diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,15112 @@
+{
+  "best_global_step": 12882,
+  "best_metric": 0.49088114500045776,
+  "best_model_checkpoint": "./results_shuffled_deep128_heads_r16_seq128_3e-5_roberta_newData/checkpoint-12882",
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 21470,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.002328830926874709,
+      "grad_norm": 7.533487796783447,
+      "learning_rate": 2.99874213836478e-05,
+      "loss": 3.2426,
+      "step": 10
+    },
+    {
+      "epoch": 0.004657661853749418,
+      "grad_norm": 5.052584171295166,
+      "learning_rate": 2.9973445143256466e-05,
+      "loss": 3.3433,
+      "step": 20
+    },
+    {
+      "epoch": 0.0069864927806241265,
+      "grad_norm": 3.4427478313446045,
+      "learning_rate": 2.995946890286513e-05,
+      "loss": 3.0442,
+      "step": 30
+    },
+    {
+      "epoch": 0.009315323707498836,
+      "grad_norm": 4.24840784072876,
+      "learning_rate": 2.9945492662473797e-05,
+      "loss": 3.1555,
+      "step": 40
+    },
+    {
+      "epoch": 0.011644154634373545,
+      "grad_norm": 4.173481464385986,
+      "learning_rate": 2.9931516422082464e-05,
+      "loss": 2.3163,
+      "step": 50
+    },
+    {
+      "epoch": 0.013972985561248253,
+      "grad_norm": 2.706721544265747,
+      "learning_rate": 2.9917540181691124e-05,
+      "loss": 2.9514,
+      "step": 60
+    },
+    {
+      "epoch": 0.016301816488122962,
+      "grad_norm": 3.585742712020874,
+      "learning_rate": 2.990356394129979e-05,
+      "loss": 2.84,
+      "step": 70
+    },
+    {
+      "epoch": 0.018630647414997672,
+      "grad_norm": 7.899261951446533,
+      "learning_rate": 2.9889587700908455e-05,
+      "loss": 3.0852,
+      "step": 80
+    },
+    {
+      "epoch": 0.02095947834187238,
+      "grad_norm": 3.6982789039611816,
+      "learning_rate": 2.987561146051712e-05,
+      "loss": 2.4326,
+      "step": 90
+    },
+    {
+      "epoch": 0.02328830926874709,
+      "grad_norm": 3.14382004737854,
+      "learning_rate": 2.9861635220125785e-05,
+      "loss": 2.6978,
+      "step": 100
+    },
+    {
+      "epoch": 0.025617140195621797,
+      "grad_norm": 6.083728313446045,
+      "learning_rate": 2.9847658979734452e-05,
+      "loss": 2.4978,
+      "step": 110
+    },
+    {
+      "epoch": 0.027945971122496506,
+      "grad_norm": 4.235475540161133,
+      "learning_rate": 2.9833682739343116e-05,
+      "loss": 2.6458,
+      "step": 120
+    },
+    {
+      "epoch": 0.030274802049371215,
+      "grad_norm": 6.495112895965576,
+      "learning_rate": 2.9819706498951783e-05,
+      "loss": 2.6736,
+      "step": 130
+    },
+    {
+      "epoch": 0.032603632976245925,
+      "grad_norm": 3.866530418395996,
+      "learning_rate": 2.9805730258560446e-05,
+      "loss": 2.355,
+      "step": 140
+    },
+    {
+      "epoch": 0.03493246390312063,
+      "grad_norm": 8.10139274597168,
+      "learning_rate": 2.9791754018169113e-05,
+      "loss": 2.7388,
+      "step": 150
+    },
+    {
+      "epoch": 0.037261294829995344,
+      "grad_norm": 9.420899391174316,
+      "learning_rate": 2.9777777777777777e-05,
+      "loss": 2.6467,
+      "step": 160
+    },
+    {
+      "epoch": 0.03959012575687005,
+      "grad_norm": 10.79451847076416,
+      "learning_rate": 2.9763801537386444e-05,
+      "loss": 2.8556,
+      "step": 170
+    },
+    {
+      "epoch": 0.04191895668374476,
+      "grad_norm": 4.488949775695801,
+      "learning_rate": 2.9749825296995107e-05,
+      "loss": 2.5636,
+      "step": 180
+    },
+    {
+      "epoch": 0.04424778761061947,
+      "grad_norm": 13.844695091247559,
+      "learning_rate": 2.9735849056603774e-05,
+      "loss": 2.3088,
+      "step": 190
+    },
+    {
+      "epoch": 0.04657661853749418,
+      "grad_norm": 5.569024562835693,
+      "learning_rate": 2.972187281621244e-05,
+      "loss": 2.7515,
+      "step": 200
+    },
+    {
+      "epoch": 0.04890544946436889,
+      "grad_norm": 3.9889473915100098,
+      "learning_rate": 2.9707896575821105e-05,
+      "loss": 2.0546,
+      "step": 210
+    },
+    {
+      "epoch": 0.05123428039124359,
+      "grad_norm": 7.793086051940918,
+      "learning_rate": 2.9693920335429772e-05,
+      "loss": 2.5026,
+      "step": 220
+    },
+    {
+      "epoch": 0.053563111318118306,
+      "grad_norm": 10.837105751037598,
+      "learning_rate": 2.9679944095038436e-05,
+      "loss": 2.7304,
+      "step": 230
+    },
+    {
+      "epoch": 0.05589194224499301,
+      "grad_norm": 5.149703025817871,
+      "learning_rate": 2.9665967854647103e-05,
+      "loss": 2.4211,
+      "step": 240
+    },
+    {
+      "epoch": 0.058220773171867725,
+      "grad_norm": 9.660852432250977,
+      "learning_rate": 2.9651991614255766e-05,
+      "loss": 2.2913,
+      "step": 250
+    },
+    {
+      "epoch": 0.06054960409874243,
+      "grad_norm": 13.585944175720215,
+      "learning_rate": 2.9638015373864433e-05,
+      "loss": 2.0449,
+      "step": 260
+    },
+    {
+      "epoch": 0.06287843502561714,
+      "grad_norm": 6.299298286437988,
+      "learning_rate": 2.9624039133473097e-05,
+      "loss": 2.2918,
+      "step": 270
+    },
+    {
+      "epoch": 0.06520726595249185,
+      "grad_norm": 3.201357364654541,
+      "learning_rate": 2.9610062893081764e-05,
+      "loss": 2.2325,
+      "step": 280
+    },
+    {
+      "epoch": 0.06753609687936656,
+      "grad_norm": 4.173748016357422,
+      "learning_rate": 2.9596086652690427e-05,
+      "loss": 2.3109,
+      "step": 290
+    },
+    {
+      "epoch": 0.06986492780624126,
+      "grad_norm": 5.973798751831055,
+      "learning_rate": 2.9582110412299094e-05,
+      "loss": 2.2817,
+      "step": 300
+    },
+    {
+      "epoch": 0.07219375873311598,
+      "grad_norm": 3.6076810359954834,
+      "learning_rate": 2.9568134171907758e-05,
+      "loss": 2.3076,
+      "step": 310
+    },
+    {
+      "epoch": 0.07452258965999069,
+      "grad_norm": 5.779540538787842,
+      "learning_rate": 2.955415793151642e-05,
+      "loss": 2.3035,
+      "step": 320
+    },
+    {
+      "epoch": 0.0768514205868654,
+      "grad_norm": 27.188243865966797,
+      "learning_rate": 2.9540181691125085e-05,
+      "loss": 2.5776,
+      "step": 330
+    },
+    {
+      "epoch": 0.0791802515137401,
+      "grad_norm": 8.513686180114746,
+      "learning_rate": 2.9526205450733752e-05,
+      "loss": 2.4484,
+      "step": 340
+    },
+    {
+      "epoch": 0.0815090824406148,
+      "grad_norm": 5.144784927368164,
+      "learning_rate": 2.951222921034242e-05,
+      "loss": 1.9469,
+      "step": 350
+    },
+    {
+      "epoch": 0.08383791336748952,
+      "grad_norm": 10.846701622009277,
+      "learning_rate": 2.9498252969951083e-05,
+      "loss": 2.301,
+      "step": 360
+    },
+    {
+      "epoch": 0.08616674429436423,
+      "grad_norm": 3.0678811073303223,
+      "learning_rate": 2.948427672955975e-05,
+      "loss": 2.4328,
+      "step": 370
+    },
+    {
+      "epoch": 0.08849557522123894,
+      "grad_norm": 4.1828413009643555,
+      "learning_rate": 2.9470300489168413e-05,
+      "loss": 1.8278,
+      "step": 380
+    },
+    {
+      "epoch": 0.09082440614811364,
+      "grad_norm": 7.489015579223633,
+      "learning_rate": 2.945632424877708e-05,
+      "loss": 1.7735,
+      "step": 390
+    },
+    {
+      "epoch": 0.09315323707498836,
+      "grad_norm": 8.380903244018555,
+      "learning_rate": 2.9442348008385744e-05,
+      "loss": 2.6727,
+      "step": 400
+    },
+    {
+      "epoch": 0.09548206800186307,
+      "grad_norm": 10.518119812011719,
+      "learning_rate": 2.942837176799441e-05,
+      "loss": 2.2046,
+      "step": 410
+    },
+    {
+      "epoch": 0.09781089892873777,
+      "grad_norm": 18.78498077392578,
+      "learning_rate": 2.9414395527603074e-05,
+      "loss": 2.3064,
+      "step": 420
+    },
+    {
+      "epoch": 0.10013972985561248,
+      "grad_norm": 4.508672714233398,
+      "learning_rate": 2.940041928721174e-05,
+      "loss": 2.1385,
+      "step": 430
+    },
+    {
+      "epoch": 0.10246856078248719,
+      "grad_norm": 12.739822387695312,
+      "learning_rate": 2.9386443046820405e-05,
+      "loss": 2.3654,
+      "step": 440
+    },
+    {
+      "epoch": 0.1047973917093619,
+      "grad_norm": 9.474522590637207,
+      "learning_rate": 2.9372466806429072e-05,
+      "loss": 1.8915,
+      "step": 450
+    },
+    {
+      "epoch": 0.10712622263623661,
+      "grad_norm": 8.7893705368042,
+      "learning_rate": 2.9358490566037735e-05,
+      "loss": 2.0838,
+      "step": 460
+    },
+    {
+      "epoch": 0.10945505356311132,
+      "grad_norm": 8.382100105285645,
+      "learning_rate": 2.9344514325646402e-05,
+      "loss": 2.2598,
+      "step": 470
+    },
+    {
+      "epoch": 0.11178388448998602,
+      "grad_norm": 10.099526405334473,
+      "learning_rate": 2.9330538085255066e-05,
+      "loss": 1.815,
+      "step": 480
+    },
+    {
+      "epoch": 0.11411271541686073,
+      "grad_norm": 7.139161109924316,
+      "learning_rate": 2.9316561844863733e-05,
+      "loss": 1.8358,
+      "step": 490
+    },
+    {
+      "epoch": 0.11644154634373545,
+      "grad_norm": 6.610865116119385,
+      "learning_rate": 2.93025856044724e-05,
+      "loss": 2.021,
+      "step": 500
+    },
+    {
+      "epoch": 0.11877037727061016,
+      "grad_norm": 8.798666000366211,
+      "learning_rate": 2.9288609364081064e-05,
+      "loss": 2.2395,
+      "step": 510
+    },
+    {
+      "epoch": 0.12109920819748486,
+      "grad_norm": 5.463034152984619,
+      "learning_rate": 2.927463312368973e-05,
+      "loss": 2.0574,
+      "step": 520
+    },
+    {
+      "epoch": 0.12342803912435957,
+      "grad_norm": 9.88456916809082,
+      "learning_rate": 2.9260656883298394e-05,
+      "loss": 2.0195,
+      "step": 530
+    },
+    {
+      "epoch": 0.1257568700512343,
+      "grad_norm": 6.4875078201293945,
+      "learning_rate": 2.924668064290706e-05,
+      "loss": 2.0221,
+      "step": 540
+    },
+    {
+      "epoch": 0.12808570097810898,
+      "grad_norm": 7.904644966125488,
+      "learning_rate": 2.9232704402515725e-05,
+      "loss": 1.9474,
+      "step": 550
+    },
+    {
+      "epoch": 0.1304145319049837,
+      "grad_norm": 6.517539024353027,
+      "learning_rate": 2.921872816212439e-05,
+      "loss": 1.9497,
+      "step": 560
+    },
+    {
+      "epoch": 0.13274336283185842,
+      "grad_norm": 13.083559036254883,
+      "learning_rate": 2.9204751921733055e-05,
+      "loss": 2.3732,
+      "step": 570
+    },
+    {
+      "epoch": 0.1350721937587331,
+      "grad_norm": 14.06116008758545,
+      "learning_rate": 2.9190775681341722e-05,
+      "loss": 2.0368,
+      "step": 580
+    },
+    {
+      "epoch": 0.13740102468560783,
+      "grad_norm": 6.1035308837890625,
+      "learning_rate": 2.9176799440950382e-05,
+      "loss": 1.7297,
+      "step": 590
+    },
+    {
+      "epoch": 0.13972985561248252,
+      "grad_norm": 22.444040298461914,
+      "learning_rate": 2.916282320055905e-05,
+      "loss": 1.9281,
+      "step": 600
+    },
+    {
+      "epoch": 0.14205868653935724,
+      "grad_norm": 8.657206535339355,
+      "learning_rate": 2.9148846960167713e-05,
+      "loss": 2.1254,
+      "step": 610
+    },
+    {
+      "epoch": 0.14438751746623196,
+      "grad_norm": 15.154228210449219,
+      "learning_rate": 2.913487071977638e-05,
+      "loss": 1.7197,
+      "step": 620
+    },
+    {
+      "epoch": 0.14671634839310665,
+      "grad_norm": 19.55249786376953,
+      "learning_rate": 2.9120894479385044e-05,
+      "loss": 2.0123,
+      "step": 630
+    },
+    {
+      "epoch": 0.14904517931998137,
+      "grad_norm": 13.391355514526367,
+      "learning_rate": 2.910691823899371e-05,
+      "loss": 1.8662,
+      "step": 640
+    },
+    {
+      "epoch": 0.15137401024685607,
+      "grad_norm": 11.850015640258789,
+      "learning_rate": 2.9092941998602377e-05,
+      "loss": 1.4906,
+      "step": 650
+    },
+    {
+      "epoch": 0.1537028411737308,
+      "grad_norm": 7.1012043952941895,
+      "learning_rate": 2.907896575821104e-05,
+      "loss": 2.0575,
+      "step": 660
+    },
+    {
+      "epoch": 0.1560316721006055,
+      "grad_norm": 6.259739875793457,
+      "learning_rate": 2.9064989517819708e-05,
+      "loss": 2.1558,
+      "step": 670
+    },
+    {
+      "epoch": 0.1583605030274802,
+      "grad_norm": 13.080730438232422,
+      "learning_rate": 2.905101327742837e-05,
+      "loss": 2.1056,
+      "step": 680
+    },
+    {
+      "epoch": 0.16068933395435492,
+      "grad_norm": 7.734439373016357,
+      "learning_rate": 2.903703703703704e-05,
+      "loss": 2.0659,
+      "step": 690
+    },
+    {
+      "epoch": 0.1630181648812296,
+      "grad_norm": 9.50960636138916,
+      "learning_rate": 2.9023060796645702e-05,
+      "loss": 2.1,
+      "step": 700
+    },
+    {
+      "epoch": 0.16534699580810433,
+      "grad_norm": 18.45827293395996,
+      "learning_rate": 2.900908455625437e-05,
+      "loss": 2.1444,
+      "step": 710
+    },
+    {
+      "epoch": 0.16767582673497905,
+      "grad_norm": 10.426290512084961,
+      "learning_rate": 2.8995108315863033e-05,
+      "loss": 2.2937,
+      "step": 720
+    },
+    {
+      "epoch": 0.17000465766185374,
+      "grad_norm": 15.96475601196289,
+      "learning_rate": 2.89811320754717e-05,
+      "loss": 2.0853,
+      "step": 730
+    },
+    {
+      "epoch": 0.17233348858872846,
+      "grad_norm": 7.1626410484313965,
+      "learning_rate": 2.8967155835080363e-05,
+      "loss": 2.1589,
+      "step": 740
+    },
+    {
+      "epoch": 0.17466231951560315,
+      "grad_norm": 16.46662139892578,
+      "learning_rate": 2.895317959468903e-05,
+      "loss": 1.5098,
+      "step": 750
+    },
+    {
+      "epoch": 0.17699115044247787,
+      "grad_norm": 12.333754539489746,
+      "learning_rate": 2.8939203354297694e-05,
+      "loss": 2.0206,
+      "step": 760
+    },
+    {
+      "epoch": 0.1793199813693526,
+      "grad_norm": 17.24032974243164,
+      "learning_rate": 2.892522711390636e-05,
+      "loss": 2.3082,
+      "step": 770
+    },
+    {
+      "epoch": 0.18164881229622729,
+      "grad_norm": 13.411188125610352,
+      "learning_rate": 2.8911250873515024e-05,
+      "loss": 1.6792,
+      "step": 780
+    },
+    {
+      "epoch": 0.183977643223102,
+      "grad_norm": 7.877953052520752,
+      "learning_rate": 2.889727463312369e-05,
+      "loss": 2.1541,
+      "step": 790
+    },
+    {
+      "epoch": 0.18630647414997673,
+      "grad_norm": 24.286130905151367,
+      "learning_rate": 2.888329839273236e-05,
+      "loss": 2.1689,
+      "step": 800
+    },
+    {
+      "epoch": 0.18863530507685142,
+      "grad_norm": 10.445568084716797,
+      "learning_rate": 2.8869322152341022e-05,
+      "loss": 1.8713,
+      "step": 810
+    },
+    {
+      "epoch": 0.19096413600372614,
+      "grad_norm": 6.439265251159668,
+      "learning_rate": 2.885534591194969e-05,
+      "loss": 1.9121,
+      "step": 820
+    },
+    {
+      "epoch": 0.19329296693060083,
+      "grad_norm": 24.311752319335938,
+      "learning_rate": 2.8841369671558353e-05,
+      "loss": 2.0249,
+      "step": 830
+    },
+    {
+      "epoch": 0.19562179785747555,
+      "grad_norm": 24.98615264892578,
+      "learning_rate": 2.882739343116702e-05,
+      "loss": 2.0608,
+      "step": 840
+    },
+    {
+      "epoch": 0.19795062878435027,
+      "grad_norm": 11.190683364868164,
+      "learning_rate": 2.881341719077568e-05,
+      "loss": 1.7668,
+      "step": 850
+    },
+    {
+      "epoch": 0.20027945971122496,
+      "grad_norm": 10.343644142150879,
+      "learning_rate": 2.8799440950384347e-05,
+      "loss": 1.8736,
+      "step": 860
+    },
+    {
+      "epoch": 0.20260829063809968,
+      "grad_norm": 4.8723530769348145,
+      "learning_rate": 2.878546470999301e-05,
+      "loss": 1.3761,
+      "step": 870
+    },
+    {
+      "epoch": 0.20493712156497437,
+      "grad_norm": 13.744318962097168,
+      "learning_rate": 2.8771488469601677e-05,
+      "loss": 1.487,
+      "step": 880
+    },
+    {
+      "epoch": 0.2072659524918491,
+      "grad_norm": 8.8036470413208,
+      "learning_rate": 2.875751222921034e-05,
+      "loss": 1.5511,
+      "step": 890
+    },
+    {
+      "epoch": 0.2095947834187238,
+      "grad_norm": 16.776226043701172,
+      "learning_rate": 2.8743535988819008e-05,
+      "loss": 1.6082,
+      "step": 900
+    },
+    {
+      "epoch": 0.2119236143455985,
+      "grad_norm": 15.255518913269043,
+      "learning_rate": 2.872955974842767e-05,
+      "loss": 1.6329,
+      "step": 910
+    },
+    {
+      "epoch": 0.21425244527247322,
+      "grad_norm": 16.834640502929688,
+      "learning_rate": 2.871558350803634e-05,
+      "loss": 1.8672,
+      "step": 920
+    },
+    {
+      "epoch": 0.21658127619934792,
+      "grad_norm": 10.923068046569824,
+      "learning_rate": 2.8701607267645002e-05,
+      "loss": 1.6011,
+      "step": 930
+    },
+    {
+      "epoch": 0.21891010712622264,
+      "grad_norm": 4.949512958526611,
+      "learning_rate": 2.868763102725367e-05,
+      "loss": 1.8381,
+      "step": 940
+    },
+    {
+      "epoch": 0.22123893805309736,
+      "grad_norm": 14.78408145904541,
+      "learning_rate": 2.8673654786862336e-05,
+      "loss": 1.7178,
+      "step": 950
+    },
+    {
+      "epoch": 0.22356776897997205,
+      "grad_norm": 10.509323120117188,
+      "learning_rate": 2.8659678546471e-05,
+      "loss": 1.5696,
+      "step": 960
+    },
+    {
+      "epoch": 0.22589659990684677,
+      "grad_norm": 13.359964370727539,
+      "learning_rate": 2.8645702306079667e-05,
+      "loss": 2.3818,
+      "step": 970
+    },
+    {
+      "epoch": 0.22822543083372146,
+      "grad_norm": 7.7604660987854,
+      "learning_rate": 2.863172606568833e-05,
+      "loss": 1.9007,
+      "step": 980
+    },
+    {
+      "epoch": 0.23055426176059618,
+      "grad_norm": 25.51626205444336,
+      "learning_rate": 2.8617749825296997e-05,
+      "loss": 1.6906,
+      "step": 990
+    },
+    {
+      "epoch": 0.2328830926874709,
+      "grad_norm": 53.45542907714844,
+      "learning_rate": 2.860377358490566e-05,
+      "loss": 2.3295,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2352119236143456,
+      "grad_norm": 6.25486421585083,
+      "learning_rate": 2.8589797344514328e-05,
+      "loss": 1.3573,
+      "step": 1010
+    },
+    {
+      "epoch": 0.2375407545412203,
+      "grad_norm": 16.020435333251953,
+      "learning_rate": 2.857582110412299e-05,
+      "loss": 1.5215,
+      "step": 1020
+    },
+    {
+      "epoch": 0.239869585468095,
+      "grad_norm": 19.388914108276367,
+      "learning_rate": 2.8561844863731658e-05,
+      "loss": 2.046,
+      "step": 1030
+    },
+    {
+      "epoch": 0.24219841639496972,
+      "grad_norm": 12.028767585754395,
+      "learning_rate": 2.8547868623340322e-05,
+      "loss": 2.0784,
+      "step": 1040
+    },
+    {
+      "epoch": 0.24452724732184444,
+      "grad_norm": 20.426069259643555,
+      "learning_rate": 2.853389238294899e-05,
+      "loss": 1.8761,
+      "step": 1050
+    },
+    {
+      "epoch": 0.24685607824871914,
+      "grad_norm": 8.661053657531738,
+      "learning_rate": 2.8519916142557652e-05,
+      "loss": 1.4791,
+      "step": 1060
+    },
+    {
+      "epoch": 0.24918490917559386,
+      "grad_norm": 5.779150009155273,
+      "learning_rate": 2.850593990216632e-05,
+      "loss": 1.741,
+      "step": 1070
+    },
+    {
+      "epoch": 0.2515137401024686,
+      "grad_norm": 8.086243629455566,
+      "learning_rate": 2.8491963661774983e-05,
+      "loss": 1.5324,
+      "step": 1080
+    },
+    {
+      "epoch": 0.25384257102934327,
+      "grad_norm": 15.885953903198242,
+      "learning_rate": 2.847798742138365e-05,
+      "loss": 1.5208,
+      "step": 1090
+    },
+    {
+      "epoch": 0.25617140195621796,
+      "grad_norm": 15.936205863952637,
+      "learning_rate": 2.8464011180992317e-05,
+      "loss": 2.0363,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2585002328830927,
+      "grad_norm": 17.394638061523438,
+      "learning_rate": 2.8450034940600977e-05,
+      "loss": 2.1736,
+      "step": 1110
+    },
+    {
+      "epoch": 0.2608290638099674,
+      "grad_norm": 32.21560287475586,
+      "learning_rate": 2.8436058700209644e-05,
+      "loss": 1.9567,
+      "step": 1120
+    },
+    {
+      "epoch": 0.2631578947368421,
+      "grad_norm": 21.898767471313477,
+      "learning_rate": 2.8422082459818308e-05,
+      "loss": 2.0504,
+      "step": 1130
+    },
+    {
+      "epoch": 0.26548672566371684,
+      "grad_norm": 18.948341369628906,
+      "learning_rate": 2.8408106219426975e-05,
+      "loss": 1.8841,
+      "step": 1140
+    },
+    {
+      "epoch": 0.26781555659059153,
+      "grad_norm": 14.407012939453125,
+      "learning_rate": 2.8394129979035638e-05,
+      "loss": 1.7384,
+      "step": 1150
+    },
+    {
+      "epoch": 0.2701443875174662,
+      "grad_norm": 7.401731967926025,
+      "learning_rate": 2.8380153738644305e-05,
+      "loss": 1.2154,
+      "step": 1160
+    },
+    {
+      "epoch": 0.2724732184443409,
+      "grad_norm": 10.56177806854248,
+      "learning_rate": 2.836617749825297e-05,
+      "loss": 1.4734,
+      "step": 1170
+    },
+    {
+      "epoch": 0.27480204937121566,
+      "grad_norm": 10.783820152282715,
+      "learning_rate": 2.8352201257861636e-05,
+      "loss": 1.4753,
+      "step": 1180
+    },
+    {
+      "epoch": 0.27713088029809035,
+      "grad_norm": 27.66096305847168,
+      "learning_rate": 2.83382250174703e-05,
+      "loss": 1.4075,
+      "step": 1190
+    },
+    {
+      "epoch": 0.27945971122496505,
+      "grad_norm": 39.09516143798828,
+      "learning_rate": 2.8324248777078966e-05,
+      "loss": 2.2883,
+      "step": 1200
+    },
+    {
+      "epoch": 0.2817885421518398,
+      "grad_norm": 20.759464263916016,
+      "learning_rate": 2.831027253668763e-05,
+      "loss": 2.2767,
+      "step": 1210
+    },
+    {
+      "epoch": 0.2841173730787145,
+      "grad_norm": 14.715803146362305,
+      "learning_rate": 2.8296296296296297e-05,
+      "loss": 1.7704,
+      "step": 1220
+    },
+    {
+      "epoch": 0.2864462040055892,
+      "grad_norm": 9.463580131530762,
+      "learning_rate": 2.828232005590496e-05,
+      "loss": 1.8876,
+      "step": 1230
+    },
+    {
+      "epoch": 0.2887750349324639,
+      "grad_norm": 20.55774688720703,
+      "learning_rate": 2.8268343815513627e-05,
+      "loss": 1.7778,
+      "step": 1240
+    },
+    {
+      "epoch": 0.2911038658593386,
+      "grad_norm": 4.738845348358154,
+      "learning_rate": 2.8254367575122294e-05,
+      "loss": 1.3079,
+      "step": 1250
+    },
+    {
+      "epoch": 0.2934326967862133,
+      "grad_norm": 33.04350280761719,
+      "learning_rate": 2.8240391334730958e-05,
+      "loss": 1.8545,
+      "step": 1260
+    },
+    {
+      "epoch": 0.29576152771308806,
+      "grad_norm": 10.747074127197266,
+      "learning_rate": 2.8226415094339625e-05,
+      "loss": 1.5794,
+      "step": 1270
+    },
+    {
+      "epoch": 0.29809035863996275,
+      "grad_norm": 20.86418914794922,
+      "learning_rate": 2.821243885394829e-05,
+      "loss": 1.7533,
+      "step": 1280
+    },
+    {
+      "epoch": 0.30041918956683744,
+      "grad_norm": 17.990188598632812,
+      "learning_rate": 2.8198462613556956e-05,
+      "loss": 1.7932,
+      "step": 1290
+    },
+    {
+      "epoch": 0.30274802049371213,
+      "grad_norm": 15.999037742614746,
+      "learning_rate": 2.818448637316562e-05,
+      "loss": 1.6424,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3050768514205869,
+      "grad_norm": 10.242752075195312,
+      "learning_rate": 2.8170510132774286e-05,
+      "loss": 1.5599,
+      "step": 1310
+    },
+    {
+      "epoch": 0.3074056823474616,
+      "grad_norm": 56.81425857543945,
+      "learning_rate": 2.815653389238295e-05,
+      "loss": 1.7664,
+      "step": 1320
+    },
+    {
+      "epoch": 0.30973451327433627,
+      "grad_norm": 5.3695502281188965,
+      "learning_rate": 2.8142557651991617e-05,
+      "loss": 1.6969,
+      "step": 1330
+    },
+    {
+      "epoch": 0.312063344201211,
+      "grad_norm": 20.256746292114258,
+      "learning_rate": 2.812858141160028e-05,
+      "loss": 1.8104,
+      "step": 1340
+    },
+    {
+      "epoch": 0.3143921751280857,
+      "grad_norm": 14.850010871887207,
+      "learning_rate": 2.8114605171208947e-05,
+      "loss": 1.5393,
+      "step": 1350
+    },
+    {
+      "epoch": 0.3167210060549604,
+      "grad_norm": 29.18364715576172,
+      "learning_rate": 2.810062893081761e-05,
+      "loss": 2.1662,
+      "step": 1360
+    },
+    {
+      "epoch": 0.31904983698183514,
+      "grad_norm": 10.589908599853516,
+      "learning_rate": 2.8086652690426278e-05,
+      "loss": 1.3533,
+      "step": 1370
+    },
+    {
+      "epoch": 0.32137866790870984,
+      "grad_norm": 21.720638275146484,
+      "learning_rate": 2.8072676450034938e-05,
+      "loss": 1.7474,
+      "step": 1380
+    },
+    {
+      "epoch": 0.32370749883558453,
+      "grad_norm": 21.47283172607422,
+      "learning_rate": 2.8058700209643605e-05,
+      "loss": 1.7673,
+      "step": 1390
+    },
+    {
+      "epoch": 0.3260363297624592,
+      "grad_norm": 6.568681716918945,
+      "learning_rate": 2.8044723969252272e-05,
+      "loss": 1.4598,
+      "step": 1400
+    },
+    {
+      "epoch": 0.32836516068933397,
+      "grad_norm": 19.102869033813477,
+      "learning_rate": 2.8030747728860936e-05,
+      "loss": 1.6002,
+      "step": 1410
+    },
+    {
+      "epoch": 0.33069399161620866,
+      "grad_norm": 18.268447875976562,
+      "learning_rate": 2.8016771488469603e-05,
+      "loss": 1.3573,
+      "step": 1420
+    },
+    {
+      "epoch": 0.33302282254308335,
+      "grad_norm": 17.070833206176758,
+      "learning_rate": 2.8002795248078266e-05,
+      "loss": 2.402,
+      "step": 1430
+    },
+    {
+      "epoch": 0.3353516534699581,
+      "grad_norm": 9.851959228515625,
+      "learning_rate": 2.7988819007686933e-05,
+      "loss": 1.481,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3376804843968328,
+      "grad_norm": 27.05587387084961,
+      "learning_rate": 2.7974842767295597e-05,
+      "loss": 1.9486,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3400093153237075,
+      "grad_norm": 17.53416633605957,
+      "learning_rate": 2.7960866526904264e-05,
+      "loss": 1.6455,
+      "step": 1460
+    },
+    {
+      "epoch": 0.34233814625058223,
+      "grad_norm": 23.41783332824707,
+      "learning_rate": 2.7946890286512927e-05,
+      "loss": 1.5375,
+      "step": 1470
+    },
+    {
+      "epoch": 0.3446669771774569,
+      "grad_norm": 8.158493995666504,
+      "learning_rate": 2.7932914046121594e-05,
+      "loss": 1.3204,
+      "step": 1480
+    },
+    {
+      "epoch": 0.3469958081043316,
+      "grad_norm": 18.967121124267578,
+      "learning_rate": 2.7918937805730258e-05,
+      "loss": 1.7482,
+      "step": 1490
+    },
+    {
+      "epoch": 0.3493246390312063,
+      "grad_norm": 15.774516105651855,
+      "learning_rate": 2.7904961565338925e-05,
+      "loss": 1.6003,
+      "step": 1500
+    },
+    {
+      "epoch": 0.35165346995808106,
+      "grad_norm": 10.410409927368164,
+      "learning_rate": 2.789098532494759e-05,
+      "loss": 1.6872,
+      "step": 1510
+    },
+    {
+      "epoch": 0.35398230088495575,
+      "grad_norm": 12.958462715148926,
+      "learning_rate": 2.7877009084556255e-05,
+      "loss": 1.6192,
+      "step": 1520
+    },
+    {
+      "epoch": 0.35631113181183044,
+      "grad_norm": 8.600435256958008,
+      "learning_rate": 2.786303284416492e-05,
+      "loss": 1.2578,
+      "step": 1530
+    },
+    {
+      "epoch": 0.3586399627387052,
+      "grad_norm": 12.038568496704102,
+      "learning_rate": 2.7849056603773586e-05,
+      "loss": 1.5104,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3609687936655799,
+      "grad_norm": 23.397693634033203,
+      "learning_rate": 2.7835080363382253e-05,
+      "loss": 1.5705,
+      "step": 1550
+    },
+    {
+      "epoch": 0.36329762459245457,
+      "grad_norm": 19.64638328552246,
+      "learning_rate": 2.7821104122990917e-05,
+      "loss": 1.2922,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3656264555193293,
+      "grad_norm": 11.045112609863281,
+      "learning_rate": 2.7807127882599584e-05,
+      "loss": 1.354,
+      "step": 1570
+    },
+    {
+      "epoch": 0.367955286446204,
+      "grad_norm": 8.17792797088623,
+      "learning_rate": 2.7793151642208247e-05,
+      "loss": 1.4809,
+      "step": 1580
+    },
+    {
+      "epoch": 0.3702841173730787,
+      "grad_norm": 15.052375793457031,
+      "learning_rate": 2.7779175401816914e-05,
+      "loss": 1.3746,
+      "step": 1590
+    },
+    {
+      "epoch": 0.37261294829995345,
+      "grad_norm": 7.82668399810791,
+      "learning_rate": 2.7765199161425578e-05,
+      "loss": 1.3388,
+      "step": 1600
+    },
+    {
+      "epoch": 0.37494177922682814,
+      "grad_norm": 24.902265548706055,
+      "learning_rate": 2.7751222921034245e-05,
+      "loss": 1.447,
+      "step": 1610
+    },
+    {
+      "epoch": 0.37727061015370283,
+      "grad_norm": 21.929048538208008,
+      "learning_rate": 2.7737246680642908e-05,
+      "loss": 1.529,
+      "step": 1620
+    },
+    {
+      "epoch": 0.3795994410805775,
+      "grad_norm": 13.071199417114258,
+      "learning_rate": 2.7723270440251575e-05,
+      "loss": 1.7533,
+      "step": 1630
+    },
+    {
+      "epoch": 0.3819282720074523,
+      "grad_norm": 23.845232009887695,
+      "learning_rate": 2.7709294199860235e-05,
+      "loss": 1.697,
+      "step": 1640
+    },
+    {
+      "epoch": 0.38425710293432697,
+      "grad_norm": 16.655656814575195,
+      "learning_rate": 2.7695317959468902e-05,
+      "loss": 1.3866,
+      "step": 1650
+    },
+    {
+      "epoch": 0.38658593386120166,
+      "grad_norm": 5.929255962371826,
+      "learning_rate": 2.7681341719077566e-05,
+      "loss": 1.5702,
+      "step": 1660
+    },
+    {
+      "epoch": 0.3889147647880764,
+      "grad_norm": 5.026130676269531,
+      "learning_rate": 2.7667365478686233e-05,
+      "loss": 1.0261,
+      "step": 1670
+    },
+    {
+      "epoch": 0.3912435957149511,
+      "grad_norm": 48.83590316772461,
+      "learning_rate": 2.76533892382949e-05,
+      "loss": 1.9335,
+      "step": 1680
+    },
+    {
+      "epoch": 0.3935724266418258,
+      "grad_norm": 6.456787109375,
+      "learning_rate": 2.7639412997903564e-05,
+      "loss": 1.8191,
+      "step": 1690
+    },
+    {
+      "epoch": 0.39590125756870054,
+      "grad_norm": 7.918270111083984,
+      "learning_rate": 2.762543675751223e-05,
+      "loss": 1.4435,
+      "step": 1700
+    },
+    {
+      "epoch": 0.39823008849557523,
+      "grad_norm": 7.082935810089111,
+      "learning_rate": 2.7611460517120894e-05,
+      "loss": 1.1776,
+      "step": 1710
+    },
+    {
+      "epoch": 0.4005589194224499,
+      "grad_norm": 15.370290756225586,
+      "learning_rate": 2.759748427672956e-05,
+      "loss": 1.7205,
+      "step": 1720
+    },
+    {
+      "epoch": 0.4028877503493246,
+      "grad_norm": 4.2508392333984375,
+      "learning_rate": 2.7583508036338225e-05,
+      "loss": 1.4412,
+      "step": 1730
+    },
+    {
+      "epoch": 0.40521658127619936,
+      "grad_norm": 21.323366165161133,
+      "learning_rate": 2.756953179594689e-05,
+      "loss": 1.6467,
+      "step": 1740
+    },
+    {
+      "epoch": 0.40754541220307405,
+      "grad_norm": 6.784306049346924,
+      "learning_rate": 2.7555555555555555e-05,
+      "loss": 2.0458,
+      "step": 1750
+    },
+    {
+      "epoch": 0.40987424312994875,
+      "grad_norm": 5.1630988121032715,
+      "learning_rate": 2.7541579315164222e-05,
+      "loss": 1.3938,
+      "step": 1760
+    },
+    {
+      "epoch": 0.4122030740568235,
+      "grad_norm": 14.7197847366333,
+      "learning_rate": 2.7527603074772886e-05,
+      "loss": 1.5101,
+      "step": 1770
+    },
+    {
+      "epoch": 0.4145319049836982,
+      "grad_norm": 28.54730987548828,
+      "learning_rate": 2.7513626834381553e-05,
+      "loss": 1.5704,
+      "step": 1780
+    },
+    {
+      "epoch": 0.4168607359105729,
+      "grad_norm": 13.938275337219238,
+      "learning_rate": 2.7499650593990216e-05,
+      "loss": 1.446,
+      "step": 1790
+    },
+    {
+      "epoch": 0.4191895668374476,
+      "grad_norm": 45.821048736572266,
+      "learning_rate": 2.7485674353598883e-05,
+      "loss": 2.2999,
+      "step": 1800
+    },
+    {
+      "epoch": 0.4215183977643223,
+      "grad_norm": 4.642112731933594,
+      "learning_rate": 2.7471698113207547e-05,
+      "loss": 1.0337,
+      "step": 1810
+    },
+    {
+      "epoch": 0.423847228691197,
+      "grad_norm": 85.71287536621094,
+      "learning_rate": 2.7457721872816214e-05,
+      "loss": 1.5382,
+      "step": 1820
+    },
+    {
+      "epoch": 0.4261760596180717,
+      "grad_norm": 12.570047378540039,
+      "learning_rate": 2.744374563242488e-05,
+      "loss": 1.6143,
+      "step": 1830
+    },
+    {
+      "epoch": 0.42850489054494645,
+      "grad_norm": 44.61286544799805,
+      "learning_rate": 2.7429769392033544e-05,
+      "loss": 1.9973,
+      "step": 1840
+    },
+    {
+      "epoch": 0.43083372147182114,
+      "grad_norm": 10.40942096710205,
+      "learning_rate": 2.741579315164221e-05,
+      "loss": 1.6417,
+      "step": 1850
+    },
+    {
+      "epoch": 0.43316255239869583,
+      "grad_norm": 14.36320686340332,
+      "learning_rate": 2.7401816911250875e-05,
+      "loss": 1.5154,
+      "step": 1860
+    },
+    {
+      "epoch": 0.4354913833255706,
+      "grad_norm": 8.61720085144043,
+      "learning_rate": 2.7387840670859542e-05,
+      "loss": 1.4092,
+      "step": 1870
+    },
+    {
+      "epoch": 0.43782021425244527,
+      "grad_norm": 6.423306941986084,
+      "learning_rate": 2.7373864430468206e-05,
+      "loss": 1.7349,
+      "step": 1880
+    },
+    {
+      "epoch": 0.44014904517931996,
+      "grad_norm": 27.920555114746094,
+      "learning_rate": 2.7359888190076873e-05,
+      "loss": 1.2185,
+      "step": 1890
+    },
+    {
+      "epoch": 0.4424778761061947,
+      "grad_norm": 17.56697654724121,
+      "learning_rate": 2.7345911949685533e-05,
+      "loss": 2.2574,
+      "step": 1900
+    },
+    {
+      "epoch": 0.4448067070330694,
+      "grad_norm": 11.816027641296387,
+      "learning_rate": 2.73319357092942e-05,
+      "loss": 1.112,
+      "step": 1910
+    },
+    {
+      "epoch": 0.4471355379599441,
+      "grad_norm": 18.829771041870117,
+      "learning_rate": 2.7317959468902863e-05,
+      "loss": 1.5212,
+      "step": 1920
+    },
+    {
+      "epoch": 0.44946436888681884,
+      "grad_norm": 20.83513832092285,
+      "learning_rate": 2.730398322851153e-05,
+      "loss": 1.9292,
+      "step": 1930
+    },
+    {
+      "epoch": 0.45179319981369354,
+      "grad_norm": 6.374683380126953,
+      "learning_rate": 2.7290006988120194e-05,
+      "loss": 1.5452,
+      "step": 1940
+    },
+    {
+      "epoch": 0.4541220307405682,
+      "grad_norm": 10.145646095275879,
+      "learning_rate": 2.727603074772886e-05,
+      "loss": 1.779,
+      "step": 1950
+    },
+    {
+      "epoch": 0.4564508616674429,
+      "grad_norm": 15.401958465576172,
+      "learning_rate": 2.7262054507337525e-05,
+      "loss": 1.4472,
+      "step": 1960
+    },
+    {
+      "epoch": 0.45877969259431767,
+      "grad_norm": 70.80525970458984,
+      "learning_rate": 2.724807826694619e-05,
+      "loss": 1.7908,
+      "step": 1970
+    },
+    {
+      "epoch": 0.46110852352119236,
