+      "grad_norm": 1.4803285598754883,
+      "learning_rate": 6.268812328945961e-06,
+      "loss": 0.4071,
+      "step": 5955
+    },
+    {
+      "epoch": 3.8776837996096294,
+      "grad_norm": 1.5369083881378174,
+      "learning_rate": 6.234667115131382e-06,
+      "loss": 0.4205,
+      "step": 5960
+    },
+    {
+      "epoch": 3.880936890045543,
+      "grad_norm": 2.9555954933166504,
+      "learning_rate": 6.200601894847324e-06,
+      "loss": 0.4054,
+      "step": 5965
+    },
+    {
+      "epoch": 3.8841899804814575,
+      "grad_norm": 1.5368090867996216,
+      "learning_rate": 6.166616813308221e-06,
+      "loss": 0.3972,
+      "step": 5970
+    },
+    {
+      "epoch": 3.8874430709173717,
+      "grad_norm": 1.577239990234375,
+      "learning_rate": 6.132712015386902e-06,
+      "loss": 0.4253,
+      "step": 5975
+    },
+    {
+      "epoch": 3.8906961613532856,
+      "grad_norm": 1.5454272031784058,
+      "learning_rate": 6.098887645613943e-06,
+      "loss": 0.4081,
+      "step": 5980
+    },
+    {
+      "epoch": 3.8939492517892,
+      "grad_norm": 1.506635069847107,
+      "learning_rate": 6.065143848177066e-06,
+      "loss": 0.4041,
+      "step": 5985
+    },
+    {
+      "epoch": 3.897202342225114,
+      "grad_norm": 5.971263408660889,
+      "learning_rate": 6.03148076692055e-06,
+      "loss": 0.4031,
+      "step": 5990
+    },
+    {
+      "epoch": 3.900455432661028,
+      "grad_norm": 2.4734601974487305,
+      "learning_rate": 5.997898545344571e-06,
+      "loss": 0.43,
+      "step": 5995
+    },
+    {
+      "epoch": 3.903708523096942,
+      "grad_norm": 2.6702654361724854,
+      "learning_rate": 5.9643973266046145e-06,
+      "loss": 0.416,
+      "step": 6000
+    },
+    {
+      "epoch": 3.9069616135328564,
+      "grad_norm": 1.439784288406372,
+      "learning_rate": 5.930977253510886e-06,
+      "loss": 0.4059,
+      "step": 6005
+    },
+    {
+      "epoch": 3.9102147039687702,
+      "grad_norm": 1.8066974878311157,
+      "learning_rate": 5.897638468527653e-06,
+      "loss": 0.4147,
+      "step": 6010
+    },
+    {
+      "epoch": 3.9134677944046845,
+      "grad_norm": 1.603287935256958,
+      "learning_rate": 5.864381113772685e-06,
+      "loss": 0.4168,
+      "step": 6015
+    },
+    {
+      "epoch": 3.9167208848405988,
+      "grad_norm": 1.6566202640533447,
+      "learning_rate": 5.831205331016612e-06,
+      "loss": 0.4095,
+      "step": 6020
+    },
+    {
+      "epoch": 3.9199739752765126,
+      "grad_norm": 2.173736095428467,
+      "learning_rate": 5.798111261682357e-06,
+      "loss": 0.3955,
+      "step": 6025
+    },
+    {
+      "epoch": 3.923227065712427,
+      "grad_norm": 1.4487274885177612,
+      "learning_rate": 5.765099046844491e-06,
+      "loss": 0.4048,
+      "step": 6030
+    },
+    {
+      "epoch": 3.926480156148341,
+      "grad_norm": 1.4893627166748047,
+      "learning_rate": 5.7321688272286596e-06,
+      "loss": 0.3908,
+      "step": 6035
+    },
+    {
+      "epoch": 3.929733246584255,
+      "grad_norm": 1.9365359544754028,
+      "learning_rate": 5.699320743210984e-06,
+      "loss": 0.4141,
+      "step": 6040
+    },
+    {
+      "epoch": 3.932986337020169,
+      "grad_norm": 1.6186310052871704,
+      "learning_rate": 5.666554934817447e-06,
+      "loss": 0.3887,
+      "step": 6045
+    },
+    {
+      "epoch": 3.9362394274560835,
+      "grad_norm": 7.201216697692871,
+      "learning_rate": 5.633871541723295e-06,
+      "loss": 0.4115,
+      "step": 6050
+    },
+    {
+      "epoch": 3.9394925178919973,
+      "grad_norm": 1.8057231903076172,
+      "learning_rate": 5.601270703252481e-06,
+      "loss": 0.3989,
+      "step": 6055
+    },
+    {
+      "epoch": 3.9427456083279115,
+      "grad_norm": 2.119253158569336,
+      "learning_rate": 5.5687525583770135e-06,
+      "loss": 0.4295,
+      "step": 6060
+    },
+    {
+      "epoch": 3.945998698763826,
+      "grad_norm": 1.8245245218276978,
+      "learning_rate": 5.536317245716391e-06,
+      "loss": 0.4213,
+      "step": 6065
+    },
+    {
+      "epoch": 3.9492517891997396,
+      "grad_norm": 1.69149649143219,
+      "learning_rate": 5.503964903537037e-06,
+      "loss": 0.4128,
+      "step": 6070
+    },
+    {
+      "epoch": 3.952504879635654,
+      "grad_norm": 1.5030890703201294,
+      "learning_rate": 5.471695669751664e-06,
+      "loss": 0.3899,
+      "step": 6075
+    },
+    {
+      "epoch": 3.955757970071568,
+      "grad_norm": 1.764691710472107,
+      "learning_rate": 5.439509681918703e-06,
+      "loss": 0.4024,
+      "step": 6080
+    },
+    {
+      "epoch": 3.959011060507482,
+      "grad_norm": 1.7484804391860962,
+      "learning_rate": 5.407407077241749e-06,
+      "loss": 0.4088,
+      "step": 6085
+    },
+    {
+      "epoch": 3.9622641509433962,
+      "grad_norm": 1.4906771183013916,
+      "learning_rate": 5.381785121346411e-06,
+      "loss": 0.4077,
+      "step": 6090
+    },
+    {
+      "epoch": 3.9655172413793105,
+      "grad_norm": 2.1942009925842285,
+      "learning_rate": 5.349832950968298e-06,
+      "loss": 0.4025,
+      "step": 6095
+    },
+    {
+      "epoch": 3.9687703318152243,
+      "grad_norm": 1.9535316228866577,
+      "learning_rate": 5.3179645460233574e-06,
+      "loss": 0.4079,
+      "step": 6100
+    },
+    {
+      "epoch": 3.9720234222511386,
+      "grad_norm": 1.6382977962493896,
+      "learning_rate": 5.286180042361361e-06,
+      "loss": 0.4072,
+      "step": 6105
+    },
+    {
+      "epoch": 3.975276512687053,
+      "grad_norm": 2.194603204727173,
+      "learning_rate": 5.254479575474411e-06,
+      "loss": 0.4147,
+      "step": 6110
+    },
+    {
+      "epoch": 3.9785296031229667,
+      "grad_norm": 2.237362861633301,
+      "learning_rate": 5.222863280496406e-06,
+      "loss": 0.4174,
+      "step": 6115
+    },
+    {
+      "epoch": 3.981782693558881,
+      "grad_norm": 1.5425729751586914,
+      "learning_rate": 5.191331292202409e-06,
+      "loss": 0.3867,
+      "step": 6120
+    },
+    {
+      "epoch": 3.985035783994795,
+      "grad_norm": 3.286719799041748,
+      "learning_rate": 5.159883745008099e-06,
+      "loss": 0.4134,
+      "step": 6125
+    },
+    {
+      "epoch": 3.988288874430709,
+      "grad_norm": 1.8072408437728882,
+      "learning_rate": 5.1285207729692146e-06,
+      "loss": 0.409,
+      "step": 6130
+    },
+    {
+      "epoch": 3.9915419648666233,
+      "grad_norm": 2.3535733222961426,
+      "learning_rate": 5.097242509780945e-06,
+      "loss": 0.3986,
+      "step": 6135
+    },
+    {
+      "epoch": 3.9947950553025375,
+      "grad_norm": 2.018533706665039,
+      "learning_rate": 5.06604908877738e-06,
+      "loss": 0.4115,
+      "step": 6140
+    },
+    {
+      "epoch": 3.9980481457384514,
+      "grad_norm": 1.9201387166976929,
+      "learning_rate": 5.03494064293096e-06,
+      "loss": 0.4206,
+      "step": 6145
+    },
+    {
+      "epoch": 4.0,
+      "eval_f1": 0.8072647932831747,
+      "eval_loss": 0.55810546875,
+      "eval_precision": 0.8068495156761535,
+      "eval_recall": 0.8078495084242743,
+      "eval_runtime": 352.5087,
+      "eval_samples_per_second": 1116.1,
+      "eval_steps_per_second": 1.092,
+      "step": 6148
+    },
+    {
+      "epoch": 4.001301236174366,
+      "grad_norm": 1.1841827630996704,
+      "learning_rate": 5.003917304851868e-06,
+      "loss": 0.3531,
+      "step": 6150
+    },
+    {
+      "epoch": 4.00455432661028,
+      "grad_norm": 1.870800495147705,
+      "learning_rate": 4.972979206787503e-06,
+      "loss": 0.2658,
+      "step": 6155
+    },
+    {
+      "epoch": 4.007807417046194,
+      "grad_norm": 6.779404163360596,
+      "learning_rate": 4.9421264806218865e-06,
+      "loss": 0.2545,
+      "step": 6160
+    },
+    {
+      "epoch": 4.011060507482108,
+      "grad_norm": 1.2459709644317627,
+      "learning_rate": 4.911359257875131e-06,
+      "loss": 0.2436,
+      "step": 6165
+    },
+    {
+      "epoch": 4.014313597918022,
+      "grad_norm": 1.5694559812545776,
+      "learning_rate": 4.880677669702846e-06,
+      "loss": 0.2433,
+      "step": 6170
+    },
+    {
+      "epoch": 4.017566688353936,
+      "grad_norm": 1.46892249584198,
+      "learning_rate": 4.850081846895591e-06,
+      "loss": 0.2388,
+      "step": 6175
+    },
+    {
+      "epoch": 4.020819778789851,
+      "grad_norm": 32.99214553833008,
+      "learning_rate": 4.819571919878346e-06,
+      "loss": 0.2353,
+      "step": 6180
+    },
+    {
+      "epoch": 4.024072869225765,
+      "grad_norm": 1.623369574546814,
+      "learning_rate": 4.78914801870991e-06,
+      "loss": 0.2389,
+      "step": 6185
+    },
+    {
+      "epoch": 4.027325959661678,
+      "grad_norm": 1.8111308813095093,
+      "learning_rate": 4.7588102730823676e-06,
+      "loss": 0.2362,
+      "step": 6190
+    },
+    {
+      "epoch": 4.030579050097593,
+      "grad_norm": 1.494247555732727,
+      "learning_rate": 4.7285588123205546e-06,
+      "loss": 0.2411,
+      "step": 6195
+    },
+    {
+      "epoch": 4.033832140533507,
+      "grad_norm": 1.8853955268859863,
+      "learning_rate": 4.698393765381473e-06,
+      "loss": 0.2434,
+      "step": 6200
+    },
+    {
+      "epoch": 4.037085230969421,
+      "grad_norm": 2.163872241973877,
+      "learning_rate": 4.668315260853753e-06,
+      "loss": 0.2634,
+      "step": 6205
+    },
+    {
+      "epoch": 4.040338321405335,
+      "grad_norm": 2.3891079425811768,
+      "learning_rate": 4.6383234269571305e-06,
+      "loss": 0.2346,
+      "step": 6210
+    },
+    {
+      "epoch": 4.043591411841249,
+      "grad_norm": 3.2441794872283936,
+      "learning_rate": 4.608418391541861e-06,
+      "loss": 0.25,
+      "step": 6215
+    },
+    {
+      "epoch": 4.046844502277163,
+      "grad_norm": 2.0283098220825195,
+      "learning_rate": 4.578600282088186e-06,
+      "loss": 0.241,
+      "step": 6220
+    },
+    {
+      "epoch": 4.050097592713078,
+      "grad_norm": 1.6591633558273315,
+      "learning_rate": 4.548869225705821e-06,
+      "loss": 0.2366,
+      "step": 6225
+    },
+    {
+      "epoch": 4.053350683148992,
+      "grad_norm": 1.9975348711013794,
+      "learning_rate": 4.5192253491333656e-06,
+      "loss": 0.2287,
+      "step": 6230
+    },
+    {
+      "epoch": 4.056603773584905,
+      "grad_norm": 1.8785626888275146,
+      "learning_rate": 4.489668778737793e-06,
+      "loss": 0.2329,
+      "step": 6235
+    },
+    {
+      "epoch": 4.05985686402082,
+      "grad_norm": 1.7227510213851929,
+      "learning_rate": 4.460199640513912e-06,
+      "loss": 0.2456,
+      "step": 6240
+    },
+    {
+      "epoch": 4.063109954456734,
+      "grad_norm": 2.2654736042022705,
+      "learning_rate": 4.430818060083816e-06,
+      "loss": 0.2271,
+      "step": 6245
+    },
+    {
+      "epoch": 4.066363044892648,
+      "grad_norm": 1.5479950904846191,
+      "learning_rate": 4.4015241626963436e-06,
+      "loss": 0.2404,
+      "step": 6250
+    },
+    {
+      "epoch": 4.0696161353285625,
+      "grad_norm": 2.179954767227173,
+      "learning_rate": 4.372318073226583e-06,
+      "loss": 0.2358,
+      "step": 6255
+    },
+    {
+      "epoch": 4.072869225764476,
+      "grad_norm": 1.5730780363082886,
+      "learning_rate": 4.343199916175284e-06,
+      "loss": 0.2256,
+      "step": 6260
+    },
+    {
+      "epoch": 4.07612231620039,
+      "grad_norm": 1.6954352855682373,
+      "learning_rate": 4.3141698156683645e-06,
+      "loss": 0.2302,
+      "step": 6265
+    },
+    {
+      "epoch": 4.079375406636305,
+      "grad_norm": 2.26914644241333,
+      "learning_rate": 4.285227895456373e-06,
+      "loss": 0.2262,
+      "step": 6270
+    },
+    {
+      "epoch": 4.082628497072219,
+      "grad_norm": 1.8111881017684937,
+      "learning_rate": 4.2563742789139635e-06,
+      "loss": 0.2368,
+      "step": 6275
+    },
+    {
+      "epoch": 4.0858815875081325,
+      "grad_norm": 2.6787023544311523,
+      "learning_rate": 4.227609089039361e-06,
+      "loss": 0.2299,
+      "step": 6280
+    },
+    {
+      "epoch": 4.089134677944047,
+      "grad_norm": 1.6558953523635864,
+      "learning_rate": 4.198932448453832e-06,
+      "loss": 0.2502,
+      "step": 6285
+    },
+    {
+      "epoch": 4.092387768379961,
+      "grad_norm": 2.19753360748291,
+      "learning_rate": 4.170344479401203e-06,
+      "loss": 0.2229,
+      "step": 6290
+    },
+    {
+      "epoch": 4.095640858815875,
+      "grad_norm": 1.825623631477356,
+      "learning_rate": 4.14184530374728e-06,
+      "loss": 0.2334,
+      "step": 6295
+    },
+    {
+      "epoch": 4.0988939492517895,
+      "grad_norm": 5.066878795623779,
+      "learning_rate": 4.113435042979357e-06,
+      "loss": 0.2323,
+      "step": 6300
+    },
+    {
+      "epoch": 4.102147039687703,
+      "grad_norm": 1.8815892934799194,
+      "learning_rate": 4.085113818205724e-06,
+      "loss": 0.2439,
+      "step": 6305
+    },
+    {
+      "epoch": 4.105400130123617,
+      "grad_norm": 1.726470708847046,
+      "learning_rate": 4.056881750155095e-06,
+      "loss": 0.2469,
+      "step": 6310
+    },
+    {
+      "epoch": 4.108653220559532,
+      "grad_norm": 2.11625075340271,
+      "learning_rate": 4.028738959176143e-06,
+      "loss": 0.234,
+      "step": 6315
+    },
+    {
+      "epoch": 4.111906310995446,
+      "grad_norm": 1.8043195009231567,
+      "learning_rate": 4.000685565236953e-06,
+      "loss": 0.235,
+      "step": 6320
+    },
+    {
+      "epoch": 4.1151594014313595,
+      "grad_norm": 1.5731197595596313,
+      "learning_rate": 3.972721687924546e-06,
+      "loss": 0.2369,
+      "step": 6325
+    },
+    {
+      "epoch": 4.118412491867274,
+      "grad_norm": 2.3870668411254883,
+      "learning_rate": 3.94484744644433e-06,
+      "loss": 0.2354,
+      "step": 6330
+    },
+    {
+      "epoch": 4.121665582303188,
+      "grad_norm": 2.9168872833251953,
+      "learning_rate": 3.917062959619611e-06,
+      "loss": 0.2391,
+      "step": 6335
+    },
+    {
+      "epoch": 4.124918672739102,
+      "grad_norm": 2.253788709640503,
+      "learning_rate": 3.889368345891101e-06,
+      "loss": 0.2414,
+      "step": 6340
+    },
+    {
+      "epoch": 4.1281717631750166,
+      "grad_norm": 1.9010400772094727,
+      "learning_rate": 3.86176372331638e-06,
+      "loss": 0.2385,
+      "step": 6345
+    },
+    {
+      "epoch": 4.13142485361093,
+      "grad_norm": 2.5451037883758545,
+      "learning_rate": 3.834249209569415e-06,
+      "loss": 0.2434,
+      "step": 6350
+    },
+    {
+      "epoch": 4.134677944046844,
+      "grad_norm": 1.9200527667999268,
+      "learning_rate": 3.806824921940069e-06,
+      "loss": 0.2327,
+      "step": 6355
+    },
+    {
+      "epoch": 4.137931034482759,
+      "grad_norm": 1.741262435913086,
+      "learning_rate": 3.7794909773335664e-06,
+      "loss": 0.2218,
+      "step": 6360
+    },
+    {
+      "epoch": 4.141184124918673,
+      "grad_norm": 2.2827839851379395,
+      "learning_rate": 3.752247492270017e-06,
+      "loss": 0.238,
+      "step": 6365
+    },
+    {
+      "epoch": 4.1444372153545865,
+      "grad_norm": 2.077956199645996,
+      "learning_rate": 3.7250945828839286e-06,
+      "loss": 0.2371,
+      "step": 6370
+    },
+    {
+      "epoch": 4.147690305790501,
+      "grad_norm": 1.9408165216445923,
+      "learning_rate": 3.6980323649236925e-06,
+      "loss": 0.2241,
+      "step": 6375
+    },
+    {
+      "epoch": 4.150943396226415,
+      "grad_norm": 2.407376766204834,
+      "learning_rate": 3.671060953751085e-06,
+      "loss": 0.2444,
+      "step": 6380
+    },
+    {
+      "epoch": 4.154196486662329,
+      "grad_norm": 1.6587666273117065,
+      "learning_rate": 3.6441804643408156e-06,