+      "grad_norm": 15.792387008666992,
+      "learning_rate": 2.723410202655486e-05,
+      "loss": 1.408,
+      "step": 1980
+    },
+    {
+      "epoch": 0.46343735444806705,
+      "grad_norm": 10.236753463745117,
+      "learning_rate": 2.7220125786163522e-05,
+      "loss": 1.2109,
+      "step": 1990
+    },
+    {
+      "epoch": 0.4657661853749418,
+      "grad_norm": 39.971900939941406,
+      "learning_rate": 2.720614954577219e-05,
+      "loss": 1.4968,
+      "step": 2000
+    },
+    {
+      "epoch": 0.4680950163018165,
+      "grad_norm": 13.20219612121582,
+      "learning_rate": 2.7192173305380853e-05,
+      "loss": 1.3401,
+      "step": 2010
+    },
+    {
+      "epoch": 0.4704238472286912,
+      "grad_norm": 20.595693588256836,
+      "learning_rate": 2.717819706498952e-05,
+      "loss": 1.3456,
+      "step": 2020
+    },
+    {
+      "epoch": 0.47275267815556593,
+      "grad_norm": 5.41182804107666,
+      "learning_rate": 2.7164220824598183e-05,
+      "loss": 1.4178,
+      "step": 2030
+    },
+    {
+      "epoch": 0.4750815090824406,
+      "grad_norm": 10.29110336303711,
+      "learning_rate": 2.715024458420685e-05,
+      "loss": 1.1662,
+      "step": 2040
+    },
+    {
+      "epoch": 0.4774103400093153,
+      "grad_norm": 8.8546781539917,
+      "learning_rate": 2.7136268343815514e-05,
+      "loss": 1.3032,
+      "step": 2050
+    },
+    {
+      "epoch": 0.47973917093619,
+      "grad_norm": 16.56825828552246,
+      "learning_rate": 2.712229210342418e-05,
+      "loss": 1.6524,
+      "step": 2060
+    },
+    {
+      "epoch": 0.48206800186306475,
+      "grad_norm": 8.039105415344238,
+      "learning_rate": 2.7108315863032844e-05,
+      "loss": 1.3149,
+      "step": 2070
+    },
+    {
+      "epoch": 0.48439683278993945,
+      "grad_norm": 7.804418087005615,
+      "learning_rate": 2.709433962264151e-05,
+      "loss": 1.2711,
+      "step": 2080
+    },
+    {
+      "epoch": 0.48672566371681414,
+      "grad_norm": 14.90645980834961,
+      "learning_rate": 2.7080363382250175e-05,
+      "loss": 1.6439,
+      "step": 2090
+    },
+    {
+      "epoch": 0.4890544946436889,
+      "grad_norm": 18.745508193969727,
+      "learning_rate": 2.7066387141858842e-05,
+      "loss": 1.3836,
+      "step": 2100
+    },
+    {
+      "epoch": 0.4913833255705636,
+      "grad_norm": 13.745448112487793,
+      "learning_rate": 2.7052410901467505e-05,
+      "loss": 1.7098,
+      "step": 2110
+    },
+    {
+      "epoch": 0.49371215649743827,
+      "grad_norm": 8.893570899963379,
+      "learning_rate": 2.7038434661076172e-05,
+      "loss": 1.5233,
+      "step": 2120
+    },
+    {
+      "epoch": 0.496040987424313,
+      "grad_norm": 5.876834869384766,
+      "learning_rate": 2.702445842068484e-05,
+      "loss": 1.7379,
+      "step": 2130
+    },
+    {
+      "epoch": 0.4983698183511877,
+      "grad_norm": 36.34437561035156,
+      "learning_rate": 2.7010482180293503e-05,
+      "loss": 1.2833,
+      "step": 2140
+    },
+    {
+      "epoch": 0.5006986492780624,
+      "grad_norm": 7.839767932891846,
+      "learning_rate": 2.699650593990217e-05,
+      "loss": 1.3774,
+      "step": 2150
+    },
+    {
+      "epoch": 0.5030274802049371,
+      "grad_norm": 20.03778648376465,
+      "learning_rate": 2.6982529699510834e-05,
+      "loss": 1.1518,
+      "step": 2160
+    },
+    {
+      "epoch": 0.5053563111318118,
+      "grad_norm": 16.76922607421875,
+      "learning_rate": 2.6968553459119497e-05,
+      "loss": 1.4198,
+      "step": 2170
+    },
+    {
+      "epoch": 0.5076851420586865,
+      "grad_norm": 15.900551795959473,
+      "learning_rate": 2.695457721872816e-05,
+      "loss": 1.8107,
+      "step": 2180
+    },
+    {
+      "epoch": 0.5100139729855613,
+      "grad_norm": 39.533077239990234,
+      "learning_rate": 2.6940600978336828e-05,
+      "loss": 1.5067,
+      "step": 2190
+    },
+    {
+      "epoch": 0.5123428039124359,
+      "grad_norm": 8.290783882141113,
+      "learning_rate": 2.692662473794549e-05,
+      "loss": 1.2505,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5146716348393107,
+      "grad_norm": 11.7975492477417,
+      "learning_rate": 2.6912648497554158e-05,
+      "loss": 1.2652,
+      "step": 2210
+    },
+    {
+      "epoch": 0.5170004657661854,
+      "grad_norm": 35.9605712890625,
+      "learning_rate": 2.6898672257162822e-05,
+      "loss": 1.5008,
+      "step": 2220
+    },
+    {
+      "epoch": 0.51932929669306,
+      "grad_norm": 8.939215660095215,
+      "learning_rate": 2.688469601677149e-05,
+      "loss": 1.3159,
+      "step": 2230
+    },
+    {
+      "epoch": 0.5216581276199348,
+      "grad_norm": 29.108440399169922,
+      "learning_rate": 2.6870719776380152e-05,
+      "loss": 1.835,
+      "step": 2240
+    },
+    {
+      "epoch": 0.5239869585468095,
+      "grad_norm": 41.85350036621094,
+      "learning_rate": 2.685674353598882e-05,
+      "loss": 1.4529,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5263157894736842,
+      "grad_norm": 15.356282234191895,
+      "learning_rate": 2.6842767295597483e-05,
+      "loss": 1.4199,
+      "step": 2260
+    },
+    {
+      "epoch": 0.5286446204005589,
+      "grad_norm": 10.252715110778809,
+      "learning_rate": 2.682879105520615e-05,
+      "loss": 1.171,
+      "step": 2270
+    },
+    {
+      "epoch": 0.5309734513274337,
+      "grad_norm": 23.096691131591797,
+      "learning_rate": 2.6814814814814817e-05,
+      "loss": 1.7865,
+      "step": 2280
+    },
+    {
+      "epoch": 0.5333022822543083,
+      "grad_norm": 6.9515910148620605,
+      "learning_rate": 2.680083857442348e-05,
+      "loss": 1.3458,
+      "step": 2290
+    },
+    {
+      "epoch": 0.5356311131811831,
+      "grad_norm": 39.11237335205078,
+      "learning_rate": 2.6786862334032148e-05,
+      "loss": 1.8611,
+      "step": 2300
+    },
+    {
+      "epoch": 0.5379599441080578,
+      "grad_norm": 6.406713485717773,
+      "learning_rate": 2.677288609364081e-05,
+      "loss": 1.3962,
+      "step": 2310
+    },
+    {
+      "epoch": 0.5402887750349324,
+      "grad_norm": 8.728299140930176,
+      "learning_rate": 2.6758909853249478e-05,
+      "loss": 1.7573,
+      "step": 2320
+    },
+    {
+      "epoch": 0.5426176059618072,
+      "grad_norm": 14.61226749420166,
+      "learning_rate": 2.674493361285814e-05,
+      "loss": 1.5703,
+      "step": 2330
+    },
+    {
+      "epoch": 0.5449464368886818,
+      "grad_norm": 13.667352676391602,
+      "learning_rate": 2.673095737246681e-05,
+      "loss": 1.0584,
+      "step": 2340
+    },
+    {
+      "epoch": 0.5472752678155566,
+      "grad_norm": 24.96540069580078,
+      "learning_rate": 2.6716981132075472e-05,
+      "loss": 1.4208,
+      "step": 2350
+    },
+    {
+      "epoch": 0.5496040987424313,
+      "grad_norm": 17.21762466430664,
+      "learning_rate": 2.670300489168414e-05,
+      "loss": 1.2686,
+      "step": 2360
+    },
+    {
+      "epoch": 0.551932929669306,
+      "grad_norm": 8.74630355834961,
+      "learning_rate": 2.6689028651292803e-05,
+      "loss": 1.3365,
+      "step": 2370
+    },
+    {
+      "epoch": 0.5542617605961807,
+      "grad_norm": 81.46407318115234,
+      "learning_rate": 2.667505241090147e-05,
+      "loss": 1.7805,
+      "step": 2380
+    },
+    {
+      "epoch": 0.5565905915230555,
+      "grad_norm": 7.569358825683594,
+      "learning_rate": 2.6661076170510133e-05,
+      "loss": 1.2882,
+      "step": 2390
+    },
+    {
+      "epoch": 0.5589194224499301,
+      "grad_norm": 9.290304183959961,
+      "learning_rate": 2.66470999301188e-05,
+      "loss": 1.3211,
+      "step": 2400
+    },
+    {
+      "epoch": 0.5612482533768048,
+      "grad_norm": 9.045026779174805,
+      "learning_rate": 2.6633123689727464e-05,
+      "loss": 1.4446,
+      "step": 2410
+    },
+    {
+      "epoch": 0.5635770843036796,
+      "grad_norm": 98.28316497802734,
+      "learning_rate": 2.661914744933613e-05,
+      "loss": 1.3712,
+      "step": 2420
+    },
+    {
+      "epoch": 0.5659059152305542,
+      "grad_norm": 15.944232940673828,
+      "learning_rate": 2.6605171208944795e-05,
+      "loss": 1.5916,
+      "step": 2430
+    },
+    {
+      "epoch": 0.568234746157429,
+      "grad_norm": 11.774263381958008,
+      "learning_rate": 2.6591194968553458e-05,
+      "loss": 1.3679,
+      "step": 2440
+    },
+    {
+      "epoch": 0.5705635770843037,
+      "grad_norm": 13.816254615783691,
+      "learning_rate": 2.6577218728162125e-05,
+      "loss": 1.1344,
+      "step": 2450
+    },
+    {
+      "epoch": 0.5728924080111784,
+      "grad_norm": 95.79457092285156,
+      "learning_rate": 2.656324248777079e-05,
+      "loss": 1.7998,
+      "step": 2460
+    },
+    {
+      "epoch": 0.5752212389380531,
+      "grad_norm": 9.518484115600586,
+      "learning_rate": 2.6549266247379456e-05,
+      "loss": 1.6957,
+      "step": 2470
+    },
+    {
+      "epoch": 0.5775500698649279,
+      "grad_norm": 7.290702819824219,
+      "learning_rate": 2.653529000698812e-05,
+      "loss": 1.2882,
+      "step": 2480
+    },
+    {
+      "epoch": 0.5798789007918025,
+      "grad_norm": 10.004209518432617,
+      "learning_rate": 2.6521313766596786e-05,
+      "loss": 0.9912,
+      "step": 2490
+    },
+    {
+      "epoch": 0.5822077317186772,
+      "grad_norm": 19.958898544311523,
+      "learning_rate": 2.650733752620545e-05,
+      "loss": 1.4569,
+      "step": 2500
+    },
+    {
+      "epoch": 0.584536562645552,
+      "grad_norm": 100.88062286376953,
+      "learning_rate": 2.6493361285814117e-05,
+      "loss": 1.8738,
+      "step": 2510
+    },
+    {
+      "epoch": 0.5868653935724266,
+      "grad_norm": 16.205934524536133,
+      "learning_rate": 2.647938504542278e-05,
+      "loss": 1.2735,
+      "step": 2520
+    },
+    {
+      "epoch": 0.5891942244993014,
+      "grad_norm": 26.231815338134766,
+      "learning_rate": 2.6465408805031447e-05,
+      "loss": 1.284,
+      "step": 2530
+    },
+    {
+      "epoch": 0.5915230554261761,
+      "grad_norm": 43.43105697631836,
+      "learning_rate": 2.645143256464011e-05,
+      "loss": 1.5655,
+      "step": 2540
+    },
+    {
+      "epoch": 0.5938518863530508,
+      "grad_norm": 14.103740692138672,
+      "learning_rate": 2.6437456324248778e-05,
+      "loss": 1.9522,
+      "step": 2550
+    },
+    {
+      "epoch": 0.5961807172799255,
+      "grad_norm": 24.451862335205078,
+      "learning_rate": 2.642348008385744e-05,
+      "loss": 1.9858,
+      "step": 2560
+    },
+    {
+      "epoch": 0.5985095482068001,
+      "grad_norm": 11.756208419799805,
+      "learning_rate": 2.640950384346611e-05,
+      "loss": 1.5646,
+      "step": 2570
+    },
+    {
+      "epoch": 0.6008383791336749,
+      "grad_norm": 22.268434524536133,
+      "learning_rate": 2.6395527603074775e-05,
+      "loss": 1.2753,
+      "step": 2580
+    },
+    {
+      "epoch": 0.6031672100605496,
+      "grad_norm": 10.877938270568848,
+      "learning_rate": 2.638155136268344e-05,
+      "loss": 1.7155,
+      "step": 2590
+    },
+    {
+      "epoch": 0.6054960409874243,
+      "grad_norm": 19.884037017822266,
+      "learning_rate": 2.6367575122292106e-05,
+      "loss": 1.0979,
+      "step": 2600
+    },
+    {
+      "epoch": 0.607824871914299,
+      "grad_norm": 9.810844421386719,
+      "learning_rate": 2.635359888190077e-05,
+      "loss": 1.3613,
+      "step": 2610
+    },
+    {
+      "epoch": 0.6101537028411738,
+      "grad_norm": 10.557251930236816,
+      "learning_rate": 2.6339622641509437e-05,
+      "loss": 1.3459,
+      "step": 2620
+    },
+    {
+      "epoch": 0.6124825337680484,
+      "grad_norm": 13.84223461151123,
+      "learning_rate": 2.63256464011181e-05,
+      "loss": 1.315,
+      "step": 2630
+    },
+    {
+      "epoch": 0.6148113646949231,
+      "grad_norm": 15.142064094543457,
+      "learning_rate": 2.6311670160726767e-05,
+      "loss": 1.5922,
+      "step": 2640
+    },
+    {
+      "epoch": 0.6171401956217979,
+      "grad_norm": 29.648056030273438,
+      "learning_rate": 2.629769392033543e-05,
+      "loss": 1.3471,
+      "step": 2650
+    },
+    {
+      "epoch": 0.6194690265486725,
+      "grad_norm": 5.652649879455566,
+      "learning_rate": 2.6283717679944098e-05,
+      "loss": 1.5202,
+      "step": 2660
+    },
+    {
+      "epoch": 0.6217978574755473,
+      "grad_norm": 7.965731143951416,
+      "learning_rate": 2.626974143955276e-05,
+      "loss": 1.0326,
+      "step": 2670
+    },
+    {
+      "epoch": 0.624126688402422,
+      "grad_norm": 60.36565017700195,
+      "learning_rate": 2.6255765199161428e-05,
+      "loss": 1.3599,
+      "step": 2680
+    },
+    {
+      "epoch": 0.6264555193292967,
+      "grad_norm": 12.119543075561523,
+      "learning_rate": 2.624178895877009e-05,
+      "loss": 1.4352,
+      "step": 2690
+    },
+    {
+      "epoch": 0.6287843502561714,
+      "grad_norm": 18.028051376342773,
+      "learning_rate": 2.6227812718378755e-05,
+      "loss": 1.7631,
+      "step": 2700
+    },
+    {
+      "epoch": 0.6311131811830462,
+      "grad_norm": 14.728964805603027,
+      "learning_rate": 2.621383647798742e-05,
+      "loss": 1.5176,
+      "step": 2710
+    },
+    {
+      "epoch": 0.6334420121099208,
+      "grad_norm": 46.449974060058594,
+      "learning_rate": 2.6199860237596086e-05,
+      "loss": 1.5507,
+      "step": 2720
+    },
+    {
+      "epoch": 0.6357708430367955,
+      "grad_norm": 5.887063026428223,
+      "learning_rate": 2.6185883997204753e-05,
+      "loss": 1.3399,
+      "step": 2730
+    },
+    {
+      "epoch": 0.6380996739636703,
+      "grad_norm": 9.698420524597168,
+      "learning_rate": 2.6171907756813417e-05,
+      "loss": 1.0447,
+      "step": 2740
+    },
+    {
+      "epoch": 0.6404285048905449,
+      "grad_norm": 8.20369815826416,
+      "learning_rate": 2.6157931516422084e-05,
+      "loss": 1.4727,
+      "step": 2750
+    },
+    {
+      "epoch": 0.6427573358174197,
+      "grad_norm": 17.19390869140625,
+      "learning_rate": 2.6143955276030747e-05,
+      "loss": 1.4083,
+      "step": 2760
+    },
+    {
+      "epoch": 0.6450861667442943,
+      "grad_norm": 17.035465240478516,
+      "learning_rate": 2.6129979035639414e-05,
+      "loss": 1.1995,
+      "step": 2770
+    },
+    {
+      "epoch": 0.6474149976711691,
+      "grad_norm": 7.173000812530518,
+      "learning_rate": 2.6116002795248078e-05,
+      "loss": 1.1313,
+      "step": 2780
+    },
+    {
+      "epoch": 0.6497438285980438,
+      "grad_norm": 20.684951782226562,
+      "learning_rate": 2.6102026554856745e-05,
+      "loss": 1.5742,
+      "step": 2790
+    },
+    {
+      "epoch": 0.6520726595249184,
+      "grad_norm": 9.083842277526855,
+      "learning_rate": 2.6088050314465408e-05,
+      "loss": 1.4098,
+      "step": 2800
+    },
+    {
+      "epoch": 0.6544014904517932,
+      "grad_norm": 7.859493732452393,
+      "learning_rate": 2.6074074074074075e-05,
+      "loss": 1.3814,
+      "step": 2810
+    },
+    {
+      "epoch": 0.6567303213786679,
+      "grad_norm": 5.692265033721924,
+      "learning_rate": 2.606009783368274e-05,
+      "loss": 1.7696,
+      "step": 2820
+    },
+    {
+      "epoch": 0.6590591523055426,
+      "grad_norm": 46.135520935058594,
+      "learning_rate": 2.6046121593291406e-05,
+      "loss": 2.3788,
+      "step": 2830
+    },
+    {
+      "epoch": 0.6613879832324173,
+      "grad_norm": 13.176454544067383,
+      "learning_rate": 2.603214535290007e-05,
+      "loss": 1.8681,
+      "step": 2840
+    },
+    {
+      "epoch": 0.6637168141592921,
+      "grad_norm": 13.691810607910156,
+      "learning_rate": 2.6018169112508736e-05,
+      "loss": 1.5054,
+      "step": 2850
+    },
+    {
+      "epoch": 0.6660456450861667,
+      "grad_norm": 115.35124969482422,
+      "learning_rate": 2.60041928721174e-05,
+      "loss": 1.4528,
+      "step": 2860
+    },
+    {
+      "epoch": 0.6683744760130415,
+      "grad_norm": 4.287045478820801,
+      "learning_rate": 2.5990216631726067e-05,
+      "loss": 1.1515,
+      "step": 2870
+    },
+    {
+      "epoch": 0.6707033069399162,
+      "grad_norm": 6.276763439178467,
+      "learning_rate": 2.5976240391334734e-05,
+      "loss": 1.3343,
+      "step": 2880
+    },
+    {
+      "epoch": 0.6730321378667908,
+      "grad_norm": 54.19549560546875,
+      "learning_rate": 2.5962264150943398e-05,
+      "loss": 1.9168,
+      "step": 2890
+    },
+    {
+      "epoch": 0.6753609687936656,
+      "grad_norm": 24.059030532836914,
+      "learning_rate": 2.5948287910552065e-05,
+      "loss": 1.4515,
+      "step": 2900
+    },
+    {
+      "epoch": 0.6776897997205403,
+      "grad_norm": 25.165706634521484,
+      "learning_rate": 2.5934311670160728e-05,
+      "loss": 1.6255,
+      "step": 2910
+    },
+    {
+      "epoch": 0.680018630647415,
+      "grad_norm": 16.515329360961914,
+      "learning_rate": 2.5920335429769395e-05,
+      "loss": 1.3936,
+      "step": 2920
+    },
+    {
+      "epoch": 0.6823474615742897,
+      "grad_norm": 21.545387268066406,
+      "learning_rate": 2.590635918937806e-05,
+      "loss": 1.5411,
+      "step": 2930
+    },
+    {
+      "epoch": 0.6846762925011645,
+      "grad_norm": 12.141153335571289,
+      "learning_rate": 2.5892382948986726e-05,
+      "loss": 2.0825,
+      "step": 2940
+    },
+    {
+      "epoch": 0.6870051234280391,
+      "grad_norm": 11.131927490234375,
+      "learning_rate": 2.587840670859539e-05,
+      "loss": 1.2421,
+      "step": 2950
+    },
+    {
+      "epoch": 0.6893339543549138,
+      "grad_norm": 8.020845413208008,
+      "learning_rate": 2.5864430468204053e-05,
+      "loss": 1.2091,
+      "step": 2960
+    },
+    {
+      "epoch": 0.6916627852817886,
+      "grad_norm": 26.62967300415039,
+      "learning_rate": 2.5850454227812716e-05,
+      "loss": 1.3636,
+      "step": 2970
+    },
+    {
+      "epoch": 0.6939916162086632,
+      "grad_norm": 15.849566459655762,
+      "learning_rate": 2.5836477987421383e-05,
+      "loss": 1.1627,
+      "step": 2980
+    },
+    {
+      "epoch": 0.696320447135538,
+      "grad_norm": 12.134406089782715,
+      "learning_rate": 2.5822501747030047e-05,
+      "loss": 1.4755,
+      "step": 2990
+    },
+    {
+      "epoch": 0.6986492780624126,
+      "grad_norm": 29.160348892211914,
+      "learning_rate": 2.5808525506638714e-05,
+      "loss": 1.3963,
+      "step": 3000
+    },
+    {
+      "epoch": 0.7009781089892874,
+      "grad_norm": 5.01591157913208,
+      "learning_rate": 2.5794549266247378e-05,
+      "loss": 1.144,
+      "step": 3010
+    },
+    {
+      "epoch": 0.7033069399161621,
+      "grad_norm": 14.045069694519043,
+      "learning_rate": 2.5780573025856045e-05,
+      "loss": 1.002,
+      "step": 3020
+    },
+    {
+      "epoch": 0.7056357708430367,
+      "grad_norm": 23.710153579711914,
+      "learning_rate": 2.576659678546471e-05,
+      "loss": 1.3336,
+      "step": 3030
+    },
+    {
+      "epoch": 0.7079646017699115,
+      "grad_norm": 25.294391632080078,
+      "learning_rate": 2.5752620545073375e-05,
+      "loss": 1.3291,
+      "step": 3040
+    },
+    {
+      "epoch": 0.7102934326967862,
+      "grad_norm": 15.712854385375977,
+      "learning_rate": 2.5738644304682042e-05,
+      "loss": 1.5373,
+      "step": 3050
+    },
+    {
+      "epoch": 0.7126222636236609,
+      "grad_norm": 19.031715393066406,
+      "learning_rate": 2.5724668064290706e-05,
+      "loss": 1.4451,
+      "step": 3060
+    },
+    {
+      "epoch": 0.7149510945505356,
+      "grad_norm": 15.004422187805176,
+      "learning_rate": 2.5710691823899373e-05,
+      "loss": 1.6015,
+      "step": 3070
+    },
+    {
+      "epoch": 0.7172799254774104,
+      "grad_norm": 15.895539283752441,
+      "learning_rate": 2.5696715583508036e-05,
+      "loss": 1.3032,
+      "step": 3080
+    },
+    {
+      "epoch": 0.719608756404285,
+      "grad_norm": 30.933595657348633,
+      "learning_rate": 2.5682739343116703e-05,
+      "loss": 1.2197,
+      "step": 3090
+    },
+    {
+      "epoch": 0.7219375873311598,
+      "grad_norm": 56.714134216308594,
+      "learning_rate": 2.5668763102725367e-05,
+      "loss": 1.5058,
+      "step": 3100
+    },
+    {
+      "epoch": 0.7242664182580345,
+      "grad_norm": 7.623404502868652,
+      "learning_rate": 2.5654786862334034e-05,
+      "loss": 1.1573,
+      "step": 3110
+    },
+    {
+      "epoch": 0.7265952491849091,
+      "grad_norm": 18.51123809814453,
+      "learning_rate": 2.5640810621942697e-05,
+      "loss": 1.1029,
+      "step": 3120
+    },
+    {
+      "epoch": 0.7289240801117839,
+      "grad_norm": 48.92169189453125,
+      "learning_rate": 2.5626834381551364e-05,
+      "loss": 1.245,
+      "step": 3130
+    },
+    {
+      "epoch": 0.7312529110386586,
+      "grad_norm": 11.136435508728027,
+      "learning_rate": 2.5612858141160028e-05,
+      "loss": 1.1932,
+      "step": 3140
+    },
+    {
+      "epoch": 0.7335817419655333,
+      "grad_norm": 44.02711868286133,
+      "learning_rate": 2.5598881900768695e-05,
+      "loss": 1.6498,
+      "step": 3150
+    },
+    {
+      "epoch": 0.735910572892408,
+      "grad_norm": 15.315217018127441,
+      "learning_rate": 2.558490566037736e-05,
+      "loss": 1.5206,
+      "step": 3160
+    },
+    {
+      "epoch": 0.7382394038192828,
+      "grad_norm": 13.75929069519043,
+      "learning_rate": 2.5570929419986025e-05,
+      "loss": 1.4698,
+      "step": 3170
+    },
+    {
+      "epoch": 0.7405682347461574,
+      "grad_norm": 14.253317832946777,
+      "learning_rate": 2.5556953179594692e-05,
+      "loss": 1.2942,
+      "step": 3180
+    },
+    {
+      "epoch": 0.7428970656730322,
+      "grad_norm": 54.78810501098633,
+      "learning_rate": 2.5542976939203356e-05,
+      "loss": 1.0529,
+      "step": 3190
+    },
+    {
+      "epoch": 0.7452258965999069,
+      "grad_norm": 3.501985788345337,
+      "learning_rate": 2.5529000698812023e-05,
+      "loss": 1.5216,
+      "step": 3200
+    },
+    {
+      "epoch": 0.7475547275267815,
+      "grad_norm": 10.592144966125488,
+      "learning_rate": 2.5515024458420687e-05,
+      "loss": 0.8294,
+      "step": 3210
+    },
+    {
+      "epoch": 0.7498835584536563,
+      "grad_norm": 26.526243209838867,
+      "learning_rate": 2.550104821802935e-05,
+      "loss": 1.5043,
+      "step": 3220
+    },
+    {
+      "epoch": 0.7522123893805309,
+      "grad_norm": 18.082988739013672,
+      "learning_rate": 2.5487071977638014e-05,
+      "loss": 0.922,
+      "step": 3230
+    },
+    {
+      "epoch": 0.7545412203074057,
+      "grad_norm": 9.192100524902344,
+      "learning_rate": 2.547309573724668e-05,
+      "loss": 1.5649,
+      "step": 3240
+    },
+    {
+      "epoch": 0.7568700512342804,
+      "grad_norm": 16.617460250854492,
+      "learning_rate": 2.5459119496855344e-05,
+      "loss": 1.109,
+      "step": 3250
+    },
+    {
+      "epoch": 0.759198882161155,
+      "grad_norm": 33.06241989135742,
+      "learning_rate": 2.544514325646401e-05,
+      "loss": 1.6267,
+      "step": 3260
+    },
+    {
+      "epoch": 0.7615277130880298,
+      "grad_norm": 9.304107666015625,
+      "learning_rate": 2.5431167016072675e-05,
+      "loss": 1.359,
+      "step": 3270
+    },
+    {
+      "epoch": 0.7638565440149045,
+      "grad_norm": 37.228939056396484,
+      "learning_rate": 2.5417190775681342e-05,
+      "loss": 2.4155,
+      "step": 3280
+    },
+    {
+      "epoch": 0.7661853749417792,
+      "grad_norm": 13.651851654052734,
+      "learning_rate": 2.5403214535290005e-05,
+      "loss": 1.1901,
+      "step": 3290
+    },
+    {
+      "epoch": 0.7685142058686539,
+      "grad_norm": 9.29826831817627,
+      "learning_rate": 2.5389238294898672e-05,
+      "loss": 1.3085,
+      "step": 3300
+    },
+    {
+      "epoch": 0.7708430367955287,
+      "grad_norm": 9.485541343688965,
+      "learning_rate": 2.5375262054507336e-05,
+      "loss": 1.4252,
+      "step": 3310
+    },
+    {
+      "epoch": 0.7731718677224033,
+      "grad_norm": 61.758750915527344,
+      "learning_rate": 2.5361285814116003e-05,
+      "loss": 1.7734,
+      "step": 3320
+    },
+    {
+      "epoch": 0.7755006986492781,
+      "grad_norm": 23.328594207763672,
+      "learning_rate": 2.534730957372467e-05,
+      "loss": 1.4627,
+      "step": 3330
+    },
+    {
+      "epoch": 0.7778295295761528,
+      "grad_norm": 8.944003105163574,
+      "learning_rate": 2.5333333333333334e-05,
+      "loss": 1.4042,
+      "step": 3340
+    },
+    {
+      "epoch": 0.7801583605030274,
+      "grad_norm": 10.560107231140137,
+      "learning_rate": 2.5319357092942e-05,
+      "loss": 1.7253,
+      "step": 3350
+    },
+    {
+      "epoch": 0.7824871914299022,
+      "grad_norm": 9.760632514953613,
+      "learning_rate": 2.5305380852550664e-05,
+      "loss": 1.6111,
+      "step": 3360
+    },
+    {
+      "epoch": 0.7848160223567769,
+      "grad_norm": 14.777600288391113,
+      "learning_rate": 2.529140461215933e-05,
+      "loss": 1.1902,
+      "step": 3370
+    },
+    {
+      "epoch": 0.7871448532836516,
+      "grad_norm": 7.539825916290283,
+      "learning_rate": 2.5277428371767995e-05,
+      "loss": 1.2884,
+      "step": 3380
+    },
+    {
+      "epoch": 0.7894736842105263,
+      "grad_norm": 15.00157642364502,
+      "learning_rate": 2.5263452131376662e-05,
+      "loss": 1.1476,
+      "step": 3390
+    },
+    {
+      "epoch": 0.7918025151374011,
+      "grad_norm": 17.299943923950195,
+      "learning_rate": 2.5249475890985325e-05,
+      "loss": 1.3168,
+      "step": 3400
+    },
+    {
+      "epoch": 0.7941313460642757,
+      "grad_norm": 118.13330841064453,
+      "learning_rate": 2.5235499650593992e-05,
+      "loss": 2.3259,
+      "step": 3410
+    },
+    {
+      "epoch": 0.7964601769911505,
+      "grad_norm": 14.524819374084473,
+      "learning_rate": 2.5221523410202656e-05,
+      "loss": 1.364,
+      "step": 3420
+    },
+    {
+      "epoch": 0.7987890079180252,
+      "grad_norm": 13.984453201293945,
+      "learning_rate": 2.5207547169811323e-05,
+      "loss": 1.0494,
+      "step": 3430
+    },
+    {
+      "epoch": 0.8011178388448998,
+      "grad_norm": 24.410167694091797,
+      "learning_rate": 2.5193570929419986e-05,
+      "loss": 1.1634,
+      "step": 3440
+    },
+    {
+      "epoch": 0.8034466697717746,
+      "grad_norm": 9.531725883483887,
+      "learning_rate": 2.5179594689028653e-05,
+      "loss": 1.4717,
+      "step": 3450
+    },
+    {
+      "epoch": 0.8057755006986492,
+      "grad_norm": 8.417661666870117,
+      "learning_rate": 2.5165618448637317e-05,
+      "loss": 1.006,
+      "step": 3460
+    },
+    {
+      "epoch": 0.808104331625524,
+      "grad_norm": 16.109172821044922,
+      "learning_rate": 2.5151642208245984e-05,
+      "loss": 1.1106,
+      "step": 3470
+    },
+    {
+      "epoch": 0.8104331625523987,
+      "grad_norm": 12.241450309753418,
+      "learning_rate": 2.5137665967854648e-05,
+      "loss": 1.2066,
+      "step": 3480
+    },
+    {
+      "epoch": 0.8127619934792734,
+      "grad_norm": 13.93542194366455,
+      "learning_rate": 2.512368972746331e-05,
+      "loss": 1.6023,
+      "step": 3490
+    },
+    {
+      "epoch": 0.8150908244061481,
+      "grad_norm": 33.94037628173828,
+      "learning_rate": 2.5109713487071978e-05,
+      "loss": 1.5284,
+      "step": 3500
+    },
+    {
+      "epoch": 0.8174196553330229,
+      "grad_norm": 13.690526008605957,
+      "learning_rate": 2.5095737246680642e-05,
+      "loss": 1.6503,
+      "step": 3510
+    },
+    {
+      "epoch": 0.8197484862598975,
+      "grad_norm": 11.894718170166016,
+      "learning_rate": 2.508176100628931e-05,
+      "loss": 1.118,
+      "step": 3520
+    },
+    {
+      "epoch": 0.8220773171867722,
+      "grad_norm": 18.130245208740234,
+      "learning_rate": 2.5067784765897972e-05,
+      "loss": 1.1148,
+      "step": 3530
+    },
+    {
+      "epoch": 0.824406148113647,
+      "grad_norm": 10.06002140045166,
+      "learning_rate": 2.505380852550664e-05,
+      "loss": 1.2269,
+      "step": 3540
+    },
+    {
+      "epoch": 0.8267349790405216,
+      "grad_norm": 17.191099166870117,
+      "learning_rate": 2.5039832285115303e-05,
+      "loss": 1.3745,
+      "step": 3550
+    },
+    {
+      "epoch": 0.8290638099673964,
+      "grad_norm": 23.614267349243164,
+      "learning_rate": 2.502585604472397e-05,
+      "loss": 1.2424,
+      "step": 3560
+    },
+    {
+      "epoch": 0.8313926408942711,
+      "grad_norm": 6.034205436706543,
+      "learning_rate": 2.5011879804332633e-05,
+      "loss": 1.0718,
+      "step": 3570
+    },
+    {
+      "epoch": 0.8337214718211458,
+      "grad_norm": 17.800487518310547,
+      "learning_rate": 2.49979035639413e-05,
+      "loss": 1.6048,
+      "step": 3580
+    },
+    {
+      "epoch": 0.8360503027480205,
+      "grad_norm": 5.585227012634277,
+      "learning_rate": 2.4983927323549964e-05,
+      "loss": 1.153,
+      "step": 3590
+    },
+    {
+      "epoch": 0.8383791336748952,
+      "grad_norm": 37.87883377075195,
+      "learning_rate": 2.496995108315863e-05,
+      "loss": 1.4075,
+      "step": 3600
+    },
+    {
+      "epoch": 0.8407079646017699,
+      "grad_norm": 10.529804229736328,
+      "learning_rate": 2.4955974842767295e-05,
+      "loss": 1.215,
+      "step": 3610
+    },
+    {
+      "epoch": 0.8430367955286446,
+      "grad_norm": 16.147066116333008,
+      "learning_rate": 2.494199860237596e-05,
+      "loss": 1.3325,
+      "step": 3620
+    },
+    {
+      "epoch": 0.8453656264555194,
+      "grad_norm": 8.722766876220703,
+      "learning_rate": 2.492802236198463e-05,
+      "loss": 1.0484,
+      "step": 3630
+    },
+    {
+      "epoch": 0.847694457382394,
+      "grad_norm": 2.9982659816741943,
+      "learning_rate": 2.4914046121593292e-05,
+      "loss": 1.0521,
+      "step": 3640
+    },
+    {
+      "epoch": 0.8500232883092688,
+      "grad_norm": 64.77198791503906,
+      "learning_rate": 2.490006988120196e-05,
+      "loss": 1.4048,
+      "step": 3650
+    },
+    {
+      "epoch": 0.8523521192361434,
+      "grad_norm": 10.345009803771973,
+      "learning_rate": 2.4886093640810623e-05,
+      "loss": 1.3138,
+      "step": 3660
+    },
+    {
+      "epoch": 0.8546809501630181,
+      "grad_norm": 43.743690490722656,
+      "learning_rate": 2.487211740041929e-05,
+      "loss": 1.3493,
+      "step": 3670
+    },
+    {
+      "epoch": 0.8570097810898929,
+      "grad_norm": 19.04966163635254,
+      "learning_rate": 2.4858141160027953e-05,
+      "loss": 1.2579,
+      "step": 3680
+    },
+    {
+      "epoch": 0.8593386120167675,
+      "grad_norm": 19.782512664794922,
+      "learning_rate": 2.484416491963662e-05,
+      "loss": 1.6133,
+      "step": 3690
+    },
+    {
+      "epoch": 0.8616674429436423,
+      "grad_norm": 48.4581298828125,
+      "learning_rate": 2.4830188679245284e-05,
+      "loss": 1.6982,
+      "step": 3700
+    },
+    {
+      "epoch": 0.863996273870517,
+      "grad_norm": 7.717815399169922,
+      "learning_rate": 2.481621243885395e-05,
+      "loss": 1.4523,
+      "step": 3710
+    },
+    {
+      "epoch": 0.8663251047973917,
+      "grad_norm": 22.847097396850586,
+      "learning_rate": 2.4802236198462614e-05,
+      "loss": 1.4004,
+      "step": 3720
+    },
+    {
+      "epoch": 0.8686539357242664,
+      "grad_norm": 118.66568756103516,
+      "learning_rate": 2.478825995807128e-05,
+      "loss": 1.4469,
+      "step": 3730
+    },
+    {
+      "epoch": 0.8709827666511412,
+      "grad_norm": 17.42070198059082,
+      "learning_rate": 2.4774283717679945e-05,
+      "loss": 1.8226,
+      "step": 3740
+    },
+    {
+      "epoch": 0.8733115975780158,
+      "grad_norm": 20.360017776489258,
+      "learning_rate": 2.476030747728861e-05,
+      "loss": 1.8337,
+      "step": 3750
+    },
+    {
+      "epoch": 0.8756404285048905,
+      "grad_norm": 20.167600631713867,
+      "learning_rate": 2.4746331236897275e-05,
+      "loss": 1.4437,
+      "step": 3760
+    },
+    {
+      "epoch": 0.8779692594317653,
+      "grad_norm": 9.806801795959473,
+      "learning_rate": 2.473235499650594e-05,
+      "loss": 1.0921,
+      "step": 3770
+    },
+    {
+      "epoch": 0.8802980903586399,
+      "grad_norm": 19.070938110351562,
+      "learning_rate": 2.4718378756114606e-05,
+      "loss": 1.0835,
+      "step": 3780
+    },
+    {
+      "epoch": 0.8826269212855147,
+      "grad_norm": 52.30155944824219,
+      "learning_rate": 2.470440251572327e-05,
+      "loss": 1.1373,
+      "step": 3790
+    },
+    {
+      "epoch": 0.8849557522123894,
+      "grad_norm": 31.602767944335938,
+      "learning_rate": 2.4690426275331937e-05,
+      "loss": 1.4057,
+      "step": 3800
+    },
+    {
+      "epoch": 0.8872845831392641,
+      "grad_norm": 6.680765628814697,
+      "learning_rate": 2.46764500349406e-05,
+      "loss": 1.8083,
+      "step": 3810
+    },
+    {
+      "epoch": 0.8896134140661388,
+      "grad_norm": 14.354714393615723,
+      "learning_rate": 2.4662473794549267e-05,
+      "loss": 1.1566,
+      "step": 3820
+    },
+    {
+      "epoch": 0.8919422449930136,
+      "grad_norm": 5.738077640533447,
+      "learning_rate": 2.464849755415793e-05,
+      "loss": 1.0986,
+      "step": 3830
+    },
+    {
+      "epoch": 0.8942710759198882,
+      "grad_norm": 12.508646011352539,
+      "learning_rate": 2.4634521313766598e-05,
+      "loss": 0.9977,
+      "step": 3840
+    },
+    {
+      "epoch": 0.8965999068467629,
+      "grad_norm": 12.393445014953613,
+      "learning_rate": 2.462054507337526e-05,
+      "loss": 1.6198,
+      "step": 3850
+    },
+    {
+      "epoch": 0.8989287377736377,
+      "grad_norm": 12.13697624206543,
+      "learning_rate": 2.460656883298393e-05,
+      "loss": 1.8646,
+      "step": 3860
+    },
+    {
+      "epoch": 0.9012575687005123,
+      "grad_norm": 29.515411376953125,
+      "learning_rate": 2.4592592592592592e-05,
+      "loss": 1.4014,
+      "step": 3870
+    },
+    {
+      "epoch": 0.9035863996273871,
+      "grad_norm": 9.969564437866211,
+      "learning_rate": 2.457861635220126e-05,
+      "loss": 0.9157,
+      "step": 3880
+    },
+    {
+      "epoch": 0.9059152305542617,
+      "grad_norm": 70.15406799316406,
+      "learning_rate": 2.4564640111809922e-05,
+      "loss": 1.3138,
+      "step": 3890
+    },
+    {
+      "epoch": 0.9082440614811365,
+      "grad_norm": 16.15462875366211,
+      "learning_rate": 2.455066387141859e-05,
+      "loss": 1.439,
+      "step": 3900
+    },
+    {
+      "epoch": 0.9105728924080112,
+      "grad_norm": 23.706735610961914,
+      "learning_rate": 2.4536687631027256e-05,
+      "loss": 1.3809,
+      "step": 3910
+    },
+    {
+      "epoch": 0.9129017233348858,
+      "grad_norm": 11.798118591308594,
+      "learning_rate": 2.452271139063592e-05,
+      "loss": 1.2233,
+      "step": 3920
+    },
+    {
+      "epoch": 0.9152305542617606,
+      "grad_norm": 6.833698749542236,
+      "learning_rate": 2.4508735150244587e-05,
+      "loss": 1.09,
+      "step": 3930
+    },
+    {
+      "epoch": 0.9175593851886353,
+      "grad_norm": 13.652518272399902,
+      "learning_rate": 2.449475890985325e-05,
+      "loss": 1.2642,
+      "step": 3940
+    },
+    {
+      "epoch": 0.91988821611551,
+      "grad_norm": 10.70895004272461,
+      "learning_rate": 2.4480782669461918e-05,
+      "loss": 1.7104,
+      "step": 3950
+    },
+    {
+      "epoch": 0.9222170470423847,
+      "grad_norm": 4.455838203430176,
+      "learning_rate": 2.446680642907058e-05,
+      "loss": 1.2201,
+      "step": 3960
+    },
+    {
+      "epoch": 0.9245458779692595,
+      "grad_norm": 20.450454711914062,