+      "loss": 0.2388,
+      "step": 6385
+    },
+    {
+      "epoch": 4.157449577098244,
+      "grad_norm": 1.826478362083435,
+      "learning_rate": 3.617391011279986e-06,
+      "loss": 0.2361,
+      "step": 6390
+    },
+    {
+      "epoch": 4.160702667534157,
+      "grad_norm": 1.628882646560669,
+      "learning_rate": 3.590692708767626e-06,
+      "loss": 0.2246,
+      "step": 6395
+    },
+    {
+      "epoch": 4.163955757970071,
+      "grad_norm": 1.8005129098892212,
+      "learning_rate": 3.5640856706142283e-06,
+      "loss": 0.2435,
+      "step": 6400
+    },
+    {
+      "epoch": 4.167208848405986,
+      "grad_norm": 1.8075467348098755,
+      "learning_rate": 3.5375700102412118e-06,
+      "loss": 0.2283,
+      "step": 6405
+    },
+    {
+      "epoch": 4.1704619388419,
+      "grad_norm": 1.7112232446670532,
+      "learning_rate": 3.51114584068048e-06,
+      "loss": 0.2428,
+      "step": 6410
+    },
+    {
+      "epoch": 4.173715029277814,
+      "grad_norm": 3.4661080837249756,
+      "learning_rate": 3.484813274573931e-06,
+      "loss": 0.2221,
+      "step": 6415
+    },
+    {
+      "epoch": 4.176968119713728,
+      "grad_norm": 1.970160961151123,
+      "learning_rate": 3.458572424172962e-06,
+      "loss": 0.2409,
+      "step": 6420
+    },
+    {
+      "epoch": 4.180221210149642,
+      "grad_norm": 2.032346248626709,
+      "learning_rate": 3.432423401338014e-06,
+      "loss": 0.2259,
+      "step": 6425
+    },
+    {
+      "epoch": 4.183474300585556,
+      "grad_norm": 2.329317808151245,
+      "learning_rate": 3.4063663175380622e-06,
+      "loss": 0.2277,
+      "step": 6430
+    },
+    {
+      "epoch": 4.186727391021471,
+      "grad_norm": 2.001601457595825,
+      "learning_rate": 3.3804012838501877e-06,
+      "loss": 0.2392,
+      "step": 6435
+    },
+    {
+      "epoch": 4.189980481457384,
+      "grad_norm": 2.017449140548706,
+      "learning_rate": 3.354528410959054e-06,
+      "loss": 0.2223,
+      "step": 6440
+    },
+    {
+      "epoch": 4.193233571893298,
+      "grad_norm": 2.0057036876678467,
+      "learning_rate": 3.3287478091564628e-06,
+      "loss": 0.2282,
+      "step": 6445
+    },
+    {
+      "epoch": 4.196486662329213,
+      "grad_norm": 1.7296228408813477,
+      "learning_rate": 3.3030595883408953e-06,
+      "loss": 0.2234,
+      "step": 6450
+    },
+    {
+      "epoch": 4.199739752765127,
+      "grad_norm": 1.9821724891662598,
+      "learning_rate": 3.2774638580170075e-06,
+      "loss": 0.2491,
+      "step": 6455
+    },
+    {
+      "epoch": 4.202992843201041,
+      "grad_norm": 1.769161581993103,
+      "learning_rate": 3.2519607272951862e-06,
+      "loss": 0.2342,
+      "step": 6460
+    },
+    {
+      "epoch": 4.206245933636955,
+      "grad_norm": 1.7646567821502686,
+      "learning_rate": 3.226550304891099e-06,
+      "loss": 0.231,
+      "step": 6465
+    },
+    {
+      "epoch": 4.209499024072869,
+      "grad_norm": 1.8771380186080933,
+      "learning_rate": 3.201232699125198e-06,
+      "loss": 0.2246,
+      "step": 6470
+    },
+    {
+      "epoch": 4.212752114508783,
+      "grad_norm": 7.158396244049072,
+      "learning_rate": 3.1760080179222663e-06,
+      "loss": 0.2281,
+      "step": 6475
+    },
+    {
+      "epoch": 4.216005204944698,
+      "grad_norm": 2.6318135261535645,
+      "learning_rate": 3.15087636881099e-06,
+      "loss": 0.2326,
+      "step": 6480
+    },
+    {
+      "epoch": 4.2192582953806115,
+      "grad_norm": 1.8173600435256958,
+      "learning_rate": 3.125837858923453e-06,
+      "loss": 0.2236,
+      "step": 6485
+    },
+    {
+      "epoch": 4.222511385816525,
+      "grad_norm": 1.9394645690917969,
+      "learning_rate": 3.100892594994706e-06,
+      "loss": 0.2356,
+      "step": 6490
+    },
+    {
+      "epoch": 4.22576447625244,
+      "grad_norm": 2.29449725151062,
+      "learning_rate": 3.076040683362308e-06,
+      "loss": 0.2336,
+      "step": 6495
+    },
+    {
+      "epoch": 4.229017566688354,
+      "grad_norm": 1.8967347145080566,
+      "learning_rate": 3.0512822299658824e-06,
+      "loss": 0.2251,
+      "step": 6500
+    },
+    {
+      "epoch": 4.232270657124268,
+      "grad_norm": 1.6016850471496582,
+      "learning_rate": 3.0266173403466438e-06,
+      "loss": 0.2119,
+      "step": 6505
+    },
+    {
+      "epoch": 4.235523747560182,
+      "grad_norm": 1.7070870399475098,
+      "learning_rate": 3.002046119646959e-06,
+      "loss": 0.2228,
+      "step": 6510
+    },
+    {
+      "epoch": 4.238776837996096,
+      "grad_norm": 2.9714415073394775,
+      "learning_rate": 2.977568672609915e-06,
+      "loss": 0.2317,
+      "step": 6515
+    },
+    {
+      "epoch": 4.24202992843201,
+      "grad_norm": 1.7705700397491455,
+      "learning_rate": 2.95318510357884e-06,
+      "loss": 0.2135,
+      "step": 6520
+    },
+    {
+      "epoch": 4.245283018867925,
+      "grad_norm": 1.5718425512313843,
+      "learning_rate": 2.9288955164968766e-06,
+      "loss": 0.2309,
+      "step": 6525
+    },
+    {
+      "epoch": 4.2485361093038385,
+      "grad_norm": 6.268550872802734,
+      "learning_rate": 2.904700014906553e-06,
+      "loss": 0.2259,
+      "step": 6530
+    },
+    {
+      "epoch": 4.251789199739752,
+      "grad_norm": 10.964041709899902,
+      "learning_rate": 2.8805987019493137e-06,
+      "loss": 0.2255,
+      "step": 6535
+    },
+    {
+      "epoch": 4.255042290175667,
+      "grad_norm": 6.053678512573242,
+      "learning_rate": 2.8565916803650866e-06,
+      "loss": 0.2171,
+      "step": 6540
+    },
+    {
+      "epoch": 4.258295380611581,
+      "grad_norm": 1.9141877889633179,
+      "learning_rate": 2.8326790524918765e-06,
+      "loss": 0.2229,
+      "step": 6545
+    },
+    {
+      "epoch": 4.261548471047496,
+      "grad_norm": 1.7668460607528687,
+      "learning_rate": 2.8088609202652742e-06,
+      "loss": 0.2281,
+      "step": 6550
+    },
+    {
+      "epoch": 4.264801561483409,
+      "grad_norm": 4.9379048347473145,
+      "learning_rate": 2.7851373852180617e-06,
+      "loss": 0.228,
+      "step": 6555
+    },
+    {
+      "epoch": 4.268054651919323,
+      "grad_norm": 1.8191304206848145,
+      "learning_rate": 2.761508548479777e-06,
+      "loss": 0.2223,
+      "step": 6560
+    },
+    {
+      "epoch": 4.271307742355237,
+      "grad_norm": 3.2812600135803223,
+      "learning_rate": 2.7379745107762726e-06,
+      "loss": 0.2254,
+      "step": 6565
+    },
+    {
+      "epoch": 4.274560832791152,
+      "grad_norm": 1.9550873041152954,
+      "learning_rate": 2.7145353724292776e-06,
+      "loss": 0.2245,
+      "step": 6570
+    },
+    {
+      "epoch": 4.2778139232270656,
+      "grad_norm": 1.6830195188522339,
+      "learning_rate": 2.691191233355986e-06,
+      "loss": 0.2343,
+      "step": 6575
+    },
+    {
+      "epoch": 4.28106701366298,
+      "grad_norm": 1.9676400423049927,
+      "learning_rate": 2.6679421930686317e-06,
+      "loss": 0.2418,
+      "step": 6580
+    },
+    {
+      "epoch": 4.284320104098894,
+      "grad_norm": 2.293408155441284,
+      "learning_rate": 2.644788350674049e-06,
+      "loss": 0.2352,
+      "step": 6585
+    },
+    {
+      "epoch": 4.287573194534808,
+      "grad_norm": 2.1475830078125,
+      "learning_rate": 2.6217298048732604e-06,
+      "loss": 0.2263,
+      "step": 6590
+    },
+    {
+      "epoch": 4.290826284970722,
+      "grad_norm": 1.7658709287643433,
+      "learning_rate": 2.598766653961068e-06,
+      "loss": 0.2369,
+      "step": 6595
+    },
+    {
+      "epoch": 4.294079375406636,
+      "grad_norm": 1.8443782329559326,
+      "learning_rate": 2.5758989958256043e-06,
+      "loss": 0.2362,
+      "step": 6600
+    },
+    {
+      "epoch": 4.29733246584255,
+      "grad_norm": 1.6136929988861084,
+      "learning_rate": 2.5531269279479325e-06,
+      "loss": 0.2475,
+      "step": 6605