+      "learning_rate": 2.4452830188679248e-05,
+      "loss": 0.9957,
+      "step": 3970
+    },
+    {
+      "epoch": 0.9268747088961341,
+      "grad_norm": 40.74632263183594,
+      "learning_rate": 2.4438853948287912e-05,
+      "loss": 1.3495,
+      "step": 3980
+    },
+    {
+      "epoch": 0.9292035398230089,
+      "grad_norm": 11.148481369018555,
+      "learning_rate": 2.442487770789658e-05,
+      "loss": 1.1178,
+      "step": 3990
+    },
+    {
+      "epoch": 0.9315323707498836,
+      "grad_norm": 7.56870174407959,
+      "learning_rate": 2.4410901467505242e-05,
+      "loss": 0.9451,
+      "step": 4000
+    },
+    {
+      "epoch": 0.9338612016767582,
+      "grad_norm": 38.12516403198242,
+      "learning_rate": 2.4396925227113906e-05,
+      "loss": 1.1355,
+      "step": 4010
+    },
+    {
+      "epoch": 0.936190032603633,
+      "grad_norm": 49.5767707824707,
+      "learning_rate": 2.438294898672257e-05,
+      "loss": 1.7586,
+      "step": 4020
+    },
+    {
+      "epoch": 0.9385188635305077,
+      "grad_norm": 13.015122413635254,
+      "learning_rate": 2.4368972746331236e-05,
+      "loss": 1.873,
+      "step": 4030
+    },
+    {
+      "epoch": 0.9408476944573824,
+      "grad_norm": 26.832674026489258,
+      "learning_rate": 2.43549965059399e-05,
+      "loss": 1.553,
+      "step": 4040
+    },
+    {
+      "epoch": 0.9431765253842571,
+      "grad_norm": 13.10566520690918,
+      "learning_rate": 2.4341020265548567e-05,
+      "loss": 1.5644,
+      "step": 4050
+    },
+    {
+      "epoch": 0.9455053563111319,
+      "grad_norm": 12.45157527923584,
+      "learning_rate": 2.4327044025157234e-05,
+      "loss": 1.6215,
+      "step": 4060
+    },
+    {
+      "epoch": 0.9478341872380065,
+      "grad_norm": 28.12361717224121,
+      "learning_rate": 2.4313067784765898e-05,
+      "loss": 1.126,
+      "step": 4070
+    },
+    {
+      "epoch": 0.9501630181648812,
+      "grad_norm": 11.773497581481934,
+      "learning_rate": 2.4299091544374565e-05,
+      "loss": 1.106,
+      "step": 4080
+    },
+    {
+      "epoch": 0.952491849091756,
+      "grad_norm": 15.833434104919434,
+      "learning_rate": 2.4285115303983228e-05,
+      "loss": 1.3659,
+      "step": 4090
+    },
+    {
+      "epoch": 0.9548206800186306,
+      "grad_norm": 7.810037612915039,
+      "learning_rate": 2.4271139063591895e-05,
+      "loss": 1.4978,
+      "step": 4100
+    },
+    {
+      "epoch": 0.9571495109455054,
+      "grad_norm": 14.155339241027832,
+      "learning_rate": 2.425716282320056e-05,
+      "loss": 1.0855,
+      "step": 4110
+    },
+    {
+      "epoch": 0.95947834187238,
+      "grad_norm": 7.536425590515137,
+      "learning_rate": 2.4243186582809226e-05,
+      "loss": 1.6742,
+      "step": 4120
+    },
+    {
+      "epoch": 0.9618071727992548,
+      "grad_norm": 41.47824478149414,
+      "learning_rate": 2.422921034241789e-05,
+      "loss": 1.7182,
+      "step": 4130
+    },
+    {
+      "epoch": 0.9641360037261295,
+      "grad_norm": 14.603498458862305,
+      "learning_rate": 2.4215234102026556e-05,
+      "loss": 0.9493,
+      "step": 4140
+    },
+    {
+      "epoch": 0.9664648346530041,
+      "grad_norm": 9.240854263305664,
+      "learning_rate": 2.420125786163522e-05,
+      "loss": 0.8127,
+      "step": 4150
+    },
+    {
+      "epoch": 0.9687936655798789,
+      "grad_norm": 11.538430213928223,
+      "learning_rate": 2.4187281621243887e-05,
+      "loss": 1.1822,
+      "step": 4160
+    },
+    {
+      "epoch": 0.9711224965067536,
+      "grad_norm": 10.658085823059082,
+      "learning_rate": 2.417330538085255e-05,
+      "loss": 1.4719,
+      "step": 4170
+    },
+    {
+      "epoch": 0.9734513274336283,
+      "grad_norm": 20.83334732055664,
+      "learning_rate": 2.4159329140461217e-05,
+      "loss": 1.1781,
+      "step": 4180
+    },
+    {
+      "epoch": 0.975780158360503,
+      "grad_norm": 8.09494400024414,
+      "learning_rate": 2.414535290006988e-05,
+      "loss": 1.3265,
+      "step": 4190
+    },
+    {
+      "epoch": 0.9781089892873778,
+      "grad_norm": 8.501059532165527,
+      "learning_rate": 2.4131376659678548e-05,
+      "loss": 0.9679,
+      "step": 4200
+    },
+    {
+      "epoch": 0.9804378202142524,
+      "grad_norm": 21.922080993652344,
+      "learning_rate": 2.4117400419287215e-05,
+      "loss": 1.3786,
+      "step": 4210
+    },
+    {
+      "epoch": 0.9827666511411272,
+      "grad_norm": 8.116072654724121,
+      "learning_rate": 2.410342417889588e-05,
+      "loss": 1.0864,
+      "step": 4220
+    },
+    {
+      "epoch": 0.9850954820680019,
+      "grad_norm": 39.07789611816406,
+      "learning_rate": 2.4089447938504546e-05,
+      "loss": 1.0388,
+      "step": 4230
+    },
+    {
+      "epoch": 0.9874243129948765,
+      "grad_norm": 18.8806095123291,
+      "learning_rate": 2.407547169811321e-05,
+      "loss": 1.2614,
+      "step": 4240
+    },
+    {
+      "epoch": 0.9897531439217513,
+      "grad_norm": 30.235370635986328,
+      "learning_rate": 2.4061495457721876e-05,
+      "loss": 1.0553,
+      "step": 4250
+    },
+    {
+      "epoch": 0.992081974848626,
+      "grad_norm": 11.719708442687988,
+      "learning_rate": 2.404751921733054e-05,
+      "loss": 1.7491,
+      "step": 4260
+    },
+    {
+      "epoch": 0.9944108057755007,
+      "grad_norm": 4.49058723449707,
+      "learning_rate": 2.4033542976939203e-05,
+      "loss": 1.6797,
+      "step": 4270
+    },
+    {
+      "epoch": 0.9967396367023754,
+      "grad_norm": 9.818137168884277,
+      "learning_rate": 2.4019566736547867e-05,
+      "loss": 1.5911,
+      "step": 4280
+    },
+    {
+      "epoch": 0.9990684676292502,
+      "grad_norm": 7.682907581329346,
+      "learning_rate": 2.4005590496156534e-05,
+      "loss": 1.3785,
+      "step": 4290
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.5099594593048096,
+      "eval_runtime": 57.748,
+      "eval_samples_per_second": 75.431,
+      "eval_steps_per_second": 1.195,
+      "step": 4294
+    },
+    {
+      "epoch": 1.0013972985561248,
+      "grad_norm": 16.090312957763672,
+      "learning_rate": 2.3991614255765197e-05,
+      "loss": 1.3029,
+      "step": 4300
+    },
+    {
+      "epoch": 1.0037261294829996,
+      "grad_norm": 10.43845272064209,
+      "learning_rate": 2.3977638015373864e-05,
+      "loss": 1.3984,
+      "step": 4310
+    },
+    {
+      "epoch": 1.0060549604098743,
+      "grad_norm": 7.594318866729736,
+      "learning_rate": 2.3963661774982528e-05,
+      "loss": 1.2071,
+      "step": 4320
+    },
+    {
+      "epoch": 1.008383791336749,
+      "grad_norm": 5.082920551300049,
+      "learning_rate": 2.3949685534591195e-05,
+      "loss": 1.1765,
+      "step": 4330
+    },
+    {
+      "epoch": 1.0107126222636236,
+      "grad_norm": 34.897769927978516,
+      "learning_rate": 2.393570929419986e-05,
+      "loss": 1.8144,
+      "step": 4340
+    },
+    {
+      "epoch": 1.0130414531904983,
+      "grad_norm": 20.48177719116211,
+      "learning_rate": 2.3921733053808526e-05,
+      "loss": 1.0724,
+      "step": 4350
+    },
+    {
+      "epoch": 1.015370284117373,
+      "grad_norm": 18.330974578857422,
+      "learning_rate": 2.3907756813417192e-05,
+      "loss": 1.4672,
+      "step": 4360
+    },
+    {
+      "epoch": 1.0176991150442478,
+      "grad_norm": 28.877042770385742,
+      "learning_rate": 2.3893780573025856e-05,
+      "loss": 1.4611,
+      "step": 4370
+    },
+    {
+      "epoch": 1.0200279459711226,
+      "grad_norm": 13.684222221374512,
+      "learning_rate": 2.3879804332634523e-05,
+      "loss": 0.9294,
+      "step": 4380
+    },
+    {
+      "epoch": 1.0223567768979973,
+      "grad_norm": 4.9023590087890625,
+      "learning_rate": 2.3865828092243187e-05,
+      "loss": 1.1675,
+      "step": 4390
+    },
+    {
+      "epoch": 1.0246856078248718,
+      "grad_norm": 9.912734031677246,
+      "learning_rate": 2.3851851851851854e-05,
+      "loss": 1.0157,
+      "step": 4400
+    },
+    {
+      "epoch": 1.0270144387517466,
+      "grad_norm": 13.99254035949707,
+      "learning_rate": 2.3837875611460517e-05,
+      "loss": 1.295,
+      "step": 4410
+    },
+    {
+      "epoch": 1.0293432696786213,
+      "grad_norm": 14.370241165161133,
+      "learning_rate": 2.3823899371069184e-05,
+      "loss": 1.6396,
+      "step": 4420
+    },
+    {
+      "epoch": 1.031672100605496,
+      "grad_norm": 20.884004592895508,
+      "learning_rate": 2.3809923130677848e-05,
+      "loss": 1.3973,
+      "step": 4430
+    },
+    {
+      "epoch": 1.0340009315323708,
+      "grad_norm": 12.322709083557129,
+      "learning_rate": 2.3795946890286515e-05,
+      "loss": 0.9916,
+      "step": 4440
+    },
+    {
+      "epoch": 1.0363297624592454,
+      "grad_norm": 54.580814361572266,
+      "learning_rate": 2.378197064989518e-05,
+      "loss": 1.2983,
+      "step": 4450
+    },
+    {
+      "epoch": 1.03865859338612,
+      "grad_norm": 45.35833740234375,
+      "learning_rate": 2.3767994409503845e-05,
+      "loss": 1.6227,
+      "step": 4460
+    },
+    {
+      "epoch": 1.0409874243129948,
+      "grad_norm": 58.20479965209961,
+      "learning_rate": 2.375401816911251e-05,
+      "loss": 1.4488,
+      "step": 4470
+    },
+    {
+      "epoch": 1.0433162552398696,
+      "grad_norm": 21.809900283813477,
+      "learning_rate": 2.3740041928721176e-05,
+      "loss": 1.5132,
+      "step": 4480
+    },
+    {
+      "epoch": 1.0456450861667443,
+      "grad_norm": 6.9720540046691895,
+      "learning_rate": 2.372606568832984e-05,
+      "loss": 0.904,
+      "step": 4490
+    },
+    {
+      "epoch": 1.047973917093619,
+      "grad_norm": 7.57560920715332,
+      "learning_rate": 2.3712089447938506e-05,
+      "loss": 1.0406,
+      "step": 4500
+    },
+    {
+      "epoch": 1.0503027480204936,
+      "grad_norm": 43.4541015625,
+      "learning_rate": 2.3698113207547173e-05,
+      "loss": 1.2533,
+      "step": 4510
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 17.777565002441406,
+      "learning_rate": 2.3684136967155837e-05,
+      "loss": 1.3282,
+      "step": 4520
+    },
+    {
+      "epoch": 1.054960409874243,
+      "grad_norm": 15.077924728393555,
+      "learning_rate": 2.3670160726764504e-05,
+      "loss": 1.2443,
+      "step": 4530
+    },
+    {
+      "epoch": 1.0572892408011179,
+      "grad_norm": 9.286961555480957,
+      "learning_rate": 2.3656184486373164e-05,
+      "loss": 1.0529,
+      "step": 4540
+    },
+    {
+      "epoch": 1.0596180717279926,
+      "grad_norm": 42.42253494262695,
+      "learning_rate": 2.364220824598183e-05,
+      "loss": 2.2219,
+      "step": 4550
+    },
+    {
+      "epoch": 1.0619469026548674,
+      "grad_norm": 11.062445640563965,
+      "learning_rate": 2.3628232005590495e-05,
+      "loss": 1.5294,
+      "step": 4560
+    },
+    {
+      "epoch": 1.0642757335817419,
+      "grad_norm": 15.275858879089355,
+      "learning_rate": 2.3614255765199162e-05,
+      "loss": 1.0046,
+      "step": 4570
+    },
+    {
+      "epoch": 1.0666045645086166,
+      "grad_norm": 14.448469161987305,
+      "learning_rate": 2.3600279524807825e-05,
+      "loss": 1.5815,
+      "step": 4580
+    },
+    {
+      "epoch": 1.0689333954354914,
+      "grad_norm": 4.570057392120361,
+      "learning_rate": 2.3586303284416492e-05,
+      "loss": 0.858,
+      "step": 4590
+    },
+    {
+      "epoch": 1.0712622263623661,
+      "grad_norm": 12.531164169311523,
+      "learning_rate": 2.3572327044025156e-05,
+      "loss": 0.9637,
+      "step": 4600
+    },
+    {
+      "epoch": 1.0735910572892409,
+      "grad_norm": 14.942547798156738,
+      "learning_rate": 2.3558350803633823e-05,
+      "loss": 1.1421,
+      "step": 4610
+    },
+    {
+      "epoch": 1.0759198882161156,
+      "grad_norm": 16.772188186645508,
+      "learning_rate": 2.3544374563242486e-05,
+      "loss": 1.3581,
+      "step": 4620
+    },
+    {
+      "epoch": 1.0782487191429901,
+      "grad_norm": 17.264406204223633,
+      "learning_rate": 2.3530398322851153e-05,
+      "loss": 1.8468,
+      "step": 4630
+    },
+    {
+      "epoch": 1.080577550069865,
+      "grad_norm": 50.972537994384766,
+      "learning_rate": 2.3516422082459817e-05,
+      "loss": 1.6117,
+      "step": 4640
+    },
+    {
+      "epoch": 1.0829063809967396,
+      "grad_norm": 8.686941146850586,
+      "learning_rate": 2.3502445842068484e-05,
+      "loss": 1.2004,
+      "step": 4650
+    },
+    {
+      "epoch": 1.0852352119236144,
+      "grad_norm": 5.48530387878418,
+      "learning_rate": 2.348846960167715e-05,
+      "loss": 1.0691,
+      "step": 4660
+    },
+    {
+      "epoch": 1.0875640428504891,
+      "grad_norm": 7.076744556427002,
+      "learning_rate": 2.3474493361285815e-05,
+      "loss": 1.0191,
+      "step": 4670
+    },
+    {
+      "epoch": 1.0898928737773637,
+      "grad_norm": 6.224733352661133,
+      "learning_rate": 2.346051712089448e-05,
+      "loss": 1.4263,
+      "step": 4680
+    },
+    {
+      "epoch": 1.0922217047042384,
+      "grad_norm": 7.603545665740967,
+      "learning_rate": 2.3446540880503145e-05,
+      "loss": 0.9943,
+      "step": 4690
+    },
+    {
+      "epoch": 1.0945505356311132,
+      "grad_norm": 7.480013370513916,
+      "learning_rate": 2.3432564640111812e-05,
+      "loss": 0.624,
+      "step": 4700
+    },
+    {
+      "epoch": 1.096879366557988,
+      "grad_norm": 21.913307189941406,
+      "learning_rate": 2.3418588399720476e-05,
+      "loss": 1.1937,
+      "step": 4710
+    },
+    {
+      "epoch": 1.0992081974848626,
+      "grad_norm": 8.472567558288574,
+      "learning_rate": 2.3404612159329143e-05,
+      "loss": 1.4631,
+      "step": 4720
+    },
+    {
+      "epoch": 1.1015370284117374,
+      "grad_norm": 22.3406925201416,
+      "learning_rate": 2.3390635918937806e-05,
+      "loss": 1.332,
+      "step": 4730
+    },
+    {
+      "epoch": 1.103865859338612,
+      "grad_norm": 8.574953079223633,
+      "learning_rate": 2.3376659678546473e-05,
+      "loss": 0.936,
+      "step": 4740
+    },
+    {
+      "epoch": 1.1061946902654867,
+      "grad_norm": 7.095656394958496,
+      "learning_rate": 2.3362683438155137e-05,
+      "loss": 1.1048,
+      "step": 4750
+    },
+    {
+      "epoch": 1.1085235211923614,
+      "grad_norm": 16.59130859375,
+      "learning_rate": 2.3348707197763804e-05,
+      "loss": 1.1469,
+      "step": 4760
+    },
+    {
+      "epoch": 1.1108523521192362,
+      "grad_norm": 176.99810791015625,
+      "learning_rate": 2.3334730957372467e-05,
+      "loss": 1.3763,
+      "step": 4770
+    },
+    {
+      "epoch": 1.113181183046111,
+      "grad_norm": 9.466316223144531,
+      "learning_rate": 2.3320754716981134e-05,
+      "loss": 1.5032,
+      "step": 4780
+    },
+    {
+      "epoch": 1.1155100139729857,
+      "grad_norm": 10.125219345092773,
+      "learning_rate": 2.3306778476589798e-05,
+      "loss": 1.0508,
+      "step": 4790
+    },
+    {
+      "epoch": 1.1178388448998602,
+      "grad_norm": 8.458738327026367,
+      "learning_rate": 2.329280223619846e-05,
+      "loss": 1.1744,
+      "step": 4800
+    },
+    {
+      "epoch": 1.120167675826735,
+      "grad_norm": 95.53096771240234,
+      "learning_rate": 2.327882599580713e-05,
+      "loss": 1.8102,
+      "step": 4810
+    },
+    {
+      "epoch": 1.1224965067536097,
+      "grad_norm": 7.951642036437988,
+      "learning_rate": 2.3264849755415792e-05,
+      "loss": 1.2638,
+      "step": 4820
+    },
+    {
+      "epoch": 1.1248253376804844,
+      "grad_norm": 4.9284586906433105,
+      "learning_rate": 2.325087351502446e-05,
+      "loss": 1.5309,
+      "step": 4830
+    },
+    {
+      "epoch": 1.1271541686073592,
+      "grad_norm": 23.07508659362793,
+      "learning_rate": 2.3236897274633123e-05,
+      "loss": 1.1686,
+      "step": 4840
+    },
+    {
+      "epoch": 1.129482999534234,
+      "grad_norm": 11.482251167297363,
+      "learning_rate": 2.322292103424179e-05,
+      "loss": 0.9867,
+      "step": 4850
+    },
+    {
+      "epoch": 1.1318118304611084,
+      "grad_norm": 54.271514892578125,
+      "learning_rate": 2.3208944793850453e-05,
+      "loss": 1.16,
+      "step": 4860
+    },
+    {
+      "epoch": 1.1341406613879832,
+      "grad_norm": 18.813350677490234,
+      "learning_rate": 2.319496855345912e-05,
+      "loss": 1.2557,
+      "step": 4870
+    },
+    {
+      "epoch": 1.136469492314858,
+      "grad_norm": 11.297906875610352,
+      "learning_rate": 2.3180992313067784e-05,
+      "loss": 1.4021,
+      "step": 4880
+    },
+    {
+      "epoch": 1.1387983232417327,
+      "grad_norm": 9.278247833251953,
+      "learning_rate": 2.316701607267645e-05,
+      "loss": 1.2495,
+      "step": 4890
+    },
+    {
+      "epoch": 1.1411271541686074,
+      "grad_norm": 13.483011245727539,
+      "learning_rate": 2.3153039832285114e-05,
+      "loss": 1.701,
+      "step": 4900
+    },
+    {
+      "epoch": 1.143455985095482,
+      "grad_norm": 12.479578971862793,
+      "learning_rate": 2.313906359189378e-05,
+      "loss": 1.1971,
+      "step": 4910
+    },
+    {
+      "epoch": 1.1457848160223567,
+      "grad_norm": 64.85933685302734,
+      "learning_rate": 2.3125087351502445e-05,
+      "loss": 1.8314,
+      "step": 4920
+    },
+    {
+      "epoch": 1.1481136469492315,
+      "grad_norm": 109.56636810302734,
+      "learning_rate": 2.3111111111111112e-05,
+      "loss": 1.1742,
+      "step": 4930
+    },
+    {
+      "epoch": 1.1504424778761062,
+      "grad_norm": 23.607959747314453,
+      "learning_rate": 2.3097134870719776e-05,
+      "loss": 0.9286,
+      "step": 4940
+    },
+    {
+      "epoch": 1.152771308802981,
+      "grad_norm": 7.5170207023620605,
+      "learning_rate": 2.3083158630328443e-05,
+      "loss": 1.0913,
+      "step": 4950
+    },
+    {
+      "epoch": 1.1551001397298557,
+      "grad_norm": 13.472895622253418,
+      "learning_rate": 2.306918238993711e-05,
+      "loss": 1.5781,
+      "step": 4960
+    },
+    {
+      "epoch": 1.1574289706567302,
+      "grad_norm": 73.42911529541016,
+      "learning_rate": 2.3055206149545773e-05,
+      "loss": 1.3364,
+      "step": 4970
+    },
+    {
+      "epoch": 1.159757801583605,
+      "grad_norm": 8.195212364196777,
+      "learning_rate": 2.304122990915444e-05,
+      "loss": 1.3403,
+      "step": 4980
+    },
+    {
+      "epoch": 1.1620866325104797,
+      "grad_norm": 41.242862701416016,
+      "learning_rate": 2.3027253668763104e-05,
+      "loss": 2.0296,
+      "step": 4990
+    },
+    {
+      "epoch": 1.1644154634373545,
+      "grad_norm": 15.134632110595703,
+      "learning_rate": 2.301327742837177e-05,
+      "loss": 1.4878,
+      "step": 5000
+    },
+    {
+      "epoch": 1.1667442943642292,
+      "grad_norm": 16.87616729736328,
+      "learning_rate": 2.2999301187980434e-05,
+      "loss": 1.0683,
+      "step": 5010
+    },
+    {
+      "epoch": 1.1690731252911037,
+      "grad_norm": 10.382403373718262,
+      "learning_rate": 2.29853249475891e-05,
+      "loss": 1.3711,
+      "step": 5020
+    },
+    {
+      "epoch": 1.1714019562179785,
+      "grad_norm": 10.93367862701416,
+      "learning_rate": 2.2971348707197765e-05,
+      "loss": 1.4611,
+      "step": 5030
+    },
+    {
+      "epoch": 1.1737307871448532,
+      "grad_norm": 30.08503532409668,
+      "learning_rate": 2.2957372466806432e-05,
+      "loss": 1.4648,
+      "step": 5040
+    },
+    {
+      "epoch": 1.176059618071728,
+      "grad_norm": 7.187095642089844,
+      "learning_rate": 2.2943396226415095e-05,
+      "loss": 1.407,
+      "step": 5050
+    },
+    {
+      "epoch": 1.1783884489986027,
+      "grad_norm": 22.837495803833008,
+      "learning_rate": 2.292941998602376e-05,
+      "loss": 1.284,
+      "step": 5060
+    },
+    {
+      "epoch": 1.1807172799254775,
+      "grad_norm": 34.510986328125,
+      "learning_rate": 2.2915443745632423e-05,
+      "loss": 1.0948,
+      "step": 5070
+    },
+    {
+      "epoch": 1.1830461108523522,
+      "grad_norm": 10.157007217407227,
+      "learning_rate": 2.290146750524109e-05,
+      "loss": 1.0363,
+      "step": 5080
+    },
+    {
+      "epoch": 1.1853749417792268,
+      "grad_norm": 4.127469539642334,
+      "learning_rate": 2.2887491264849753e-05,
+      "loss": 1.1315,
+      "step": 5090
+    },
+    {
+      "epoch": 1.1877037727061015,
+      "grad_norm": 8.76297378540039,
+      "learning_rate": 2.287351502445842e-05,
+      "loss": 1.2935,
+      "step": 5100
+    },
+    {
+      "epoch": 1.1900326036329762,
+      "grad_norm": 58.847904205322266,
+      "learning_rate": 2.2859538784067087e-05,
+      "loss": 2.1187,
+      "step": 5110
+    },
+    {
+      "epoch": 1.192361434559851,
+      "grad_norm": 6.2726030349731445,
+      "learning_rate": 2.284556254367575e-05,
+      "loss": 1.0023,
+      "step": 5120
+    },
+    {
+      "epoch": 1.1946902654867257,
+      "grad_norm": 13.07908821105957,
+      "learning_rate": 2.2831586303284418e-05,
+      "loss": 1.069,
+      "step": 5130
+    },
+    {
+      "epoch": 1.1970190964136003,
+      "grad_norm": 12.537034034729004,
+      "learning_rate": 2.281761006289308e-05,
+      "loss": 1.1431,
+      "step": 5140
+    },
+    {
+      "epoch": 1.199347927340475,
+      "grad_norm": 7.794961452484131,
+      "learning_rate": 2.2803633822501748e-05,
+      "loss": 1.2082,
+      "step": 5150
+    },
+    {
+      "epoch": 1.2016767582673498,
+      "grad_norm": 72.09727478027344,
+      "learning_rate": 2.2789657582110412e-05,
+      "loss": 1.2637,
+      "step": 5160
+    },
+    {
+      "epoch": 1.2040055891942245,
+      "grad_norm": 12.098929405212402,
+      "learning_rate": 2.277568134171908e-05,
+      "loss": 1.671,
+      "step": 5170
+    },
+    {
+      "epoch": 1.2063344201210993,
+      "grad_norm": 13.814184188842773,
+      "learning_rate": 2.2761705101327742e-05,
+      "loss": 1.1143,
+      "step": 5180
+    },
+    {
+      "epoch": 1.208663251047974,
+      "grad_norm": 9.710376739501953,
+      "learning_rate": 2.274772886093641e-05,
+      "loss": 1.1579,
+      "step": 5190
+    },
+    {
+      "epoch": 1.2109920819748485,
+      "grad_norm": 10.720894813537598,
+      "learning_rate": 2.2733752620545073e-05,
+      "loss": 0.9838,
+      "step": 5200
+    },
+    {
+      "epoch": 1.2133209129017233,
+      "grad_norm": 32.7349739074707,
+      "learning_rate": 2.271977638015374e-05,
+      "loss": 2.0061,
+      "step": 5210
+    },
+    {
+      "epoch": 1.215649743828598,
+      "grad_norm": 12.79881763458252,
+      "learning_rate": 2.2705800139762403e-05,
+      "loss": 1.2333,
+      "step": 5220
+    },
+    {
+      "epoch": 1.2179785747554728,
+      "grad_norm": 13.296406745910645,
+      "learning_rate": 2.269182389937107e-05,
+      "loss": 1.3174,
+      "step": 5230
+    },
+    {
+      "epoch": 1.2203074056823475,
+      "grad_norm": 9.031120300292969,
+      "learning_rate": 2.2677847658979734e-05,
+      "loss": 0.9053,
+      "step": 5240
+    },
+    {
+      "epoch": 1.222636236609222,
+      "grad_norm": 7.792122840881348,
+      "learning_rate": 2.26638714185884e-05,
+      "loss": 1.2156,
+      "step": 5250
+    },
+    {
+      "epoch": 1.2249650675360968,
+      "grad_norm": 15.26120662689209,
+      "learning_rate": 2.2649895178197068e-05,
+      "loss": 1.9811,
+      "step": 5260
+    },
+    {
+      "epoch": 1.2272938984629715,
+      "grad_norm": 18.058717727661133,
+      "learning_rate": 2.263591893780573e-05,
+      "loss": 1.1064,
+      "step": 5270
+    },
+    {
+      "epoch": 1.2296227293898463,
+      "grad_norm": 57.864341735839844,
+      "learning_rate": 2.26219426974144e-05,
+      "loss": 1.9861,
+      "step": 5280
+    },
+    {
+      "epoch": 1.231951560316721,
+      "grad_norm": 6.74455451965332,
+      "learning_rate": 2.2607966457023062e-05,
+      "loss": 1.1326,
+      "step": 5290
+    },
+    {
+      "epoch": 1.2342803912435958,
+      "grad_norm": 6.738804817199707,
+      "learning_rate": 2.259399021663173e-05,
+      "loss": 2.296,
+      "step": 5300
+    },
+    {
+      "epoch": 1.2366092221704705,
+      "grad_norm": 10.682368278503418,
+      "learning_rate": 2.2580013976240393e-05,
+      "loss": 1.7348,
+      "step": 5310
+    },
+    {
+      "epoch": 1.238938053097345,
+      "grad_norm": 9.553335189819336,
+      "learning_rate": 2.256603773584906e-05,
+      "loss": 1.1105,
+      "step": 5320
+    },
+    {
+      "epoch": 1.2412668840242198,
+      "grad_norm": 14.826432228088379,
+      "learning_rate": 2.255206149545772e-05,
+      "loss": 1.7976,
+      "step": 5330
+    },
+    {
+      "epoch": 1.2435957149510946,
+      "grad_norm": 13.372576713562012,
+      "learning_rate": 2.2538085255066387e-05,
+      "loss": 1.2633,
+      "step": 5340
+    },
+    {
+      "epoch": 1.2459245458779693,
+      "grad_norm": 14.81163215637207,
+      "learning_rate": 2.252410901467505e-05,
+      "loss": 1.2059,
+      "step": 5350
+    },
+    {
+      "epoch": 1.248253376804844,
+      "grad_norm": 9.57998275756836,
+      "learning_rate": 2.2510132774283717e-05,
+      "loss": 1.2529,
+      "step": 5360
+    },
+    {
+      "epoch": 1.2505822077317186,
+      "grad_norm": 23.91578483581543,
+      "learning_rate": 2.249615653389238e-05,
+      "loss": 1.8046,
+      "step": 5370
+    },
+    {
+      "epoch": 1.2529110386585933,
+      "grad_norm": 8.844199180603027,
+      "learning_rate": 2.2482180293501048e-05,
+      "loss": 1.0261,
+      "step": 5380
+    },
+    {
+      "epoch": 1.255239869585468,
+      "grad_norm": 7.080227851867676,
+      "learning_rate": 2.246820405310971e-05,
+      "loss": 1.0358,
+      "step": 5390
+    },
+    {
+      "epoch": 1.2575687005123428,
+      "grad_norm": 9.825217247009277,
+      "learning_rate": 2.245422781271838e-05,
+      "loss": 1.0371,
+      "step": 5400
+    },
+    {
+      "epoch": 1.2598975314392176,
+      "grad_norm": 51.84904098510742,
+      "learning_rate": 2.2440251572327046e-05,
+      "loss": 1.2432,
+      "step": 5410
+    },
+    {
+      "epoch": 1.2622263623660923,
+      "grad_norm": 6.208728313446045,
+      "learning_rate": 2.242627533193571e-05,
+      "loss": 0.8359,
+      "step": 5420
+    },
+    {
+      "epoch": 1.264555193292967,
+      "grad_norm": 8.808072090148926,
+      "learning_rate": 2.2412299091544376e-05,
+      "loss": 1.3509,
+      "step": 5430
+    },
+    {
+      "epoch": 1.2668840242198416,
+      "grad_norm": 63.0085334777832,
+      "learning_rate": 2.239832285115304e-05,
+      "loss": 1.1609,
+      "step": 5440
+    },
+    {
+      "epoch": 1.2692128551467163,
+      "grad_norm": 13.395509719848633,
+      "learning_rate": 2.2384346610761707e-05,
+      "loss": 1.8954,
+      "step": 5450
+    },
+    {
+      "epoch": 1.271541686073591,
+      "grad_norm": 42.68526840209961,
+      "learning_rate": 2.237037037037037e-05,
+      "loss": 1.2054,
+      "step": 5460
+    },
+    {
+      "epoch": 1.2738705170004658,
+      "grad_norm": 5.643659591674805,
+      "learning_rate": 2.2356394129979037e-05,
+      "loss": 1.3347,
+      "step": 5470
+    },
+    {
+      "epoch": 1.2761993479273404,
+      "grad_norm": 19.961103439331055,
+      "learning_rate": 2.23424178895877e-05,
+      "loss": 1.7996,
+      "step": 5480
+    },
+    {
+      "epoch": 1.278528178854215,
+      "grad_norm": 11.5109224319458,
+      "learning_rate": 2.2328441649196368e-05,
+      "loss": 0.8807,
+      "step": 5490
+    },
+    {
+      "epoch": 1.2808570097810899,
+      "grad_norm": 4.86500883102417,
+      "learning_rate": 2.231446540880503e-05,
+      "loss": 0.9854,
+      "step": 5500
+    },
+    {
+      "epoch": 1.2831858407079646,
+      "grad_norm": 5.852017402648926,
+      "learning_rate": 2.23004891684137e-05,
+      "loss": 1.3599,
+      "step": 5510
+    },
+    {
+      "epoch": 1.2855146716348393,
+      "grad_norm": 31.3754825592041,
+      "learning_rate": 2.2286512928022362e-05,
+      "loss": 1.3582,
+      "step": 5520
+    },
+    {
+      "epoch": 1.287843502561714,
+      "grad_norm": 16.57115364074707,
+      "learning_rate": 2.227253668763103e-05,
+      "loss": 1.4891,
+      "step": 5530
+    },
+    {
+      "epoch": 1.2901723334885888,
+      "grad_norm": 8.07555103302002,
+      "learning_rate": 2.2258560447239693e-05,
+      "loss": 1.1457,
+      "step": 5540
+    },
+    {
+      "epoch": 1.2925011644154634,
+      "grad_norm": 51.47915267944336,
+      "learning_rate": 2.224458420684836e-05,
+      "loss": 1.2228,
+      "step": 5550
+    },
+    {
+      "epoch": 1.2948299953423381,
+      "grad_norm": 61.57811737060547,
+      "learning_rate": 2.2230607966457026e-05,
+      "loss": 1.4929,
+      "step": 5560
+    },
+    {
+      "epoch": 1.2971588262692129,
+      "grad_norm": 61.08300018310547,
+      "learning_rate": 2.221663172606569e-05,
+      "loss": 1.2155,
+      "step": 5570
+    },
+    {
+      "epoch": 1.2994876571960876,
+      "grad_norm": 124.03094482421875,
+      "learning_rate": 2.2202655485674357e-05,
+      "loss": 1.7477,
+      "step": 5580
+    },
+    {
+      "epoch": 1.3018164881229621,
+      "grad_norm": 7.754411220550537,
+      "learning_rate": 2.2188679245283017e-05,
+      "loss": 1.252,
+      "step": 5590
+    },
+    {
+      "epoch": 1.3041453190498369,
+      "grad_norm": 14.382433891296387,
+      "learning_rate": 2.2174703004891684e-05,
+      "loss": 1.3252,
+      "step": 5600
+    },
+    {
+      "epoch": 1.3064741499767116,
+      "grad_norm": 15.235672950744629,
+      "learning_rate": 2.2160726764500348e-05,
+      "loss": 1.0168,
+      "step": 5610
+    },
+    {
+      "epoch": 1.3088029809035864,
+      "grad_norm": 17.89712905883789,
+      "learning_rate": 2.2146750524109015e-05,
+      "loss": 1.451,
+      "step": 5620
+    },
+    {
+      "epoch": 1.3111318118304611,
+      "grad_norm": 38.380924224853516,
+      "learning_rate": 2.213277428371768e-05,
+      "loss": 1.2927,
+      "step": 5630
+    },
+    {
+      "epoch": 1.3134606427573359,
+      "grad_norm": 13.725621223449707,
+      "learning_rate": 2.2118798043326345e-05,
+      "loss": 1.2255,
+      "step": 5640
+    },
+    {
+      "epoch": 1.3157894736842106,
+      "grad_norm": 16.346235275268555,
+      "learning_rate": 2.210482180293501e-05,
+      "loss": 1.1837,
+      "step": 5650
+    },
+    {
+      "epoch": 1.3181183046110854,
+      "grad_norm": 15.437034606933594,
+      "learning_rate": 2.2090845562543676e-05,
+      "loss": 1.4109,
+      "step": 5660
+    },
+    {
+      "epoch": 1.32044713553796,
+      "grad_norm": 13.442436218261719,
+      "learning_rate": 2.207686932215234e-05,
+      "loss": 1.2536,
+      "step": 5670
+    },
+    {
+      "epoch": 1.3227759664648346,
+      "grad_norm": 21.472286224365234,
+      "learning_rate": 2.2062893081761007e-05,
+      "loss": 1.6111,
+      "step": 5680
+    },
+    {
+      "epoch": 1.3251047973917094,
+      "grad_norm": 13.40835189819336,
+      "learning_rate": 2.2048916841369673e-05,
+      "loss": 1.5086,
+      "step": 5690
+    },
+    {
+      "epoch": 1.3274336283185841,
+      "grad_norm": 9.588897705078125,
+      "learning_rate": 2.2034940600978337e-05,
+      "loss": 1.1776,
+      "step": 5700
+    },
+    {
+      "epoch": 1.3297624592454587,
+      "grad_norm": 14.10551643371582,
+      "learning_rate": 2.2020964360587004e-05,
+      "loss": 1.3947,
+      "step": 5710
+    },
+    {
+      "epoch": 1.3320912901723334,
+      "grad_norm": 27.368440628051758,
+      "learning_rate": 2.2006988120195668e-05,
+      "loss": 1.139,
+      "step": 5720
+    },
+    {
+      "epoch": 1.3344201210992082,
+      "grad_norm": 8.45453929901123,
+      "learning_rate": 2.1993011879804335e-05,
+      "loss": 1.7411,
+      "step": 5730
+    },
+    {
+      "epoch": 1.336748952026083,
+      "grad_norm": 17.110980987548828,
+      "learning_rate": 2.1979035639412998e-05,
+      "loss": 1.9253,
+      "step": 5740
+    },
+    {
+      "epoch": 1.3390777829529577,
+      "grad_norm": 94.51256561279297,
+      "learning_rate": 2.1965059399021665e-05,
+      "loss": 1.4818,
+      "step": 5750
+    },
+    {
+      "epoch": 1.3414066138798324,
+      "grad_norm": 7.826664924621582,
+      "learning_rate": 2.195108315863033e-05,
+      "loss": 0.8781,
+      "step": 5760
+    },
+    {
+      "epoch": 1.3437354448067071,
+      "grad_norm": 7.50213623046875,
+      "learning_rate": 2.1937106918238996e-05,
+      "loss": 1.1239,
+      "step": 5770
+    },
+    {
+      "epoch": 1.3460642757335817,
+      "grad_norm": 31.289094924926758,
+      "learning_rate": 2.192313067784766e-05,
+      "loss": 1.385,
+      "step": 5780
+    },
+    {
+      "epoch": 1.3483931066604564,
+      "grad_norm": 59.284637451171875,
+      "learning_rate": 2.1909154437456326e-05,
+      "loss": 1.3212,
+      "step": 5790
+    },
+    {
+      "epoch": 1.3507219375873312,
+      "grad_norm": 6.5409040451049805,
+      "learning_rate": 2.189517819706499e-05,
+      "loss": 1.13,
+      "step": 5800
+    },
+    {
+      "epoch": 1.353050768514206,
+      "grad_norm": 5.605960369110107,
+      "learning_rate": 2.1881201956673657e-05,
+      "loss": 0.9533,
+      "step": 5810
+    },
+    {
+      "epoch": 1.3553795994410804,
+      "grad_norm": 15.074896812438965,
+      "learning_rate": 2.186722571628232e-05,
+      "loss": 1.5392,
+      "step": 5820
+    },
+    {
+      "epoch": 1.3577084303679552,
+      "grad_norm": 22.98057746887207,
+      "learning_rate": 2.1853249475890987e-05,
+      "loss": 1.467,
+      "step": 5830
+    },
+    {
+      "epoch": 1.36003726129483,
+      "grad_norm": 17.14531898498535,
+      "learning_rate": 2.183927323549965e-05,
+      "loss": 1.4293,
+      "step": 5840
+    },
+    {
+      "epoch": 1.3623660922217047,
+      "grad_norm": 11.25844955444336,
+      "learning_rate": 2.1825296995108315e-05,
+      "loss": 1.0427,
+      "step": 5850
+    },
+    {
+      "epoch": 1.3646949231485794,
+      "grad_norm": 9.811922073364258,
+      "learning_rate": 2.181132075471698e-05,
+      "loss": 1.5446,
+      "step": 5860
+    },
+    {
+      "epoch": 1.3670237540754542,
+      "grad_norm": 15.156676292419434,
+      "learning_rate": 2.1797344514325645e-05,
+      "loss": 0.7154,
+      "step": 5870
+    },
+    {
+      "epoch": 1.369352585002329,
+      "grad_norm": 10.407319068908691,
+      "learning_rate": 2.1783368273934312e-05,
+      "loss": 1.5355,
+      "step": 5880
+    },
+    {
+      "epoch": 1.3716814159292037,
+      "grad_norm": 28.042938232421875,
+      "learning_rate": 2.1769392033542976e-05,
+      "loss": 1.6673,
+      "step": 5890
+    },
+    {
+      "epoch": 1.3740102468560782,
+      "grad_norm": 6.387546062469482,
+      "learning_rate": 2.1755415793151643e-05,
+      "loss": 1.2251,
+      "step": 5900
+    },
+    {
+      "epoch": 1.376339077782953,
+      "grad_norm": 14.928718566894531,
+      "learning_rate": 2.1741439552760306e-05,
+      "loss": 1.0641,
+      "step": 5910
+    },
+    {
+      "epoch": 1.3786679087098277,
+      "grad_norm": 20.759645462036133,
+      "learning_rate": 2.1727463312368973e-05,
+      "loss": 1.6231,
+      "step": 5920
+    },
+    {
+      "epoch": 1.3809967396367024,
+      "grad_norm": 91.47080993652344,
+      "learning_rate": 2.1713487071977637e-05,
+      "loss": 1.541,
+      "step": 5930
+    },
+    {
+      "epoch": 1.383325570563577,
+      "grad_norm": 121.2904281616211,
+      "learning_rate": 2.1699510831586304e-05,
+      "loss": 1.3784,
+      "step": 5940
+    },
+    {
+      "epoch": 1.3856544014904517,
+      "grad_norm": 21.550537109375,
+      "learning_rate": 2.1685534591194967e-05,
+      "loss": 1.221,
+      "step": 5950