+    },
+    {
+      "epoch": 4.300585556278465,
+      "grad_norm": 1.7001075744628906,
+      "learning_rate": 2.530450547401647e-06,
+      "loss": 0.2257,
+      "step": 6610
+    },
+    {
+      "epoch": 4.303838646714379,
+      "grad_norm": 1.7640526294708252,
+      "learning_rate": 2.5078699508524288e-06,
+      "loss": 0.2359,
+      "step": 6615
+    },
+    {
+      "epoch": 4.307091737150293,
+      "grad_norm": 1.509757161140442,
+      "learning_rate": 2.485385234557641e-06,
+      "loss": 0.2163,
+      "step": 6620
+    },
+    {
+      "epoch": 4.310344827586207,
+      "grad_norm": 1.9168065786361694,
+      "learning_rate": 2.462996494365949e-06,
+      "loss": 0.2209,
+      "step": 6625
+    },
+    {
+      "epoch": 4.313597918022121,
+      "grad_norm": 1.701675295829773,
+      "learning_rate": 2.440703825716867e-06,
+      "loss": 0.2263,
+      "step": 6630
+    },
+    {
+      "epoch": 4.316851008458035,
+      "grad_norm": 3.188270330429077,
+      "learning_rate": 2.4185073236403707e-06,
+      "loss": 0.2286,
+      "step": 6635
+    },
+    {
+      "epoch": 4.32010409889395,
+      "grad_norm": 2.1177477836608887,
+      "learning_rate": 2.396407082756513e-06,
+      "loss": 0.2153,
+      "step": 6640
+    },
+    {
+      "epoch": 4.3233571893298635,
+      "grad_norm": 1.7593662738800049,
+      "learning_rate": 2.3744031972749826e-06,
+      "loss": 0.2378,
+      "step": 6645
+    },
+    {
+      "epoch": 4.326610279765777,
+      "grad_norm": 1.7856861352920532,
+      "learning_rate": 2.352495760994733e-06,
+      "loss": 0.2222,
+      "step": 6650
+    },
+    {
+      "epoch": 4.329863370201692,
+      "grad_norm": 2.1246399879455566,
+      "learning_rate": 2.3306848673035536e-06,
+      "loss": 0.2222,
+      "step": 6655
+    },
+    {
+      "epoch": 4.333116460637606,
+      "grad_norm": 1.8715474605560303,
+      "learning_rate": 2.308970609177713e-06,
+      "loss": 0.2283,
+      "step": 6660
+    },
+    {
+      "epoch": 4.33636955107352,
+      "grad_norm": 1.6002229452133179,
+      "learning_rate": 2.28735307918152e-06,
+      "loss": 0.2311,
+      "step": 6665
+    },
+    {
+      "epoch": 4.339622641509434,
+      "grad_norm": 1.71176016330719,
+      "learning_rate": 2.2658323694669498e-06,
+      "loss": 0.2261,
+      "step": 6670
+    },
+    {
+      "epoch": 4.342875731945348,
+      "grad_norm": 4.760283946990967,
+      "learning_rate": 2.24440857177326e-06,
+      "loss": 0.2174,
+      "step": 6675
+    },
+    {
+      "epoch": 4.346128822381262,
+      "grad_norm": 1.7307775020599365,
+      "learning_rate": 2.2230817774265724e-06,
+      "loss": 0.2235,
+      "step": 6680
+    },
+    {
+      "epoch": 4.349381912817177,
+      "grad_norm": 2.089346170425415,
+      "learning_rate": 2.201852077339506e-06,
+      "loss": 0.2387,
+      "step": 6685
+    },
+    {
+      "epoch": 4.3526350032530905,
+      "grad_norm": 1.815763235092163,
+      "learning_rate": 2.1807195620107914e-06,
+      "loss": 0.2406,
+      "step": 6690
+    },
+    {
+      "epoch": 4.355888093689004,
+      "grad_norm": 1.9279545545578003,
+      "learning_rate": 2.15968432152486e-06,
+      "loss": 0.2385,
+      "step": 6695
+    },
+    {
+      "epoch": 4.359141184124919,
+      "grad_norm": 1.6756705045700073,
+      "learning_rate": 2.1387464455514928e-06,
+      "loss": 0.2232,
+      "step": 6700
+    },
+    {
+      "epoch": 4.362394274560833,
+      "grad_norm": 1.606360912322998,
+      "learning_rate": 2.117906023345406e-06,
+      "loss": 0.2275,
+      "step": 6705
+    },
+    {
+      "epoch": 4.365647364996747,
+      "grad_norm": 1.7262459993362427,
+      "learning_rate": 2.097163143745909e-06,
+      "loss": 0.2232,
+      "step": 6710
+    },
+    {
+      "epoch": 4.368900455432661,
+      "grad_norm": 1.5949680805206299,
+      "learning_rate": 2.0765178951764774e-06,
+      "loss": 0.2306,
+      "step": 6715
+    },
+    {
+      "epoch": 4.372153545868575,
+      "grad_norm": 1.6405928134918213,
+      "learning_rate": 2.0559703656444107e-06,
+      "loss": 0.2286,
+      "step": 6720
+    },
+    {
+      "epoch": 4.375406636304489,
+      "grad_norm": 1.8739897012710571,
+      "learning_rate": 2.0355206427404626e-06,
+      "loss": 0.2128,
+      "step": 6725
+    },
+    {
+      "epoch": 4.378659726740404,
+      "grad_norm": 1.7278401851654053,
+      "learning_rate": 2.015168813638435e-06,
+      "loss": 0.2217,
+      "step": 6730
+    },
+    {
+      "epoch": 4.3819128171763175,
+      "grad_norm": 1.776465654373169,
+      "learning_rate": 1.9949149650948267e-06,
+      "loss": 0.2492,
+      "step": 6735
+    },
+    {
+      "epoch": 4.385165907612231,
+      "grad_norm": 3.8171842098236084,
+      "learning_rate": 1.974759183448477e-06,
+      "loss": 0.2262,
+      "step": 6740
+    },
+    {
+      "epoch": 4.388418998048146,
+      "grad_norm": 1.7197240591049194,
+      "learning_rate": 1.954701554620164e-06,
+      "loss": 0.2287,
+      "step": 6745
+    },
+    {
+      "epoch": 4.39167208848406,
+      "grad_norm": 3.6815900802612305,
+      "learning_rate": 1.9347421641122576e-06,
+      "loss": 0.2285,
+      "step": 6750
+    },
+    {
+      "epoch": 4.394925178919974,
+      "grad_norm": 2.5846571922302246,
+      "learning_rate": 1.9148810970083725e-06,
+      "loss": 0.2228,
+      "step": 6755
+    },
+    {
+      "epoch": 4.398178269355888,
+      "grad_norm": 1.7526190280914307,
+      "learning_rate": 1.8951184379729674e-06,
+      "loss": 0.2268,
+      "step": 6760
+    },
+    {
+      "epoch": 4.401431359791802,
+      "grad_norm": 3.6665701866149902,
+      "learning_rate": 1.8754542712510065e-06,
+      "loss": 0.2295,
+      "step": 6765
+    },
+    {
+      "epoch": 4.404684450227716,
+      "grad_norm": 2.0210907459259033,
+      "learning_rate": 1.8558886806676112e-06,
+      "loss": 0.2212,
+      "step": 6770
+    },
+    {
+      "epoch": 4.407937540663631,
+      "grad_norm": 1.6685539484024048,
+      "learning_rate": 1.8364217496276731e-06,
+      "loss": 0.2336,
+      "step": 6775
+    },
+    {
+      "epoch": 4.411190631099545,
+      "grad_norm": 1.7345695495605469,
+      "learning_rate": 1.8170535611155143e-06,
+      "loss": 0.2349,
+      "step": 6780
+    },
+    {
+      "epoch": 4.414443721535458,
+      "grad_norm": 1.5429155826568604,
+      "learning_rate": 1.797784197694552e-06,
+      "loss": 0.2206,
+      "step": 6785
+    },
+    {
+      "epoch": 4.417696811971373,
+      "grad_norm": 1.8358302116394043,
+      "learning_rate": 1.7786137415069126e-06,
+      "loss": 0.225,
+      "step": 6790
+    },
+    {
+      "epoch": 4.420949902407287,
+      "grad_norm": 1.858405590057373,
+      "learning_rate": 1.7595422742730905e-06,
+      "loss": 0.2049,
+      "step": 6795
+    },
+    {
+      "epoch": 4.424202992843201,
+      "grad_norm": 1.6918565034866333,
+      "learning_rate": 1.7405698772916313e-06,
+      "loss": 0.2224,
+      "step": 6800
+    },
+    {
+      "epoch": 4.427456083279115,
+      "grad_norm": 1.7888590097427368,
+      "learning_rate": 1.7216966314387378e-06,
+      "loss": 0.2359,
+      "step": 6805
+    },
+    {
+      "epoch": 4.430709173715029,
+      "grad_norm": 4.2224273681640625,
+      "learning_rate": 1.7029226171679542e-06,
+      "loss": 0.2278,
+      "step": 6810
+    },
+    {
+      "epoch": 4.433962264150943,
+      "grad_norm": 3.129756450653076,
+      "learning_rate": 1.684247914509826e-06,
+      "loss": 0.2317,
+      "step": 6815
+    },
+    {
+      "epoch": 4.437215354586858,
+      "grad_norm": 1.7897382974624634,
+      "learning_rate": 1.6656726030715358e-06,
+      "loss": 0.2305,
+      "step": 6820
+    },
+    {
+      "epoch": 4.440468445022772,
+      "grad_norm": 2.1024298667907715,
+      "learning_rate": 1.6471967620365846e-06,
+      "loss": 0.2433,
+      "step": 6825
+    },