+    },
+    {
+      "epoch": 1.3879832324173265,
+      "grad_norm": 22.803874969482422,
+      "learning_rate": 2.1671558350803634e-05,
+      "loss": 1.239,
+      "step": 5960
+    },
+    {
+      "epoch": 1.3903120633442012,
+      "grad_norm": 3.7256100177764893,
+      "learning_rate": 2.1657582110412298e-05,
+      "loss": 0.8911,
+      "step": 5970
+    },
+    {
+      "epoch": 1.392640894271076,
+      "grad_norm": 6.628619194030762,
+      "learning_rate": 2.1643605870020965e-05,
+      "loss": 1.2682,
+      "step": 5980
+    },
+    {
+      "epoch": 1.3949697251979507,
+      "grad_norm": 13.052042961120605,
+      "learning_rate": 2.1629629629629632e-05,
+      "loss": 1.2541,
+      "step": 5990
+    },
+    {
+      "epoch": 1.3972985561248255,
+      "grad_norm": 12.446592330932617,
+      "learning_rate": 2.1615653389238296e-05,
+      "loss": 1.2242,
+      "step": 6000
+    },
+    {
+      "epoch": 1.3996273870517,
+      "grad_norm": 15.783132553100586,
+      "learning_rate": 2.1601677148846963e-05,
+      "loss": 1.069,
+      "step": 6010
+    },
+    {
+      "epoch": 1.4019562179785747,
+      "grad_norm": 19.638019561767578,
+      "learning_rate": 2.1587700908455626e-05,
+      "loss": 1.039,
+      "step": 6020
+    },
+    {
+      "epoch": 1.4042850489054495,
+      "grad_norm": 17.819883346557617,
+      "learning_rate": 2.1573724668064293e-05,
+      "loss": 1.0184,
+      "step": 6030
+    },
+    {
+      "epoch": 1.4066138798323242,
+      "grad_norm": 14.201866149902344,
+      "learning_rate": 2.1559748427672957e-05,
+      "loss": 1.3285,
+      "step": 6040
+    },
+    {
+      "epoch": 1.4089427107591987,
+      "grad_norm": 9.433976173400879,
+      "learning_rate": 2.1545772187281624e-05,
+      "loss": 0.9471,
+      "step": 6050
+    },
+    {
+      "epoch": 1.4112715416860735,
+      "grad_norm": 19.676637649536133,
+      "learning_rate": 2.1531795946890287e-05,
+      "loss": 0.9991,
+      "step": 6060
+    },
+    {
+      "epoch": 1.4136003726129482,
+      "grad_norm": 7.89212703704834,
+      "learning_rate": 2.1517819706498954e-05,
+      "loss": 1.157,
+      "step": 6070
+    },
+    {
+      "epoch": 1.415929203539823,
+      "grad_norm": 8.475332260131836,
+      "learning_rate": 2.1503843466107618e-05,
+      "loss": 1.8082,
+      "step": 6080
+    },
+    {
+      "epoch": 1.4182580344666977,
+      "grad_norm": 30.952489852905273,
+      "learning_rate": 2.1489867225716285e-05,
+      "loss": 1.3291,
+      "step": 6090
+    },
+    {
+      "epoch": 1.4205868653935725,
+      "grad_norm": 41.84647750854492,
+      "learning_rate": 2.147589098532495e-05,
+      "loss": 1.4919,
+      "step": 6100
+    },
+    {
+      "epoch": 1.4229156963204472,
+      "grad_norm": 58.389835357666016,
+      "learning_rate": 2.1461914744933615e-05,
+      "loss": 1.8779,
+      "step": 6110
+    },
+    {
+      "epoch": 1.425244527247322,
+      "grad_norm": 73.64154052734375,
+      "learning_rate": 2.1447938504542276e-05,
+      "loss": 1.2146,
+      "step": 6120
+    },
+    {
+      "epoch": 1.4275733581741965,
+      "grad_norm": 10.781891822814941,
+      "learning_rate": 2.1433962264150943e-05,
+      "loss": 1.1101,
+      "step": 6130
+    },
+    {
+      "epoch": 1.4299021891010713,
+      "grad_norm": 12.203949928283691,
+      "learning_rate": 2.141998602375961e-05,
+      "loss": 1.1829,
+      "step": 6140
+    },
+    {
+      "epoch": 1.432231020027946,
+      "grad_norm": 8.079778671264648,
+      "learning_rate": 2.1406009783368273e-05,
+      "loss": 1.6185,
+      "step": 6150
+    },
+    {
+      "epoch": 1.4345598509548207,
+      "grad_norm": 16.477624893188477,
+      "learning_rate": 2.139203354297694e-05,
+      "loss": 1.5992,
+      "step": 6160
+    },
+    {
+      "epoch": 1.4368886818816953,
+      "grad_norm": 20.183237075805664,
+      "learning_rate": 2.1378057302585604e-05,
+      "loss": 1.7741,
+      "step": 6170
+    },
+    {
+      "epoch": 1.43921751280857,
+      "grad_norm": 14.812673568725586,
+      "learning_rate": 2.136408106219427e-05,
+      "loss": 1.2379,
+      "step": 6180
+    },
+    {
+      "epoch": 1.4415463437354448,
+      "grad_norm": 12.094216346740723,
+      "learning_rate": 2.1350104821802934e-05,
+      "loss": 1.7963,
+      "step": 6190
+    },
+    {
+      "epoch": 1.4438751746623195,
+      "grad_norm": 13.727142333984375,
+      "learning_rate": 2.13361285814116e-05,
+      "loss": 1.2336,
+      "step": 6200
+    },
+    {
+      "epoch": 1.4462040055891943,
+      "grad_norm": 12.991110801696777,
+      "learning_rate": 2.1322152341020265e-05,
+      "loss": 1.0751,
+      "step": 6210
+    },
+    {
+      "epoch": 1.448532836516069,
+      "grad_norm": 13.830179214477539,
+      "learning_rate": 2.1308176100628932e-05,
+      "loss": 1.4648,
+      "step": 6220
+    },
+    {
+      "epoch": 1.4508616674429438,
+      "grad_norm": 11.587153434753418,
+      "learning_rate": 2.1294199860237595e-05,
+      "loss": 1.4138,
+      "step": 6230
+    },
+    {
+      "epoch": 1.4531904983698183,
+      "grad_norm": 7.644782543182373,
+      "learning_rate": 2.1280223619846262e-05,
+      "loss": 1.0203,
+      "step": 6240
+    },
+    {
+      "epoch": 1.455519329296693,
+      "grad_norm": 12.32254695892334,
+      "learning_rate": 2.1266247379454926e-05,
+      "loss": 1.598,
+      "step": 6250
+    },
+    {
+      "epoch": 1.4578481602235678,
+      "grad_norm": 3.885669469833374,
+      "learning_rate": 2.1252271139063593e-05,
+      "loss": 1.4226,
+      "step": 6260
+    },
+    {
+      "epoch": 1.4601769911504425,
+      "grad_norm": 27.064685821533203,
+      "learning_rate": 2.1238294898672257e-05,
+      "loss": 1.4802,
+      "step": 6270
+    },
+    {
+      "epoch": 1.462505822077317,
+      "grad_norm": 14.793761253356934,
+      "learning_rate": 2.1224318658280923e-05,
+      "loss": 1.0699,
+      "step": 6280
+    },
+    {
+      "epoch": 1.4648346530041918,
+      "grad_norm": 9.927505493164062,
+      "learning_rate": 2.121034241788959e-05,
+      "loss": 1.0639,
+      "step": 6290
+    },
+    {
+      "epoch": 1.4671634839310665,
+      "grad_norm": 10.051878929138184,
+      "learning_rate": 2.1196366177498254e-05,
+      "loss": 0.8853,
+      "step": 6300
+    },
+    {
+      "epoch": 1.4694923148579413,
+      "grad_norm": 12.47751522064209,
+      "learning_rate": 2.118238993710692e-05,
+      "loss": 1.3319,
+      "step": 6310
+    },
+    {
+      "epoch": 1.471821145784816,
+      "grad_norm": 7.797066688537598,
+      "learning_rate": 2.1168413696715585e-05,
+      "loss": 0.914,
+      "step": 6320
+    },
+    {
+      "epoch": 1.4741499767116908,
+      "grad_norm": 14.888928413391113,
+      "learning_rate": 2.115443745632425e-05,
+      "loss": 1.3441,
+      "step": 6330
+    },
+    {
+      "epoch": 1.4764788076385655,
+      "grad_norm": 7.610679626464844,
+      "learning_rate": 2.1140461215932915e-05,
+      "loss": 1.0514,
+      "step": 6340
+    },
+    {
+      "epoch": 1.4788076385654403,
+      "grad_norm": 6.923398971557617,
+      "learning_rate": 2.1126484975541582e-05,
+      "loss": 1.6134,
+      "step": 6350
+    },
+    {
+      "epoch": 1.4811364694923148,
+      "grad_norm": 24.528564453125,
+      "learning_rate": 2.1112508735150246e-05,
+      "loss": 1.5473,
+      "step": 6360
+    },
+    {
+      "epoch": 1.4834653004191896,
+      "grad_norm": 4.077647686004639,
+      "learning_rate": 2.1098532494758913e-05,
+      "loss": 1.2879,
+      "step": 6370
+    },
+    {
+      "epoch": 1.4857941313460643,
+      "grad_norm": 13.990240097045898,
+      "learning_rate": 2.1084556254367573e-05,
+      "loss": 1.6288,
+      "step": 6380
+    },
+    {
+      "epoch": 1.488122962272939,
+      "grad_norm": 19.50718116760254,
+      "learning_rate": 2.107058001397624e-05,
+      "loss": 1.522,
+      "step": 6390
+    },
+    {
+      "epoch": 1.4904517931998136,
+      "grad_norm": 11.860418319702148,
+      "learning_rate": 2.1056603773584904e-05,
+      "loss": 1.2409,
+      "step": 6400
+    },
+    {
+      "epoch": 1.4927806241266883,
+      "grad_norm": 8.834514617919922,
+      "learning_rate": 2.104262753319357e-05,
+      "loss": 1.1322,
+      "step": 6410
+    },
+    {
+      "epoch": 1.495109455053563,
+      "grad_norm": 12.05262565612793,
+      "learning_rate": 2.1028651292802234e-05,
+      "loss": 1.2565,
+      "step": 6420
+    },
+    {
+      "epoch": 1.4974382859804378,
+      "grad_norm": 47.0871696472168,
+      "learning_rate": 2.10146750524109e-05,
+      "loss": 1.6558,
+      "step": 6430
+    },
+    {
+      "epoch": 1.4997671169073126,
+      "grad_norm": 25.88595962524414,
+      "learning_rate": 2.1000698812019568e-05,
+      "loss": 1.4326,
+      "step": 6440
+    },
+    {
+      "epoch": 1.5020959478341873,
+      "grad_norm": 21.97953224182129,
+      "learning_rate": 2.098672257162823e-05,
+      "loss": 1.4276,
+      "step": 6450
+    },
+    {
+      "epoch": 1.504424778761062,
+      "grad_norm": 6.62868595123291,
+      "learning_rate": 2.09727463312369e-05,
+      "loss": 1.4069,
+      "step": 6460
+    },
+    {
+      "epoch": 1.5067536096879368,
+      "grad_norm": 16.54710578918457,
+      "learning_rate": 2.0958770090845562e-05,
+      "loss": 1.2972,
+      "step": 6470
+    },
+    {
+      "epoch": 1.5090824406148113,
+      "grad_norm": 14.595316886901855,
+      "learning_rate": 2.094479385045423e-05,
+      "loss": 1.2964,
+      "step": 6480
+    },
+    {
+      "epoch": 1.511411271541686,
+      "grad_norm": 29.083581924438477,
+      "learning_rate": 2.0930817610062893e-05,
+      "loss": 1.6111,
+      "step": 6490
+    },
+    {
+      "epoch": 1.5137401024685608,
+      "grad_norm": 17.463119506835938,
+      "learning_rate": 2.091684136967156e-05,
+      "loss": 0.8385,
+      "step": 6500
+    },
+    {
+      "epoch": 1.5160689333954354,
+      "grad_norm": 11.917354583740234,
+      "learning_rate": 2.0902865129280223e-05,
+      "loss": 0.9631,
+      "step": 6510
+    },
+    {
+      "epoch": 1.51839776432231,
+      "grad_norm": 8.570207595825195,
+      "learning_rate": 2.088888888888889e-05,
+      "loss": 1.4749,
+      "step": 6520
+    },
+    {
+      "epoch": 1.5207265952491849,
+      "grad_norm": 6.944149971008301,
+      "learning_rate": 2.0874912648497554e-05,
+      "loss": 1.1676,
+      "step": 6530
+    },
+    {
+      "epoch": 1.5230554261760596,
+      "grad_norm": 19.165542602539062,
+      "learning_rate": 2.086093640810622e-05,
+      "loss": 1.7928,
+      "step": 6540
+    },
+    {
+      "epoch": 1.5253842571029343,
+      "grad_norm": 5.844601154327393,
+      "learning_rate": 2.0846960167714884e-05,
+      "loss": 1.0746,
+      "step": 6550
+    },
+    {
+      "epoch": 1.527713088029809,
+      "grad_norm": 79.03755187988281,
+      "learning_rate": 2.083298392732355e-05,
+      "loss": 0.9438,
+      "step": 6560
+    },
+    {
+      "epoch": 1.5300419189566838,
+      "grad_norm": 70.38703155517578,
+      "learning_rate": 2.0819007686932215e-05,
+      "loss": 1.2571,
+      "step": 6570
+    },
+    {
+      "epoch": 1.5323707498835586,
+      "grad_norm": 8.431889533996582,
+      "learning_rate": 2.0805031446540882e-05,
+      "loss": 1.53,
+      "step": 6580
+    },
+    {
+      "epoch": 1.5346995808104331,
+      "grad_norm": 13.615509033203125,
+      "learning_rate": 2.079105520614955e-05,
+      "loss": 1.0836,
+      "step": 6590
+    },
+    {
+      "epoch": 1.5370284117373079,
+      "grad_norm": 10.16684341430664,
+      "learning_rate": 2.0777078965758213e-05,
+      "loss": 1.4537,
+      "step": 6600
+    },
+    {
+      "epoch": 1.5393572426641826,
+      "grad_norm": 62.860252380371094,
+      "learning_rate": 2.076310272536688e-05,
+      "loss": 1.6445,
+      "step": 6610
+    },
+    {
+      "epoch": 1.5416860735910571,
+      "grad_norm": 10.388005256652832,
+      "learning_rate": 2.0749126484975543e-05,
+      "loss": 1.8212,
+      "step": 6620
+    },
+    {
+      "epoch": 1.5440149045179319,
+      "grad_norm": 15.161338806152344,
+      "learning_rate": 2.073515024458421e-05,
+      "loss": 1.5165,
+      "step": 6630
+    },
+    {
+      "epoch": 1.5463437354448066,
+      "grad_norm": 10.338879585266113,
+      "learning_rate": 2.072117400419287e-05,
+      "loss": 0.9325,
+      "step": 6640
+    },
+    {
+      "epoch": 1.5486725663716814,
+      "grad_norm": 50.81422424316406,
+      "learning_rate": 2.0707197763801537e-05,
+      "loss": 1.3275,
+      "step": 6650
+    },
+    {
+      "epoch": 1.5510013972985561,
+      "grad_norm": 63.32241439819336,
+      "learning_rate": 2.06932215234102e-05,
+      "loss": 1.6739,
+      "step": 6660
+    },
+    {
+      "epoch": 1.5533302282254309,
+      "grad_norm": 12.45587158203125,
+      "learning_rate": 2.0679245283018868e-05,
+      "loss": 1.6098,
+      "step": 6670
+    },
+    {
+      "epoch": 1.5556590591523056,
+      "grad_norm": 17.617839813232422,
+      "learning_rate": 2.066526904262753e-05,
+      "loss": 1.3134,
+      "step": 6680
+    },
+    {
+      "epoch": 1.5579878900791804,
+      "grad_norm": 4.559420108795166,
+      "learning_rate": 2.06512928022362e-05,
+      "loss": 1.0789,
+      "step": 6690
+    },
+    {
+      "epoch": 1.5603167210060551,
+      "grad_norm": 5.019169330596924,
+      "learning_rate": 2.0637316561844862e-05,
+      "loss": 1.186,
+      "step": 6700
+    },
+    {
+      "epoch": 1.5626455519329296,
+      "grad_norm": 47.28404998779297,
+      "learning_rate": 2.062334032145353e-05,
+      "loss": 2.23,
+      "step": 6710
+    },
+    {
+      "epoch": 1.5649743828598044,
+      "grad_norm": 8.204998970031738,
+      "learning_rate": 2.0609364081062193e-05,
+      "loss": 1.5562,
+      "step": 6720
+    },
+    {
+      "epoch": 1.5673032137866791,
+      "grad_norm": 12.214827537536621,
+      "learning_rate": 2.059538784067086e-05,
+      "loss": 1.2355,
+      "step": 6730
+    },
+    {
+      "epoch": 1.5696320447135537,
+      "grad_norm": 92.05329895019531,
+      "learning_rate": 2.0581411600279527e-05,
+      "loss": 1.6341,
+      "step": 6740
+    },
+    {
+      "epoch": 1.5719608756404284,
+      "grad_norm": 9.725508689880371,
+      "learning_rate": 2.056743535988819e-05,
+      "loss": 1.6041,
+      "step": 6750
+    },
+    {
+      "epoch": 1.5742897065673032,
+      "grad_norm": 5.3793134689331055,
+      "learning_rate": 2.0553459119496857e-05,
+      "loss": 0.8478,
+      "step": 6760
+    },
+    {
+      "epoch": 1.576618537494178,
+      "grad_norm": 7.88288688659668,
+      "learning_rate": 2.053948287910552e-05,
+      "loss": 1.0917,
+      "step": 6770
+    },
+    {
+      "epoch": 1.5789473684210527,
+      "grad_norm": 59.44280242919922,
+      "learning_rate": 2.0525506638714188e-05,
+      "loss": 1.3453,
+      "step": 6780
+    },
+    {
+      "epoch": 1.5812761993479274,
+      "grad_norm": 18.248361587524414,
+      "learning_rate": 2.051153039832285e-05,
+      "loss": 1.3135,
+      "step": 6790
+    },
+    {
+      "epoch": 1.5836050302748022,
+      "grad_norm": 4.527297496795654,
+      "learning_rate": 2.0497554157931518e-05,
+      "loss": 1.6539,
+      "step": 6800
+    },
+    {
+      "epoch": 1.585933861201677,
+      "grad_norm": 6.267810821533203,
+      "learning_rate": 2.0483577917540182e-05,
+      "loss": 1.2551,
+      "step": 6810
+    },
+    {
+      "epoch": 1.5882626921285514,
+      "grad_norm": 9.992630958557129,
+      "learning_rate": 2.046960167714885e-05,
+      "loss": 0.78,
+      "step": 6820
+    },
+    {
+      "epoch": 1.5905915230554262,
+      "grad_norm": 6.590550899505615,
+      "learning_rate": 2.0455625436757512e-05,
+      "loss": 1.3155,
+      "step": 6830
+    },
+    {
+      "epoch": 1.592920353982301,
+      "grad_norm": 5.96619176864624,
+      "learning_rate": 2.044164919636618e-05,
+      "loss": 1.5254,
+      "step": 6840
+    },
+    {
+      "epoch": 1.5952491849091754,
+      "grad_norm": 15.359123229980469,
+      "learning_rate": 2.0427672955974843e-05,
+      "loss": 1.7816,
+      "step": 6850
+    },
+    {
+      "epoch": 1.5975780158360502,
+      "grad_norm": 7.312784671783447,
+      "learning_rate": 2.041369671558351e-05,
+      "loss": 1.0621,
+      "step": 6860
+    },
+    {
+      "epoch": 1.599906846762925,
+      "grad_norm": 3.967820167541504,
+      "learning_rate": 2.0399720475192174e-05,
+      "loss": 1.3336,
+      "step": 6870
+    },
+    {
+      "epoch": 1.6022356776897997,
+      "grad_norm": 10.185931205749512,
+      "learning_rate": 2.038574423480084e-05,
+      "loss": 1.0566,
+      "step": 6880
+    },
+    {
+      "epoch": 1.6045645086166744,
+      "grad_norm": 7.750170707702637,
+      "learning_rate": 2.0371767994409507e-05,
+      "loss": 1.0334,
+      "step": 6890
+    },
+    {
+      "epoch": 1.6068933395435492,
+      "grad_norm": 11.731595993041992,
+      "learning_rate": 2.035779175401817e-05,
+      "loss": 0.8421,
+      "step": 6900
+    },
+    {
+      "epoch": 1.609222170470424,
+      "grad_norm": 17.454936981201172,
+      "learning_rate": 2.0343815513626835e-05,
+      "loss": 1.2394,
+      "step": 6910
+    },
+    {
+      "epoch": 1.6115510013972987,
+      "grad_norm": 4.5072712898254395,
+      "learning_rate": 2.0329839273235498e-05,
+      "loss": 1.1167,
+      "step": 6920
+    },
+    {
+      "epoch": 1.6138798323241734,
+      "grad_norm": 5.470495700836182,
+      "learning_rate": 2.0315863032844165e-05,
+      "loss": 1.2338,
+      "step": 6930
+    },
+    {
+      "epoch": 1.616208663251048,
+      "grad_norm": 12.268614768981934,
+      "learning_rate": 2.030188679245283e-05,
+      "loss": 1.5808,
+      "step": 6940
+    },
+    {
+      "epoch": 1.6185374941779227,
+      "grad_norm": 11.641376495361328,
+      "learning_rate": 2.0287910552061496e-05,
+      "loss": 1.1396,
+      "step": 6950
+    },
+    {
+      "epoch": 1.6208663251047974,
+      "grad_norm": 12.721565246582031,
+      "learning_rate": 2.027393431167016e-05,
+      "loss": 0.9674,
+      "step": 6960
+    },
+    {
+      "epoch": 1.623195156031672,
+      "grad_norm": 18.911191940307617,
+      "learning_rate": 2.0259958071278826e-05,
+      "loss": 1.4093,
+      "step": 6970
+    },
+    {
+      "epoch": 1.6255239869585467,
+      "grad_norm": 16.1356201171875,
+      "learning_rate": 2.024598183088749e-05,
+      "loss": 1.2923,
+      "step": 6980
+    },
+    {
+      "epoch": 1.6278528178854215,
+      "grad_norm": 5.580846786499023,
+      "learning_rate": 2.0232005590496157e-05,
+      "loss": 1.4068,
+      "step": 6990
+    },
+    {
+      "epoch": 1.6301816488122962,
+      "grad_norm": 10.02719783782959,
+      "learning_rate": 2.021802935010482e-05,
+      "loss": 1.3046,
+      "step": 7000
+    },
+    {
+      "epoch": 1.632510479739171,
+      "grad_norm": 10.013179779052734,
+      "learning_rate": 2.0204053109713487e-05,
+      "loss": 1.3374,
+      "step": 7010
+    },
+    {
+      "epoch": 1.6348393106660457,
+      "grad_norm": 7.2384724617004395,
+      "learning_rate": 2.019007686932215e-05,
+      "loss": 0.9352,
+      "step": 7020
+    },
+    {
+      "epoch": 1.6371681415929205,
+      "grad_norm": 15.23076343536377,
+      "learning_rate": 2.0176100628930818e-05,
+      "loss": 1.7616,
+      "step": 7030
+    },
+    {
+      "epoch": 1.6394969725197952,
+      "grad_norm": 35.012516021728516,
+      "learning_rate": 2.0162124388539485e-05,
+      "loss": 1.6217,
+      "step": 7040
+    },
+    {
+      "epoch": 1.6418258034466697,
+      "grad_norm": 7.518862724304199,
+      "learning_rate": 2.014814814814815e-05,
+      "loss": 0.77,
+      "step": 7050
+    },
+    {
+      "epoch": 1.6441546343735445,
+      "grad_norm": 8.740734100341797,
+      "learning_rate": 2.0134171907756816e-05,
+      "loss": 0.945,
+      "step": 7060
+    },
+    {
+      "epoch": 1.6464834653004192,
+      "grad_norm": 14.992202758789062,
+      "learning_rate": 2.012019566736548e-05,
+      "loss": 1.4889,
+      "step": 7070
+    },
+    {
+      "epoch": 1.6488122962272937,
+      "grad_norm": 9.385587692260742,
+      "learning_rate": 2.0106219426974146e-05,
+      "loss": 1.4533,
+      "step": 7080
+    },
+    {
+      "epoch": 1.6511411271541685,
+      "grad_norm": 3.7752981185913086,
+      "learning_rate": 2.009224318658281e-05,
+      "loss": 1.1922,
+      "step": 7090
+    },
+    {
+      "epoch": 1.6534699580810432,
+      "grad_norm": 9.964385986328125,
+      "learning_rate": 2.0078266946191477e-05,
+      "loss": 1.009,
+      "step": 7100
+    },
+    {
+      "epoch": 1.655798789007918,
+      "grad_norm": 22.4150333404541,
+      "learning_rate": 2.006429070580014e-05,
+      "loss": 1.3856,
+      "step": 7110
+    },
+    {
+      "epoch": 1.6581276199347927,
+      "grad_norm": 6.669826984405518,
+      "learning_rate": 2.0050314465408807e-05,
+      "loss": 1.1092,
+      "step": 7120
+    },
+    {
+      "epoch": 1.6604564508616675,
+      "grad_norm": 9.804742813110352,
+      "learning_rate": 2.003633822501747e-05,
+      "loss": 1.3718,
+      "step": 7130
+    },
+    {
+      "epoch": 1.6627852817885422,
+      "grad_norm": 9.104567527770996,
+      "learning_rate": 2.0022361984626138e-05,
+      "loss": 0.7321,
+      "step": 7140
+    },
+    {
+      "epoch": 1.665114112715417,
+      "grad_norm": 64.06851196289062,
+      "learning_rate": 2.00083857442348e-05,
+      "loss": 1.7271,
+      "step": 7150
+    },
+    {
+      "epoch": 1.6674429436422917,
+      "grad_norm": 12.741191864013672,
+      "learning_rate": 1.999440950384347e-05,
+      "loss": 1.6974,
+      "step": 7160
+    },
+    {
+      "epoch": 1.6697717745691663,
+      "grad_norm": 9.56039810180664,
+      "learning_rate": 1.998043326345213e-05,
+      "loss": 1.2862,
+      "step": 7170
+    },
+    {
+      "epoch": 1.672100605496041,
+      "grad_norm": 4.528858661651611,
+      "learning_rate": 1.9966457023060796e-05,
+      "loss": 1.1691,
+      "step": 7180
+    },
+    {
+      "epoch": 1.6744294364229158,
+      "grad_norm": 17.700504302978516,
+      "learning_rate": 1.9952480782669463e-05,
+      "loss": 1.1343,
+      "step": 7190
+    },
+    {
+      "epoch": 1.6767582673497903,
+      "grad_norm": 10.610898971557617,
+      "learning_rate": 1.9938504542278126e-05,
+      "loss": 1.5065,
+      "step": 7200
+    },
+    {
+      "epoch": 1.679087098276665,
+      "grad_norm": 4.640142917633057,
+      "learning_rate": 1.9924528301886793e-05,
+      "loss": 1.9951,
+      "step": 7210
+    },
+    {
+      "epoch": 1.6814159292035398,
+      "grad_norm": 8.147830963134766,
+      "learning_rate": 1.9910552061495457e-05,
+      "loss": 1.1698,
+      "step": 7220
+    },
+    {
+      "epoch": 1.6837447601304145,
+      "grad_norm": 36.66588592529297,
+      "learning_rate": 1.9896575821104124e-05,
+      "loss": 2.1156,
+      "step": 7230
+    },
+    {
+      "epoch": 1.6860735910572893,
+      "grad_norm": 18.592336654663086,
+      "learning_rate": 1.9882599580712787e-05,
+      "loss": 1.8785,
+      "step": 7240
+    },
+    {
+      "epoch": 1.688402421984164,
+      "grad_norm": 53.298561096191406,
+      "learning_rate": 1.9868623340321454e-05,
+      "loss": 1.2684,
+      "step": 7250
+    },
+    {
+      "epoch": 1.6907312529110388,
+      "grad_norm": 16.278640747070312,
+      "learning_rate": 1.9854647099930118e-05,
+      "loss": 1.2625,
+      "step": 7260
+    },
+    {
+      "epoch": 1.6930600838379135,
+      "grad_norm": 14.667279243469238,
+      "learning_rate": 1.9840670859538785e-05,
+      "loss": 1.2811,
+      "step": 7270
+    },
+    {
+      "epoch": 1.695388914764788,
+      "grad_norm": 8.197450637817383,
+      "learning_rate": 1.982669461914745e-05,
+      "loss": 1.451,
+      "step": 7280
+    },
+    {
+      "epoch": 1.6977177456916628,
+      "grad_norm": 43.80754852294922,
+      "learning_rate": 1.9812718378756115e-05,
+      "loss": 1.1298,
+      "step": 7290
+    },
+    {
+      "epoch": 1.7000465766185375,
+      "grad_norm": 7.743311405181885,
+      "learning_rate": 1.979874213836478e-05,
+      "loss": 1.2866,
+      "step": 7300
+    },
+    {
+      "epoch": 1.702375407545412,
+      "grad_norm": 51.69382858276367,
+      "learning_rate": 1.9784765897973446e-05,
+      "loss": 1.0107,
+      "step": 7310
+    },
+    {
+      "epoch": 1.7047042384722868,
+      "grad_norm": 7.2719573974609375,
+      "learning_rate": 1.977078965758211e-05,
+      "loss": 1.162,
+      "step": 7320
+    },
+    {
+      "epoch": 1.7070330693991616,
+      "grad_norm": 7.873298645019531,
+      "learning_rate": 1.9756813417190777e-05,
+      "loss": 1.7486,
+      "step": 7330
+    },
+    {
+      "epoch": 1.7093619003260363,
+      "grad_norm": 16.609840393066406,
+      "learning_rate": 1.9742837176799444e-05,
+      "loss": 1.2125,
+      "step": 7340
+    },
+    {
+      "epoch": 1.711690731252911,
+      "grad_norm": 6.402132034301758,
+      "learning_rate": 1.9728860936408107e-05,
+      "loss": 0.9309,
+      "step": 7350
+    },
+    {
+      "epoch": 1.7140195621797858,
+      "grad_norm": 3.3988287448883057,
+      "learning_rate": 1.9714884696016774e-05,
+      "loss": 1.2475,
+      "step": 7360
+    },
+    {
+      "epoch": 1.7163483931066605,
+      "grad_norm": 13.434524536132812,
+      "learning_rate": 1.9700908455625438e-05,
+      "loss": 1.1006,
+      "step": 7370
+    },
+    {
+      "epoch": 1.7186772240335353,
+      "grad_norm": 54.3780632019043,
+      "learning_rate": 1.9686932215234105e-05,
+      "loss": 1.2499,
+      "step": 7380
+    },
+    {
+      "epoch": 1.72100605496041,
+      "grad_norm": 5.294698715209961,
+      "learning_rate": 1.9672955974842768e-05,
+      "loss": 1.5747,
+      "step": 7390
+    },
+    {
+      "epoch": 1.7233348858872846,
+      "grad_norm": 11.490097045898438,
+      "learning_rate": 1.9658979734451435e-05,
+      "loss": 1.0938,
+      "step": 7400
+    },
+    {
+      "epoch": 1.7256637168141593,
+      "grad_norm": 19.641559600830078,
+      "learning_rate": 1.96450034940601e-05,
+      "loss": 2.3794,
+      "step": 7410
+    },
+    {
+      "epoch": 1.7279925477410338,
+      "grad_norm": 14.349048614501953,
+      "learning_rate": 1.9631027253668766e-05,
+      "loss": 1.4,
+      "step": 7420
+    },
+    {
+      "epoch": 1.7303213786679086,
+      "grad_norm": 5.6987528800964355,
+      "learning_rate": 1.9617051013277426e-05,
+      "loss": 1.1306,
+      "step": 7430
+    },
+    {
+      "epoch": 1.7326502095947833,
+      "grad_norm": 5.87094259262085,
+      "learning_rate": 1.9603074772886093e-05,
+      "loss": 1.3163,
+      "step": 7440
+    },
+    {
+      "epoch": 1.734979040521658,
+      "grad_norm": 16.617454528808594,
+      "learning_rate": 1.9589098532494757e-05,
+      "loss": 1.575,
+      "step": 7450
+    },
+    {
+      "epoch": 1.7373078714485328,
+      "grad_norm": 9.811037063598633,
+      "learning_rate": 1.9575122292103424e-05,
+      "loss": 1.4052,
+      "step": 7460
+    },
+    {
+      "epoch": 1.7396367023754076,
+      "grad_norm": 12.65397834777832,
+      "learning_rate": 1.9561146051712087e-05,
+      "loss": 1.4929,
+      "step": 7470
+    },
+    {
+      "epoch": 1.7419655333022823,
+      "grad_norm": 5.476722717285156,
+      "learning_rate": 1.9547169811320754e-05,
+      "loss": 0.8258,
+      "step": 7480
+    },
+    {
+      "epoch": 1.744294364229157,
+      "grad_norm": 13.89340877532959,
+      "learning_rate": 1.953319357092942e-05,
+      "loss": 1.4055,
+      "step": 7490
+    },
+    {
+      "epoch": 1.7466231951560318,
+      "grad_norm": 8.496129035949707,
+      "learning_rate": 1.9519217330538085e-05,
+      "loss": 1.3281,
+      "step": 7500
+    },
+    {
+      "epoch": 1.7489520260829063,
+      "grad_norm": 10.355660438537598,
+      "learning_rate": 1.950524109014675e-05,
+      "loss": 0.9894,
+      "step": 7510
+    },
+    {
+      "epoch": 1.751280857009781,
+      "grad_norm": 7.4711012840271,
+      "learning_rate": 1.9491264849755415e-05,
+      "loss": 1.3505,
+      "step": 7520
+    },
+    {
+      "epoch": 1.7536096879366558,
+      "grad_norm": 23.025426864624023,
+      "learning_rate": 1.9477288609364082e-05,
+      "loss": 1.0166,
+      "step": 7530
+    },
+    {
+      "epoch": 1.7559385188635304,
+      "grad_norm": 8.058009147644043,
+      "learning_rate": 1.9463312368972746e-05,
+      "loss": 1.1184,
+      "step": 7540
+    },
+    {
+      "epoch": 1.758267349790405,
+      "grad_norm": 23.26887321472168,
+      "learning_rate": 1.9449336128581413e-05,
+      "loss": 1.0111,
+      "step": 7550
+    },
+    {
+      "epoch": 1.7605961807172799,
+      "grad_norm": 7.433112144470215,
+      "learning_rate": 1.9435359888190076e-05,
+      "loss": 0.9874,
+      "step": 7560
+    },
+    {
+      "epoch": 1.7629250116441546,
+      "grad_norm": 10.770955085754395,
+      "learning_rate": 1.9421383647798743e-05,
+      "loss": 0.6591,
+      "step": 7570
+    },
+    {
+      "epoch": 1.7652538425710294,
+      "grad_norm": 7.087027549743652,
+      "learning_rate": 1.9407407407407407e-05,
+      "loss": 1.2804,
+      "step": 7580
+    },
+    {
+      "epoch": 1.767582673497904,
+      "grad_norm": 7.6507086753845215,
+      "learning_rate": 1.9393431167016074e-05,
+      "loss": 1.3398,
+      "step": 7590
+    },
+    {
+      "epoch": 1.7699115044247788,
+      "grad_norm": 6.744988918304443,
+      "learning_rate": 1.9379454926624738e-05,
+      "loss": 1.0609,
+      "step": 7600
+    },
+    {
+      "epoch": 1.7722403353516536,
+      "grad_norm": 63.14842224121094,
+      "learning_rate": 1.9365478686233404e-05,
+      "loss": 1.5501,
+      "step": 7610
+    },
+    {
+      "epoch": 1.7745691662785283,
+      "grad_norm": 28.729961395263672,
+      "learning_rate": 1.9351502445842068e-05,
+      "loss": 2.0076,
+      "step": 7620
+    },
+    {
+      "epoch": 1.7768979972054029,
+      "grad_norm": 16.338687896728516,
+      "learning_rate": 1.9337526205450735e-05,
+      "loss": 1.2513,
+      "step": 7630
+    },
+    {
+      "epoch": 1.7792268281322776,
+      "grad_norm": 8.51009750366211,
+      "learning_rate": 1.9323549965059402e-05,
+      "loss": 1.2085,
+      "step": 7640
+    },
+    {
+      "epoch": 1.7815556590591521,
+      "grad_norm": 145.62452697753906,
+      "learning_rate": 1.9309573724668066e-05,
+      "loss": 1.2988,
+      "step": 7650
+    },
+    {
+      "epoch": 1.7838844899860269,
+      "grad_norm": 20.35512924194336,
+      "learning_rate": 1.9295597484276733e-05,
+      "loss": 1.6377,
+      "step": 7660
+    },
+    {
+      "epoch": 1.7862133209129016,
+      "grad_norm": 14.52749252319336,
+      "learning_rate": 1.9281621243885396e-05,
+      "loss": 1.4134,
+      "step": 7670
+    },
+    {
+      "epoch": 1.7885421518397764,
+      "grad_norm": 17.46294403076172,
+      "learning_rate": 1.9267645003494063e-05,
+      "loss": 1.3532,
+      "step": 7680
+    },
+    {
+      "epoch": 1.7908709827666511,
+      "grad_norm": 43.280277252197266,
+      "learning_rate": 1.9253668763102727e-05,
+      "loss": 1.7841,
+      "step": 7690
+    },
+    {
+      "epoch": 1.7931998136935259,
+      "grad_norm": 15.846488952636719,
+      "learning_rate": 1.923969252271139e-05,
+      "loss": 2.242,
+      "step": 7700
+    },
+    {
+      "epoch": 1.7955286446204006,
+      "grad_norm": 73.0659408569336,
+      "learning_rate": 1.9225716282320054e-05,
+      "loss": 1.3953,
+      "step": 7710
+    },
+    {
+      "epoch": 1.7978574755472754,
+      "grad_norm": 38.85818862915039,
+      "learning_rate": 1.921174004192872e-05,
+      "loss": 1.8649,
+      "step": 7720
+    },
+    {
+      "epoch": 1.8001863064741501,
+      "grad_norm": 5.484123706817627,
+      "learning_rate": 1.9197763801537384e-05,
+      "loss": 1.4726,
+      "step": 7730
+    },
+    {
+      "epoch": 1.8025151374010246,
+      "grad_norm": 7.886444091796875,
+      "learning_rate": 1.918378756114605e-05,
+      "loss": 1.3963,
+      "step": 7740
+    },
+    {
+      "epoch": 1.8048439683278994,
+      "grad_norm": 13.587664604187012,
+      "learning_rate": 1.9169811320754715e-05,
+      "loss": 1.5654,
+      "step": 7750
+    },
+    {
+      "epoch": 1.8071727992547741,
+      "grad_norm": 23.31873893737793,
+      "learning_rate": 1.9155835080363382e-05,
+      "loss": 1.0611,
+      "step": 7760
+    },
+    {
+      "epoch": 1.8095016301816487,
+      "grad_norm": 85.84972381591797,
+      "learning_rate": 1.914185883997205e-05,
+      "loss": 1.3399,
+      "step": 7770
+    },
+    {
+      "epoch": 1.8118304611085234,
+      "grad_norm": 12.845376014709473,
+      "learning_rate": 1.9127882599580713e-05,
+      "loss": 2.0852,
+      "step": 7780
+    },
+    {
+      "epoch": 1.8141592920353982,
+      "grad_norm": 8.165783882141113,
+      "learning_rate": 1.911390635918938e-05,
+      "loss": 1.0608,
+      "step": 7790
+    },
+    {
+      "epoch": 1.816488122962273,
+      "grad_norm": 18.634197235107422,
+      "learning_rate": 1.9099930118798043e-05,
+      "loss": 1.2293,
+      "step": 7800
+    },
+    {
+      "epoch": 1.8188169538891477,
+      "grad_norm": 4.013707160949707,
+      "learning_rate": 1.908595387840671e-05,
+      "loss": 1.1989,
+      "step": 7810
+    },
+    {
+      "epoch": 1.8211457848160224,
+      "grad_norm": 9.955910682678223,
+      "learning_rate": 1.9071977638015374e-05,
+      "loss": 1.7447,
+      "step": 7820
+    },
+    {
+      "epoch": 1.8234746157428972,
+      "grad_norm": 3.6328682899475098,
+      "learning_rate": 1.905800139762404e-05,
+      "loss": 1.3645,
+      "step": 7830
+    },
+    {
+      "epoch": 1.825803446669772,
+      "grad_norm": 5.54978609085083,
+      "learning_rate": 1.9044025157232704e-05,
+      "loss": 1.3699,
+      "step": 7840
+    },
+    {
+      "epoch": 1.8281322775966466,
+      "grad_norm": 19.170997619628906,
+      "learning_rate": 1.903004891684137e-05,
+      "loss": 1.2573,
+      "step": 7850
+    },
+    {
+      "epoch": 1.8304611085235212,
+      "grad_norm": 13.860367774963379,
+      "learning_rate": 1.9016072676450035e-05,
+      "loss": 1.1799,
+      "step": 7860
+    },
+    {
+      "epoch": 1.832789939450396,
+      "grad_norm": 10.46110725402832,
+      "learning_rate": 1.9002096436058702e-05,
+      "loss": 2.0828,
+      "step": 7870
+    },
+    {
+      "epoch": 1.8351187703772704,
+      "grad_norm": 63.465240478515625,
+      "learning_rate": 1.8988120195667365e-05,
+      "loss": 1.0285,
+      "step": 7880
+    },
+    {
+      "epoch": 1.8374476013041452,
+      "grad_norm": 10.297215461730957,
+      "learning_rate": 1.8974143955276032e-05,
+      "loss": 1.0443,
+      "step": 7890
+    },
+    {
+      "epoch": 1.83977643223102,
+      "grad_norm": 7.570961952209473,
+      "learning_rate": 1.8960167714884696e-05,
+      "loss": 1.3184,
+      "step": 7900
+    },
+    {
+      "epoch": 1.8421052631578947,
+      "grad_norm": 9.8710298538208,
+      "learning_rate": 1.8946191474493363e-05,
+      "loss": 1.0637,
+      "step": 7910
+    },
+    {
+      "epoch": 1.8444340940847694,
+      "grad_norm": 11.42401123046875,
+      "learning_rate": 1.8932215234102027e-05,
+      "loss": 1.2406,
+      "step": 7920
+    },
+    {
+      "epoch": 1.8467629250116442,
+      "grad_norm": 15.829558372497559,
+      "learning_rate": 1.8918238993710694e-05,
+      "loss": 0.9227,
+      "step": 7930
+    },
+    {
+      "epoch": 1.849091755938519,
+      "grad_norm": 20.00259017944336,
+      "learning_rate": 1.890426275331936e-05,
+      "loss": 1.1393,
+      "step": 7940
+    },
+    {
+      "epoch": 1.8514205868653937,
+      "grad_norm": 5.527038097381592,