+    {
+      "epoch": 4.443721535458685,
+      "grad_norm": 1.8943312168121338,
+      "learning_rate": 1.6288204701644382e-06,
+      "loss": 0.2301,
+      "step": 6830
+    },
+    {
+      "epoch": 4.4469746258946,
+      "grad_norm": 1.8938087224960327,
+      "learning_rate": 1.6105438057902295e-06,
+      "loss": 0.2245,
+      "step": 6835
+    },
+    {
+      "epoch": 4.450227716330514,
+      "grad_norm": 1.9831109046936035,
+      "learning_rate": 1.592366846824364e-06,
+      "loss": 0.2471,
+      "step": 6840
+    },
+    {
+      "epoch": 4.453480806766428,
+      "grad_norm": 4.050066947937012,
+      "learning_rate": 1.5742896707522242e-06,
+      "loss": 0.2437,
+      "step": 6845
+    },
+    {
+      "epoch": 4.4567338972023425,
+      "grad_norm": 1.9161465167999268,
+      "learning_rate": 1.5563123546338572e-06,
+      "loss": 0.2232,
+      "step": 6850
+    },
+    {
+      "epoch": 4.459986987638256,
+      "grad_norm": 2.1633474826812744,
+      "learning_rate": 1.5384349751035948e-06,
+      "loss": 0.2387,
+      "step": 6855
+    },
+    {
+      "epoch": 4.46324007807417,
+      "grad_norm": 1.729988694190979,
+      "learning_rate": 1.5206576083697687e-06,
+      "loss": 0.2387,
+      "step": 6860
+    },
+    {
+      "epoch": 4.466493168510085,
+      "grad_norm": 3.217106342315674,
+      "learning_rate": 1.502980330214379e-06,
+      "loss": 0.222,
+      "step": 6865
+    },
+    {
+      "epoch": 4.469746258945999,
+      "grad_norm": 3.8715853691101074,
+      "learning_rate": 1.4854032159927562e-06,
+      "loss": 0.2256,
+      "step": 6870
+    },
+    {
+      "epoch": 4.4729993493819125,
+      "grad_norm": 1.8818020820617676,
+      "learning_rate": 1.4679263406332467e-06,
+      "loss": 0.2282,
+      "step": 6875
+    },
+    {
+      "epoch": 4.476252439817827,
+      "grad_norm": 1.7817474603652954,
+      "learning_rate": 1.450549778636895e-06,
+      "loss": 0.2336,
+      "step": 6880
+    },
+    {
+      "epoch": 4.479505530253741,
+      "grad_norm": 1.9050745964050293,
+      "learning_rate": 1.43327360407714e-06,
+      "loss": 0.2239,
+      "step": 6885
+    },
+    {
+      "epoch": 4.482758620689655,
+      "grad_norm": 1.8388644456863403,
+      "learning_rate": 1.416097890599466e-06,
+      "loss": 0.2231,
+      "step": 6890
+    },
+    {
+      "epoch": 4.4860117111255695,
+      "grad_norm": 1.8944287300109863,
+      "learning_rate": 1.3990227114211191e-06,
+      "loss": 0.2252,
+      "step": 6895
+    },
+    {
+      "epoch": 4.489264801561483,
+      "grad_norm": 1.994055986404419,
+      "learning_rate": 1.3820481393307855e-06,
+      "loss": 0.2351,
+      "step": 6900
+    },
+    {
+      "epoch": 4.492517891997397,
+      "grad_norm": 1.8318955898284912,
+      "learning_rate": 1.365174246688275e-06,
+      "loss": 0.2328,
+      "step": 6905
+    },
+    {
+      "epoch": 4.495770982433312,
+      "grad_norm": 1.8265342712402344,
+      "learning_rate": 1.3484011054242157e-06,
+      "loss": 0.2202,
+      "step": 6910
+    },
+    {
+      "epoch": 4.499024072869226,
+      "grad_norm": 2.5501973628997803,
+      "learning_rate": 1.3317287870397572e-06,
+      "loss": 0.2283,
+      "step": 6915
+    },
+    {
+      "epoch": 4.5022771633051395,
+      "grad_norm": 2.1136600971221924,
+      "learning_rate": 1.3151573626062535e-06,
+      "loss": 0.2362,
+      "step": 6920
+    },
+    {
+      "epoch": 4.505530253741054,
+      "grad_norm": 2.1394035816192627,
+      "learning_rate": 1.298686902764959e-06,
+      "loss": 0.2262,
+      "step": 6925
+    },
+    {
+      "epoch": 4.508783344176968,
+      "grad_norm": 2.5853805541992188,
+      "learning_rate": 1.2823174777267439e-06,
+      "loss": 0.2235,
+      "step": 6930
+    },
+    {
+      "epoch": 4.512036434612883,
+      "grad_norm": 11.947797775268555,
+      "learning_rate": 1.266049157271773e-06,
+      "loss": 0.2135,
+      "step": 6935
+    },
+    {
+      "epoch": 4.5152895250487965,
+      "grad_norm": 1.589674472808838,
+      "learning_rate": 1.2498820107492204e-06,
+      "loss": 0.232,
+      "step": 6940
+    },
+    {
+      "epoch": 4.51854261548471,
+      "grad_norm": 2.020092248916626,
+      "learning_rate": 1.233816107076985e-06,
+      "loss": 0.2373,
+      "step": 6945
+    },
+    {
+      "epoch": 4.521795705920624,
+      "grad_norm": 1.9216246604919434,
+      "learning_rate": 1.2178515147413665e-06,
+      "loss": 0.2275,
+      "step": 6950
+    },
+    {
+      "epoch": 4.525048796356539,
+      "grad_norm": 2.0308332443237305,
+      "learning_rate": 1.2019883017967943e-06,
+      "loss": 0.225,
+      "step": 6955
+    },
+    {
+      "epoch": 4.528301886792453,
+      "grad_norm": 1.7747068405151367,
+      "learning_rate": 1.1862265358655505e-06,
+      "loss": 0.2252,
+      "step": 6960
+    },
+    {
+      "epoch": 4.531554977228367,
+      "grad_norm": 3.725299119949341,
+      "learning_rate": 1.170566284137442e-06,
+      "loss": 0.2343,
+      "step": 6965
+    },
+    {
+      "epoch": 4.534808067664281,
+      "grad_norm": 2.894679069519043,
+      "learning_rate": 1.1550076133695604e-06,
+      "loss": 0.2436,
+      "step": 6970
+    },
+    {
+      "epoch": 4.538061158100195,
+      "grad_norm": 1.8062316179275513,
+      "learning_rate": 1.1395505898859487e-06,
+      "loss": 0.237,
+      "step": 6975
+    },
+    {
+      "epoch": 4.541314248536109,
+      "grad_norm": 1.745668888092041,
+      "learning_rate": 1.1241952795773697e-06,
+      "loss": 0.2324,
+      "step": 6980
+    },
+    {
+      "epoch": 4.544567338972024,
+      "grad_norm": 1.908809781074524,
+      "learning_rate": 1.108941747900985e-06,
+      "loss": 0.2356,
+      "step": 6985
+    },
+    {
+      "epoch": 4.547820429407937,
+      "grad_norm": 6.190516948699951,
+      "learning_rate": 1.0937900598800872e-06,
+      "loss": 0.23,
+      "step": 6990
+    },
+    {
+      "epoch": 4.551073519843852,
+      "grad_norm": 2.0687854290008545,
+      "learning_rate": 1.0787402801038405e-06,
+      "loss": 0.2376,
+      "step": 6995
+    },
+    {
+      "epoch": 4.554326610279766,
+      "grad_norm": 2.2060608863830566,
+      "learning_rate": 1.0637924727269822e-06,
+      "loss": 0.2179,
+      "step": 7000
+    },
+    {
+      "epoch": 4.55757970071568,
+      "grad_norm": 1.8114817142486572,
+      "learning_rate": 1.0489467014695526e-06,
+      "loss": 0.2445,
+      "step": 7005
+    },
+    {
+      "epoch": 4.560832791151594,
+      "grad_norm": 1.8267011642456055,
+      "learning_rate": 1.0342030296166428e-06,
+      "loss": 0.2249,
+      "step": 7010
+    },
+    {
+      "epoch": 4.564085881587508,
+      "grad_norm": 1.8274999856948853,
+      "learning_rate": 1.0195615200180974e-06,
+      "loss": 0.2198,
+      "step": 7015
+    },
+    {
+      "epoch": 4.567338972023422,
+      "grad_norm": 4.727597713470459,
+      "learning_rate": 1.0050222350882682e-06,
+      "loss": 0.23,
+      "step": 7020
+    },
+    {
+      "epoch": 4.570592062459337,
+      "grad_norm": 17.39756965637207,
+      "learning_rate": 9.905852368057383e-07,
+      "loss": 0.2217,
+      "step": 7025
+    },
+    {
+      "epoch": 4.573845152895251,
+      "grad_norm": 1.8855489492416382,
+      "learning_rate": 9.762505867130594e-07,
+      "loss": 0.2233,
+      "step": 7030
+    },
+    {
+      "epoch": 4.577098243331164,
+      "grad_norm": 1.946487545967102,
+      "learning_rate": 9.620183459164878e-07,
+      "loss": 0.2398,
+      "step": 7035
+    },
+    {
+      "epoch": 4.580351333767078,
+      "grad_norm": 5.050466060638428,
+      "learning_rate": 9.478885750857285e-07,
+      "loss": 0.2437,
+      "step": 7040
+    },
+    {
+      "epoch": 4.583604424202993,
+      "grad_norm": 2.065145254135132,
+      "learning_rate": 9.338613344536701e-07,
+      "loss": 0.235,
+      "step": 7045
+    },
+    {
+      "epoch": 4.586857514638907,