+      "learning_rate": 1.8890286512928024e-05,
+      "loss": 1.1476,
+      "step": 7950
+    },
+    {
+      "epoch": 1.8537494177922684,
+      "grad_norm": 7.4063334465026855,
+      "learning_rate": 1.8876310272536688e-05,
+      "loss": 1.1472,
+      "step": 7960
+    },
+    {
+      "epoch": 1.856078248719143,
+      "grad_norm": 6.894715785980225,
+      "learning_rate": 1.886233403214535e-05,
+      "loss": 1.1841,
+      "step": 7970
+    },
+    {
+      "epoch": 1.8584070796460177,
+      "grad_norm": 53.53019714355469,
+      "learning_rate": 1.8848357791754018e-05,
+      "loss": 1.1964,
+      "step": 7980
+    },
+    {
+      "epoch": 1.8607359105728924,
+      "grad_norm": 50.521297454833984,
+      "learning_rate": 1.8834381551362682e-05,
+      "loss": 1.3668,
+      "step": 7990
+    },
+    {
+      "epoch": 1.863064741499767,
+      "grad_norm": 8.40929889678955,
+      "learning_rate": 1.882040531097135e-05,
+      "loss": 0.8395,
+      "step": 8000
+    },
+    {
+      "epoch": 1.8653935724266417,
+      "grad_norm": 14.613615989685059,
+      "learning_rate": 1.8806429070580012e-05,
+      "loss": 1.3538,
+      "step": 8010
+    },
+    {
+      "epoch": 1.8677224033535165,
+      "grad_norm": 5.63895320892334,
+      "learning_rate": 1.879245283018868e-05,
+      "loss": 1.0303,
+      "step": 8020
+    },
+    {
+      "epoch": 1.8700512342803912,
+      "grad_norm": 8.055392265319824,
+      "learning_rate": 1.8778476589797343e-05,
+      "loss": 1.5969,
+      "step": 8030
+    },
+    {
+      "epoch": 1.872380065207266,
+      "grad_norm": 16.90934181213379,
+      "learning_rate": 1.876450034940601e-05,
+      "loss": 1.3653,
+      "step": 8040
+    },
+    {
+      "epoch": 1.8747088961341407,
+      "grad_norm": 20.84347915649414,
+      "learning_rate": 1.8750524109014674e-05,
+      "loss": 0.9698,
+      "step": 8050
+    },
+    {
+      "epoch": 1.8770377270610155,
+      "grad_norm": 4.819990634918213,
+      "learning_rate": 1.873654786862334e-05,
+      "loss": 1.1794,
+      "step": 8060
+    },
+    {
+      "epoch": 1.8793665579878902,
+      "grad_norm": 30.6875057220459,
+      "learning_rate": 1.8722571628232008e-05,
+      "loss": 1.5583,
+      "step": 8070
+    },
+    {
+      "epoch": 1.8816953889147647,
+      "grad_norm": 5.91504430770874,
+      "learning_rate": 1.870859538784067e-05,
+      "loss": 0.9559,
+      "step": 8080
+    },
+    {
+      "epoch": 1.8840242198416395,
+      "grad_norm": 58.85013961791992,
+      "learning_rate": 1.8694619147449338e-05,
+      "loss": 1.8267,
+      "step": 8090
+    },
+    {
+      "epoch": 1.8863530507685142,
+      "grad_norm": 8.03494644165039,
+      "learning_rate": 1.8680642907058e-05,
+      "loss": 1.4692,
+      "step": 8100
+    },
+    {
+      "epoch": 1.8886818816953888,
+      "grad_norm": 7.644827842712402,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 1.3284,
+      "step": 8110
+    },
+    {
+      "epoch": 1.8910107126222635,
+      "grad_norm": 7.554988861083984,
+      "learning_rate": 1.8652690426275332e-05,
+      "loss": 1.5628,
+      "step": 8120
+    },
+    {
+      "epoch": 1.8933395435491382,
+      "grad_norm": 28.42096519470215,
+      "learning_rate": 1.8638714185884e-05,
+      "loss": 1.6219,
+      "step": 8130
+    },
+    {
+      "epoch": 1.895668374476013,
+      "grad_norm": 4.3279032707214355,
+      "learning_rate": 1.8624737945492663e-05,
+      "loss": 0.9164,
+      "step": 8140
+    },
+    {
+      "epoch": 1.8979972054028877,
+      "grad_norm": 15.174659729003906,
+      "learning_rate": 1.861076170510133e-05,
+      "loss": 1.5619,
+      "step": 8150
+    },
+    {
+      "epoch": 1.9003260363297625,
+      "grad_norm": 5.442610740661621,
+      "learning_rate": 1.8596785464709993e-05,
+      "loss": 0.9886,
+      "step": 8160
+    },
+    {
+      "epoch": 1.9026548672566372,
+      "grad_norm": 4.958669662475586,
+      "learning_rate": 1.858280922431866e-05,
+      "loss": 1.652,
+      "step": 8170
+    },
+    {
+      "epoch": 1.904983698183512,
+      "grad_norm": 17.52564811706543,
+      "learning_rate": 1.8568832983927324e-05,
+      "loss": 0.9101,
+      "step": 8180
+    },
+    {
+      "epoch": 1.9073125291103867,
+      "grad_norm": 11.019481658935547,
+      "learning_rate": 1.855485674353599e-05,
+      "loss": 1.5395,
+      "step": 8190
+    },
+    {
+      "epoch": 1.9096413600372613,
+      "grad_norm": 7.791713714599609,
+      "learning_rate": 1.8540880503144655e-05,
+      "loss": 1.3154,
+      "step": 8200
+    },
+    {
+      "epoch": 1.911970190964136,
+      "grad_norm": 19.8372802734375,
+      "learning_rate": 1.852690426275332e-05,
+      "loss": 1.1015,
+      "step": 8210
+    },
+    {
+      "epoch": 1.9142990218910108,
+      "grad_norm": 8.076462745666504,
+      "learning_rate": 1.8512928022361985e-05,
+      "loss": 1.1082,
+      "step": 8220
+    },
+    {
+      "epoch": 1.9166278528178853,
+      "grad_norm": 5.504517555236816,
+      "learning_rate": 1.849895178197065e-05,
+      "loss": 1.5135,
+      "step": 8230
+    },
+    {
+      "epoch": 1.91895668374476,
+      "grad_norm": 9.928620338439941,
+      "learning_rate": 1.8484975541579316e-05,
+      "loss": 0.9919,
+      "step": 8240
+    },
+    {
+      "epoch": 1.9212855146716348,
+      "grad_norm": 12.56198787689209,
+      "learning_rate": 1.847099930118798e-05,
+      "loss": 1.756,
+      "step": 8250
+    },
+    {
+      "epoch": 1.9236143455985095,
+      "grad_norm": 6.634976863861084,
+      "learning_rate": 1.8457023060796646e-05,
+      "loss": 1.1167,
+      "step": 8260
+    },
+    {
+      "epoch": 1.9259431765253843,
+      "grad_norm": 9.223847389221191,
+      "learning_rate": 1.844304682040531e-05,
+      "loss": 1.0864,
+      "step": 8270
+    },
+    {
+      "epoch": 1.928272007452259,
+      "grad_norm": 62.113494873046875,
+      "learning_rate": 1.8429070580013977e-05,
+      "loss": 1.324,
+      "step": 8280
+    },
+    {
+      "epoch": 1.9306008383791338,
+      "grad_norm": 8.178874969482422,
+      "learning_rate": 1.841509433962264e-05,
+      "loss": 1.0045,
+      "step": 8290
+    },
+    {
+      "epoch": 1.9329296693060085,
+      "grad_norm": 16.192407608032227,
+      "learning_rate": 1.8401118099231307e-05,
+      "loss": 1.2798,
+      "step": 8300
+    },
+    {
+      "epoch": 1.935258500232883,
+      "grad_norm": 6.274266242980957,
+      "learning_rate": 1.838714185883997e-05,
+      "loss": 1.9747,
+      "step": 8310
+    },
+    {
+      "epoch": 1.9375873311597578,
+      "grad_norm": 21.142480850219727,
+      "learning_rate": 1.8373165618448638e-05,
+      "loss": 0.9294,
+      "step": 8320
+    },
+    {
+      "epoch": 1.9399161620866325,
+      "grad_norm": 14.333216667175293,
+      "learning_rate": 1.83591893780573e-05,
+      "loss": 1.1272,
+      "step": 8330
+    },
+    {
+      "epoch": 1.942244993013507,
+      "grad_norm": 6.282688140869141,
+      "learning_rate": 1.834521313766597e-05,
+      "loss": 0.9866,
+      "step": 8340
+    },
+    {
+      "epoch": 1.9445738239403818,
+      "grad_norm": 11.034943580627441,
+      "learning_rate": 1.8331236897274632e-05,
+      "loss": 1.4079,
+      "step": 8350
+    },
+    {
+      "epoch": 1.9469026548672566,
+      "grad_norm": 6.519465923309326,
+      "learning_rate": 1.83172606568833e-05,
+      "loss": 0.7809,
+      "step": 8360
+    },
+    {
+      "epoch": 1.9492314857941313,
+      "grad_norm": 8.534490585327148,
+      "learning_rate": 1.8303284416491966e-05,
+      "loss": 1.5021,
+      "step": 8370
+    },
+    {
+      "epoch": 1.951560316721006,
+      "grad_norm": 12.68577766418457,
+      "learning_rate": 1.828930817610063e-05,
+      "loss": 1.5503,
+      "step": 8380
+    },
+    {
+      "epoch": 1.9538891476478808,
+      "grad_norm": 20.240991592407227,
+      "learning_rate": 1.8275331935709297e-05,
+      "loss": 1.5506,
+      "step": 8390
+    },
+    {
+      "epoch": 1.9562179785747555,
+      "grad_norm": 9.088332176208496,
+      "learning_rate": 1.826135569531796e-05,
+      "loss": 1.1122,
+      "step": 8400
+    },
+    {
+      "epoch": 1.9585468095016303,
+      "grad_norm": 22.946273803710938,
+      "learning_rate": 1.8247379454926627e-05,
+      "loss": 1.1421,
+      "step": 8410
+    },
+    {
+      "epoch": 1.960875640428505,
+      "grad_norm": 23.892812728881836,
+      "learning_rate": 1.823340321453529e-05,
+      "loss": 1.2118,
+      "step": 8420
+    },
+    {
+      "epoch": 1.9632044713553796,
+      "grad_norm": 9.210332870483398,
+      "learning_rate": 1.8219426974143958e-05,
+      "loss": 1.4426,
+      "step": 8430
+    },
+    {
+      "epoch": 1.9655333022822543,
+      "grad_norm": 16.855960845947266,
+      "learning_rate": 1.820545073375262e-05,
+      "loss": 1.6925,
+      "step": 8440
+    },
+    {
+      "epoch": 1.967862133209129,
+      "grad_norm": 63.36180877685547,
+      "learning_rate": 1.8191474493361288e-05,
+      "loss": 1.7084,
+      "step": 8450
+    },
+    {
+      "epoch": 1.9701909641360036,
+      "grad_norm": 7.800633430480957,
+      "learning_rate": 1.8177498252969952e-05,
+      "loss": 1.0854,
+      "step": 8460
+    },
+    {
+      "epoch": 1.9725197950628783,
+      "grad_norm": 6.054851055145264,
+      "learning_rate": 1.816352201257862e-05,
+      "loss": 1.6178,
+      "step": 8470
+    },
+    {
+      "epoch": 1.974848625989753,
+      "grad_norm": 7.2113938331604,
+      "learning_rate": 1.8149545772187282e-05,
+      "loss": 1.0403,
+      "step": 8480
+    },
+    {
+      "epoch": 1.9771774569166278,
+      "grad_norm": 13.045158386230469,
+      "learning_rate": 1.8135569531795946e-05,
+      "loss": 1.1769,
+      "step": 8490
+    },
+    {
+      "epoch": 1.9795062878435026,
+      "grad_norm": 12.509069442749023,
+      "learning_rate": 1.812159329140461e-05,
+      "loss": 1.4351,
+      "step": 8500
+    },
+    {
+      "epoch": 1.9818351187703773,
+      "grad_norm": 16.71681022644043,
+      "learning_rate": 1.8107617051013277e-05,
+      "loss": 0.8778,
+      "step": 8510
+    },
+    {
+      "epoch": 1.984163949697252,
+      "grad_norm": 10.629281997680664,
+      "learning_rate": 1.8093640810621944e-05,
+      "loss": 0.7607,
+      "step": 8520
+    },
+    {
+      "epoch": 1.9864927806241268,
+      "grad_norm": 6.857609272003174,
+      "learning_rate": 1.8079664570230607e-05,
+      "loss": 0.9464,
+      "step": 8530
+    },
+    {
+      "epoch": 1.9888216115510013,
+      "grad_norm": 6.874720573425293,
+      "learning_rate": 1.8065688329839274e-05,
+      "loss": 1.3631,
+      "step": 8540
+    },
+    {
+      "epoch": 1.991150442477876,
+      "grad_norm": 7.804827690124512,
+      "learning_rate": 1.8051712089447938e-05,
+      "loss": 0.9221,
+      "step": 8550
+    },
+    {
+      "epoch": 1.9934792734047508,
+      "grad_norm": 84.84910583496094,
+      "learning_rate": 1.8037735849056605e-05,
+      "loss": 1.2762,
+      "step": 8560
+    },
+    {
+      "epoch": 1.9958081043316254,
+      "grad_norm": 24.175487518310547,
+      "learning_rate": 1.8023759608665268e-05,
+      "loss": 1.3768,
+      "step": 8570
+    },
+    {
+      "epoch": 1.9981369352585001,
+      "grad_norm": 5.0154900550842285,
+      "learning_rate": 1.8009783368273935e-05,
+      "loss": 1.106,
+      "step": 8580
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 0.502763032913208,
+      "eval_runtime": 57.2795,
+      "eval_samples_per_second": 76.048,
+      "eval_steps_per_second": 1.205,
+      "step": 8588
+    },
+    {
+      "epoch": 2.000465766185375,
+      "grad_norm": 9.15768814086914,
+      "learning_rate": 1.7997204751921734e-05,
+      "loss": 0.8738,
+      "step": 8590
+    },
+    {
+      "epoch": 2.0027945971122496,
+      "grad_norm": 19.934829711914062,
+      "learning_rate": 5.960857409133271e-06,
+      "loss": 0.9064,
+      "step": 8600
+    },
+    {
+      "epoch": 2.0051234280391244,
+      "grad_norm": 7.445651054382324,
+      "learning_rate": 5.932898415657036e-06,
+      "loss": 0.9088,
+      "step": 8610
+    },
+    {
+      "epoch": 2.007452258965999,
+      "grad_norm": 11.872694969177246,
+      "learning_rate": 5.904939422180802e-06,
+      "loss": 1.0579,
+      "step": 8620
+    },
+    {
+      "epoch": 2.009781089892874,
+      "grad_norm": 23.22488784790039,
+      "learning_rate": 5.8797763280521905e-06,
+      "loss": 1.1115,
+      "step": 8630
+    },
+    {
+      "epoch": 2.0121099208197486,
+      "grad_norm": 16.743192672729492,
+      "learning_rate": 5.854613233923579e-06,
+      "loss": 0.9759,
+      "step": 8640
+    },
+    {
+      "epoch": 2.0144387517466233,
+      "grad_norm": 6.448403358459473,
+      "learning_rate": 5.826654240447344e-06,
+      "loss": 0.9698,
+      "step": 8650
+    },
+    {
+      "epoch": 2.016767582673498,
+      "grad_norm": 22.09251594543457,
+      "learning_rate": 5.79869524697111e-06,
+      "loss": 1.0631,
+      "step": 8660
+    },
+    {
+      "epoch": 2.0190964136003724,
+      "grad_norm": 8.56945514678955,
+      "learning_rate": 5.770736253494874e-06,
+      "loss": 1.2748,
+      "step": 8670
+    },
+    {
+      "epoch": 2.021425244527247,
+      "grad_norm": 21.083589553833008,
+      "learning_rate": 5.742777260018639e-06,
+      "loss": 1.6963,
+      "step": 8680
+    },
+    {
+      "epoch": 2.023754075454122,
+      "grad_norm": 7.669532775878906,
+      "learning_rate": 5.714818266542404e-06,
+      "loss": 1.0708,
+      "step": 8690
+    },
+    {
+      "epoch": 2.0260829063809966,
+      "grad_norm": 10.819165229797363,
+      "learning_rate": 5.68685927306617e-06,
+      "loss": 0.9201,
+      "step": 8700
+    },
+    {
+      "epoch": 2.0284117373078714,
+      "grad_norm": 7.748441696166992,
+      "learning_rate": 5.6589002795899356e-06,
+      "loss": 1.4279,
+      "step": 8710
+    },
+    {
+      "epoch": 2.030740568234746,
+      "grad_norm": 5.020903587341309,
+      "learning_rate": 5.6309412861136996e-06,
+      "loss": 1.7287,
+      "step": 8720
+    },
+    {
+      "epoch": 2.033069399161621,
+      "grad_norm": 13.887042045593262,
+      "learning_rate": 5.602982292637465e-06,
+      "loss": 1.1915,
+      "step": 8730
+    },
+    {
+      "epoch": 2.0353982300884956,
+      "grad_norm": 69.59370422363281,
+      "learning_rate": 5.57502329916123e-06,
+      "loss": 1.2851,
+      "step": 8740
+    },
+    {
+      "epoch": 2.0377270610153704,
+      "grad_norm": 46.7410888671875,
+      "learning_rate": 5.547064305684996e-06,
+      "loss": 1.6063,
+      "step": 8750
+    },
+    {
+      "epoch": 2.040055891942245,
+      "grad_norm": 6.513312339782715,
+      "learning_rate": 5.519105312208761e-06,
+      "loss": 1.3997,
+      "step": 8760
+    },
+    {
+      "epoch": 2.04238472286912,
+      "grad_norm": 34.405757904052734,
+      "learning_rate": 5.491146318732526e-06,
+      "loss": 1.4069,
+      "step": 8770
+    },
+    {
+      "epoch": 2.0447135537959946,
+      "grad_norm": 18.17137336730957,
+      "learning_rate": 5.46318732525629e-06,
+      "loss": 1.1842,
+      "step": 8780
+    },
+    {
+      "epoch": 2.047042384722869,
+      "grad_norm": 50.68307876586914,
+      "learning_rate": 5.435228331780056e-06,
+      "loss": 1.3923,
+      "step": 8790
+    },
+    {
+      "epoch": 2.0493712156497437,
+      "grad_norm": 5.154820919036865,
+      "learning_rate": 5.407269338303822e-06,
+      "loss": 1.1868,
+      "step": 8800
+    },
+    {
+      "epoch": 2.0517000465766184,
+      "grad_norm": 57.07107162475586,
+      "learning_rate": 5.3793103448275865e-06,
+      "loss": 1.1506,
+      "step": 8810
+    },
+    {
+      "epoch": 2.054028877503493,
+      "grad_norm": 5.840853691101074,
+      "learning_rate": 5.351351351351352e-06,
+      "loss": 0.9782,
+      "step": 8820
+    },
+    {
+      "epoch": 2.056357708430368,
+      "grad_norm": 12.97612190246582,
+      "learning_rate": 5.323392357875116e-06,
+      "loss": 1.4827,
+      "step": 8830
+    },
+    {
+      "epoch": 2.0586865393572427,
+      "grad_norm": 16.05446434020996,
+      "learning_rate": 5.295433364398882e-06,
+      "loss": 1.2806,
+      "step": 8840
+    },
+    {
+      "epoch": 2.0610153702841174,
+      "grad_norm": 13.387228012084961,
+      "learning_rate": 5.267474370922647e-06,
+      "loss": 1.1232,
+      "step": 8850
+    },
+    {
+      "epoch": 2.063344201210992,
+      "grad_norm": 6.215972423553467,
+      "learning_rate": 5.239515377446412e-06,
+      "loss": 1.3148,
+      "step": 8860
+    },
+    {
+      "epoch": 2.065673032137867,
+      "grad_norm": 13.655588150024414,
+      "learning_rate": 5.211556383970177e-06,
+      "loss": 0.9524,
+      "step": 8870
+    },
+    {
+      "epoch": 2.0680018630647417,
+      "grad_norm": 6.583061695098877,
+      "learning_rate": 5.183597390493942e-06,
+      "loss": 1.0593,
+      "step": 8880
+    },
+    {
+      "epoch": 2.0703306939916164,
+      "grad_norm": 7.793154716491699,
+      "learning_rate": 5.155638397017708e-06,
+      "loss": 1.2788,
+      "step": 8890
+    },
+    {
+      "epoch": 2.0726595249184907,
+      "grad_norm": 17.533212661743164,
+      "learning_rate": 5.1276794035414726e-06,
+      "loss": 0.9179,
+      "step": 8900
+    },
+    {
+      "epoch": 2.0749883558453655,
+      "grad_norm": 8.611387252807617,
+      "learning_rate": 5.099720410065238e-06,
+      "loss": 1.1824,
+      "step": 8910
+    },
+    {
+      "epoch": 2.07731718677224,
+      "grad_norm": 8.162396430969238,
+      "learning_rate": 5.071761416589003e-06,
+      "loss": 0.7685,
+      "step": 8920
+    },
+    {
+      "epoch": 2.079646017699115,
+      "grad_norm": 14.802386283874512,
+      "learning_rate": 5.043802423112768e-06,
+      "loss": 0.9294,
+      "step": 8930
+    },
+    {
+      "epoch": 2.0819748486259897,
+      "grad_norm": 14.819314956665039,
+      "learning_rate": 5.015843429636533e-06,
+      "loss": 1.0837,
+      "step": 8940
+    },
+    {
+      "epoch": 2.0843036795528644,
+      "grad_norm": 45.875030517578125,
+      "learning_rate": 4.9878844361602984e-06,
+      "loss": 1.8533,
+      "step": 8950
+    },
+    {
+      "epoch": 2.086632510479739,
+      "grad_norm": 4.293564796447754,
+      "learning_rate": 4.959925442684063e-06,
+      "loss": 1.0036,
+      "step": 8960
+    },
+    {
+      "epoch": 2.088961341406614,
+      "grad_norm": 56.804134368896484,
+      "learning_rate": 4.931966449207829e-06,
+      "loss": 1.1895,
+      "step": 8970
+    },
+    {
+      "epoch": 2.0912901723334887,
+      "grad_norm": 14.47994327545166,
+      "learning_rate": 4.904007455731594e-06,
+      "loss": 1.5637,
+      "step": 8980
+    },
+    {
+      "epoch": 2.0936190032603634,
+      "grad_norm": 19.151981353759766,
+      "learning_rate": 4.876048462255359e-06,
+      "loss": 0.885,
+      "step": 8990
+    },
+    {
+      "epoch": 2.095947834187238,
+      "grad_norm": 30.888566970825195,
+      "learning_rate": 4.848089468779124e-06,
+      "loss": 1.1562,
+      "step": 9000
+    },
+    {
+      "epoch": 2.098276665114113,
+      "grad_norm": 7.732218265533447,
+      "learning_rate": 4.820130475302889e-06,
+      "loss": 0.9851,
+      "step": 9010
+    },
+    {
+      "epoch": 2.1006054960409872,
+      "grad_norm": 8.628619194030762,
+      "learning_rate": 4.792171481826655e-06,
+      "loss": 1.1773,
+      "step": 9020
+    },
+    {
+      "epoch": 2.102934326967862,
+      "grad_norm": 5.685352802276611,
+      "learning_rate": 4.76421248835042e-06,
+      "loss": 0.8583,
+      "step": 9030
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 59.909934997558594,
+      "learning_rate": 4.7362534948741845e-06,
+      "loss": 0.8674,
+      "step": 9040
+    },
+    {
+      "epoch": 2.1075919888216115,
+      "grad_norm": 10.20012378692627,
+      "learning_rate": 4.708294501397949e-06,
+      "loss": 1.5069,
+      "step": 9050
+    },
+    {
+      "epoch": 2.109920819748486,
+      "grad_norm": 11.948861122131348,
+      "learning_rate": 4.680335507921715e-06,
+      "loss": 0.8768,
+      "step": 9060
+    },
+    {
+      "epoch": 2.112249650675361,
+      "grad_norm": 22.48712158203125,
+      "learning_rate": 4.65237651444548e-06,
+      "loss": 1.3001,
+      "step": 9070
+    },
+    {
+      "epoch": 2.1145784816022357,
+      "grad_norm": 7.2206315994262695,
+      "learning_rate": 4.6244175209692456e-06,
+      "loss": 1.2205,
+      "step": 9080
+    },
+    {
+      "epoch": 2.1169073125291105,
+      "grad_norm": 8.07583236694336,
+      "learning_rate": 4.59645852749301e-06,
+      "loss": 1.4863,
+      "step": 9090
+    },
+    {
+      "epoch": 2.119236143455985,
+      "grad_norm": 11.95862102508545,
+      "learning_rate": 4.568499534016775e-06,
+      "loss": 1.0672,
+      "step": 9100
+    },
+    {
+      "epoch": 2.12156497438286,
+      "grad_norm": 60.69342041015625,
+      "learning_rate": 4.540540540540541e-06,
+      "loss": 1.4529,
+      "step": 9110
+    },
+    {
+      "epoch": 2.1238938053097347,
+      "grad_norm": 11.049464225769043,
+      "learning_rate": 4.512581547064306e-06,
+      "loss": 0.9201,
+      "step": 9120
+    },
+    {
+      "epoch": 2.1262226362366095,
+      "grad_norm": 15.80212116241455,
+      "learning_rate": 4.4846225535880715e-06,
+      "loss": 1.3178,
+      "step": 9130
+    },
+    {
+      "epoch": 2.1285514671634838,
+      "grad_norm": 12.804340362548828,
+      "learning_rate": 4.4566635601118354e-06,
+      "loss": 1.0052,
+      "step": 9140
+    },
+    {
+      "epoch": 2.1308802980903585,
+      "grad_norm": 52.227439880371094,
+      "learning_rate": 4.428704566635601e-06,
+      "loss": 0.6626,
+      "step": 9150
+    },
+    {
+      "epoch": 2.1332091290172333,
+      "grad_norm": 4.820531368255615,
+      "learning_rate": 4.400745573159366e-06,
+      "loss": 1.1978,
+      "step": 9160
+    },
+    {
+      "epoch": 2.135537959944108,
+      "grad_norm": 9.708754539489746,
+      "learning_rate": 4.372786579683132e-06,
+      "loss": 1.1169,
+      "step": 9170
+    },
+    {
+      "epoch": 2.1378667908709827,
+      "grad_norm": 23.4190731048584,
+      "learning_rate": 4.344827586206897e-06,
+      "loss": 0.9696,
+      "step": 9180
+    },
+    {
+      "epoch": 2.1401956217978575,
+      "grad_norm": 9.562813758850098,
+      "learning_rate": 4.316868592730661e-06,
+      "loss": 0.9916,
+      "step": 9190
+    },
+    {
+      "epoch": 2.1425244527247322,
+      "grad_norm": 51.41989517211914,
+      "learning_rate": 4.288909599254427e-06,
+      "loss": 1.6923,
+      "step": 9200
+    },
+    {
+      "epoch": 2.144853283651607,
+      "grad_norm": 18.753562927246094,
+      "learning_rate": 4.260950605778192e-06,
+      "loss": 1.283,
+      "step": 9210
+    },
+    {
+      "epoch": 2.1471821145784817,
+      "grad_norm": 6.791256904602051,
+      "learning_rate": 4.2329916123019575e-06,
+      "loss": 1.1738,
+      "step": 9220
+    },
+    {
+      "epoch": 2.1495109455053565,
+      "grad_norm": 79.94830322265625,
+      "learning_rate": 4.205032618825722e-06,
+      "loss": 0.9406,
+      "step": 9230
+    },
+    {
+      "epoch": 2.1518397764322312,
+      "grad_norm": 6.553360939025879,
+      "learning_rate": 4.177073625349487e-06,
+      "loss": 0.8378,
+      "step": 9240
+    },
+    {
+      "epoch": 2.1541686073591055,
+      "grad_norm": 5.851924419403076,
+      "learning_rate": 4.149114631873252e-06,
+      "loss": 0.9363,
+      "step": 9250
+    },
+    {
+      "epoch": 2.1564974382859803,
+      "grad_norm": 24.280319213867188,
+      "learning_rate": 4.121155638397018e-06,
+      "loss": 1.1103,
+      "step": 9260
+    },
+    {
+      "epoch": 2.158826269212855,
+      "grad_norm": 4.203216075897217,
+      "learning_rate": 4.093196644920783e-06,
+      "loss": 1.2514,
+      "step": 9270
+    },
+    {
+      "epoch": 2.16115510013973,
+      "grad_norm": 7.374353408813477,
+      "learning_rate": 4.065237651444548e-06,
+      "loss": 0.9512,
+      "step": 9280
+    },
+    {
+      "epoch": 2.1634839310666045,
+      "grad_norm": 17.918182373046875,
+      "learning_rate": 4.037278657968314e-06,
+      "loss": 0.9938,
+      "step": 9290
+    },
+    {
+      "epoch": 2.1658127619934793,
+      "grad_norm": 10.405567169189453,
+      "learning_rate": 4.009319664492078e-06,
+      "loss": 1.0328,
+      "step": 9300
+    },
+    {
+      "epoch": 2.168141592920354,
+      "grad_norm": 56.74263381958008,
+      "learning_rate": 3.981360671015844e-06,
+      "loss": 1.0014,
+      "step": 9310
+    },
+    {
+      "epoch": 2.1704704238472288,
+      "grad_norm": 67.22490692138672,
+      "learning_rate": 3.9534016775396084e-06,
+      "loss": 1.6543,
+      "step": 9320
+    },
+    {
+      "epoch": 2.1727992547741035,
+      "grad_norm": 18.301307678222656,
+      "learning_rate": 3.925442684063374e-06,
+      "loss": 1.19,
+      "step": 9330
+    },
+    {
+      "epoch": 2.1751280857009783,
+      "grad_norm": 4.8976287841796875,
+      "learning_rate": 3.897483690587139e-06,
+      "loss": 1.3139,
+      "step": 9340
+    },
+    {
+      "epoch": 2.177456916627853,
+      "grad_norm": 5.022180557250977,
+      "learning_rate": 3.869524697110904e-06,
+      "loss": 0.9576,
+      "step": 9350
+    },
+    {
+      "epoch": 2.1797857475547273,
+      "grad_norm": 17.034072875976562,
+      "learning_rate": 3.8415657036346695e-06,
+      "loss": 1.6219,
+      "step": 9360
+    },
+    {
+      "epoch": 2.182114578481602,
+      "grad_norm": 17.095535278320312,
+      "learning_rate": 3.8136067101584343e-06,
+      "loss": 1.2369,
+      "step": 9370
+    },
+    {
+      "epoch": 2.184443409408477,
+      "grad_norm": 8.286972045898438,
+      "learning_rate": 3.7856477166822e-06,
+      "loss": 1.7725,
+      "step": 9380
+    },
+    {
+      "epoch": 2.1867722403353516,
+      "grad_norm": 40.874053955078125,
+      "learning_rate": 3.7576887232059644e-06,
+      "loss": 1.0996,
+      "step": 9390
+    },
+    {
+      "epoch": 2.1891010712622263,
+      "grad_norm": 7.332772254943848,
+      "learning_rate": 3.72972972972973e-06,
+      "loss": 1.4552,
+      "step": 9400
+    },
+    {
+      "epoch": 2.191429902189101,
+      "grad_norm": 19.27867889404297,
+      "learning_rate": 3.701770736253495e-06,
+      "loss": 0.8538,
+      "step": 9410
+    },
+    {
+      "epoch": 2.193758733115976,
+      "grad_norm": 9.231902122497559,
+      "learning_rate": 3.67381174277726e-06,
+      "loss": 1.0024,
+      "step": 9420
+    },
+    {
+      "epoch": 2.1960875640428505,
+      "grad_norm": 10.635993003845215,
+      "learning_rate": 3.6458527493010255e-06,
+      "loss": 0.8558,
+      "step": 9430
+    },
+    {
+      "epoch": 2.1984163949697253,
+      "grad_norm": 8.209709167480469,
+      "learning_rate": 3.6178937558247903e-06,
+      "loss": 1.0147,
+      "step": 9440
+    },
+    {
+      "epoch": 2.2007452258966,
+      "grad_norm": 108.34497833251953,
+      "learning_rate": 3.5899347623485556e-06,
+      "loss": 1.5607,
+      "step": 9450
+    },
+    {
+      "epoch": 2.203074056823475,
+      "grad_norm": 42.276737213134766,
+      "learning_rate": 3.561975768872321e-06,
+      "loss": 1.2168,
+      "step": 9460
+    },
+    {
+      "epoch": 2.205402887750349,
+      "grad_norm": 11.989846229553223,
+      "learning_rate": 3.5340167753960857e-06,
+      "loss": 1.2395,
+      "step": 9470
+    },
+    {
+      "epoch": 2.207731718677224,
+      "grad_norm": 63.10080337524414,
+      "learning_rate": 3.506057781919851e-06,
+      "loss": 1.2162,
+      "step": 9480
+    },
+    {
+      "epoch": 2.2100605496040986,
+      "grad_norm": 5.683536529541016,
+      "learning_rate": 3.4780987884436158e-06,
+      "loss": 1.3208,
+      "step": 9490
+    },
+    {
+      "epoch": 2.2123893805309733,
+      "grad_norm": 5.439475059509277,
+      "learning_rate": 3.4501397949673814e-06,
+      "loss": 0.8779,
+      "step": 9500
+    },
+    {
+      "epoch": 2.214718211457848,
+      "grad_norm": 15.528427124023438,
+      "learning_rate": 3.4221808014911467e-06,
+      "loss": 1.3149,
+      "step": 9510
+    },
+    {
+      "epoch": 2.217047042384723,
+      "grad_norm": 16.126054763793945,
+      "learning_rate": 3.3942218080149115e-06,
+      "loss": 1.3207,
+      "step": 9520
+    },
+    {
+      "epoch": 2.2193758733115976,
+      "grad_norm": 6.91264533996582,
+      "learning_rate": 3.366262814538677e-06,
+      "loss": 1.3919,
+      "step": 9530
+    },
+    {
+      "epoch": 2.2217047042384723,
+      "grad_norm": 16.272396087646484,
+      "learning_rate": 3.3383038210624416e-06,
+      "loss": 1.2776,
+      "step": 9540
+    },
+    {
+      "epoch": 2.224033535165347,
+      "grad_norm": 6.406686782836914,
+      "learning_rate": 3.310344827586207e-06,
+      "loss": 0.7831,
+      "step": 9550
+    },
+    {
+      "epoch": 2.226362366092222,
+      "grad_norm": 17.45789909362793,
+      "learning_rate": 3.282385834109972e-06,
+      "loss": 1.6462,
+      "step": 9560
+    },
+    {
+      "epoch": 2.2286911970190966,
+      "grad_norm": 30.968273162841797,
+      "learning_rate": 3.254426840633737e-06,
+      "loss": 1.1366,
+      "step": 9570
+    },
+    {
+      "epoch": 2.2310200279459713,
+      "grad_norm": 13.682144165039062,
+      "learning_rate": 3.2264678471575023e-06,
+      "loss": 1.1012,
+      "step": 9580
+    },
+    {
+      "epoch": 2.233348858872846,
+      "grad_norm": 9.104609489440918,
+      "learning_rate": 3.198508853681268e-06,
+      "loss": 1.3075,
+      "step": 9590
+    },
+    {
+      "epoch": 2.2356776897997204,
+      "grad_norm": 79.64781188964844,
+      "learning_rate": 3.170549860205033e-06,
+      "loss": 1.2202,
+      "step": 9600
+    },
+    {
+      "epoch": 2.238006520726595,
+      "grad_norm": 6.4605631828308105,
+      "learning_rate": 3.142590866728798e-06,
+      "loss": 0.6628,
+      "step": 9610
+    },
+    {
+      "epoch": 2.24033535165347,
+      "grad_norm": 47.32363510131836,
+      "learning_rate": 3.114631873252563e-06,
+      "loss": 1.0585,
+      "step": 9620
+    },
+    {
+      "epoch": 2.2426641825803446,
+      "grad_norm": 8.3875732421875,
+      "learning_rate": 3.086672879776328e-06,
+      "loss": 1.4663,
+      "step": 9630
+    },
+    {
+      "epoch": 2.2449930135072194,
+      "grad_norm": 14.244718551635742,
+      "learning_rate": 3.0587138863000934e-06,
+      "loss": 1.2015,
+      "step": 9640
+    },
+    {
+      "epoch": 2.247321844434094,
+      "grad_norm": 22.4896297454834,
+      "learning_rate": 3.0307548928238582e-06,
+      "loss": 1.9219,
+      "step": 9650
+    },
+    {
+      "epoch": 2.249650675360969,
+      "grad_norm": 10.090749740600586,
+      "learning_rate": 3.0027958993476235e-06,
+      "loss": 0.8067,
+      "step": 9660
+    },
+    {
+      "epoch": 2.2519795062878436,
+      "grad_norm": 8.410811424255371,
+      "learning_rate": 2.9748369058713883e-06,
+      "loss": 1.0709,
+      "step": 9670
+    },
+    {
+      "epoch": 2.2543083372147183,
+      "grad_norm": 25.9169921875,
+      "learning_rate": 2.946877912395154e-06,
+      "loss": 1.3688,
+      "step": 9680
+    },
+    {
+      "epoch": 2.256637168141593,
+      "grad_norm": 24.173946380615234,
+      "learning_rate": 2.9189189189189193e-06,
+      "loss": 1.3964,
+      "step": 9690
+    },
+    {
+      "epoch": 2.258965999068468,
+      "grad_norm": 15.42428207397461,
+      "learning_rate": 2.890959925442684e-06,
+      "loss": 1.2085,
+      "step": 9700
+    },
+    {
+      "epoch": 2.261294829995342,
+      "grad_norm": 16.04413604736328,
+      "learning_rate": 2.8630009319664494e-06,
+      "loss": 1.2269,
+      "step": 9710
+    },
+    {
+      "epoch": 2.263623660922217,
+      "grad_norm": 17.178184509277344,
+      "learning_rate": 2.8350419384902146e-06,
+      "loss": 1.4462,
+      "step": 9720
+    },
+    {
+      "epoch": 2.2659524918490916,
+      "grad_norm": 6.437899589538574,
+      "learning_rate": 2.8070829450139795e-06,
+      "loss": 1.521,
+      "step": 9730
+    },
+    {
+      "epoch": 2.2682813227759664,
+      "grad_norm": 10.40018081665039,
+      "learning_rate": 2.7791239515377447e-06,
+      "loss": 1.0889,
+      "step": 9740
+    },
+    {
+      "epoch": 2.270610153702841,
+      "grad_norm": 6.229658603668213,
+      "learning_rate": 2.7511649580615096e-06,
+      "loss": 1.0799,
+      "step": 9750
+    },
+    {
+      "epoch": 2.272938984629716,
+      "grad_norm": 35.4992790222168,
+      "learning_rate": 2.723205964585275e-06,
+      "loss": 0.956,
+      "step": 9760
+    },
+    {
+      "epoch": 2.2752678155565906,
+      "grad_norm": 15.053701400756836,
+      "learning_rate": 2.69524697110904e-06,
+      "loss": 1.033,
+      "step": 9770
+    },
+    {
+      "epoch": 2.2775966464834654,
+      "grad_norm": 33.243473052978516,
+      "learning_rate": 2.670083876980429e-06,
+      "loss": 1.4346,
+      "step": 9780
+    },
+    {
+      "epoch": 2.27992547741034,
+      "grad_norm": 6.714288234710693,
+      "learning_rate": 2.6421248835041936e-06,
+      "loss": 1.1813,
+      "step": 9790
+    },
+    {
+      "epoch": 2.282254308337215,
+      "grad_norm": 11.203160285949707,
+      "learning_rate": 2.614165890027959e-06,
+      "loss": 0.7792,
+      "step": 9800
+    },
+    {
+      "epoch": 2.2845831392640896,
+      "grad_norm": 146.86062622070312,
+      "learning_rate": 2.5862068965517246e-06,
+      "loss": 1.2079,
+      "step": 9810
+    },
+    {
+      "epoch": 2.286911970190964,
+      "grad_norm": 9.571805000305176,
+      "learning_rate": 2.5582479030754894e-06,
+      "loss": 0.8902,
+      "step": 9820
+    },
+    {
+      "epoch": 2.2892408011178387,
+      "grad_norm": 15.292964935302734,
+      "learning_rate": 2.5302889095992547e-06,
+      "loss": 1.2055,
+      "step": 9830
+    },
+    {
+      "epoch": 2.2915696320447134,
+      "grad_norm": 85.07530975341797,
+      "learning_rate": 2.5023299161230195e-06,
+      "loss": 1.6654,
+      "step": 9840
+    },
+    {
+      "epoch": 2.293898462971588,
+      "grad_norm": 11.542017936706543,
+      "learning_rate": 2.4743709226467848e-06,
+      "loss": 1.3289,
+      "step": 9850
+    },
+    {
+      "epoch": 2.296227293898463,
+      "grad_norm": 7.581414222717285,
+      "learning_rate": 2.44641192917055e-06,
+      "loss": 1.1882,
+      "step": 9860
+    },
+    {
+      "epoch": 2.2985561248253377,
+      "grad_norm": 12.976031303405762,
+      "learning_rate": 2.418452935694315e-06,
+      "loss": 1.2158,
+      "step": 9870
+    },
+    {
+      "epoch": 2.3008849557522124,
+      "grad_norm": 6.565402507781982,
+      "learning_rate": 2.39049394221808e-06,
+      "loss": 1.0179,
+      "step": 9880
+    },
+    {
+      "epoch": 2.303213786679087,
+      "grad_norm": 17.8703556060791,
+      "learning_rate": 2.362534948741845e-06,
+      "loss": 1.0429,
+      "step": 9890
+    },
+    {
+      "epoch": 2.305542617605962,
+      "grad_norm": 24.56272315979004,
+      "learning_rate": 2.3345759552656107e-06,
+      "loss": 1.3841,
+      "step": 9900
+    },
+    {
+      "epoch": 2.3078714485328367,
+      "grad_norm": 13.09598159790039,
+      "learning_rate": 2.306616961789376e-06,
+      "loss": 0.9365,
+      "step": 9910
+    },
+    {
+      "epoch": 2.3102002794597114,
+      "grad_norm": 26.23815155029297,
+      "learning_rate": 2.2786579683131408e-06,
+      "loss": 1.4342,
+      "step": 9920
+    },
+    {
+      "epoch": 2.3125291103865857,
+      "grad_norm": 9.41688060760498,
+      "learning_rate": 2.250698974836906e-06,
+      "loss": 1.1212,
+      "step": 9930
+    },
+    {
+      "epoch": 2.3148579413134605,
+      "grad_norm": 7.139398097991943,
+      "learning_rate": 2.2227399813606713e-06,
+      "loss": 0.6761,
+      "step": 9940
+    },
+    {
+      "epoch": 2.317186772240335,
+      "grad_norm": 13.8255615234375,
+      "learning_rate": 2.194780987884436e-06,
+      "loss": 1.2559,
+      "step": 9950
+    },
+    {
+      "epoch": 2.31951560316721,
+      "grad_norm": 28.836498260498047,
+      "learning_rate": 2.1668219944082014e-06,
+      "loss": 1.3978,
+      "step": 9960
+    },
+    {
+      "epoch": 2.3218444340940847,
+      "grad_norm": 11.870185852050781,
+      "learning_rate": 2.1388630009319662e-06,
+      "loss": 1.5667,
+      "step": 9970
+    },
+    {
+      "epoch": 2.3241732650209594,
+      "grad_norm": 55.986881256103516,
+      "learning_rate": 2.1109040074557315e-06,
+      "loss": 1.238,
+      "step": 9980
+    },
+    {
+      "epoch": 2.326502095947834,
+      "grad_norm": 40.541297912597656,
+      "learning_rate": 2.082945013979497e-06,
+      "loss": 1.1823,
+      "step": 9990
+    },
+    {
+      "epoch": 2.328830926874709,
+      "grad_norm": 10.11244010925293,
+      "learning_rate": 2.054986020503262e-06,
+      "loss": 1.3272,
+      "step": 10000
+    },
+    {
+      "epoch": 2.3311597578015837,
+      "grad_norm": 15.212204933166504,
+      "learning_rate": 2.0270270270270273e-06,
+      "loss": 0.8434,
+      "step": 10010
+    },
+    {
+      "epoch": 2.3334885887284584,
+      "grad_norm": 7.350932598114014,
+      "learning_rate": 1.999068033550792e-06,
+      "loss": 1.1032,
+      "step": 10020
+    },
+    {
+      "epoch": 2.335817419655333,
+      "grad_norm": 32.078880310058594,
+      "learning_rate": 1.9711090400745574e-06,
+      "loss": 1.5609,
+      "step": 10030
+    },
+    {
+      "epoch": 2.3381462505822075,
+      "grad_norm": 9.110737800598145,
+      "learning_rate": 1.9431500465983226e-06,
+      "loss": 1.3581,
+      "step": 10040
+    },
+    {
+      "epoch": 2.3404750815090827,
+      "grad_norm": 43.029335021972656,
+      "learning_rate": 1.9151910531220875e-06,
+      "loss": 1.1401,
+      "step": 10050
+    },
+    {
+      "epoch": 2.342803912435957,
+      "grad_norm": 9.121164321899414,
+      "learning_rate": 1.8872320596458527e-06,
+      "loss": 1.1575,
+      "step": 10060
+    },
+    {
+      "epoch": 2.3451327433628317,
+      "grad_norm": 10.35621452331543,
+      "learning_rate": 1.859273066169618e-06,
+      "loss": 0.918,
+      "step": 10070
+    },
+    {
+      "epoch": 2.3474615742897065,
+      "grad_norm": 15.430923461914062,
+      "learning_rate": 1.831314072693383e-06,
+      "loss": 0.9122,
+      "step": 10080
+    },
+    {
+      "epoch": 2.3497904052165812,
+      "grad_norm": 58.956695556640625,
+      "learning_rate": 1.8033550792171483e-06,
+      "loss": 1.055,
+      "step": 10090
+    },
+    {
+      "epoch": 2.352119236143456,
+      "grad_norm": 4.652696132659912,
+      "learning_rate": 1.7753960857409133e-06,
+      "loss": 2.0394,
+      "step": 10100
+    },
+    {
+      "epoch": 2.3544480670703307,
+      "grad_norm": 27.606590270996094,
+      "learning_rate": 1.7474370922646786e-06,
+      "loss": 1.7843,
+      "step": 10110
+    },
+    {
+      "epoch": 2.3567768979972055,
+      "grad_norm": 13.248260498046875,
+      "learning_rate": 1.7194780987884436e-06,
+      "loss": 1.7327,
+      "step": 10120
+    },
+    {
+      "epoch": 2.35910572892408,
+      "grad_norm": 38.81512451171875,
+      "learning_rate": 1.6915191053122087e-06,
+      "loss": 1.8256,
+      "step": 10130
+    },
+    {
+      "epoch": 2.361434559850955,
+      "grad_norm": 18.427837371826172,
+      "learning_rate": 1.663560111835974e-06,
+      "loss": 0.8146,
+      "step": 10140
+    },
+    {
+      "epoch": 2.3637633907778297,
+      "grad_norm": 21.866954803466797,
+      "learning_rate": 1.635601118359739e-06,
+      "loss": 1.3319,
+      "step": 10150
+    },
+    {
+      "epoch": 2.3660922217047045,
+      "grad_norm": 12.13359546661377,
+      "learning_rate": 1.6076421248835043e-06,
+      "loss": 0.9745,
+      "step": 10160
+    },
+    {
+      "epoch": 2.3684210526315788,
+      "grad_norm": 16.050683975219727,
+      "learning_rate": 1.5796831314072693e-06,
+      "loss": 1.3594,
+      "step": 10170
+    },
+    {
+      "epoch": 2.3707498835584535,
+      "grad_norm": 9.901156425476074,
+      "learning_rate": 1.5517241379310346e-06,
+      "loss": 1.2522,
+      "step": 10180
+    },
+    {
+      "epoch": 2.3730787144853283,
+      "grad_norm": 10.21035385131836,
+      "learning_rate": 1.5237651444547996e-06,
+      "loss": 1.0524,
+      "step": 10190
+    },
+    {
+      "epoch": 2.375407545412203,
+      "grad_norm": 12.119868278503418,
+      "learning_rate": 1.4958061509785649e-06,
+      "loss": 1.3232,
+      "step": 10200
+    },
+    {
+      "epoch": 2.3777363763390778,
+      "grad_norm": 15.788361549377441,
+      "learning_rate": 1.46784715750233e-06,
+      "loss": 1.1433,
+      "step": 10210
+    },
+    {
+      "epoch": 2.3800652072659525,
+      "grad_norm": 14.479697227478027,
+      "learning_rate": 1.4398881640260952e-06,
+      "loss": 0.6549,
+      "step": 10220
+    },
+    {
+      "epoch": 2.3823940381928272,
+      "grad_norm": 5.601083278656006,
+      "learning_rate": 1.4119291705498603e-06,
+      "loss": 0.8918,
+      "step": 10230
+    },
+    {
+      "epoch": 2.384722869119702,
+      "grad_norm": 7.322697639465332,
+      "learning_rate": 1.3839701770736253e-06,
+      "loss": 1.4818,
+      "step": 10240
+    },
+    {
+      "epoch": 2.3870517000465767,
+      "grad_norm": 11.558350563049316,
+      "learning_rate": 1.3560111835973906e-06,
+      "loss": 0.9127,
+      "step": 10250
+    },
+    {
+      "epoch": 2.3893805309734515,
+      "grad_norm": 102.35279846191406,
+      "learning_rate": 1.3280521901211556e-06,
+      "loss": 1.1501,
+      "step": 10260
+    },
+    {
+      "epoch": 2.3917093619003262,
+      "grad_norm": 32.1492805480957,
+      "learning_rate": 1.3000931966449209e-06,
+      "loss": 1.2799,
+      "step": 10270
+    },
+    {
+      "epoch": 2.3940381928272005,
+      "grad_norm": 34.78493881225586,
+      "learning_rate": 1.272134203168686e-06,
+      "loss": 1.5782,
+      "step": 10280
+    },
+    {
+      "epoch": 2.3963670237540753,
+      "grad_norm": 100.30006408691406,
+      "learning_rate": 1.244175209692451e-06,
+      "loss": 1.5619,
+      "step": 10290
+    },
+    {
+      "epoch": 2.39869585468095,
+      "grad_norm": 10.30715274810791,
+      "learning_rate": 1.2162162162162162e-06,
+      "loss": 1.1595,
+      "step": 10300
+    },
+    {
+      "epoch": 2.401024685607825,
+      "grad_norm": 7.528810977935791,
+      "learning_rate": 1.1882572227399815e-06,
+      "loss": 0.8228,
+      "step": 10310
+    },
+    {
+      "epoch": 2.4033535165346995,
+      "grad_norm": 9.433389663696289,
+      "learning_rate": 1.1602982292637465e-06,
+      "loss": 1.124,
+      "step": 10320
+    },
+    {
+      "epoch": 2.4056823474615743,
+      "grad_norm": 50.19340133666992,
+      "learning_rate": 1.1323392357875116e-06,
+      "loss": 1.3402,
+      "step": 10330
+    },
+    {
+      "epoch": 2.408011178388449,
+      "grad_norm": 28.69070816040039,
+      "learning_rate": 1.1043802423112769e-06,
+      "loss": 1.1887,
+      "step": 10340
+    },
+    {
+      "epoch": 2.4103400093153238,
+      "grad_norm": 9.501694679260254,
+      "learning_rate": 1.0764212488350421e-06,
+      "loss": 1.0498,
+      "step": 10350
+    },
+    {
+      "epoch": 2.4126688402421985,
+      "grad_norm": 18.250404357910156,
+      "learning_rate": 1.0484622553588072e-06,
+      "loss": 1.7923,
+      "step": 10360
+    },
+    {
+      "epoch": 2.4149976711690733,
+      "grad_norm": 6.274858474731445,
+      "learning_rate": 1.0205032618825722e-06,
+      "loss": 1.1322,
+      "step": 10370
+    },
+    {
+      "epoch": 2.417326502095948,
+      "grad_norm": 46.66001892089844,
+      "learning_rate": 9.925442684063373e-07,
+      "loss": 1.7829,
+      "step": 10380
+    },
+    {
+      "epoch": 2.4196553330228223,
+      "grad_norm": 6.775675296783447,
+      "learning_rate": 9.645852749301025e-07,
+      "loss": 1.2486,
+      "step": 10390
+    },
+    {
+      "epoch": 2.421984163949697,
+      "grad_norm": 7.135899066925049,
+      "learning_rate": 9.366262814538677e-07,
+      "loss": 0.8247,
+      "step": 10400
+    },
+    {
+      "epoch": 2.424312994876572,
+      "grad_norm": 9.592740058898926,
+      "learning_rate": 9.086672879776328e-07,
+      "loss": 0.9868,
+      "step": 10410
+    },
+    {
+      "epoch": 2.4266418258034466,
+      "grad_norm": 8.71237564086914,
+      "learning_rate": 8.807082945013979e-07,
+      "loss": 1.6636,
+      "step": 10420
+    },
+    {
+      "epoch": 2.4289706567303213,
+      "grad_norm": 11.005659103393555,
+      "learning_rate": 8.527493010251631e-07,
+      "loss": 1.2524,
+      "step": 10430
+    },
+    {
+      "epoch": 2.431299487657196,
+      "grad_norm": 5.304808616638184,
+      "learning_rate": 8.247903075489282e-07,
+      "loss": 1.1702,
+      "step": 10440
+    },
+    {
+      "epoch": 2.433628318584071,
+      "grad_norm": 16.233470916748047,
+      "learning_rate": 7.968313140726935e-07,
+      "loss": 1.2214,
+      "step": 10450
+    },
+    {
+      "epoch": 2.4359571495109456,
+      "grad_norm": 7.547863960266113,
+      "learning_rate": 7.688723205964585e-07,
+      "loss": 0.954,
+      "step": 10460
+    },
+    {
+      "epoch": 2.4382859804378203,
+      "grad_norm": 10.951506614685059,
+      "learning_rate": 7.409133271202238e-07,
+      "loss": 1.1177,
+      "step": 10470
+    },
+    {
+      "epoch": 2.440614811364695,
+      "grad_norm": 6.370845317840576,
+      "learning_rate": 7.129543336439888e-07,
+      "loss": 0.897,
+      "step": 10480
+    },
+    {
+      "epoch": 2.44294364229157,
+      "grad_norm": 9.310226440429688,
+      "learning_rate": 6.84995340167754e-07,
+      "loss": 1.4497,
+      "step": 10490
+    },
+    {
+      "epoch": 2.445272473218444,
+      "grad_norm": 79.52318572998047,
+      "learning_rate": 6.570363466915191e-07,
+      "loss": 1.5888,
+      "step": 10500
+    },
+    {
+      "epoch": 2.4476013041453193,
+      "grad_norm": 6.686119556427002,
+      "learning_rate": 6.290773532152843e-07,
+      "loss": 1.0586,
+      "step": 10510
+    },
+    {
+      "epoch": 2.4499301350721936,
+      "grad_norm": 16.480131149291992,
+      "learning_rate": 6.011183597390494e-07,
+      "loss": 1.2573,
+      "step": 10520
+    },
+    {
+      "epoch": 2.4522589659990683,
+      "grad_norm": 18.26215362548828,
+      "learning_rate": 5.731593662628145e-07,
+      "loss": 1.4545,
+      "step": 10530
+    },
+    {
+      "epoch": 2.454587796925943,
+      "grad_norm": 6.198437213897705,
+      "learning_rate": 5.452003727865797e-07,
+      "loss": 1.0308,
+      "step": 10540
+    },
+    {
+      "epoch": 2.456916627852818,
+      "grad_norm": 8.964813232421875,
+      "learning_rate": 5.172413793103448e-07,
+      "loss": 1.3532,
+      "step": 10550
+    },
+    {
+      "epoch": 2.4592454587796926,
+      "grad_norm": 11.396198272705078,
+      "learning_rate": 4.892823858341099e-07,
+      "loss": 1.2104,
+      "step": 10560
+    },
+    {
+      "epoch": 2.4615742897065673,
+      "grad_norm": 10.318989753723145,
+      "learning_rate": 4.6132339235787515e-07,
+      "loss": 0.9936,
+      "step": 10570
+    },
+    {
+      "epoch": 2.463903120633442,
+      "grad_norm": 84.06649780273438,
+      "learning_rate": 4.333643988816403e-07,
+      "loss": 1.2898,
+      "step": 10580
+    },
+    {
+      "epoch": 2.466231951560317,
+      "grad_norm": 9.45693588256836,
+      "learning_rate": 4.0540540540540546e-07,
+      "loss": 1.3486,
+      "step": 10590
+    },
+    {
+      "epoch": 2.4685607824871916,
+      "grad_norm": 15.41531753540039,
+      "learning_rate": 3.774464119291705e-07,
+      "loss": 1.9033,
+      "step": 10600
+    },
+    {
+      "epoch": 2.4708896134140663,
+      "grad_norm": 23.50023078918457,
+      "learning_rate": 3.4948741845293567e-07,
+      "loss": 0.8877,
+      "step": 10610
+    },
+    {
+      "epoch": 2.473218444340941,
+      "grad_norm": 11.658283233642578,
+      "learning_rate": 3.215284249767008e-07,
+      "loss": 1.4179,
+      "step": 10620
+    },
+    {
+      "epoch": 2.4755472752678154,
+      "grad_norm": 9.004804611206055,
+      "learning_rate": 2.93569431500466e-07,
+      "loss": 1.254,
+      "step": 10630
+    },
+    {
+      "epoch": 2.47787610619469,
+      "grad_norm": 11.310983657836914,
+      "learning_rate": 2.6561043802423113e-07,
+      "loss": 0.9785,
+      "step": 10640
+    },
+    {
+      "epoch": 2.480204937121565,
+      "grad_norm": 24.459308624267578,
+      "learning_rate": 2.376514445479963e-07,
+      "loss": 1.8078,
+      "step": 10650
+    },
+    {
+      "epoch": 2.4825337680484396,
+      "grad_norm": 14.287947654724121,
+      "learning_rate": 2.0969245107176142e-07,
+      "loss": 1.2669,
+      "step": 10660
+    },
+    {
+      "epoch": 2.4848625989753144,
+      "grad_norm": 9.133045196533203,
+      "learning_rate": 1.8173345759552657e-07,
+      "loss": 0.8908,
+      "step": 10670
+    },
+    {
+      "epoch": 2.487191429902189,
+      "grad_norm": 45.257755279541016,
+      "learning_rate": 1.5377446411929173e-07,
+      "loss": 1.939,
+      "step": 10680
+    },
+    {
+      "epoch": 2.489520260829064,
+      "grad_norm": 14.624387741088867,
+      "learning_rate": 1.2581547064305686e-07,
+      "loss": 1.2464,
+      "step": 10690
+    },
+    {
+      "epoch": 2.4918490917559386,
+      "grad_norm": 14.175239562988281,
+      "learning_rate": 9.7856477166822e-08,
+      "loss": 1.524,
+      "step": 10700
+    },
+    {
+      "epoch": 2.4941779226828134,
+      "grad_norm": 10.873252868652344,
+      "learning_rate": 6.989748369058714e-08,
+      "loss": 1.2533,
+      "step": 10710
+    },
+    {
+      "epoch": 2.496506753609688,
+      "grad_norm": 9.596633911132812,
+      "learning_rate": 4.193849021435229e-08,
+      "loss": 1.5095,
+      "step": 10720
+    },
+    {
+      "epoch": 2.498835584536563,
+      "grad_norm": 80.26763153076172,
+      "learning_rate": 1.3979496738117427e-08,
+      "loss": 2.6379,
+      "step": 10730
+    },
+    {
+      "epoch": 2.501164415463437,
+      "grad_norm": 9.910447120666504,
+      "learning_rate": 0.0,
+      "loss": 0.9697,
+      "step": 10740
+    },
+    {
+      "epoch": 2.503493246390312,
+      "grad_norm": 11.675068855285645,
+      "learning_rate": 0.0,
+      "loss": 1.0056,
+      "step": 10750
+    },
+    {
+      "epoch": 2.5058220773171866,
+      "grad_norm": 11.493444442749023,
+      "learning_rate": 0.0,
+      "loss": 1.5176,
+      "step": 10760
+    },
+    {
+      "epoch": 2.5081509082440614,
+      "grad_norm": 12.753838539123535,
+      "learning_rate": 0.0,
+      "loss": 1.1031,
+      "step": 10770
+    },
+    {
+      "epoch": 2.510479739170936,
+      "grad_norm": 14.464014053344727,
+      "learning_rate": 0.0,
+      "loss": 0.9649,
+      "step": 10780
+    },
+    {
+      "epoch": 2.512808570097811,
+      "grad_norm": 15.377732276916504,
+      "learning_rate": 0.0,
+      "loss": 1.4322,
+      "step": 10790
+    },
+    {
+      "epoch": 2.5151374010246856,
+      "grad_norm": 9.88819694519043,
+      "learning_rate": 0.0,
+      "loss": 0.9797,
+      "step": 10800
+    },
+    {
+      "epoch": 2.5174662319515604,
+      "grad_norm": 6.533481121063232,
+      "learning_rate": 0.0,
+      "loss": 1.2964,
+      "step": 10810
+    },
+    {
+      "epoch": 2.519795062878435,
+      "grad_norm": 20.36925506591797,
+      "learning_rate": 0.0,
+      "loss": 1.6195,
+      "step": 10820
+    },
+    {
+      "epoch": 2.52212389380531,
+      "grad_norm": 6.780529022216797,
+      "learning_rate": 0.0,
+      "loss": 1.039,
+      "step": 10830
+    },
+    {
+      "epoch": 2.5244527247321846,
+      "grad_norm": 36.58348083496094,
+      "learning_rate": 0.0,
+      "loss": 1.3896,
+      "step": 10840
+    },
+    {
+      "epoch": 2.526781555659059,
+      "grad_norm": 7.8758039474487305,
+      "learning_rate": 0.0,
+      "loss": 1.4437,
+      "step": 10850
+    },
+    {
+      "epoch": 2.529110386585934,
+      "grad_norm": 8.424162864685059,
+      "learning_rate": 0.0,
+      "loss": 1.5715,
+      "step": 10860
+    },
+    {
+      "epoch": 2.5314392175128084,
+      "grad_norm": 15.89005184173584,
+      "learning_rate": 0.0,
+      "loss": 1.1113,
+      "step": 10870
+    },
+    {
+      "epoch": 2.533768048439683,
+      "grad_norm": 10.675698280334473,
+      "learning_rate": 0.0,
+      "loss": 1.5257,
+      "step": 10880
+    },
+    {
+      "epoch": 2.536096879366558,
+      "grad_norm": 36.62575912475586,
+      "learning_rate": 0.0,
+      "loss": 0.8961,
+      "step": 10890
+    },
+    {
+      "epoch": 2.5384257102934327,
+      "grad_norm": 5.308418273925781,
+      "learning_rate": 0.0,
+      "loss": 0.6685,
+      "step": 10900
+    },
+    {
+      "epoch": 2.5407545412203074,
+      "grad_norm": 29.720823287963867,
+      "learning_rate": 0.0,
+      "loss": 0.9164,
+      "step": 10910
+    },
+    {
+      "epoch": 2.543083372147182,
+      "grad_norm": 77.38935852050781,
+      "learning_rate": 0.0,
+      "loss": 2.2807,
+      "step": 10920
+    },
+    {
+      "epoch": 2.545412203074057,
+      "grad_norm": 86.90229034423828,
+      "learning_rate": 0.0,
+      "loss": 1.4119,
+      "step": 10930
+    },
+    {
+      "epoch": 2.5477410340009317,
+      "grad_norm": 7.217380046844482,
+      "learning_rate": 0.0,
+      "loss": 0.8429,
+      "step": 10940
+    },
+    {
+      "epoch": 2.5500698649278064,
+      "grad_norm": 15.611220359802246,
+      "learning_rate": 0.0,
+      "loss": 0.918,
+      "step": 10950
+    },
+    {
+      "epoch": 2.5523986958546807,
+      "grad_norm": 7.818775177001953,
+      "learning_rate": 0.0,
+      "loss": 1.1341,
+      "step": 10960
+    },
+    {
+      "epoch": 2.554727526781556,
+      "grad_norm": 10.395451545715332,
+      "learning_rate": 0.0,
+      "loss": 1.0161,
+      "step": 10970
+    },
+    {
+      "epoch": 2.55705635770843,
+      "grad_norm": 9.080485343933105,
+      "learning_rate": 0.0,
+      "loss": 0.9479,
+      "step": 10980
+    },
+    {
+      "epoch": 2.559385188635305,
+      "grad_norm": 19.17698860168457,
+      "learning_rate": 0.0,
+      "loss": 1.0788,
+      "step": 10990
+    },
+    {
+      "epoch": 2.5617140195621797,
+      "grad_norm": 9.98133659362793,
+      "learning_rate": 0.0,
+      "loss": 1.4046,
+      "step": 11000
+    },
+    {
+      "epoch": 2.5640428504890544,
+      "grad_norm": 12.91927433013916,
+      "learning_rate": 0.0,
+      "loss": 2.1021,
+      "step": 11010
+    },
+    {
+      "epoch": 2.566371681415929,
+      "grad_norm": 10.550889015197754,
+      "learning_rate": 0.0,
+      "loss": 1.4063,
+      "step": 11020
+    },
+    {
+      "epoch": 2.568700512342804,
+      "grad_norm": 20.967208862304688,
+      "learning_rate": 0.0,
+      "loss": 0.9094,
+      "step": 11030
+    },
+    {
+      "epoch": 2.5710293432696787,
+      "grad_norm": 9.758930206298828,
+      "learning_rate": 0.0,
+      "loss": 0.9526,
+      "step": 11040
+    },
+    {
+      "epoch": 2.5733581741965534,
+      "grad_norm": 10.604878425598145,
+      "learning_rate": 0.0,
+      "loss": 0.8984,
+      "step": 11050
+    },
+    {
+      "epoch": 2.575687005123428,
+      "grad_norm": 68.84051513671875,
+      "learning_rate": 0.0,
+      "loss": 1.647,
+      "step": 11060
+    },
+    {
+      "epoch": 2.5780158360503025,
+      "grad_norm": 25.205806732177734,
+      "learning_rate": 0.0,
+      "loss": 1.0534,
+      "step": 11070
+    },
+    {
+      "epoch": 2.5803446669771777,
+      "grad_norm": 31.481040954589844,
+      "learning_rate": 0.0,
+      "loss": 1.038,
+      "step": 11080
+    },
+    {
+      "epoch": 2.582673497904052,
+      "grad_norm": 8.334689140319824,
+      "learning_rate": 0.0,
+      "loss": 1.0556,
+      "step": 11090
+    },
+    {
+      "epoch": 2.5850023288309267,
+      "grad_norm": 12.273963928222656,
+      "learning_rate": 0.0,
+      "loss": 1.6612,
+      "step": 11100
+    },
+    {
+      "epoch": 2.5873311597578015,
+      "grad_norm": 29.150667190551758,
+      "learning_rate": 0.0,
+      "loss": 1.0504,
+      "step": 11110
+    },
+    {
+      "epoch": 2.5896599906846762,
+      "grad_norm": 11.735727310180664,
+      "learning_rate": 0.0,
+      "loss": 0.7439,
+      "step": 11120
+    },
+    {
+      "epoch": 2.591988821611551,
+      "grad_norm": 23.498836517333984,
+      "learning_rate": 0.0,
+      "loss": 1.6355,
+      "step": 11130
+    },
+    {
+      "epoch": 2.5943176525384257,
+      "grad_norm": 117.8069076538086,
+      "learning_rate": 0.0,
+      "loss": 1.2481,
+      "step": 11140
+    },
+    {
+      "epoch": 2.5966464834653005,
+      "grad_norm": 10.995659828186035,
+      "learning_rate": 0.0,
+      "loss": 1.3976,
+      "step": 11150
+    },
+    {
+      "epoch": 2.598975314392175,
+      "grad_norm": 5.801515102386475,
+      "learning_rate": 0.0,
+      "loss": 1.3652,
+      "step": 11160
+    },
+    {
+      "epoch": 2.60130414531905,
+      "grad_norm": 5.238818168640137,
+      "learning_rate": 0.0,
+      "loss": 1.6681,
+      "step": 11170
+    },
+    {
+      "epoch": 2.6036329762459243,
+      "grad_norm": 28.656795501708984,
+      "learning_rate": 0.0,
+      "loss": 0.8382,
+      "step": 11180
+    },
+    {
+      "epoch": 2.6059618071727995,
+      "grad_norm": 8.944382667541504,
+      "learning_rate": 0.0,
+      "loss": 1.1245,
+      "step": 11190
+    },
+    {
+      "epoch": 2.6082906380996738,
+      "grad_norm": 18.96076774597168,
+      "learning_rate": 0.0,
+      "loss": 1.0715,
+      "step": 11200
+    },
+    {
+      "epoch": 2.6106194690265485,
+      "grad_norm": 6.842470169067383,
+      "learning_rate": 0.0,
+      "loss": 1.1061,
+      "step": 11210
+    },
+    {
+      "epoch": 2.6129482999534233,
+      "grad_norm": 5.037587642669678,
+      "learning_rate": 0.0,
+      "loss": 0.8291,
+      "step": 11220
+    },
+    {
+      "epoch": 2.615277130880298,
+      "grad_norm": 33.31461715698242,
+      "learning_rate": 0.0,
+      "loss": 1.1437,
+      "step": 11230
+    },
+    {
+      "epoch": 2.6176059618071728,
+      "grad_norm": 5.52734375,
+      "learning_rate": 0.0,
+      "loss": 1.1537,
+      "step": 11240
+    },
+    {
+      "epoch": 2.6199347927340475,
+      "grad_norm": 5.281623363494873,
+      "learning_rate": 0.0,
+      "loss": 1.483,
+      "step": 11250
+    },
+    {
+      "epoch": 2.6222636236609222,
+      "grad_norm": 6.426329135894775,
+      "learning_rate": 0.0,
+      "loss": 1.2722,
+      "step": 11260
+    },
+    {
+      "epoch": 2.624592454587797,
+      "grad_norm": 8.326626777648926,
+      "learning_rate": 0.0,
+      "loss": 1.2723,
+      "step": 11270
+    },
+    {
+      "epoch": 2.6269212855146717,
+      "grad_norm": 10.69015121459961,
+      "learning_rate": 0.0,
+      "loss": 1.4028,
+      "step": 11280
+    },
+    {
+      "epoch": 2.6292501164415465,
+      "grad_norm": 73.84967803955078,
+      "learning_rate": 0.0,
+      "loss": 1.2928,
+      "step": 11290
+    },
+    {
+      "epoch": 2.6315789473684212,
+      "grad_norm": 8.578020095825195,
+      "learning_rate": 0.0,
+      "loss": 1.137,
+      "step": 11300
+    },
+    {
+      "epoch": 2.6339077782952955,
+      "grad_norm": 84.88838195800781,
+      "learning_rate": 0.0,
+      "loss": 1.819,
+      "step": 11310
+    },
+    {
+      "epoch": 2.6362366092221707,
+      "grad_norm": 10.036768913269043,
+      "learning_rate": 0.0,
+      "loss": 1.3546,
+      "step": 11320
+    },
+    {
+      "epoch": 2.638565440149045,
+      "grad_norm": 8.671600341796875,
+      "learning_rate": 0.0,
+      "loss": 1.06,
+      "step": 11330
+    },
+    {
+      "epoch": 2.64089427107592,
+      "grad_norm": 9.737736701965332,
+      "learning_rate": 0.0,
+      "loss": 1.3624,
+      "step": 11340
+    },
+    {
+      "epoch": 2.6432231020027945,
+      "grad_norm": 27.788164138793945,
+      "learning_rate": 0.0,
+      "loss": 1.4631,
+      "step": 11350
+    },
+    {
+      "epoch": 2.6455519329296693,
+      "grad_norm": 15.201400756835938,
+      "learning_rate": 0.0,
+      "loss": 0.9118,
+      "step": 11360
+    },
+    {
+      "epoch": 2.647880763856544,
+      "grad_norm": 23.824310302734375,
+      "learning_rate": 0.0,
+      "loss": 1.4152,
+      "step": 11370
+    },
+    {
+      "epoch": 2.6502095947834188,
+      "grad_norm": 14.977895736694336,
+      "learning_rate": 0.0,
+      "loss": 1.2898,
+      "step": 11380
+    },
+    {
+      "epoch": 2.6525384257102935,
+      "grad_norm": 12.554730415344238,
+      "learning_rate": 0.0,
+      "loss": 1.6473,
+      "step": 11390
+    },
+    {
+      "epoch": 2.6548672566371683,
+      "grad_norm": 11.867571830749512,
+      "learning_rate": 0.0,
+      "loss": 0.9147,
+      "step": 11400
+    },
+    {
+      "epoch": 2.657196087564043,
+      "grad_norm": 23.419448852539062,
+      "learning_rate": 0.0,
+      "loss": 1.6672,
+      "step": 11410
+    },
+    {
+      "epoch": 2.6595249184909173,
+      "grad_norm": 10.939473152160645,
+      "learning_rate": 0.0,
+      "loss": 1.1738,
+      "step": 11420
+    },
+    {
+      "epoch": 2.6618537494177925,
+      "grad_norm": 47.00886535644531,
+      "learning_rate": 0.0,
+      "loss": 1.4211,
+      "step": 11430
+    },
+    {
+      "epoch": 2.664182580344667,
+      "grad_norm": 9.351873397827148,
+      "learning_rate": 0.0,
+      "loss": 1.0582,
+      "step": 11440
+    },
+    {
+      "epoch": 2.6665114112715416,
+      "grad_norm": 19.46343994140625,
+      "learning_rate": 0.0,
+      "loss": 1.6338,
+      "step": 11450
+    },
+    {
+      "epoch": 2.6688402421984163,
+      "grad_norm": 28.46370506286621,
+      "learning_rate": 0.0,
+      "loss": 0.9037,
+      "step": 11460
+    },
+    {
+      "epoch": 2.671169073125291,
+      "grad_norm": 15.675468444824219,
+      "learning_rate": 0.0,
+      "loss": 1.2858,
+      "step": 11470
+    },
+    {
+      "epoch": 2.673497904052166,
+      "grad_norm": 38.96082305908203,
+      "learning_rate": 0.0,
+      "loss": 1.2323,
+      "step": 11480
+    },
+    {
+      "epoch": 2.6758267349790406,
+      "grad_norm": 77.10676574707031,
+      "learning_rate": 0.0,
+      "loss": 1.2334,
+      "step": 11490
+    },
+    {
+      "epoch": 2.6781555659059153,
+      "grad_norm": 10.799468994140625,
+      "learning_rate": 0.0,
+      "loss": 1.2866,
+      "step": 11500
+    },
+    {
+      "epoch": 2.68048439683279,
+      "grad_norm": 5.765465259552002,
+      "learning_rate": 0.0,
+      "loss": 0.9091,
+      "step": 11510
+    },
+    {
+      "epoch": 2.682813227759665,
+      "grad_norm": 17.54412078857422,
+      "learning_rate": 0.0,
+      "loss": 1.2162,
+      "step": 11520
+    },
+    {
+      "epoch": 2.685142058686539,
+      "grad_norm": 9.990829467773438,
+      "learning_rate": 0.0,
+      "loss": 0.9272,
+      "step": 11530
+    },
+    {
+      "epoch": 2.6874708896134143,
+      "grad_norm": 11.573451042175293,
+      "learning_rate": 0.0,
+      "loss": 1.3428,
+      "step": 11540
+    },
+    {
+      "epoch": 2.6897997205402886,
+      "grad_norm": 7.813601016998291,
+      "learning_rate": 0.0,
+      "loss": 1.3985,
+      "step": 11550
+    },
+    {
+      "epoch": 2.6921285514671633,
+      "grad_norm": 44.93555450439453,
+      "learning_rate": 0.0,
+      "loss": 1.5427,
+      "step": 11560
+    },
+    {
+      "epoch": 2.694457382394038,
+      "grad_norm": 66.35843658447266,
+      "learning_rate": 0.0,
+      "loss": 1.8103,
+      "step": 11570
+    },
+    {
+      "epoch": 2.696786213320913,
+      "grad_norm": 12.34382152557373,
+      "learning_rate": 0.0,
+      "loss": 1.2642,
+      "step": 11580
+    },
+    {
+      "epoch": 2.6991150442477876,
+      "grad_norm": 13.233556747436523,
+      "learning_rate": 0.0,
+      "loss": 0.9407,
+      "step": 11590
+    },
+    {
+      "epoch": 2.7014438751746623,
+      "grad_norm": 11.109960556030273,
+      "learning_rate": 0.0,
+      "loss": 1.0398,
+      "step": 11600
+    },
+    {
+      "epoch": 2.703772706101537,
+      "grad_norm": 18.077089309692383,
+      "learning_rate": 0.0,
+      "loss": 0.9129,
+      "step": 11610
+    },
+    {
+      "epoch": 2.706101537028412,
+      "grad_norm": 14.020467758178711,
+      "learning_rate": 0.0,
+      "loss": 1.0783,
+      "step": 11620
+    },
+    {
+      "epoch": 2.7084303679552866,
+      "grad_norm": 11.337139129638672,
+      "learning_rate": 0.0,
+      "loss": 1.5425,
+      "step": 11630
+    },
+    {
+      "epoch": 2.710759198882161,
+      "grad_norm": 20.209257125854492,
+      "learning_rate": 0.0,
+      "loss": 1.4743,
+      "step": 11640
+    },
+    {
+      "epoch": 2.713088029809036,
+      "grad_norm": 10.780107498168945,
+      "learning_rate": 0.0,
+      "loss": 1.8225,
+      "step": 11650
+    },
+    {
+      "epoch": 2.7154168607359104,
+      "grad_norm": 6.723607540130615,
+      "learning_rate": 0.0,
+      "loss": 1.4223,
+      "step": 11660
+    },
+    {
+      "epoch": 2.717745691662785,
+      "grad_norm": 28.61166000366211,
+      "learning_rate": 0.0,
+      "loss": 1.0589,
+      "step": 11670
+    },
+    {
+      "epoch": 2.72007452258966,
+      "grad_norm": 11.463347434997559,
+      "learning_rate": 0.0,
+      "loss": 1.1931,
+      "step": 11680
+    },
+    {
+      "epoch": 2.7224033535165346,
+      "grad_norm": 14.94326400756836,
+      "learning_rate": 0.0,
+      "loss": 1.0032,
+      "step": 11690
+    },
+    {
+      "epoch": 2.7247321844434094,
+      "grad_norm": 13.956268310546875,
+      "learning_rate": 0.0,
+      "loss": 0.8667,
+      "step": 11700
+    },
+    {
+      "epoch": 2.727061015370284,
+      "grad_norm": 50.78772735595703,
+      "learning_rate": 0.0,
+      "loss": 1.3266,
+      "step": 11710
+    },
+    {
+      "epoch": 2.729389846297159,
+      "grad_norm": 3.0896995067596436,
+      "learning_rate": 0.0,
+      "loss": 0.7148,
+      "step": 11720
+    },
+    {
+      "epoch": 2.7317186772240336,
+      "grad_norm": 27.836071014404297,
+      "learning_rate": 0.0,
+      "loss": 1.3137,
+      "step": 11730
+    },
+    {
+      "epoch": 2.7340475081509084,
+      "grad_norm": 7.0214715003967285,
+      "learning_rate": 0.0,
+      "loss": 0.9149,
+      "step": 11740
+    },
+    {
+      "epoch": 2.736376339077783,
+      "grad_norm": 21.97905158996582,
+      "learning_rate": 0.0,
+      "loss": 1.5909,
+      "step": 11750
+    },
+    {
+      "epoch": 2.738705170004658,
+      "grad_norm": 6.0453338623046875,
+      "learning_rate": 0.0,
+      "loss": 1.1069,
+      "step": 11760
+    },
+    {
+      "epoch": 2.741034000931532,
+      "grad_norm": 15.303627967834473,
+      "learning_rate": 0.0,
+      "loss": 1.1191,
+      "step": 11770
+    },
+    {
+      "epoch": 2.7433628318584073,
+      "grad_norm": 25.093414306640625,
+      "learning_rate": 0.0,
+      "loss": 1.4118,
+      "step": 11780
+    },
+    {
+      "epoch": 2.7456916627852817,
+      "grad_norm": 20.566513061523438,
+      "learning_rate": 0.0,
+      "loss": 1.7382,
+      "step": 11790
+    },
+    {
+      "epoch": 2.7480204937121564,
+      "grad_norm": 44.23474884033203,
+      "learning_rate": 0.0,
+      "loss": 1.36,
+      "step": 11800
+    },
+    {
+      "epoch": 2.750349324639031,
+      "grad_norm": 8.558969497680664,
+      "learning_rate": 0.0,
+      "loss": 1.4381,
+      "step": 11810
+    },
+    {
+      "epoch": 2.752678155565906,
+      "grad_norm": 15.822858810424805,
+      "learning_rate": 0.0,
+      "loss": 1.1273,
+      "step": 11820
+    },
+    {
+      "epoch": 2.7550069864927806,
+      "grad_norm": 7.357201099395752,
+      "learning_rate": 0.0,
+      "loss": 1.1263,
+      "step": 11830
+    },
+    {
+      "epoch": 2.7573358174196554,
+      "grad_norm": 74.23464965820312,
+      "learning_rate": 0.0,
+      "loss": 1.187,
+      "step": 11840
+    },
+    {
+      "epoch": 2.75966464834653,
+      "grad_norm": 23.02132797241211,
+      "learning_rate": 0.0,
+      "loss": 1.0388,
+      "step": 11850
+    },
+    {
+      "epoch": 2.761993479273405,
+      "grad_norm": 8.519207954406738,
+      "learning_rate": 0.0,
+      "loss": 0.9178,
+      "step": 11860
+    },
+    {
+      "epoch": 2.7643223102002796,
+      "grad_norm": 8.386043548583984,
+      "learning_rate": 0.0,
+      "loss": 1.3762,
+      "step": 11870
+    },
+    {
+      "epoch": 2.766651141127154,
+      "grad_norm": 12.414632797241211,
+      "learning_rate": 0.0,
+      "loss": 2.2633,
+      "step": 11880
+    },
+    {
+      "epoch": 2.768979972054029,
+      "grad_norm": 10.087637901306152,
+      "learning_rate": 0.0,
+      "loss": 1.366,
+      "step": 11890
+    },
+    {
+      "epoch": 2.7713088029809034,
+      "grad_norm": 93.13560485839844,
+      "learning_rate": 0.0,
+      "loss": 1.6345,
+      "step": 11900
+    },
+    {
+      "epoch": 2.773637633907778,
+      "grad_norm": 51.69975280761719,
+      "learning_rate": 0.0,
+      "loss": 1.0775,
+      "step": 11910
+    },
+    {
+      "epoch": 2.775966464834653,
+      "grad_norm": 47.3151741027832,
+      "learning_rate": 0.0,
+      "loss": 1.3134,
+      "step": 11920
+    },
+    {
+      "epoch": 2.7782952957615277,
+      "grad_norm": 16.13167381286621,
+      "learning_rate": 0.0,
+      "loss": 1.72,
+      "step": 11930
+    },
+    {
+      "epoch": 2.7806241266884024,
+      "grad_norm": 60.05651092529297,
+      "learning_rate": 0.0,
+      "loss": 1.5407,
+      "step": 11940
+    },
+    {
+      "epoch": 2.782952957615277,
+      "grad_norm": 7.127047538757324,
+      "learning_rate": 0.0,
+      "loss": 0.826,
+      "step": 11950
+    },
+    {
+      "epoch": 2.785281788542152,
+      "grad_norm": 8.19342041015625,
+      "learning_rate": 0.0,
+      "loss": 1.2436,
+      "step": 11960
+    },
+    {
+      "epoch": 2.7876106194690267,
+      "grad_norm": 17.038658142089844,
+      "learning_rate": 0.0,
+      "loss": 1.3329,
+      "step": 11970
+    },
+    {
+      "epoch": 2.7899394503959014,
+      "grad_norm": 45.72443771362305,
+      "learning_rate": 0.0,
+      "loss": 1.092,
+      "step": 11980
+    },
+    {
+      "epoch": 2.7922682813227757,
+      "grad_norm": 8.771135330200195,
+      "learning_rate": 0.0,
+      "loss": 1.1903,
+      "step": 11990
+    },
+    {
+      "epoch": 2.794597112249651,
+      "grad_norm": 8.184369087219238,
+      "learning_rate": 0.0,
+      "loss": 1.4032,
+      "step": 12000
+    },
+    {
+      "epoch": 2.796925943176525,
+      "grad_norm": 22.36016082763672,
+      "learning_rate": 0.0,
+      "loss": 1.0558,
+      "step": 12010
+    },
+    {
+      "epoch": 2.7992547741034,
+      "grad_norm": 6.70443058013916,
+      "learning_rate": 0.0,
+      "loss": 0.9546,
+      "step": 12020
+    },
+    {
+      "epoch": 2.8015836050302747,
+      "grad_norm": 10.176080703735352,
+      "learning_rate": 0.0,
+      "loss": 1.1962,
+      "step": 12030
+    },
+    {
+      "epoch": 2.8039124359571495,
+      "grad_norm": 17.690227508544922,
+      "learning_rate": 0.0,
+      "loss": 1.1685,
+      "step": 12040
+    },
+    {
+      "epoch": 2.806241266884024,
+      "grad_norm": 17.06946563720703,
+      "learning_rate": 0.0,
+      "loss": 1.1656,
+      "step": 12050
+    },
+    {
+      "epoch": 2.808570097810899,
+      "grad_norm": 6.972208023071289,
+      "learning_rate": 0.0,
+      "loss": 1.434,
+      "step": 12060
+    },