+      "grad_norm": 1.8699109554290771,
+      "learning_rate": 9.199366838161389e-07,
+      "loss": 0.2314,
+      "step": 7050
+    },
+    {
+      "epoch": 4.5901106050748215,
+      "grad_norm": 1.8447792530059814,
+      "learning_rate": 9.06114682531628e-07,
+      "loss": 0.2261,
+      "step": 7055
+    },
+    {
+      "epoch": 4.593363695510735,
+      "grad_norm": 7.244560241699219,
+      "learning_rate": 8.923953895210612e-07,
+      "loss": 0.2094,
+      "step": 7060
+    },
+    {
+      "epoch": 4.596616785946649,
+      "grad_norm": 1.6492129564285278,
+      "learning_rate": 8.787788632675293e-07,
+      "loss": 0.2185,
+      "step": 7065
+    },
+    {
+      "epoch": 4.599869876382563,
+      "grad_norm": 1.8190096616744995,
+      "learning_rate": 8.652651618160424e-07,
+      "loss": 0.2368,
+      "step": 7070
+    },
+    {
+      "epoch": 4.603122966818478,
+      "grad_norm": 1.6560496091842651,
+      "learning_rate": 8.51854342773295e-07,
+      "loss": 0.2297,
+      "step": 7075
+    },
+    {
+      "epoch": 4.6063760572543915,
+      "grad_norm": 1.6347981691360474,
+      "learning_rate": 8.385464633074019e-07,
+      "loss": 0.2228,
+      "step": 7080
+    },
+    {
+      "epoch": 4.609629147690306,
+      "grad_norm": 2.087787628173828,
+      "learning_rate": 8.25341580147665e-07,
+      "loss": 0.2277,
+      "step": 7085
+    },
+    {
+      "epoch": 4.61288223812622,
+      "grad_norm": 1.9914604425430298,
+      "learning_rate": 8.122397495843343e-07,
+      "loss": 0.2332,
+      "step": 7090
+    },
+    {
+      "epoch": 4.616135328562134,
+      "grad_norm": 2.1197216510772705,
+      "learning_rate": 7.992410274683615e-07,
+      "loss": 0.2218,
+      "step": 7095
+    },
+    {
+      "epoch": 4.6193884189980485,
+      "grad_norm": 3.4777309894561768,
+      "learning_rate": 7.863454692111583e-07,
+      "loss": 0.224,
+      "step": 7100
+    },
+    {
+      "epoch": 4.622641509433962,
+      "grad_norm": 1.6859432458877563,
+      "learning_rate": 7.735531297843713e-07,
+      "loss": 0.2434,
+      "step": 7105
+    },
+    {
+      "epoch": 4.625894599869876,
+      "grad_norm": 3.000272035598755,
+      "learning_rate": 7.60864063719649e-07,
+      "loss": 0.2296,
+      "step": 7110
+    },
+    {
+      "epoch": 4.629147690305791,
+      "grad_norm": 1.8207471370697021,
+      "learning_rate": 7.482783251083869e-07,
+      "loss": 0.2252,
+      "step": 7115
+    },
+    {
+      "epoch": 4.632400780741705,
+      "grad_norm": 2.1532914638519287,
+      "learning_rate": 7.357959676015214e-07,
+      "loss": 0.2275,
+      "step": 7120
+    },
+    {
+      "epoch": 4.6356538711776185,
+      "grad_norm": 1.644968867301941,
+      "learning_rate": 7.234170444092942e-07,
+      "loss": 0.2384,
+      "step": 7125
+    },
+    {
+      "epoch": 4.638906961613533,
+      "grad_norm": 1.7306429147720337,
+      "learning_rate": 7.11141608301022e-07,
+      "loss": 0.2202,
+      "step": 7130
+    },
+    {
+      "epoch": 4.642160052049447,
+      "grad_norm": 2.0186638832092285,
+      "learning_rate": 6.989697116048633e-07,
+      "loss": 0.2242,
+      "step": 7135
+    },
+    {
+      "epoch": 4.645413142485361,
+      "grad_norm": 1.7209802865982056,
+      "learning_rate": 6.86901406207624e-07,
+      "loss": 0.2152,
+      "step": 7140
+    },
+    {
+      "epoch": 4.648666232921276,
+      "grad_norm": 1.6327604055404663,
+      "learning_rate": 6.749367435545024e-07,
+      "loss": 0.2384,
+      "step": 7145
+    },
+    {
+      "epoch": 4.651919323357189,
+      "grad_norm": 2.552006721496582,
+      "learning_rate": 6.630757746488886e-07,
+      "loss": 0.2287,
+      "step": 7150
+    },
+    {
+      "epoch": 4.655172413793103,
+      "grad_norm": 1.8012171983718872,
+      "learning_rate": 6.513185500521463e-07,
+      "loss": 0.2148,
+      "step": 7155
+    },
+    {
+      "epoch": 4.658425504229018,
+      "grad_norm": 1.9249675273895264,
+      "learning_rate": 6.396651198833897e-07,
+      "loss": 0.2249,
+      "step": 7160
+    },
+    {
+      "epoch": 4.661678594664932,
+      "grad_norm": 1.640886664390564,
+      "learning_rate": 6.281155338192762e-07,
+      "loss": 0.2167,
+      "step": 7165
+    },
+    {
+      "epoch": 4.6649316851008455,
+      "grad_norm": 1.9455459117889404,
+      "learning_rate": 6.166698410937949e-07,
+      "loss": 0.2264,
+      "step": 7170
+    },
+    {
+      "epoch": 4.66818477553676,
+      "grad_norm": 2.079929828643799,
+      "learning_rate": 6.053280904980557e-07,
+      "loss": 0.2249,
+      "step": 7175
+    },
+    {
+      "epoch": 4.671437865972674,
+      "grad_norm": 1.5457408428192139,
+      "learning_rate": 5.940903303800705e-07,
+      "loss": 0.218,
+      "step": 7180
+    },
+    {
+      "epoch": 4.674690956408588,
+      "grad_norm": 2.4226558208465576,
+      "learning_rate": 5.829566086445721e-07,
+      "loss": 0.2229,
+      "step": 7185
+    },
+    {
+      "epoch": 4.677944046844503,
+      "grad_norm": 2.8459372520446777,
+      "learning_rate": 5.719269727527843e-07,
+      "loss": 0.2206,
+      "step": 7190
+    },
+    {
+      "epoch": 4.681197137280416,
+      "grad_norm": 1.8545024394989014,
+      "learning_rate": 5.610014697222249e-07,
+      "loss": 0.2157,
+      "step": 7195
+    },
+    {
+      "epoch": 4.68445022771633,
+      "grad_norm": 2.315584421157837,
+      "learning_rate": 5.501801461265304e-07,
+      "loss": 0.2179,
+      "step": 7200
+    },
+    {
+      "epoch": 4.687703318152245,
+      "grad_norm": 1.7075843811035156,
+      "learning_rate": 5.394630480952178e-07,
+      "loss": 0.2147,
+      "step": 7205
+    },
+    {
+      "epoch": 4.690956408588159,
+      "grad_norm": 1.7120945453643799,
+      "learning_rate": 5.288502213135149e-07,
+      "loss": 0.2329,
+      "step": 7210
+    },
+    {
+      "epoch": 4.694209499024073,
+      "grad_norm": 1.6173934936523438,
+      "learning_rate": 5.183417110221606e-07,
+      "loss": 0.2338,
+      "step": 7215
+    },
+    {
+      "epoch": 4.697462589459987,
+      "grad_norm": 1.8131818771362305,
+      "learning_rate": 5.07937562017205e-07,
+      "loss": 0.231,
+      "step": 7220
+    },
+    {
+      "epoch": 4.700715679895901,
+      "grad_norm": 2.225557327270508,
+      "learning_rate": 4.976378186498293e-07,
+      "loss": 0.2303,
+      "step": 7225
+    },
+    {
+      "epoch": 4.703968770331815,
+      "grad_norm": 1.9588440656661987,
+      "learning_rate": 4.874425248261428e-07,
+      "loss": 0.2254,
+      "step": 7230
+    },
+    {
+      "epoch": 4.70722186076773,
+      "grad_norm": 1.5929067134857178,
+      "learning_rate": 4.773517240070108e-07,
+      "loss": 0.2177,
+      "step": 7235
+    },
+    {
+      "epoch": 4.7104749512036435,
+      "grad_norm": 1.794873833656311,
+      "learning_rate": 4.67365459207858e-07,
+      "loss": 0.2268,
+      "step": 7240
+    },
+    {
+      "epoch": 4.713728041639557,
+      "grad_norm": 2.0999152660369873,
+      "learning_rate": 4.5748377299849045e-07,
+      "loss": 0.2201,
+      "step": 7245
+    },
+    {
+      "epoch": 4.716981132075472,
+      "grad_norm": 1.7107945680618286,
+      "learning_rate": 4.477067075029123e-07,
+      "loss": 0.2272,
+      "step": 7250
+    },
+    {
+      "epoch": 4.720234222511386,
+      "grad_norm": 1.8921961784362793,
+      "learning_rate": 4.3803430439915137e-07,
+      "loss": 0.2176,
+      "step": 7255
+    },
+    {
+      "epoch": 4.7234873129473,
+      "grad_norm": 2.4301459789276123,
+      "learning_rate": 4.284666049190644e-07,
+      "loss": 0.2146,
+      "step": 7260
+    },
+    {
+      "epoch": 4.726740403383214,
+      "grad_norm": 1.8662853240966797,
+      "learning_rate": 4.1900364984818754e-07,
+      "loss": 0.2483,
+      "step": 7265
+    },
+    {
+      "epoch": 4.729993493819128,
+      "grad_norm": 1.5476515293121338,