+    {
+      "epoch": 2.8108989287377737,
+      "grad_norm": 17.20701026916504,
+      "learning_rate": 0.0,
+      "loss": 1.1218,
+      "step": 12070
+    },
+    {
+      "epoch": 2.8132277596646484,
+      "grad_norm": 9.323033332824707,
+      "learning_rate": 0.0,
+      "loss": 1.1368,
+      "step": 12080
+    },
+    {
+      "epoch": 2.815556590591523,
+      "grad_norm": 26.10862159729004,
+      "learning_rate": 0.0,
+      "loss": 1.1541,
+      "step": 12090
+    },
+    {
+      "epoch": 2.8178854215183975,
+      "grad_norm": 12.300597190856934,
+      "learning_rate": 0.0,
+      "loss": 1.3575,
+      "step": 12100
+    },
+    {
+      "epoch": 2.8202142524452727,
+      "grad_norm": 7.925255298614502,
+      "learning_rate": 0.0,
+      "loss": 2.2617,
+      "step": 12110
+    },
+    {
+      "epoch": 2.822543083372147,
+      "grad_norm": 5.388205528259277,
+      "learning_rate": 0.0,
+      "loss": 0.9115,
+      "step": 12120
+    },
+    {
+      "epoch": 2.8248719142990217,
+      "grad_norm": 9.119213104248047,
+      "learning_rate": 0.0,
+      "loss": 1.5579,
+      "step": 12130
+    },
+    {
+      "epoch": 2.8272007452258965,
+      "grad_norm": 127.05992126464844,
+      "learning_rate": 0.0,
+      "loss": 1.0463,
+      "step": 12140
+    },
+    {
+      "epoch": 2.8295295761527712,
+      "grad_norm": 4.77928352355957,
+      "learning_rate": 0.0,
+      "loss": 0.8053,
+      "step": 12150
+    },
+    {
+      "epoch": 2.831858407079646,
+      "grad_norm": 10.260478973388672,
+      "learning_rate": 0.0,
+      "loss": 1.3177,
+      "step": 12160
+    },
+    {
+      "epoch": 2.8341872380065207,
+      "grad_norm": 6.844449996948242,
+      "learning_rate": 0.0,
+      "loss": 1.3177,
+      "step": 12170
+    },
+    {
+      "epoch": 2.8365160689333955,
+      "grad_norm": 65.62310791015625,
+      "learning_rate": 0.0,
+      "loss": 1.1177,
+      "step": 12180
+    },
+    {
+      "epoch": 2.83884489986027,
+      "grad_norm": 19.49369239807129,
+      "learning_rate": 0.0,
+      "loss": 1.2459,
+      "step": 12190
+    },
+    {
+      "epoch": 2.841173730787145,
+      "grad_norm": 17.212875366210938,
+      "learning_rate": 0.0,
+      "loss": 1.4008,
+      "step": 12200
+    },
+    {
+      "epoch": 2.8435025617140197,
+      "grad_norm": 31.808021545410156,
+      "learning_rate": 0.0,
+      "loss": 1.2806,
+      "step": 12210
+    },
+    {
+      "epoch": 2.8458313926408945,
+      "grad_norm": 9.024837493896484,
+      "learning_rate": 0.0,
+      "loss": 1.629,
+      "step": 12220
+    },
+    {
+      "epoch": 2.8481602235677688,
+      "grad_norm": 26.87761116027832,
+      "learning_rate": 0.0,
+      "loss": 1.2044,
+      "step": 12230
+    },
+    {
+      "epoch": 2.850489054494644,
+      "grad_norm": 19.435070037841797,
+      "learning_rate": 0.0,
+      "loss": 1.1629,
+      "step": 12240
+    },
+    {
+      "epoch": 2.8528178854215183,
+      "grad_norm": 15.5372953414917,
+      "learning_rate": 0.0,
+      "loss": 1.684,
+      "step": 12250
+    },
+    {
+      "epoch": 2.855146716348393,
+      "grad_norm": 4.47623872756958,
+      "learning_rate": 0.0,
+      "loss": 1.0723,
+      "step": 12260
+    },
+    {
+      "epoch": 2.8574755472752678,
+      "grad_norm": 75.65230560302734,
+      "learning_rate": 0.0,
+      "loss": 1.2155,
+      "step": 12270
+    },
+    {
+      "epoch": 2.8598043782021425,
+      "grad_norm": 8.398263931274414,
+      "learning_rate": 0.0,
+      "loss": 1.2201,
+      "step": 12280
+    },
+    {
+      "epoch": 2.8621332091290173,
+      "grad_norm": 9.25592041015625,
+      "learning_rate": 0.0,
+      "loss": 1.1723,
+      "step": 12290
+    },
+    {
+      "epoch": 2.864462040055892,
+      "grad_norm": 10.627755165100098,
+      "learning_rate": 0.0,
+      "loss": 1.167,
+      "step": 12300
+    },
+    {
+      "epoch": 2.8667908709827667,
+      "grad_norm": 13.692951202392578,
+      "learning_rate": 0.0,
+      "loss": 1.0071,
+      "step": 12310
+    },
+    {
+      "epoch": 2.8691197019096415,
+      "grad_norm": 11.026177406311035,
+      "learning_rate": 0.0,
+      "loss": 1.0744,
+      "step": 12320
+    },
+    {
+      "epoch": 2.8714485328365162,
+      "grad_norm": 5.878359317779541,
+      "learning_rate": 0.0,
+      "loss": 1.3973,
+      "step": 12330
+    },
+    {
+      "epoch": 2.8737773637633905,
+      "grad_norm": 41.004493713378906,
+      "learning_rate": 0.0,
+      "loss": 1.2759,
+      "step": 12340
+    },
+    {
+      "epoch": 2.8761061946902657,
+      "grad_norm": 11.279379844665527,
+      "learning_rate": 0.0,
+      "loss": 1.3207,
+      "step": 12350
+    },
+    {
+      "epoch": 2.87843502561714,
+      "grad_norm": 8.147149085998535,
+      "learning_rate": 0.0,
+      "loss": 1.1631,
+      "step": 12360
+    },
+    {
+      "epoch": 2.880763856544015,
+      "grad_norm": 10.414231300354004,
+      "learning_rate": 0.0,
+      "loss": 1.0849,
+      "step": 12370
+    },
+    {
+      "epoch": 2.8830926874708895,
+      "grad_norm": 16.238142013549805,
+      "learning_rate": 0.0,
+      "loss": 0.9335,
+      "step": 12380
+    },
+    {
+      "epoch": 2.8854215183977643,
+      "grad_norm": 12.68599796295166,
+      "learning_rate": 0.0,
+      "loss": 1.0928,
+      "step": 12390
+    },
+    {
+      "epoch": 2.887750349324639,
+      "grad_norm": 33.59490966796875,
+      "learning_rate": 0.0,
+      "loss": 1.1499,
+      "step": 12400
+    },
+    {
+      "epoch": 2.890079180251514,
+      "grad_norm": 4.714087963104248,
+      "learning_rate": 0.0,
+      "loss": 1.404,
+      "step": 12410
+    },
+    {
+      "epoch": 2.8924080111783885,
+      "grad_norm": 107.86946105957031,
+      "learning_rate": 0.0,
+      "loss": 1.5526,
+      "step": 12420
+    },
+    {
+      "epoch": 2.8947368421052633,
+      "grad_norm": 6.616478443145752,
+      "learning_rate": 0.0,
+      "loss": 1.0008,
+      "step": 12430
+    },
+    {
+      "epoch": 2.897065673032138,
+      "grad_norm": 10.684284210205078,
+      "learning_rate": 0.0,
+      "loss": 1.5301,
+      "step": 12440
+    },
+    {
+      "epoch": 2.8993945039590123,
+      "grad_norm": 41.24277114868164,
+      "learning_rate": 0.0,
+      "loss": 1.2296,
+      "step": 12450
+    },
+    {
+      "epoch": 2.9017233348858875,
+      "grad_norm": 72.13574981689453,
+      "learning_rate": 0.0,
+      "loss": 1.2495,
+      "step": 12460
+    },
+    {
+      "epoch": 2.904052165812762,
+      "grad_norm": 11.254340171813965,
+      "learning_rate": 0.0,
+      "loss": 1.2704,
+      "step": 12470
+    },
+    {
+      "epoch": 2.9063809967396366,
+      "grad_norm": 9.927990913391113,
+      "learning_rate": 0.0,
+      "loss": 1.4381,
+      "step": 12480
+    },
+    {
+      "epoch": 2.9087098276665113,
+      "grad_norm": 8.379936218261719,
+      "learning_rate": 0.0,
+      "loss": 1.3127,
+      "step": 12490
+    },
+    {
+      "epoch": 2.911038658593386,
+      "grad_norm": 6.6258955001831055,
+      "learning_rate": 0.0,
+      "loss": 1.0506,
+      "step": 12500
+    },
+    {
+      "epoch": 2.913367489520261,
+      "grad_norm": 13.267959594726562,
+      "learning_rate": 0.0,
+      "loss": 1.4729,
+      "step": 12510
+    },
+    {
+      "epoch": 2.9156963204471356,
+      "grad_norm": 14.169622421264648,
+      "learning_rate": 0.0,
+      "loss": 1.485,
+      "step": 12520
+    },
+    {
+      "epoch": 2.9180251513740103,
+      "grad_norm": 14.324849128723145,
+      "learning_rate": 0.0,
+      "loss": 1.3262,
+      "step": 12530
+    },
+    {
+      "epoch": 2.920353982300885,
+      "grad_norm": 139.77540588378906,
+      "learning_rate": 0.0,
+      "loss": 1.1871,
+      "step": 12540
+    },
+    {
+      "epoch": 2.92268281322776,
+      "grad_norm": 19.438474655151367,
+      "learning_rate": 0.0,
+      "loss": 1.0036,
+      "step": 12550
+    },
+    {
+      "epoch": 2.925011644154634,
+      "grad_norm": 5.162458419799805,
+      "learning_rate": 0.0,
+      "loss": 1.3947,
+      "step": 12560
+    },
+    {
+      "epoch": 2.9273404750815093,
+      "grad_norm": 9.66451644897461,
+      "learning_rate": 0.0,
+      "loss": 1.1011,
+      "step": 12570
+    },
+    {
+      "epoch": 2.9296693060083836,
+      "grad_norm": 10.375886917114258,
+      "learning_rate": 0.0,
+      "loss": 1.0098,
+      "step": 12580
+    },
+    {
+      "epoch": 2.9319981369352583,
+      "grad_norm": 13.053080558776855,
+      "learning_rate": 0.0,
+      "loss": 1.1087,
+      "step": 12590
+    },
+    {
+      "epoch": 2.934326967862133,
+      "grad_norm": 29.2081298828125,
+      "learning_rate": 0.0,
+      "loss": 1.0568,
+      "step": 12600
+    },
+    {
+      "epoch": 2.936655798789008,
+      "grad_norm": 8.402816772460938,
+      "learning_rate": 0.0,
+      "loss": 1.3496,
+      "step": 12610
+    },
+    {
+      "epoch": 2.9389846297158826,
+      "grad_norm": 12.343842506408691,
+      "learning_rate": 0.0,
+      "loss": 1.1508,
+      "step": 12620
+    },
+    {
+      "epoch": 2.9413134606427573,
+      "grad_norm": 6.9314494132995605,
+      "learning_rate": 0.0,
+      "loss": 1.6717,
+      "step": 12630
+    },
+    {
+      "epoch": 2.943642291569632,
+      "grad_norm": 18.752248764038086,
+      "learning_rate": 0.0,
+      "loss": 1.0485,
+      "step": 12640
+    },
+    {
+      "epoch": 2.945971122496507,
+      "grad_norm": 27.50632667541504,
+      "learning_rate": 0.0,
+      "loss": 1.3482,
+      "step": 12650
+    },
+    {
+      "epoch": 2.9482999534233816,
+      "grad_norm": 9.300987243652344,
+      "learning_rate": 0.0,
+      "loss": 1.2977,
+      "step": 12660
+    },
+    {
+      "epoch": 2.9506287843502563,
+      "grad_norm": 9.483819961547852,
+      "learning_rate": 0.0,
+      "loss": 1.2739,
+      "step": 12670
+    },
+    {
+      "epoch": 2.952957615277131,
+      "grad_norm": 14.791114807128906,
+      "learning_rate": 0.0,
+      "loss": 1.9703,
+      "step": 12680
+    },
+    {
+      "epoch": 2.9552864462040054,
+      "grad_norm": 14.320032119750977,
+      "learning_rate": 0.0,
+      "loss": 1.1498,
+      "step": 12690
+    },
+    {
+      "epoch": 2.9576152771308806,
+      "grad_norm": 10.36143684387207,
+      "learning_rate": 0.0,
+      "loss": 1.3839,
+      "step": 12700
+    },
+    {
+      "epoch": 2.959944108057755,
+      "grad_norm": 4.280786037445068,
+      "learning_rate": 0.0,
+      "loss": 1.1626,
+      "step": 12710
+    },
+    {
+      "epoch": 2.9622729389846296,
+      "grad_norm": 35.886634826660156,
+      "learning_rate": 0.0,
+      "loss": 1.6616,
+      "step": 12720
+    },
+    {
+      "epoch": 2.9646017699115044,
+      "grad_norm": 17.06854248046875,
+      "learning_rate": 0.0,
+      "loss": 1.0649,
+      "step": 12730
+    },
+    {
+      "epoch": 2.966930600838379,
+      "grad_norm": 11.25122356414795,
+      "learning_rate": 0.0,
+      "loss": 0.9763,
+      "step": 12740
+    },
+    {
+      "epoch": 2.969259431765254,
+      "grad_norm": 11.460931777954102,
+      "learning_rate": 0.0,
+      "loss": 1.4561,
+      "step": 12750
+    },
+    {
+      "epoch": 2.9715882626921286,
+      "grad_norm": 26.476194381713867,
+      "learning_rate": 0.0,
+      "loss": 1.814,
+      "step": 12760
+    },
+    {
+      "epoch": 2.9739170936190034,
+      "grad_norm": 19.238082885742188,
+      "learning_rate": 0.0,
+      "loss": 1.1605,
+      "step": 12770
+    },
+    {
+      "epoch": 2.976245924545878,
+      "grad_norm": 44.26625061035156,
+      "learning_rate": 0.0,
+      "loss": 1.1757,
+      "step": 12780
+    },
+    {
+      "epoch": 2.978574755472753,
+      "grad_norm": 98.12911224365234,
+      "learning_rate": 0.0,
+      "loss": 1.7095,
+      "step": 12790
+    },
+    {
+      "epoch": 2.980903586399627,
+      "grad_norm": 14.261837005615234,
+      "learning_rate": 0.0,
+      "loss": 1.3246,
+      "step": 12800
+    },
+    {
+      "epoch": 2.9832324173265024,
+      "grad_norm": 6.469521522521973,
+      "learning_rate": 0.0,
+      "loss": 1.2186,
+      "step": 12810
+    },
+    {
+      "epoch": 2.9855612482533767,
+      "grad_norm": 21.546524047851562,
+      "learning_rate": 0.0,
+      "loss": 1.3713,
+      "step": 12820
+    },
+    {
+      "epoch": 2.9878900791802514,
+      "grad_norm": 15.523602485656738,
+      "learning_rate": 0.0,
+      "loss": 1.46,
+      "step": 12830
+    },
+    {
+      "epoch": 2.990218910107126,
+      "grad_norm": 22.233688354492188,
+      "learning_rate": 0.0,
+      "loss": 1.3171,
+      "step": 12840
+    },
+    {
+      "epoch": 2.992547741034001,
+      "grad_norm": 10.238198280334473,
+      "learning_rate": 0.0,
+      "loss": 1.1856,
+      "step": 12850
+    },
+    {
+      "epoch": 2.9948765719608756,
+      "grad_norm": 32.13593292236328,
+      "learning_rate": 0.0,
+      "loss": 1.3003,
+      "step": 12860
+    },
+    {
+      "epoch": 2.9972054028877504,
+      "grad_norm": 13.810074806213379,
+      "learning_rate": 0.0,
+      "loss": 0.9724,
+      "step": 12870
+    },
+    {
+      "epoch": 2.999534233814625,
+      "grad_norm": 9.147926330566406,
+      "learning_rate": 0.0,
+      "loss": 1.052,
+      "step": 12880
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 0.49088114500045776,
+      "eval_runtime": 15.2585,
+      "eval_samples_per_second": 285.48,
+      "eval_steps_per_second": 2.294,
+      "step": 12882
+    },
+    {
+      "epoch": 3.0018630647415,
+      "grad_norm": 21.883949279785156,
+      "learning_rate": 0.0,
+      "loss": 1.7736,
+      "step": 12890
+    },
+    {
+      "epoch": 3.0041918956683746,
+      "grad_norm": 7.900768280029297,
+      "learning_rate": 0.0,
+      "loss": 1.0961,
+      "step": 12900
+    },
+    {
+      "epoch": 3.0065207265952494,
+      "grad_norm": 70.86416625976562,
+      "learning_rate": 0.0,
+      "loss": 0.9499,
+      "step": 12910
+    },
+    {
+      "epoch": 3.0088495575221237,
+      "grad_norm": 9.933052062988281,
+      "learning_rate": 0.0,
+      "loss": 0.9614,
+      "step": 12920
+    },
+    {
+      "epoch": 3.0111783884489984,
+      "grad_norm": 15.00586986541748,
+      "learning_rate": 0.0,
+      "loss": 1.4427,
+      "step": 12930
+    },
+    {
+      "epoch": 3.013507219375873,
+      "grad_norm": 31.774045944213867,
+      "learning_rate": 0.0,
+      "loss": 1.567,
+      "step": 12940
+    },
+    {
+      "epoch": 3.015836050302748,
+      "grad_norm": 6.541329860687256,
+      "learning_rate": 0.0,
+      "loss": 1.0219,
+      "step": 12950
+    },
+    {
+      "epoch": 3.0181648812296227,
+      "grad_norm": 9.174057006835938,
+      "learning_rate": 0.0,
+      "loss": 1.3925,
+      "step": 12960
+    },
+    {
+      "epoch": 3.0204937121564974,
+      "grad_norm": 5.683773517608643,
+      "learning_rate": 0.0,
+      "loss": 1.072,
+      "step": 12970
+    },
+    {
+      "epoch": 3.022822543083372,
+      "grad_norm": 10.569469451904297,
+      "learning_rate": 0.0,
+      "loss": 1.3213,
+      "step": 12980
+    },
+    {
+      "epoch": 3.025151374010247,
+      "grad_norm": 11.428179740905762,
+      "learning_rate": 0.0,
+      "loss": 1.0445,
+      "step": 12990
+    },
+    {
+      "epoch": 3.0274802049371217,
+      "grad_norm": 5.285568714141846,
+      "learning_rate": 0.0,
+      "loss": 1.3264,
+      "step": 13000
+    },
+    {
+      "epoch": 3.0298090358639964,
+      "grad_norm": 24.80561065673828,
+      "learning_rate": 0.0,
+      "loss": 1.1625,
+      "step": 13010
+    },
+    {
+      "epoch": 3.032137866790871,
+      "grad_norm": 61.148956298828125,
+      "learning_rate": 0.0,
+      "loss": 1.0419,
+      "step": 13020
+    },
+    {
+      "epoch": 3.034466697717746,
+      "grad_norm": 20.153419494628906,
+      "learning_rate": 0.0,
+      "loss": 1.4652,
+      "step": 13030
+    },
+    {
+      "epoch": 3.03679552864462,
+      "grad_norm": 6.97303581237793,
+      "learning_rate": 0.0,
+      "loss": 1.2172,
+      "step": 13040
+    },
+    {
+      "epoch": 3.039124359571495,
+      "grad_norm": 8.286640167236328,
+      "learning_rate": 0.0,
+      "loss": 0.8866,
+      "step": 13050
+    },
+    {
+      "epoch": 3.0414531904983697,
+      "grad_norm": 6.162503242492676,
+      "learning_rate": 0.0,
+      "loss": 0.9947,
+      "step": 13060
+    },
+    {
+      "epoch": 3.0437820214252445,
+      "grad_norm": 32.659934997558594,
+      "learning_rate": 0.0,
+      "loss": 1.163,
+      "step": 13070
+    },
+    {
+      "epoch": 3.046110852352119,
+      "grad_norm": 24.297115325927734,
+      "learning_rate": 0.0,
+      "loss": 1.4345,
+      "step": 13080
+    },
+    {
+      "epoch": 3.048439683278994,
+      "grad_norm": 15.972943305969238,
+      "learning_rate": 0.0,
+      "loss": 0.8504,
+      "step": 13090
+    },
+    {
+      "epoch": 3.0507685142058687,
+      "grad_norm": 62.08342742919922,
+      "learning_rate": 0.0,
+      "loss": 1.1412,
+      "step": 13100
+    },
+    {
+      "epoch": 3.0530973451327434,
+      "grad_norm": 5.202879905700684,
+      "learning_rate": 0.0,
+      "loss": 1.2152,
+      "step": 13110
+    },
+    {
+      "epoch": 3.055426176059618,
+      "grad_norm": 5.640526294708252,
+      "learning_rate": 0.0,
+      "loss": 1.5043,
+      "step": 13120
+    },
+    {
+      "epoch": 3.057755006986493,
+      "grad_norm": 40.53738021850586,
+      "learning_rate": 0.0,
+      "loss": 1.3858,
+      "step": 13130
+    },
+    {
+      "epoch": 3.0600838379133677,
+      "grad_norm": 4.845460414886475,
+      "learning_rate": 0.0,
+      "loss": 1.0233,
+      "step": 13140
+    },
+    {
+      "epoch": 3.062412668840242,
+      "grad_norm": 4.9804158210754395,
+      "learning_rate": 0.0,
+      "loss": 1.4265,
+      "step": 13150
+    },
+    {
+      "epoch": 3.0647414997671167,
+      "grad_norm": 4.985668659210205,
+      "learning_rate": 0.0,
+      "loss": 1.4086,
+      "step": 13160
+    },
+    {
+      "epoch": 3.0670703306939915,
+      "grad_norm": 19.393199920654297,
+      "learning_rate": 0.0,
+      "loss": 0.9075,
+      "step": 13170
+    },
+    {
+      "epoch": 3.0693991616208662,
+      "grad_norm": 16.305225372314453,
+      "learning_rate": 0.0,
+      "loss": 1.5215,
+      "step": 13180
+    },
+    {
+      "epoch": 3.071727992547741,
+      "grad_norm": 4.610240936279297,
+      "learning_rate": 0.0,
+      "loss": 0.9681,
+      "step": 13190
+    },
+    {
+      "epoch": 3.0740568234746157,
+      "grad_norm": 3.354733467102051,
+      "learning_rate": 0.0,
+      "loss": 1.1587,
+      "step": 13200
+    },
+    {
+      "epoch": 3.0763856544014905,
+      "grad_norm": 4.949262619018555,
+      "learning_rate": 0.0,
+      "loss": 0.9106,
+      "step": 13210
+    },
+    {
+      "epoch": 3.0787144853283652,
+      "grad_norm": 12.779787063598633,
+      "learning_rate": 0.0,
+      "loss": 0.9381,
+      "step": 13220
+    },
+    {
+      "epoch": 3.08104331625524,
+      "grad_norm": 5.218024253845215,
+      "learning_rate": 0.0,
+      "loss": 0.8552,
+      "step": 13230
+    },
+    {
+      "epoch": 3.0833721471821147,
+      "grad_norm": 32.33806228637695,
+      "learning_rate": 0.0,
+      "loss": 1.4397,
+      "step": 13240
+    },
+    {
+      "epoch": 3.0857009781089895,
+      "grad_norm": 11.891190528869629,
+      "learning_rate": 0.0,
+      "loss": 1.3937,
+      "step": 13250
+    },
+    {
+      "epoch": 3.0880298090358638,
+      "grad_norm": 9.155755996704102,
+      "learning_rate": 0.0,
+      "loss": 1.0606,
+      "step": 13260
+    },
+    {
+      "epoch": 3.0903586399627385,
+      "grad_norm": 13.458362579345703,
+      "learning_rate": 0.0,
+      "loss": 1.6445,
+      "step": 13270
+    },
+    {
+      "epoch": 3.0926874708896133,
+      "grad_norm": 6.68931770324707,
+      "learning_rate": 0.0,
+      "loss": 1.1771,
+      "step": 13280
+    },
+    {
+      "epoch": 3.095016301816488,
+      "grad_norm": 19.457380294799805,
+      "learning_rate": 0.0,
+      "loss": 1.6591,
+      "step": 13290
+    },
+    {
+      "epoch": 3.0973451327433628,
+      "grad_norm": 7.39839506149292,
+      "learning_rate": 0.0,
+      "loss": 1.587,
+      "step": 13300
+    },
+    {
+      "epoch": 3.0996739636702375,
+      "grad_norm": 11.434883117675781,
+      "learning_rate": 0.0,
+      "loss": 1.0876,
+      "step": 13310
+    },
+    {
+      "epoch": 3.1020027945971123,
+      "grad_norm": 42.85976028442383,
+      "learning_rate": 0.0,
+      "loss": 1.3607,
+      "step": 13320
+    },
+    {
+      "epoch": 3.104331625523987,
+      "grad_norm": 7.510742664337158,
+      "learning_rate": 0.0,
+      "loss": 1.5981,
+      "step": 13330
+    },
+    {
+      "epoch": 3.1066604564508618,
+      "grad_norm": 9.284518241882324,
+      "learning_rate": 0.0,
+      "loss": 1.1812,
+      "step": 13340
+    },
+    {
+      "epoch": 3.1089892873777365,
+      "grad_norm": 8.393990516662598,
+      "learning_rate": 0.0,
+      "loss": 1.3099,
+      "step": 13350
+    },
+    {
+      "epoch": 3.1113181183046112,
+      "grad_norm": 8.153864860534668,
+      "learning_rate": 0.0,
+      "loss": 0.8854,
+      "step": 13360
+    },
+    {
+      "epoch": 3.113646949231486,
+      "grad_norm": 15.750825881958008,
+      "learning_rate": 0.0,
+      "loss": 1.7133,
+      "step": 13370
+    },
+    {
+      "epoch": 3.1159757801583603,
+      "grad_norm": 52.85032272338867,
+      "learning_rate": 0.0,
+      "loss": 1.0101,
+      "step": 13380
+    },
+    {
+      "epoch": 3.118304611085235,
+      "grad_norm": 6.499885559082031,
+      "learning_rate": 0.0,
+      "loss": 1.5199,
+      "step": 13390
+    },
+    {
+      "epoch": 3.12063344201211,
+      "grad_norm": 63.261356353759766,
+      "learning_rate": 0.0,
+      "loss": 1.3876,
+      "step": 13400
+    },
+    {
+      "epoch": 3.1229622729389845,
+      "grad_norm": 14.717015266418457,
+      "learning_rate": 0.0,
+      "loss": 1.735,
+      "step": 13410
+    },
+    {
+      "epoch": 3.1252911038658593,
+      "grad_norm": 5.445131778717041,
+      "learning_rate": 0.0,
+      "loss": 1.337,
+      "step": 13420
+    },
+    {
+      "epoch": 3.127619934792734,
+      "grad_norm": 35.080650329589844,
+      "learning_rate": 0.0,
+      "loss": 1.2368,
+      "step": 13430
+    },
+    {
+      "epoch": 3.129948765719609,
+      "grad_norm": 9.115435600280762,
+      "learning_rate": 0.0,
+      "loss": 0.7256,
+      "step": 13440
+    },
+    {
+      "epoch": 3.1322775966464835,
+      "grad_norm": 5.406123638153076,
+      "learning_rate": 0.0,
+      "loss": 0.8733,
+      "step": 13450
+    },
+    {
+      "epoch": 3.1346064275733583,
+      "grad_norm": 5.928583145141602,
+      "learning_rate": 0.0,
+      "loss": 0.9502,
+      "step": 13460
+    },
+    {
+      "epoch": 3.136935258500233,
+      "grad_norm": 4.724367618560791,
+      "learning_rate": 0.0,
+      "loss": 1.0116,
+      "step": 13470
+    },
+    {
+      "epoch": 3.1392640894271078,
+      "grad_norm": 22.139331817626953,
+      "learning_rate": 0.0,
+      "loss": 0.7872,
+      "step": 13480
+    },
+    {
+      "epoch": 3.1415929203539825,
+      "grad_norm": 6.987404823303223,
+      "learning_rate": 0.0,
+      "loss": 1.1415,
+      "step": 13490
+    },
+    {
+      "epoch": 3.143921751280857,
+      "grad_norm": 23.272977828979492,
+      "learning_rate": 0.0,
+      "loss": 1.1225,
+      "step": 13500
+    },
+    {
+      "epoch": 3.1462505822077316,
+      "grad_norm": 14.455467224121094,
+      "learning_rate": 0.0,
+      "loss": 1.449,
+      "step": 13510
+    },
+    {
+      "epoch": 3.1485794131346063,
+      "grad_norm": 23.383798599243164,
+      "learning_rate": 0.0,
+      "loss": 1.2972,
+      "step": 13520
+    },
+    {
+      "epoch": 3.150908244061481,
+      "grad_norm": 6.212050437927246,
+      "learning_rate": 0.0,
+      "loss": 1.0594,
+      "step": 13530
+    },
+    {
+      "epoch": 3.153237074988356,
+      "grad_norm": 5.7028422355651855,
+      "learning_rate": 0.0,
+      "loss": 0.9708,
+      "step": 13540
+    },
+    {
+      "epoch": 3.1555659059152306,
+      "grad_norm": 23.22449493408203,
+      "learning_rate": 0.0,
+      "loss": 0.9153,
+      "step": 13550
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 11.004061698913574,
+      "learning_rate": 0.0,
+      "loss": 0.8146,
+      "step": 13560
+    },
+    {
+      "epoch": 3.16022356776898,
+      "grad_norm": 12.053899765014648,
+      "learning_rate": 0.0,
+      "loss": 1.0754,
+      "step": 13570
+    },
+    {
+      "epoch": 3.162552398695855,
+      "grad_norm": 6.76538610458374,
+      "learning_rate": 0.0,
+      "loss": 1.383,
+      "step": 13580
+    },
+    {
+      "epoch": 3.1648812296227296,
+      "grad_norm": 17.094440460205078,
+      "learning_rate": 0.0,
+      "loss": 1.3829,
+      "step": 13590
+    },
+    {
+      "epoch": 3.1672100605496043,
+      "grad_norm": 91.07170867919922,
+      "learning_rate": 0.0,
+      "loss": 1.7515,
+      "step": 13600
+    },
+    {
+      "epoch": 3.1695388914764786,
+      "grad_norm": 5.213118076324463,
+      "learning_rate": 0.0,
+      "loss": 1.1723,
+      "step": 13610
+    },
+    {
+      "epoch": 3.1718677224033534,
+      "grad_norm": 85.04914855957031,
+      "learning_rate": 0.0,
+      "loss": 1.4981,
+      "step": 13620
+    },
+    {
+      "epoch": 3.174196553330228,
+      "grad_norm": 12.307666778564453,
+      "learning_rate": 0.0,
+      "loss": 1.4268,
+      "step": 13630
+    },
+    {
+      "epoch": 3.176525384257103,
+      "grad_norm": 9.055723190307617,
+      "learning_rate": 0.0,
+      "loss": 1.3236,
+      "step": 13640
+    },
+    {
+      "epoch": 3.1788542151839776,
+      "grad_norm": 13.058332443237305,
+      "learning_rate": 0.0,
+      "loss": 1.6913,
+      "step": 13650
+    },
+    {
+      "epoch": 3.1811830461108523,
+      "grad_norm": 12.987732887268066,
+      "learning_rate": 0.0,
+      "loss": 0.8965,
+      "step": 13660
+    },
+    {
+      "epoch": 3.183511877037727,
+      "grad_norm": 13.6783447265625,
+      "learning_rate": 0.0,
+      "loss": 1.252,
+      "step": 13670
+    },
+    {
+      "epoch": 3.185840707964602,
+      "grad_norm": 9.275946617126465,
+      "learning_rate": 0.0,
+      "loss": 1.1708,
+      "step": 13680
+    },
+    {
+      "epoch": 3.1881695388914766,
+      "grad_norm": 6.581894874572754,
+      "learning_rate": 0.0,
+      "loss": 0.994,
+      "step": 13690
+    },
+    {
+      "epoch": 3.1904983698183513,
+      "grad_norm": 9.445444107055664,
+      "learning_rate": 0.0,
+      "loss": 1.0164,
+      "step": 13700
+    },
+    {
+      "epoch": 3.192827200745226,
+      "grad_norm": 9.902006149291992,
+      "learning_rate": 0.0,
+      "loss": 1.1095,
+      "step": 13710
+    },
+    {
+      "epoch": 3.1951560316721004,
+      "grad_norm": 7.1675944328308105,
+      "learning_rate": 0.0,
+      "loss": 1.4237,
+      "step": 13720
+    },
+    {
+      "epoch": 3.197484862598975,
+      "grad_norm": 9.444397926330566,
+      "learning_rate": 0.0,
+      "loss": 1.3244,
+      "step": 13730
+    },
+    {
+      "epoch": 3.19981369352585,
+      "grad_norm": 5.586214065551758,
+      "learning_rate": 0.0,
+      "loss": 0.9574,
+      "step": 13740
+    },
+    {
+      "epoch": 3.2021425244527246,
+      "grad_norm": 5.1344218254089355,
+      "learning_rate": 0.0,
+      "loss": 1.5185,
+      "step": 13750
+    },
+    {
+      "epoch": 3.2044713553795994,
+      "grad_norm": 17.44362449645996,
+      "learning_rate": 0.0,
+      "loss": 0.8098,
+      "step": 13760
+    },
+    {
+      "epoch": 3.206800186306474,
+      "grad_norm": 13.0212984085083,
+      "learning_rate": 0.0,
+      "loss": 1.0167,
+      "step": 13770
+    },
+    {
+      "epoch": 3.209129017233349,
+      "grad_norm": 8.9563570022583,
+      "learning_rate": 0.0,
+      "loss": 1.357,
+      "step": 13780
+    },
+    {
+      "epoch": 3.2114578481602236,
+      "grad_norm": 17.59966468811035,
+      "learning_rate": 0.0,
+      "loss": 1.4481,
+      "step": 13790
+    },
+    {
+      "epoch": 3.2137866790870984,
+      "grad_norm": 46.96272659301758,
+      "learning_rate": 0.0,
+      "loss": 1.1264,
+      "step": 13800
+    },
+    {
+      "epoch": 3.216115510013973,
+      "grad_norm": 5.420903205871582,
+      "learning_rate": 0.0,
+      "loss": 0.8953,
+      "step": 13810
+    },
+    {
+      "epoch": 3.218444340940848,
+      "grad_norm": 3.29380202293396,
+      "learning_rate": 0.0,
+      "loss": 1.1815,
+      "step": 13820
+    },
+    {
+      "epoch": 3.2207731718677226,
+      "grad_norm": 5.43313455581665,
+      "learning_rate": 0.0,
+      "loss": 1.3487,
+      "step": 13830
+    },
+    {
+      "epoch": 3.223102002794597,
+      "grad_norm": 6.049877166748047,
+      "learning_rate": 0.0,
+      "loss": 0.9866,
+      "step": 13840
+    },
+    {
+      "epoch": 3.2254308337214717,
+      "grad_norm": 12.948495864868164,
+      "learning_rate": 0.0,
+      "loss": 1.2618,
+      "step": 13850
+    },
+    {
+      "epoch": 3.2277596646483464,
+      "grad_norm": 11.312189102172852,
+      "learning_rate": 0.0,
+      "loss": 1.2437,
+      "step": 13860
+    },
+    {
+      "epoch": 3.230088495575221,
+      "grad_norm": 11.164241790771484,
+      "learning_rate": 0.0,
+      "loss": 1.9565,
+      "step": 13870
+    },
+    {
+      "epoch": 3.232417326502096,
+      "grad_norm": 6.2988176345825195,
+      "learning_rate": 0.0,
+      "loss": 0.9325,
+      "step": 13880
+    },
+    {
+      "epoch": 3.2347461574289706,
+      "grad_norm": 8.579294204711914,
+      "learning_rate": 0.0,
+      "loss": 1.1787,
+      "step": 13890
+    },
+    {
+      "epoch": 3.2370749883558454,
+      "grad_norm": 7.805800437927246,
+      "learning_rate": 0.0,
+      "loss": 0.992,
+      "step": 13900
+    },
+    {
+      "epoch": 3.23940381928272,
+      "grad_norm": 8.043397903442383,
+      "learning_rate": 0.0,
+      "loss": 1.6505,
+      "step": 13910
+    },
+    {
+      "epoch": 3.241732650209595,
+      "grad_norm": 51.754547119140625,
+      "learning_rate": 0.0,
+      "loss": 1.076,
+      "step": 13920
+    },
+    {
+      "epoch": 3.2440614811364696,
+      "grad_norm": 15.905028343200684,
+      "learning_rate": 0.0,
+      "loss": 1.4288,
+      "step": 13930
+    },
+    {
+      "epoch": 3.2463903120633444,
+      "grad_norm": 8.984749794006348,
+      "learning_rate": 0.0,
+      "loss": 1.0526,
+      "step": 13940
+    },
+    {
+      "epoch": 3.248719142990219,
+      "grad_norm": 8.098136901855469,
+      "learning_rate": 0.0,
+      "loss": 1.0857,
+      "step": 13950
+    },
+    {
+      "epoch": 3.2510479739170934,
+      "grad_norm": 52.77598190307617,
+      "learning_rate": 0.0,
+      "loss": 1.1508,
+      "step": 13960
+    },
+    {
+      "epoch": 3.253376804843968,
+      "grad_norm": 9.703697204589844,
+      "learning_rate": 0.0,
+      "loss": 1.1592,
+      "step": 13970
+    },
+    {
+      "epoch": 3.255705635770843,
+      "grad_norm": 41.5420036315918,
+      "learning_rate": 0.0,
+      "loss": 1.4785,
+      "step": 13980
+    },
+    {
+      "epoch": 3.2580344666977177,
+      "grad_norm": 9.39918041229248,
+      "learning_rate": 0.0,
+      "loss": 1.4824,
+      "step": 13990
+    },
+    {
+      "epoch": 3.2603632976245924,
+      "grad_norm": 7.907922267913818,
+      "learning_rate": 0.0,
+      "loss": 0.9171,
+      "step": 14000
+    },
+    {
+      "epoch": 3.262692128551467,
+      "grad_norm": 9.973319053649902,
+      "learning_rate": 0.0,
+      "loss": 0.9349,
+      "step": 14010
+    },
+    {
+      "epoch": 3.265020959478342,
+      "grad_norm": 10.504914283752441,
+      "learning_rate": 0.0,
+      "loss": 1.0402,
+      "step": 14020
+    },
+    {
+      "epoch": 3.2673497904052167,
+      "grad_norm": 8.200088500976562,
+      "learning_rate": 0.0,
+      "loss": 1.0022,
+      "step": 14030
+    },
+    {
+      "epoch": 3.2696786213320914,
+      "grad_norm": 33.09117889404297,
+      "learning_rate": 0.0,
+      "loss": 1.3235,
+      "step": 14040
+    },
+    {
+      "epoch": 3.272007452258966,
+      "grad_norm": 9.324043273925781,
+      "learning_rate": 0.0,
+      "loss": 1.0813,
+      "step": 14050
+    },
+    {
+      "epoch": 3.274336283185841,
+      "grad_norm": 13.463679313659668,
+      "learning_rate": 0.0,
+      "loss": 0.9718,
+      "step": 14060
+    },
+    {
+      "epoch": 3.276665114112715,
+      "grad_norm": 6.217166423797607,
+      "learning_rate": 0.0,
+      "loss": 1.0963,
+      "step": 14070
+    },
+    {
+      "epoch": 3.27899394503959,
+      "grad_norm": 60.56024169921875,
+      "learning_rate": 0.0,
+      "loss": 1.0586,
+      "step": 14080
+    },
+    {
+      "epoch": 3.2813227759664647,
+      "grad_norm": 37.3182373046875,
+      "learning_rate": 0.0,
+      "loss": 1.1455,
+      "step": 14090
+    },
+    {
+      "epoch": 3.2836516068933395,
+      "grad_norm": 5.846940040588379,
+      "learning_rate": 0.0,
+      "loss": 1.5728,
+      "step": 14100
+    },
+    {
+      "epoch": 3.285980437820214,
+      "grad_norm": 33.51951217651367,
+      "learning_rate": 0.0,
+      "loss": 1.2916,
+      "step": 14110
+    },
+    {
+      "epoch": 3.288309268747089,
+      "grad_norm": 7.1149797439575195,
+      "learning_rate": 0.0,
+      "loss": 0.8535,
+      "step": 14120
+    },
+    {
+      "epoch": 3.2906380996739637,
+      "grad_norm": 9.088951110839844,
+      "learning_rate": 0.0,
+      "loss": 1.0247,
+      "step": 14130
+    },
+    {
+      "epoch": 3.2929669306008384,
+      "grad_norm": 9.655440330505371,
+      "learning_rate": 0.0,
+      "loss": 1.1175,
+      "step": 14140
+    },
+    {
+      "epoch": 3.295295761527713,
+      "grad_norm": 22.48821258544922,
+      "learning_rate": 0.0,
+      "loss": 1.1427,
+      "step": 14150
+    },
+    {
+      "epoch": 3.297624592454588,
+      "grad_norm": 51.519859313964844,
+      "learning_rate": 0.0,
+      "loss": 1.8776,
+      "step": 14160
+    },
+    {
+      "epoch": 3.2999534233814627,
+      "grad_norm": 25.946897506713867,
+      "learning_rate": 0.0,
+      "loss": 1.7691,
+      "step": 14170
+    },
+    {
+      "epoch": 3.302282254308337,
+      "grad_norm": 6.704770565032959,
+      "learning_rate": 0.0,
+      "loss": 1.6418,
+      "step": 14180
+    },
+    {
+      "epoch": 3.3046110852352117,
+      "grad_norm": 30.36494255065918,
+      "learning_rate": 0.0,
+      "loss": 1.0997,
+      "step": 14190
+    },
+    {
+      "epoch": 3.3069399161620865,
+      "grad_norm": 16.188684463500977,
+      "learning_rate": 0.0,
+      "loss": 1.4068,
+      "step": 14200
+    },
+    {
+      "epoch": 3.3092687470889612,
+      "grad_norm": 11.660002708435059,
+      "learning_rate": 0.0,
+      "loss": 0.9104,
+      "step": 14210
+    },
+    {
+      "epoch": 3.311597578015836,
+      "grad_norm": 42.77584457397461,
+      "learning_rate": 0.0,
+      "loss": 1.49,
+      "step": 14220
+    },
+    {
+      "epoch": 3.3139264089427107,
+      "grad_norm": 15.70578670501709,
+      "learning_rate": 0.0,
+      "loss": 1.204,
+      "step": 14230
+    },
+    {
+      "epoch": 3.3162552398695855,
+      "grad_norm": 28.181676864624023,
+      "learning_rate": 0.0,
+      "loss": 1.132,
+      "step": 14240
+    },
+    {
+      "epoch": 3.3185840707964602,
+      "grad_norm": 6.674837112426758,
+      "learning_rate": 0.0,
+      "loss": 1.1825,
+      "step": 14250
+    },
+    {
+      "epoch": 3.320912901723335,
+      "grad_norm": 51.54294967651367,
+      "learning_rate": 0.0,
+      "loss": 1.0609,
+      "step": 14260
+    },
+    {
+      "epoch": 3.3232417326502097,