+      "learning_rate": 4.0964547952554443e-07,
+      "loss": 0.2265,
+      "step": 7270
+    },
+    {
+      "epoch": 4.733246584255042,
+      "grad_norm": 2.1670968532562256,
+      "learning_rate": 4.0039213384347187e-07,
+      "loss": 0.2384,
+      "step": 7275
+    },
+    {
+      "epoch": 4.736499674690957,
+      "grad_norm": 1.7863905429840088,
+      "learning_rate": 3.912436522474666e-07,
+      "loss": 0.2287,
+      "step": 7280
+    },
+    {
+      "epoch": 4.7397527651268705,
+      "grad_norm": 2.369838237762451,
+      "learning_rate": 3.822000737360026e-07,
+      "loss": 0.2362,
+      "step": 7285
+    },
+    {
+      "epoch": 4.743005855562784,
+      "grad_norm": 2.5364902019500732,
+      "learning_rate": 3.7326143686036706e-07,
+      "loss": 0.2211,
+      "step": 7290
+    },
+    {
+      "epoch": 4.746258945998699,
+      "grad_norm": 2.2473394870758057,
+      "learning_rate": 3.644277797244966e-07,
+      "loss": 0.2349,
+      "step": 7295
+    },
+    {
+      "epoch": 4.749512036434613,
+      "grad_norm": 2.3497040271759033,
+      "learning_rate": 3.556991399848275e-07,
+      "loss": 0.2254,
+      "step": 7300
+    },
+    {
+      "epoch": 4.752765126870527,
+      "grad_norm": 1.744187831878662,
+      "learning_rate": 3.4707555485011533e-07,
+      "loss": 0.2231,
+      "step": 7305
+    },
+    {
+      "epoch": 4.756018217306441,
+      "grad_norm": 2.6059043407440186,
+      "learning_rate": 3.385570610812794e-07,
+      "loss": 0.2141,
+      "step": 7310
+    },
+    {
+      "epoch": 4.759271307742355,
+      "grad_norm": 2.001283645629883,
+      "learning_rate": 3.3014369499126675e-07,
+      "loss": 0.2219,
+      "step": 7315
+    },
+    {
+      "epoch": 4.762524398178269,
+      "grad_norm": 2.1158323287963867,
+      "learning_rate": 3.218354924448719e-07,
+      "loss": 0.237,
+      "step": 7320
+    },
+    {
+      "epoch": 4.765777488614184,
+      "grad_norm": 1.978641152381897,
+      "learning_rate": 3.1363248885859506e-07,
+      "loss": 0.2225,
+      "step": 7325
+    },
+    {
+      "epoch": 4.7690305790500975,
+      "grad_norm": 2.592278003692627,
+      "learning_rate": 3.055347192004954e-07,
+      "loss": 0.2235,
+      "step": 7330
+    },
+    {
+      "epoch": 4.772283669486011,
+      "grad_norm": 3.1187829971313477,
+      "learning_rate": 2.9754221799003503e-07,
+      "loss": 0.2324,
+      "step": 7335
+    },
+    {
+      "epoch": 4.775536759921926,
+      "grad_norm": 1.6826578378677368,
+      "learning_rate": 2.8965501929792695e-07,
+      "loss": 0.2309,
+      "step": 7340
+    },
+    {
+      "epoch": 4.77878985035784,
+      "grad_norm": 1.6795811653137207,
+      "learning_rate": 2.818731567460098e-07,
+      "loss": 0.2087,
+      "step": 7345
+    },
+    {
+      "epoch": 4.782042940793754,
+      "grad_norm": 1.9984930753707886,
+      "learning_rate": 2.741966635070842e-07,
+      "loss": 0.22,
+      "step": 7350
+    },
+    {
+      "epoch": 4.785296031229668,
+      "grad_norm": 2.397368907928467,
+      "learning_rate": 2.6662557230477667e-07,
+      "loss": 0.227,
+      "step": 7355
+    },
+    {
+      "epoch": 4.788549121665582,
+      "grad_norm": 1.748075008392334,
+      "learning_rate": 2.5915991541340667e-07,
+      "loss": 0.2206,
+      "step": 7360
+    },
+    {
+      "epoch": 4.791802212101496,
+      "grad_norm": 2.152040481567383,
+      "learning_rate": 2.5179972465784186e-07,
+      "loss": 0.2214,
+      "step": 7365
+    },
+    {
+      "epoch": 4.795055302537411,
+      "grad_norm": 1.829852819442749,
+      "learning_rate": 2.4454503141336513e-07,
+      "loss": 0.213,
+      "step": 7370
+    },
+    {
+      "epoch": 4.798308392973325,
+      "grad_norm": 2.7500312328338623,
+      "learning_rate": 2.3739586660554148e-07,
+      "loss": 0.2188,
+      "step": 7375
+    },
+    {
+      "epoch": 4.801561483409239,
+      "grad_norm": 2.0260448455810547,
+      "learning_rate": 2.3035226071008997e-07,
+      "loss": 0.2264,
+      "step": 7380
+    },
+    {
+      "epoch": 4.804814573845153,
+      "grad_norm": 2.0971879959106445,
+      "learning_rate": 2.2341424375274256e-07,
+      "loss": 0.2382,
+      "step": 7385
+    },
+    {
+      "epoch": 4.808067664281067,
+      "grad_norm": 1.875805139541626,
+      "learning_rate": 2.165818453091245e-07,
+      "loss": 0.2181,
+      "step": 7390
+    },
+    {
+      "epoch": 4.811320754716981,
+      "grad_norm": 2.7413530349731445,
+      "learning_rate": 2.098550945046268e-07,
+      "loss": 0.2342,
+      "step": 7395
+    },
+    {
+      "epoch": 4.814573845152895,
+      "grad_norm": 1.8814730644226074,
+      "learning_rate": 2.0323402001428682e-07,
+      "loss": 0.2251,
+      "step": 7400
+    },
+    {
+      "epoch": 4.817826935588809,
+      "grad_norm": 2.818944215774536,
+      "learning_rate": 1.9671865006265223e-07,
+      "loss": 0.2247,
+      "step": 7405
+    },
+    {
+      "epoch": 4.821080026024724,
+      "grad_norm": 1.7889087200164795,
+      "learning_rate": 1.9030901242367837e-07,
+      "loss": 0.2401,
+      "step": 7410
+    },
+    {
+      "epoch": 4.824333116460638,
+      "grad_norm": 2.2675249576568604,
+      "learning_rate": 1.8400513442059786e-07,
+      "loss": 0.2343,
+      "step": 7415
+    },
+    {
+      "epoch": 4.827586206896552,
+      "grad_norm": 2.051952362060547,
+      "learning_rate": 1.7780704292580107e-07,
+      "loss": 0.2222,
+      "step": 7420
+    },
+    {
+      "epoch": 4.830839297332465,
+      "grad_norm": 1.9150989055633545,
+      "learning_rate": 1.7292475378629936e-07,
+      "loss": 0.2197,
+      "step": 7425
+    },
+    {
+      "epoch": 4.83409238776838,
+      "grad_norm": 2.003403902053833,
+      "learning_rate": 1.6691714428535288e-07,
+      "loss": 0.2274,
+      "step": 7430
+    },
+    {
+      "epoch": 4.837345478204294,
+      "grad_norm": 1.729634165763855,
+      "learning_rate": 1.6101539413598822e-07,
+      "loss": 0.2313,
+      "step": 7435
+    },
+    {
+      "epoch": 4.840598568640209,
+      "grad_norm": 2.2372894287109375,
+      "learning_rate": 1.5521952849639476e-07,
+      "loss": 0.2227,
+      "step": 7440
+    },
+    {
+      "epoch": 4.8438516590761225,
+      "grad_norm": 1.8825215101242065,
+      "learning_rate": 1.4952957207339802e-07,
+      "loss": 0.2182,
+      "step": 7445
+    },
+    {
+      "epoch": 4.847104749512036,
+      "grad_norm": 1.7914789915084839,
+      "learning_rate": 1.439455491223457e-07,
+      "loss": 0.2104,
+      "step": 7450
+    },
+    {
+      "epoch": 4.85035783994795,
+      "grad_norm": 1.8011417388916016,
+      "learning_rate": 1.3846748344701065e-07,
+      "loss": 0.2243,
+      "step": 7455
+    },
+    {
+      "epoch": 4.853610930383865,
+      "grad_norm": 1.737289547920227,
+      "learning_rate": 1.3309539839948538e-07,
+      "loss": 0.2272,
+      "step": 7460
+    },
+    {
+      "epoch": 4.856864020819779,
+      "grad_norm": 2.1911211013793945,
+      "learning_rate": 1.2782931688008482e-07,
+      "loss": 0.2273,
+      "step": 7465
+    },
+    {
+      "epoch": 4.860117111255693,
+      "grad_norm": 6.419521331787109,
+      "learning_rate": 1.2266926133725487e-07,
+      "loss": 0.2332,
+      "step": 7470
+    },
+    {
+      "epoch": 4.863370201691607,
+      "grad_norm": 7.40968656539917,
+      "learning_rate": 1.1761525376745575e-07,
+      "loss": 0.2092,
+      "step": 7475
+    },
+    {
+      "epoch": 4.866623292127521,
+      "grad_norm": 1.6772918701171875,
+      "learning_rate": 1.1266731571509815e-07,
+      "loss": 0.222,
+      "step": 7480
+    },
+    {
+      "epoch": 4.869876382563435,
+      "grad_norm": 1.780863881111145,
+      "learning_rate": 1.0782546827242667e-07,
+      "loss": 0.2239,
+      "step": 7485
+    },
+    {
+      "epoch": 4.8731294729993495,
+      "grad_norm": 1.984481930732727,
+      "learning_rate": 1.0308973207944217e-07,
+      "loss": 0.2212,
+      "step": 7490
+    },
+    {
+      "epoch": 4.876382563435263,
+      "grad_norm": 1.720576286315918,
+      "learning_rate": 9.846012732380727e-08,
+      "loss": 0.2308,
+      "step": 7495
+    },
+    {
+      "epoch": 4.879635653871178,
+      "grad_norm": 8.550189971923828,
+      "learning_rate": 9.483287143148001e-08,
+      "loss": 0.2275,
+      "step": 7500
+    },
+    {
+      "epoch": 4.882888744307092,
+      "grad_norm": 2.3628833293914795,
+      "learning_rate": 9.039435269181384e-08,
+      "loss": 0.2137,
+      "step": 7505
+    },
+    {
+      "epoch": 4.886141834743006,
+      "grad_norm": 2.029315233230591,
+      "learning_rate": 8.606201950781267e-08,
+      "loss": 0.2229,
+      "step": 7510
+    },
+    {
+      "epoch": 4.8893949251789195,
+      "grad_norm": 4.765905857086182,
+      "learning_rate": 8.183589034750639e-08,
+      "loss": 0.2361,
+      "step": 7515
+    },
+    {
+      "epoch": 4.892648015614834,
+      "grad_norm": 1.9303253889083862,
+      "learning_rate": 7.771598322618422e-08,
+      "loss": 0.2198,
+      "step": 7520
+    },
+    {
+      "epoch": 4.895901106050748,
+      "grad_norm": 2.502507209777832,
+      "learning_rate": 7.370231570633656e-08,
+      "loss": 0.2263,
+      "step": 7525
+    },
+    {
+      "epoch": 4.899154196486663,
+      "grad_norm": 2.3058369159698486,
+      "learning_rate": 6.979490489756601e-08,
+      "loss": 0.2224,
+      "step": 7530
+    },
+    {
+      "epoch": 4.9024072869225765,
+      "grad_norm": 1.8423314094543457,
+      "learning_rate": 6.599376745652641e-08,
+      "loss": 0.2237,
+      "step": 7535
+    },
+    {
+      "epoch": 4.90566037735849,
+      "grad_norm": 1.7330564260482788,
+      "learning_rate": 6.229891958683675e-08,
+      "loss": 0.2248,
+      "step": 7540
+    },
+    {
+      "epoch": 4.908913467794405,
+      "grad_norm": 1.9083019495010376,
+      "learning_rate": 5.8710377039031264e-08,
+      "loss": 0.2174,
+      "step": 7545
+    },
+    {
+      "epoch": 4.912166558230319,
+      "grad_norm": 4.153858661651611,
+      "learning_rate": 5.52281551104733e-08,
+      "loss": 0.2279,
+      "step": 7550
+    },
+    {
+      "epoch": 4.915419648666233,
+      "grad_norm": 1.6192256212234497,
+      "learning_rate": 5.185226864530546e-08,
+      "loss": 0.2263,
+      "step": 7555
+    },
+    {
+      "epoch": 4.918672739102147,
+      "grad_norm": 1.8693500757217407,
+      "learning_rate": 4.8582732034374575e-08,
+      "loss": 0.2234,
+      "step": 7560
+    },
+    {
+      "epoch": 4.921925829538061,
+      "grad_norm": 1.9380360841751099,
+      "learning_rate": 4.541955921518182e-08,
+      "loss": 0.2182,
+      "step": 7565
+    },
+    {
+      "epoch": 4.925178919973975,
+      "grad_norm": 1.9134066104888916,
+      "learning_rate": 4.236276367180769e-08,
+      "loss": 0.2283,
+      "step": 7570
+    },
+    {
+      "epoch": 4.92843201040989,
+      "grad_norm": 1.6409050226211548,
+      "learning_rate": 3.9412358434876003e-08,
+      "loss": 0.2325,
+      "step": 7575
+    },
+    {
+      "epoch": 4.931685100845804,
+      "grad_norm": 1.7989153861999512,
+      "learning_rate": 3.6568356081473354e-08,
+      "loss": 0.2357,
+      "step": 7580
+    },
+    {
+      "epoch": 4.934938191281717,
+      "grad_norm": 2.0496795177459717,
+      "learning_rate": 3.383076873511859e-08,
+      "loss": 0.2218,
+      "step": 7585
+    },
+    {
+      "epoch": 4.938191281717632,
+      "grad_norm": 1.7194660902023315,
+      "learning_rate": 3.119960806569344e-08,
+      "loss": 0.2324,
+      "step": 7590
+    },
+    {
+      "epoch": 4.941444372153546,
+      "grad_norm": 2.1285674571990967,
+      "learning_rate": 2.867488528940643e-08,
+      "loss": 0.2339,
+      "step": 7595
+    },
+    {
+      "epoch": 4.94469746258946,
+      "grad_norm": 2.114654779434204,
+      "learning_rate": 2.6256611168734568e-08,
+      "loss": 0.2403,
+      "step": 7600
+    },
+    {
+      "epoch": 4.9479505530253745,
+      "grad_norm": 7.625702857971191,
+      "learning_rate": 2.3944796012381754e-08,
+      "loss": 0.2255,
+      "step": 7605
+    },
+    {
+      "epoch": 4.951203643461288,
+      "grad_norm": 1.6961826086044312,
+      "learning_rate": 2.173944967523711e-08,
+      "loss": 0.2197,
+      "step": 7610
+    },
+    {
+      "epoch": 4.954456733897202,
+      "grad_norm": 3.5485808849334717,
+      "learning_rate": 1.9640581558330594e-08,
+      "loss": 0.2309,
+      "step": 7615
+    },
+    {
+      "epoch": 4.957709824333117,
+      "grad_norm": 1.928830862045288,
+      "learning_rate": 1.7648200608791353e-08,
+      "loss": 0.2205,
+      "step": 7620
+    },
+    {
+      "epoch": 4.960962914769031,
+      "grad_norm": 2.2189247608184814,
+      "learning_rate": 1.5762315319814425e-08,
+      "loss": 0.2159,
+      "step": 7625
+    },
+    {
+      "epoch": 4.964216005204944,
+      "grad_norm": 1.728050947189331,
+      "learning_rate": 1.3982933730613545e-08,
+      "loss": 0.223,
+      "step": 7630
+    },
+    {
+      "epoch": 4.967469095640859,
+      "grad_norm": 7.718484878540039,
+      "learning_rate": 1.2310063426404506e-08,
+      "loss": 0.2345,
+      "step": 7635
+    },
+    {
+      "epoch": 4.970722186076773,
+      "grad_norm": 1.8739633560180664,
+      "learning_rate": 1.0743711538357959e-08,
+      "loss": 0.2354,
+      "step": 7640
+    },
+    {
+      "epoch": 4.973975276512687,
+      "grad_norm": 2.505018711090088,
+      "learning_rate": 9.28388474357167e-09,
+      "loss": 0.2341,
+      "step": 7645
+    },
+    {
+      "epoch": 4.9772283669486015,
+      "grad_norm": 2.42279314994812,
+      "learning_rate": 7.930589265051081e-09,
+      "loss": 0.2242,
+      "step": 7650
+    },
+    {
+      "epoch": 4.980481457384515,
+      "grad_norm": 1.8265007734298706,
+      "learning_rate": 6.683830871667685e-09,
+      "loss": 0.2052,
+      "step": 7655
+    },
+    {
+      "epoch": 4.983734547820429,
+      "grad_norm": 1.9440053701400757,
+      "learning_rate": 5.543614878153469e-09,
+      "loss": 0.2132,
+      "step": 7660
+    },
+    {
+      "epoch": 4.986987638256344,
+      "grad_norm": 3.599294900894165,
+      "learning_rate": 4.509946145059285e-09,
+      "loss": 0.2236,
+      "step": 7665
+    },
+    {
+      "epoch": 4.990240728692258,
+      "grad_norm": 6.0209126472473145,
+      "learning_rate": 3.5828290787437436e-09,
+      "loss": 0.2376,
+      "step": 7670
+    },
+    {
+      "epoch": 4.9934938191281715,
+      "grad_norm": 1.537784218788147,
+      "learning_rate": 2.762267631356563e-09,
+      "loss": 0.2264,
+      "step": 7675
+    },
+    {
+      "epoch": 4.996746909564086,
+      "grad_norm": 2.2779605388641357,
+      "learning_rate": 2.0482653008163654e-09,
+      "loss": 0.2452,
+      "step": 7680
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 3.292484998703003,
+      "learning_rate": 1.440825130796797e-09,
+      "loss": 0.2322,
+      "step": 7685
+    },
+    {
+      "epoch": 5.0,
+      "eval_f1": 0.801246098825816,
+      "eval_loss": 0.7490234375,
+      "eval_precision": 0.8011274883507161,
+      "eval_recall": 0.8013850778258188,
+      "eval_runtime": 255.3842,
+      "eval_samples_per_second": 1540.561,
+      "eval_steps_per_second": 1.508,
+      "step": 7685
+    },
+    {
+      "epoch": 5.0,
+      "step": 7685,
+      "total_flos": 1.664464427634026e+19,
+      "train_loss": 0.6255086388770993,
+      "train_runtime": 65052.9958,
+      "train_samples_per_second": 241.916,
+      "train_steps_per_second": 0.118
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 7685,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 5.0,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.664464427634026e+19,
+  "train_batch_size": 512,
+  "trial_name": null,
+  "trial_params": null
+}