+      "grad_norm": 11.583158493041992,
+      "learning_rate": 0.0,
+      "loss": 1.0657,
+      "step": 14270
+    },
+    {
+      "epoch": 3.3255705635770845,
+      "grad_norm": 15.019560813903809,
+      "learning_rate": 0.0,
+      "loss": 1.038,
+      "step": 14280
+    },
+    {
+      "epoch": 3.3278993945039588,
+      "grad_norm": 33.25980758666992,
+      "learning_rate": 0.0,
+      "loss": 1.1687,
+      "step": 14290
+    },
+    {
+      "epoch": 3.330228225430834,
+      "grad_norm": 28.561473846435547,
+      "learning_rate": 0.0,
+      "loss": 1.1052,
+      "step": 14300
+    },
+    {
+      "epoch": 3.3325570563577083,
+      "grad_norm": 62.97575378417969,
+      "learning_rate": 0.0,
+      "loss": 1.1649,
+      "step": 14310
+    },
+    {
+      "epoch": 3.334885887284583,
+      "grad_norm": 16.132755279541016,
+      "learning_rate": 0.0,
+      "loss": 1.1446,
+      "step": 14320
+    },
+    {
+      "epoch": 3.3372147182114578,
+      "grad_norm": 28.80936622619629,
+      "learning_rate": 0.0,
+      "loss": 1.3675,
+      "step": 14330
+    },
+    {
+      "epoch": 3.3395435491383325,
+      "grad_norm": 27.189163208007812,
+      "learning_rate": 0.0,
+      "loss": 1.2651,
+      "step": 14340
+    },
+    {
+      "epoch": 3.3418723800652073,
+      "grad_norm": 54.67479705810547,
+      "learning_rate": 0.0,
+      "loss": 1.2248,
+      "step": 14350
+    },
+    {
+      "epoch": 3.344201210992082,
+      "grad_norm": 31.085508346557617,
+      "learning_rate": 0.0,
+      "loss": 1.6074,
+      "step": 14360
+    },
+    {
+      "epoch": 3.3465300419189568,
+      "grad_norm": 4.460205078125,
+      "learning_rate": 0.0,
+      "loss": 1.4311,
+      "step": 14370
+    },
+    {
+      "epoch": 3.3488588728458315,
+      "grad_norm": 9.599722862243652,
+      "learning_rate": 0.0,
+      "loss": 1.4001,
+      "step": 14380
+    },
+    {
+      "epoch": 3.3511877037727063,
+      "grad_norm": 13.639843940734863,
+      "learning_rate": 0.0,
+      "loss": 1.2749,
+      "step": 14390
+    },
+    {
+      "epoch": 3.353516534699581,
+      "grad_norm": 34.45295715332031,
+      "learning_rate": 0.0,
+      "loss": 1.1642,
+      "step": 14400
+    },
+    {
+      "epoch": 3.3558453656264557,
+      "grad_norm": 12.180970191955566,
+      "learning_rate": 0.0,
+      "loss": 1.1084,
+      "step": 14410
+    },
+    {
+      "epoch": 3.35817419655333,
+      "grad_norm": 74.1428451538086,
+      "learning_rate": 0.0,
+      "loss": 2.0657,
+      "step": 14420
+    },
+    {
+      "epoch": 3.360503027480205,
+      "grad_norm": 9.939842224121094,
+      "learning_rate": 0.0,
+      "loss": 1.2155,
+      "step": 14430
+    },
+    {
+      "epoch": 3.3628318584070795,
+      "grad_norm": 6.518401145935059,
+      "learning_rate": 0.0,
+      "loss": 0.7601,
+      "step": 14440
+    },
+    {
+      "epoch": 3.3651606893339543,
+      "grad_norm": 64.0587387084961,
+      "learning_rate": 0.0,
+      "loss": 1.5006,
+      "step": 14450
+    },
+    {
+      "epoch": 3.367489520260829,
+      "grad_norm": 8.050665855407715,
+      "learning_rate": 0.0,
+      "loss": 1.5511,
+      "step": 14460
+    },
+    {
+      "epoch": 3.369818351187704,
+      "grad_norm": 6.347721099853516,
+      "learning_rate": 0.0,
+      "loss": 1.5936,
+      "step": 14470
+    },
+    {
+      "epoch": 3.3721471821145785,
+      "grad_norm": 16.514511108398438,
+      "learning_rate": 0.0,
+      "loss": 0.9755,
+      "step": 14480
+    },
+    {
+      "epoch": 3.3744760130414533,
+      "grad_norm": 10.741750717163086,
+      "learning_rate": 0.0,
+      "loss": 1.2707,
+      "step": 14490
+    },
+    {
+      "epoch": 3.376804843968328,
+      "grad_norm": 5.380224227905273,
+      "learning_rate": 0.0,
+      "loss": 1.6041,
+      "step": 14500
+    },
+    {
+      "epoch": 3.3791336748952028,
+      "grad_norm": 55.9947395324707,
+      "learning_rate": 0.0,
+      "loss": 1.5839,
+      "step": 14510
+    },
+    {
+      "epoch": 3.3814625058220775,
+      "grad_norm": 15.040616035461426,
+      "learning_rate": 0.0,
+      "loss": 1.0565,
+      "step": 14520
+    },
+    {
+      "epoch": 3.383791336748952,
+      "grad_norm": 7.4967522621154785,
+      "learning_rate": 0.0,
+      "loss": 1.5983,
+      "step": 14530
+    },
+    {
+      "epoch": 3.3861201676758266,
+      "grad_norm": 16.126693725585938,
+      "learning_rate": 0.0,
+      "loss": 0.9066,
+      "step": 14540
+    },
+    {
+      "epoch": 3.3884489986027013,
+      "grad_norm": 25.731449127197266,
+      "learning_rate": 0.0,
+      "loss": 0.8373,
+      "step": 14550
+    },
+    {
+      "epoch": 3.390777829529576,
+      "grad_norm": 9.732860565185547,
+      "learning_rate": 0.0,
+      "loss": 1.3208,
+      "step": 14560
+    },
+    {
+      "epoch": 3.393106660456451,
+      "grad_norm": 93.20537567138672,
+      "learning_rate": 0.0,
+      "loss": 1.3592,
+      "step": 14570
+    },
+    {
+      "epoch": 3.3954354913833256,
+      "grad_norm": 10.425060272216797,
+      "learning_rate": 0.0,
+      "loss": 1.3393,
+      "step": 14580
+    },
+    {
+      "epoch": 3.3977643223102003,
+      "grad_norm": 9.359151840209961,
+      "learning_rate": 0.0,
+      "loss": 1.4633,
+      "step": 14590
+    },
+    {
+      "epoch": 3.400093153237075,
+      "grad_norm": 19.98016357421875,
+      "learning_rate": 0.0,
+      "loss": 1.8438,
+      "step": 14600
+    },
+    {
+      "epoch": 3.40242198416395,
+      "grad_norm": 15.402310371398926,
+      "learning_rate": 0.0,
+      "loss": 1.7674,
+      "step": 14610
+    },
+    {
+      "epoch": 3.4047508150908246,
+      "grad_norm": 17.607166290283203,
+      "learning_rate": 0.0,
+      "loss": 1.1485,
+      "step": 14620
+    },
+    {
+      "epoch": 3.4070796460176993,
+      "grad_norm": 9.94525146484375,
+      "learning_rate": 0.0,
+      "loss": 0.9103,
+      "step": 14630
+    },
+    {
+      "epoch": 3.4094084769445736,
+      "grad_norm": 17.84208106994629,
+      "learning_rate": 0.0,
+      "loss": 1.1366,
+      "step": 14640
+    },
+    {
+      "epoch": 3.4117373078714484,
+      "grad_norm": 15.35504150390625,
+      "learning_rate": 0.0,
+      "loss": 1.2129,
+      "step": 14650
+    },
+    {
+      "epoch": 3.414066138798323,
+      "grad_norm": 21.50774574279785,
+      "learning_rate": 0.0,
+      "loss": 1.0228,
+      "step": 14660
+    },
+    {
+      "epoch": 3.416394969725198,
+      "grad_norm": 45.42424774169922,
+      "learning_rate": 0.0,
+      "loss": 1.6175,
+      "step": 14670
+    },
+    {
+      "epoch": 3.4187238006520726,
+      "grad_norm": 34.5196533203125,
+      "learning_rate": 0.0,
+      "loss": 1.2004,
+      "step": 14680
+    },
+    {
+      "epoch": 3.4210526315789473,
+      "grad_norm": 34.76347351074219,
+      "learning_rate": 0.0,
+      "loss": 0.9398,
+      "step": 14690
+    },
+    {
+      "epoch": 3.423381462505822,
+      "grad_norm": 12.189762115478516,
+      "learning_rate": 0.0,
+      "loss": 1.0523,
+      "step": 14700
+    },
+    {
+      "epoch": 3.425710293432697,
+      "grad_norm": 8.52271842956543,
+      "learning_rate": 0.0,
+      "loss": 0.9799,
+      "step": 14710
+    },
+    {
+      "epoch": 3.4280391243595716,
+      "grad_norm": 62.98123550415039,
+      "learning_rate": 0.0,
+      "loss": 1.4572,
+      "step": 14720
+    },
+    {
+      "epoch": 3.4303679552864463,
+      "grad_norm": 5.033658504486084,
+      "learning_rate": 0.0,
+      "loss": 1.0885,
+      "step": 14730
+    },
+    {
+      "epoch": 3.432696786213321,
+      "grad_norm": 17.944128036499023,
+      "learning_rate": 0.0,
+      "loss": 1.467,
+      "step": 14740
+    },
+    {
+      "epoch": 3.4350256171401954,
+      "grad_norm": 7.418787956237793,
+      "learning_rate": 0.0,
+      "loss": 1.0984,
+      "step": 14750
+    },
+    {
+      "epoch": 3.4373544480670706,
+      "grad_norm": 7.411730766296387,
+      "learning_rate": 0.0,
+      "loss": 0.8685,
+      "step": 14760
+    },
+    {
+      "epoch": 3.439683278993945,
+      "grad_norm": 19.966922760009766,
+      "learning_rate": 0.0,
+      "loss": 1.4455,
+      "step": 14770
+    },
+    {
+      "epoch": 3.4420121099208196,
+      "grad_norm": 10.585490226745605,
+      "learning_rate": 0.0,
+      "loss": 1.1593,
+      "step": 14780
+    },
+    {
+      "epoch": 3.4443409408476944,
+      "grad_norm": 8.054330825805664,
+      "learning_rate": 0.0,
+      "loss": 1.3982,
+      "step": 14790
+    },
+    {
+      "epoch": 3.446669771774569,
+      "grad_norm": 8.496953964233398,
+      "learning_rate": 0.0,
+      "loss": 1.1782,
+      "step": 14800
+    },
+    {
+      "epoch": 3.448998602701444,
+      "grad_norm": 9.365567207336426,
+      "learning_rate": 0.0,
+      "loss": 0.9234,
+      "step": 14810
+    },
+    {
+      "epoch": 3.4513274336283186,
+      "grad_norm": 13.222329139709473,
+      "learning_rate": 0.0,
+      "loss": 1.2217,
+      "step": 14820
+    },
+    {
+      "epoch": 3.4536562645551934,
+      "grad_norm": 28.72707748413086,
+      "learning_rate": 0.0,
+      "loss": 1.0365,
+      "step": 14830
+    },
+    {
+      "epoch": 3.455985095482068,
+      "grad_norm": 27.434940338134766,
+      "learning_rate": 0.0,
+      "loss": 1.1692,
+      "step": 14840
+    },
+    {
+      "epoch": 3.458313926408943,
+      "grad_norm": 4.920462131500244,
+      "learning_rate": 0.0,
+      "loss": 1.0483,
+      "step": 14850
+    },
+    {
+      "epoch": 3.4606427573358176,
+      "grad_norm": 12.028072357177734,
+      "learning_rate": 0.0,
+      "loss": 1.547,
+      "step": 14860
+    },
+    {
+      "epoch": 3.4629715882626924,
+      "grad_norm": 14.676789283752441,
+      "learning_rate": 0.0,
+      "loss": 1.5541,
+      "step": 14870
+    },
+    {
+      "epoch": 3.4653004191895667,
+      "grad_norm": 31.549835205078125,
+      "learning_rate": 0.0,
+      "loss": 2.1055,
+      "step": 14880
+    },
+    {
+      "epoch": 3.4676292501164414,
+      "grad_norm": 14.0533447265625,
+      "learning_rate": 0.0,
+      "loss": 1.0097,
+      "step": 14890
+    },
+    {
+      "epoch": 3.469958081043316,
+      "grad_norm": 18.530563354492188,
+      "learning_rate": 0.0,
+      "loss": 1.1465,
+      "step": 14900
+    },
+    {
+      "epoch": 3.472286911970191,
+      "grad_norm": 10.129080772399902,
+      "learning_rate": 0.0,
+      "loss": 1.3289,
+      "step": 14910
+    },
+    {
+      "epoch": 3.4746157428970657,
+      "grad_norm": 12.979293823242188,
+      "learning_rate": 0.0,
+      "loss": 1.708,
+      "step": 14920
+    },
+    {
+      "epoch": 3.4769445738239404,
+      "grad_norm": 48.22283935546875,
+      "learning_rate": 0.0,
+      "loss": 0.9649,
+      "step": 14930
+    },
+    {
+      "epoch": 3.479273404750815,
+      "grad_norm": 15.611557960510254,
+      "learning_rate": 0.0,
+      "loss": 1.407,
+      "step": 14940
+    },
+    {
+      "epoch": 3.48160223567769,
+      "grad_norm": 18.218061447143555,
+      "learning_rate": 0.0,
+      "loss": 1.6172,
+      "step": 14950
+    },
+    {
+      "epoch": 3.4839310666045646,
+      "grad_norm": 4.562121391296387,
+      "learning_rate": 0.0,
+      "loss": 0.8124,
+      "step": 14960
+    },
+    {
+      "epoch": 3.4862598975314394,
+      "grad_norm": 52.5666389465332,
+      "learning_rate": 0.0,
+      "loss": 1.391,
+      "step": 14970
+    },
+    {
+      "epoch": 3.488588728458314,
+      "grad_norm": 15.410503387451172,
+      "learning_rate": 0.0,
+      "loss": 1.4648,
+      "step": 14980
+    },
+    {
+      "epoch": 3.4909175593851884,
+      "grad_norm": 51.82364273071289,
+      "learning_rate": 0.0,
+      "loss": 1.0539,
+      "step": 14990
+    },
+    {
+      "epoch": 3.493246390312063,
+      "grad_norm": 5.764760971069336,
+      "learning_rate": 0.0,
+      "loss": 0.9975,
+      "step": 15000
+    },
+    {
+      "epoch": 3.495575221238938,
+      "grad_norm": 33.286373138427734,
+      "learning_rate": 0.0,
+      "loss": 1.4706,
+      "step": 15010
+    },
+    {
+      "epoch": 3.4979040521658127,
+      "grad_norm": 13.835527420043945,
+      "learning_rate": 0.0,
+      "loss": 1.182,
+      "step": 15020
+    },
+    {
+      "epoch": 3.5002328830926874,
+      "grad_norm": 29.695833206176758,
+      "learning_rate": 0.0,
+      "loss": 1.5128,
+      "step": 15030
+    },
+    {
+      "epoch": 3.502561714019562,
+      "grad_norm": 8.429316520690918,
+      "learning_rate": 0.0,
+      "loss": 1.1939,
+      "step": 15040
+    },
+    {
+      "epoch": 3.504890544946437,
+      "grad_norm": 34.458953857421875,
+      "learning_rate": 0.0,
+      "loss": 0.8596,
+      "step": 15050
+    },
+    {
+      "epoch": 3.5072193758733117,
+      "grad_norm": 38.27690505981445,
+      "learning_rate": 0.0,
+      "loss": 1.0778,
+      "step": 15060
+    },
+    {
+      "epoch": 3.5095482068001864,
+      "grad_norm": 11.751981735229492,
+      "learning_rate": 0.0,
+      "loss": 1.1244,
+      "step": 15070
+    },
+    {
+      "epoch": 3.511877037727061,
+      "grad_norm": 61.4968147277832,
+      "learning_rate": 0.0,
+      "loss": 1.1948,
+      "step": 15080
+    },
+    {
+      "epoch": 3.514205868653936,
+      "grad_norm": 7.1937408447265625,
+      "learning_rate": 0.0,
+      "loss": 0.9009,
+      "step": 15090
+    },
+    {
+      "epoch": 3.51653469958081,
+      "grad_norm": 116.81895446777344,
+      "learning_rate": 0.0,
+      "loss": 1.6013,
+      "step": 15100
+    },
+    {
+      "epoch": 3.5188635305076854,
+      "grad_norm": 6.943803787231445,
+      "learning_rate": 0.0,
+      "loss": 0.8098,
+      "step": 15110
+    },
+    {
+      "epoch": 3.5211923614345597,
+      "grad_norm": 12.434028625488281,
+      "learning_rate": 0.0,
+      "loss": 1.3035,
+      "step": 15120
+    },
+    {
+      "epoch": 3.5235211923614345,
+      "grad_norm": 7.836872577667236,
+      "learning_rate": 0.0,
+      "loss": 1.0687,
+      "step": 15130
+    },
+    {
+      "epoch": 3.525850023288309,
+      "grad_norm": 8.494155883789062,
+      "learning_rate": 0.0,
+      "loss": 1.2936,
+      "step": 15140
+    },
+    {
+      "epoch": 3.528178854215184,
+      "grad_norm": 27.25113868713379,
+      "learning_rate": 0.0,
+      "loss": 1.3568,
+      "step": 15150
+    },
+    {
+      "epoch": 3.5305076851420587,
+      "grad_norm": 13.736648559570312,
+      "learning_rate": 0.0,
+      "loss": 1.3675,
+      "step": 15160
+    },
+    {
+      "epoch": 3.5328365160689335,
+      "grad_norm": 8.055909156799316,
+      "learning_rate": 0.0,
+      "loss": 0.7531,
+      "step": 15170
+    },
+    {
+      "epoch": 3.535165346995808,
+      "grad_norm": 12.262843132019043,
+      "learning_rate": 0.0,
+      "loss": 1.0719,
+      "step": 15180
+    },
+    {
+      "epoch": 3.537494177922683,
+      "grad_norm": 12.301920890808105,
+      "learning_rate": 0.0,
+      "loss": 1.2898,
+      "step": 15190
+    },
+    {
+      "epoch": 3.5398230088495577,
+      "grad_norm": 22.987722396850586,
+      "learning_rate": 0.0,
+      "loss": 1.5796,
+      "step": 15200
+    },
+    {
+      "epoch": 3.542151839776432,
+      "grad_norm": 10.889913558959961,
+      "learning_rate": 0.0,
+      "loss": 0.9,
+      "step": 15210
+    },
+    {
+      "epoch": 3.544480670703307,
+      "grad_norm": 46.55030822753906,
+      "learning_rate": 0.0,
+      "loss": 1.046,
+      "step": 15220
+    },
+    {
+      "epoch": 3.5468095016301815,
+      "grad_norm": 49.14213180541992,
+      "learning_rate": 0.0,
+      "loss": 1.4466,
+      "step": 15230
+    },
+    {
+      "epoch": 3.5491383325570562,
+      "grad_norm": 9.739147186279297,
+      "learning_rate": 0.0,
+      "loss": 0.7041,
+      "step": 15240
+    },
+    {
+      "epoch": 3.551467163483931,
+      "grad_norm": 9.940231323242188,
+      "learning_rate": 0.0,
+      "loss": 1.7196,
+      "step": 15250
+    },
+    {
+      "epoch": 3.5537959944108057,
+      "grad_norm": 27.860456466674805,
+      "learning_rate": 0.0,
+      "loss": 0.9869,
+      "step": 15260
+    },
+    {
+      "epoch": 3.5561248253376805,
+      "grad_norm": 12.440192222595215,
+      "learning_rate": 0.0,
+      "loss": 0.8983,
+      "step": 15270
+    },
+    {
+      "epoch": 3.5584536562645552,
+      "grad_norm": 7.742676258087158,
+      "learning_rate": 0.0,
+      "loss": 1.1601,
+      "step": 15280
+    },
+    {
+      "epoch": 3.56078248719143,
+      "grad_norm": 10.191648483276367,
+      "learning_rate": 0.0,
+      "loss": 1.2396,
+      "step": 15290
+    },
+    {
+      "epoch": 3.5631113181183047,
+      "grad_norm": 11.127610206604004,
+      "learning_rate": 0.0,
+      "loss": 0.8596,
+      "step": 15300
+    },
+    {
+      "epoch": 3.5654401490451795,
+      "grad_norm": 22.639829635620117,
+      "learning_rate": 0.0,
+      "loss": 1.0627,
+      "step": 15310
+    },
+    {
+      "epoch": 3.5677689799720538,
+      "grad_norm": 12.301551818847656,
+      "learning_rate": 0.0,
+      "loss": 1.3009,
+      "step": 15320
+    },
+    {
+      "epoch": 3.570097810898929,
+      "grad_norm": 32.67119598388672,
+      "learning_rate": 0.0,
+      "loss": 1.1616,
+      "step": 15330
+    },
+    {
+      "epoch": 3.5724266418258033,
+      "grad_norm": 25.659578323364258,
+      "learning_rate": 0.0,
+      "loss": 1.0511,
+      "step": 15340
+    },
+    {
+      "epoch": 3.574755472752678,
+      "grad_norm": 34.240386962890625,
+      "learning_rate": 0.0,
+      "loss": 1.2238,
+      "step": 15350
+    },
+    {
+      "epoch": 3.5770843036795528,
+      "grad_norm": 8.90058422088623,
+      "learning_rate": 0.0,
+      "loss": 1.2145,
+      "step": 15360
+    },
+    {
+      "epoch": 3.5794131346064275,
+      "grad_norm": 8.673346519470215,
+      "learning_rate": 0.0,
+      "loss": 0.8024,
+      "step": 15370
+    },
+    {
+      "epoch": 3.5817419655333023,
+      "grad_norm": 4.38419246673584,
+      "learning_rate": 0.0,
+      "loss": 1.4394,
+      "step": 15380
+    },
+    {
+      "epoch": 3.584070796460177,
+      "grad_norm": 13.117408752441406,
+      "learning_rate": 0.0,
+      "loss": 1.3587,
+      "step": 15390
+    },
+    {
+      "epoch": 3.5863996273870518,
+      "grad_norm": 8.190533638000488,
+      "learning_rate": 0.0,
+      "loss": 1.2683,
+      "step": 15400
+    },
+    {
+      "epoch": 3.5887284583139265,
+      "grad_norm": 16.9916934967041,
+      "learning_rate": 0.0,
+      "loss": 1.5906,
+      "step": 15410
+    },
+    {
+      "epoch": 3.5910572892408013,
+      "grad_norm": 3.9092514514923096,
+      "learning_rate": 0.0,
+      "loss": 0.9965,
+      "step": 15420
+    },
+    {
+      "epoch": 3.5933861201676756,
+      "grad_norm": 15.621074676513672,
+      "learning_rate": 0.0,
+      "loss": 0.954,
+      "step": 15430
+    },
+    {
+      "epoch": 3.5957149510945507,
+      "grad_norm": 6.421728134155273,
+      "learning_rate": 0.0,
+      "loss": 1.135,
+      "step": 15440
+    },
+    {
+      "epoch": 3.598043782021425,
+      "grad_norm": 21.08034324645996,
+      "learning_rate": 0.0,
+      "loss": 0.9873,
+      "step": 15450
+    },
+    {
+      "epoch": 3.6003726129483,
+      "grad_norm": 27.012897491455078,
+      "learning_rate": 0.0,
+      "loss": 1.203,
+      "step": 15460
+    },
+    {
+      "epoch": 3.6027014438751745,
+      "grad_norm": 53.65129852294922,
+      "learning_rate": 0.0,
+      "loss": 1.1453,
+      "step": 15470
+    },
+    {
+      "epoch": 3.6050302748020493,
+      "grad_norm": 9.872217178344727,
+      "learning_rate": 0.0,
+      "loss": 1.0103,
+      "step": 15480
+    },
+    {
+      "epoch": 3.607359105728924,
+      "grad_norm": 6.3376569747924805,
+      "learning_rate": 0.0,
+      "loss": 0.7322,
+      "step": 15490
+    },
+    {
+      "epoch": 3.609687936655799,
+      "grad_norm": 34.383384704589844,
+      "learning_rate": 0.0,
+      "loss": 2.007,
+      "step": 15500
+    },
+    {
+      "epoch": 3.6120167675826735,
+      "grad_norm": 19.966154098510742,
+      "learning_rate": 0.0,
+      "loss": 1.0879,
+      "step": 15510
+    },
+    {
+      "epoch": 3.6143455985095483,
+      "grad_norm": 12.672451972961426,
+      "learning_rate": 0.0,
+      "loss": 1.0139,
+      "step": 15520
+    },
+    {
+      "epoch": 3.616674429436423,
+      "grad_norm": 8.703089714050293,
+      "learning_rate": 0.0,
+      "loss": 0.9467,
+      "step": 15530
+    },
+    {
+      "epoch": 3.619003260363298,
+      "grad_norm": 12.565585136413574,
+      "learning_rate": 0.0,
+      "loss": 1.2754,
+      "step": 15540
+    },
+    {
+      "epoch": 3.6213320912901725,
+      "grad_norm": 7.127060890197754,
+      "learning_rate": 0.0,
+      "loss": 1.1098,
+      "step": 15550
+    },
+    {
+      "epoch": 3.623660922217047,
+      "grad_norm": 6.133944034576416,
+      "learning_rate": 0.0,
+      "loss": 1.2271,
+      "step": 15560
+    },
+    {
+      "epoch": 3.625989753143922,
+      "grad_norm": 10.252046585083008,
+      "learning_rate": 0.0,
+      "loss": 1.0133,
+      "step": 15570
+    },
+    {
+      "epoch": 3.6283185840707963,
+      "grad_norm": 41.735809326171875,
+      "learning_rate": 0.0,
+      "loss": 1.5424,
+      "step": 15580
+    },
+    {
+      "epoch": 3.630647414997671,
+      "grad_norm": 12.260039329528809,
+      "learning_rate": 0.0,
+      "loss": 1.0404,
+      "step": 15590
+    },
+    {
+      "epoch": 3.632976245924546,
+      "grad_norm": 7.584269046783447,
+      "learning_rate": 0.0,
+      "loss": 1.5002,
+      "step": 15600
+    },
+    {
+      "epoch": 3.6353050768514206,
+      "grad_norm": 10.819036483764648,
+      "learning_rate": 0.0,
+      "loss": 1.1327,
+      "step": 15610
+    },
+    {
+      "epoch": 3.6376339077782953,
+      "grad_norm": 48.199974060058594,
+      "learning_rate": 0.0,
+      "loss": 1.1238,
+      "step": 15620
+    },
+    {
+      "epoch": 3.63996273870517,
+      "grad_norm": 35.932064056396484,
+      "learning_rate": 0.0,
+      "loss": 1.7948,
+      "step": 15630
+    },
+    {
+      "epoch": 3.642291569632045,
+      "grad_norm": 6.570189476013184,
+      "learning_rate": 0.0,
+      "loss": 1.2111,
+      "step": 15640
+    },
+    {
+      "epoch": 3.6446204005589196,
+      "grad_norm": 5.2273101806640625,
+      "learning_rate": 0.0,
+      "loss": 0.9788,
+      "step": 15650
+    },
+    {
+      "epoch": 3.6469492314857943,
+      "grad_norm": 15.28518009185791,
+      "learning_rate": 0.0,
+      "loss": 1.1353,
+      "step": 15660
+    },
+    {
+      "epoch": 3.6492780624126686,
+      "grad_norm": 11.389464378356934,
+      "learning_rate": 0.0,
+      "loss": 0.9977,
+      "step": 15670
+    },
+    {
+      "epoch": 3.651606893339544,
+      "grad_norm": 27.311847686767578,
+      "learning_rate": 0.0,
+      "loss": 1.3812,
+      "step": 15680
+    },
+    {
+      "epoch": 3.653935724266418,
+      "grad_norm": 94.494384765625,
+      "learning_rate": 0.0,
+      "loss": 1.4974,
+      "step": 15690
+    },
+    {
+      "epoch": 3.656264555193293,
+      "grad_norm": 42.005332946777344,
+      "learning_rate": 0.0,
+      "loss": 1.1392,
+      "step": 15700
+    },
+    {
+      "epoch": 3.6585933861201676,
+      "grad_norm": 40.726131439208984,
+      "learning_rate": 0.0,
+      "loss": 1.5276,
+      "step": 15710
+    },
+    {
+      "epoch": 3.6609222170470423,
+      "grad_norm": 11.148183822631836,
+      "learning_rate": 0.0,
+      "loss": 0.9784,
+      "step": 15720
+    },
+    {
+      "epoch": 3.663251047973917,
+      "grad_norm": 76.10807800292969,
+      "learning_rate": 0.0,
+      "loss": 1.2827,
+      "step": 15730
+    },
+    {
+      "epoch": 3.665579878900792,
+      "grad_norm": 198.63717651367188,
+      "learning_rate": 0.0,
+      "loss": 1.4597,
+      "step": 15740
+    },
+    {
+      "epoch": 3.6679087098276666,
+      "grad_norm": 65.95364379882812,
+      "learning_rate": 0.0,
+      "loss": 1.2488,
+      "step": 15750
+    },
+    {
+      "epoch": 3.6702375407545413,
+      "grad_norm": 31.9210262298584,
+      "learning_rate": 0.0,
+      "loss": 1.1257,
+      "step": 15760
+    },
+    {
+      "epoch": 3.672566371681416,
+      "grad_norm": 14.054140090942383,
+      "learning_rate": 0.0,
+      "loss": 0.8987,
+      "step": 15770
+    },
+    {
+      "epoch": 3.6748952026082904,
+      "grad_norm": 13.724059104919434,
+      "learning_rate": 0.0,
+      "loss": 1.4745,
+      "step": 15780
+    },
+    {
+      "epoch": 3.6772240335351656,
+      "grad_norm": 38.00988006591797,
+      "learning_rate": 0.0,
+      "loss": 1.3208,
+      "step": 15790
+    },
+    {
+      "epoch": 3.67955286446204,
+      "grad_norm": 20.750085830688477,
+      "learning_rate": 0.0,
+      "loss": 0.9753,
+      "step": 15800
+    },
+    {
+      "epoch": 3.6818816953889146,
+      "grad_norm": 6.154573440551758,
+      "learning_rate": 0.0,
+      "loss": 1.1994,
+      "step": 15810
+    },
+    {
+      "epoch": 3.6842105263157894,
+      "grad_norm": 8.296215057373047,
+      "learning_rate": 0.0,
+      "loss": 1.6545,
+      "step": 15820
+    },
+    {
+      "epoch": 3.686539357242664,
+      "grad_norm": 8.964814186096191,
+      "learning_rate": 0.0,
+      "loss": 0.8555,
+      "step": 15830
+    },
+    {
+      "epoch": 3.688868188169539,
+      "grad_norm": 72.2197494506836,
+      "learning_rate": 0.0,
+      "loss": 1.3968,
+      "step": 15840
+    },
+    {
+      "epoch": 3.6911970190964136,
+      "grad_norm": 8.642261505126953,
+      "learning_rate": 0.0,
+      "loss": 1.1964,
+      "step": 15850
+    },
+    {
+      "epoch": 3.6935258500232884,
+      "grad_norm": 9.64587688446045,
+      "learning_rate": 0.0,
+      "loss": 1.5,
+      "step": 15860
+    },
+    {
+      "epoch": 3.695854680950163,
+      "grad_norm": 9.663128852844238,
+      "learning_rate": 0.0,
+      "loss": 0.9647,
+      "step": 15870
+    },
+    {
+      "epoch": 3.698183511877038,
+      "grad_norm": 9.119319915771484,
+      "learning_rate": 0.0,
+      "loss": 0.9505,
+      "step": 15880
+    },
+    {
+      "epoch": 3.700512342803912,
+      "grad_norm": 13.707714080810547,
+      "learning_rate": 0.0,
+      "loss": 1.2409,
+      "step": 15890
+    },
+    {
+      "epoch": 3.7028411737307874,
+      "grad_norm": 9.985272407531738,
+      "learning_rate": 0.0,
+      "loss": 1.2205,
+      "step": 15900
+    },
+    {
+      "epoch": 3.7051700046576617,
+      "grad_norm": 12.141681671142578,
+      "learning_rate": 0.0,
+      "loss": 1.6891,
+      "step": 15910
+    },
+    {
+      "epoch": 3.7074988355845364,
+      "grad_norm": 12.171897888183594,
+      "learning_rate": 0.0,
+      "loss": 0.9045,
+      "step": 15920
+    },
+    {
+      "epoch": 3.709827666511411,
+      "grad_norm": 17.349260330200195,
+      "learning_rate": 0.0,
+      "loss": 1.9987,
+      "step": 15930
+    },
+    {
+      "epoch": 3.712156497438286,
+      "grad_norm": 14.914823532104492,
+      "learning_rate": 0.0,
+      "loss": 1.425,
+      "step": 15940
+    },
+    {
+      "epoch": 3.7144853283651607,
+      "grad_norm": 8.951475143432617,
+      "learning_rate": 0.0,
+      "loss": 1.1546,
+      "step": 15950
+    },
+    {
+      "epoch": 3.7168141592920354,
+      "grad_norm": 40.80722427368164,
+      "learning_rate": 0.0,
+      "loss": 1.1636,
+      "step": 15960
+    },
+    {
+      "epoch": 3.71914299021891,
+      "grad_norm": 45.113365173339844,
+      "learning_rate": 0.0,
+      "loss": 1.535,
+      "step": 15970
+    },
+    {
+      "epoch": 3.721471821145785,
+      "grad_norm": 16.32363510131836,
+      "learning_rate": 0.0,
+      "loss": 1.1573,
+      "step": 15980
+    },
+    {
+      "epoch": 3.7238006520726596,
+      "grad_norm": 14.807791709899902,
+      "learning_rate": 0.0,
+      "loss": 0.8877,
+      "step": 15990
+    },
+    {
+      "epoch": 3.7261294829995344,
+      "grad_norm": 29.387487411499023,
+      "learning_rate": 0.0,
+      "loss": 1.6075,
+      "step": 16000
+    },
+    {
+      "epoch": 3.728458313926409,
+      "grad_norm": 17.61796760559082,
+      "learning_rate": 0.0,
+      "loss": 1.144,
+      "step": 16010
+    },
+    {
+      "epoch": 3.7307871448532834,
+      "grad_norm": 9.548173904418945,
+      "learning_rate": 0.0,
+      "loss": 0.9762,
+      "step": 16020
+    },
+    {
+      "epoch": 3.7331159757801586,
+      "grad_norm": 31.526107788085938,
+      "learning_rate": 0.0,
+      "loss": 1.792,
+      "step": 16030
+    },
+    {
+      "epoch": 3.735444806707033,
+      "grad_norm": 5.914717674255371,
+      "learning_rate": 0.0,
+      "loss": 0.8789,
+      "step": 16040
+    },
+    {
+      "epoch": 3.7377736376339077,
+      "grad_norm": 16.024036407470703,
+      "learning_rate": 0.0,
+      "loss": 1.1737,
+      "step": 16050
+    },
+    {
+      "epoch": 3.7401024685607824,
+      "grad_norm": 7.287302494049072,
+      "learning_rate": 0.0,
+      "loss": 1.1577,
+      "step": 16060
+    },
+    {
+      "epoch": 3.742431299487657,
+      "grad_norm": 5.333652496337891,
+      "learning_rate": 0.0,
+      "loss": 1.2293,
+      "step": 16070
+    },
+    {
+      "epoch": 3.744760130414532,
+      "grad_norm": 18.019535064697266,
+      "learning_rate": 0.0,
+      "loss": 1.3213,
+      "step": 16080
+    },
+    {
+      "epoch": 3.7470889613414067,
+      "grad_norm": 9.020870208740234,
+      "learning_rate": 0.0,
+      "loss": 1.2686,
+      "step": 16090
+    },
+    {
+      "epoch": 3.7494177922682814,
+      "grad_norm": 5.7825822830200195,
+      "learning_rate": 0.0,
+      "loss": 1.3063,
+      "step": 16100
+    },
+    {
+      "epoch": 3.751746623195156,
+      "grad_norm": 64.28756713867188,
+      "learning_rate": 0.0,
+      "loss": 1.4442,
+      "step": 16110
+    },
+    {
+      "epoch": 3.754075454122031,
+      "grad_norm": 5.167896270751953,
+      "learning_rate": 0.0,
+      "loss": 0.839,
+      "step": 16120
+    },
+    {
+      "epoch": 3.7564042850489052,
+      "grad_norm": 10.846601486206055,
+      "learning_rate": 0.0,
+      "loss": 1.2187,
+      "step": 16130
+    },
+    {
+      "epoch": 3.7587331159757804,
+      "grad_norm": 74.83040618896484,
+      "learning_rate": 0.0,
+      "loss": 1.6439,
+      "step": 16140
+    },
+    {
+      "epoch": 3.7610619469026547,
+      "grad_norm": 8.98836612701416,
+      "learning_rate": 0.0,
+      "loss": 1.1921,
+      "step": 16150
+    },
+    {
+      "epoch": 3.7633907778295295,
+      "grad_norm": 8.449505805969238,
+      "learning_rate": 0.0,
+      "loss": 1.2248,
+      "step": 16160
+    },
+    {
+      "epoch": 3.765719608756404,
+      "grad_norm": 21.298240661621094,
+      "learning_rate": 0.0,
+      "loss": 1.1026,
+      "step": 16170
+    },
+    {
+      "epoch": 3.768048439683279,
+      "grad_norm": 7.777653217315674,
+      "learning_rate": 0.0,
+      "loss": 1.1569,
+      "step": 16180
+    },
+    {
+      "epoch": 3.7703772706101537,
+      "grad_norm": 53.60115051269531,
+      "learning_rate": 0.0,
+      "loss": 1.6678,
+      "step": 16190
+    },
+    {
+      "epoch": 3.7727061015370285,
+      "grad_norm": 14.203383445739746,
+      "learning_rate": 0.0,
+      "loss": 1.468,
+      "step": 16200
+    },
+    {
+      "epoch": 3.775034932463903,
+      "grad_norm": 24.37751007080078,
+      "learning_rate": 0.0,
+      "loss": 1.997,
+      "step": 16210
+    },
+    {
+      "epoch": 3.777363763390778,
+      "grad_norm": 17.089813232421875,
+      "learning_rate": 0.0,
+      "loss": 0.9753,
+      "step": 16220
+    },
+    {
+      "epoch": 3.7796925943176527,
+      "grad_norm": 40.838375091552734,
+      "learning_rate": 0.0,
+      "loss": 1.3751,
+      "step": 16230
+    },
+    {
+      "epoch": 3.782021425244527,
+      "grad_norm": 3.4650375843048096,
+      "learning_rate": 0.0,
+      "loss": 1.162,
+      "step": 16240
+    },
+    {
+      "epoch": 3.784350256171402,
+      "grad_norm": 20.963716506958008,
+      "learning_rate": 0.0,
+      "loss": 1.3415,
+      "step": 16250
+    },
+    {
+      "epoch": 3.7866790870982765,
+      "grad_norm": 12.60222053527832,
+      "learning_rate": 0.0,
+      "loss": 1.5009,
+      "step": 16260
+    },
+    {
+      "epoch": 3.7890079180251512,
+      "grad_norm": 16.32765007019043,
+      "learning_rate": 0.0,
+      "loss": 1.3629,
+      "step": 16270
+    },
+    {
+      "epoch": 3.791336748952026,
+      "grad_norm": 9.827703475952148,
+      "learning_rate": 0.0,
+      "loss": 0.8297,
+      "step": 16280
+    },
+    {
+      "epoch": 3.7936655798789007,
+      "grad_norm": 11.158079147338867,
+      "learning_rate": 0.0,
+      "loss": 1.051,
+      "step": 16290
+    },
+    {
+      "epoch": 3.7959944108057755,
+      "grad_norm": 7.290315628051758,
+      "learning_rate": 0.0,
+      "loss": 1.3413,
+      "step": 16300
+    },
+    {
+      "epoch": 3.7983232417326502,
+      "grad_norm": 5.4215521812438965,
+      "learning_rate": 0.0,
+      "loss": 0.7628,
+      "step": 16310
+    },
+    {
+      "epoch": 3.800652072659525,
+      "grad_norm": 8.901557922363281,
+      "learning_rate": 0.0,
+      "loss": 1.1244,
+      "step": 16320
+    },
+    {
+      "epoch": 3.8029809035863997,
+      "grad_norm": 10.666454315185547,
+      "learning_rate": 0.0,
+      "loss": 1.1982,
+      "step": 16330
+    },
+    {
+      "epoch": 3.8053097345132745,
+      "grad_norm": 32.98258590698242,
+      "learning_rate": 0.0,
+      "loss": 1.5894,
+      "step": 16340
+    },
+    {
+      "epoch": 3.807638565440149,
+      "grad_norm": 11.325212478637695,
+      "learning_rate": 0.0,
+      "loss": 1.7647,
+      "step": 16350
+    },
+    {
+      "epoch": 3.809967396367024,
+      "grad_norm": 34.25052261352539,
+      "learning_rate": 0.0,
+      "loss": 1.2847,
+      "step": 16360
+    },
+    {
+      "epoch": 3.8122962272938983,
+      "grad_norm": 12.110124588012695,
+      "learning_rate": 0.0,
+      "loss": 1.3453,
+      "step": 16370
+    },
+    {
+      "epoch": 3.814625058220773,
+      "grad_norm": 10.258769989013672,
+      "learning_rate": 0.0,
+      "loss": 0.7541,
+      "step": 16380
+    },
+    {
+      "epoch": 3.8169538891476478,
+      "grad_norm": 7.55911922454834,
+      "learning_rate": 0.0,
+      "loss": 1.1706,
+      "step": 16390
+    },
+    {
+      "epoch": 3.8192827200745225,
+      "grad_norm": 6.271566390991211,
+      "learning_rate": 0.0,
+      "loss": 0.9377,
+      "step": 16400
+    },
+    {
+      "epoch": 3.8216115510013973,
+      "grad_norm": 28.639516830444336,
+      "learning_rate": 0.0,
+      "loss": 1.2274,
+      "step": 16410
+    },
+    {
+      "epoch": 3.823940381928272,
+      "grad_norm": 5.265059471130371,
+      "learning_rate": 0.0,
+      "loss": 0.8603,
+      "step": 16420
+    },
+    {
+      "epoch": 3.8262692128551468,
+      "grad_norm": 8.896881103515625,
+      "learning_rate": 0.0,
+      "loss": 1.3604,
+      "step": 16430
+    },
+    {
+      "epoch": 3.8285980437820215,
+      "grad_norm": 8.662596702575684,
+      "learning_rate": 0.0,
+      "loss": 1.1663,
+      "step": 16440
+    },
+    {
+      "epoch": 3.8309268747088963,
+      "grad_norm": 7.28228759765625,
+      "learning_rate": 0.0,
+      "loss": 0.9696,
+      "step": 16450
+    },
+    {
+      "epoch": 3.833255705635771,
+      "grad_norm": 51.78569030761719,
+      "learning_rate": 0.0,
+      "loss": 1.6647,
+      "step": 16460
+    },
+    {
+      "epoch": 3.8355845365626458,
+      "grad_norm": 10.870829582214355,
+      "learning_rate": 0.0,
+      "loss": 1.3452,
+      "step": 16470
+    },
+    {
+      "epoch": 3.83791336748952,
+      "grad_norm": 74.10052490234375,
+      "learning_rate": 0.0,
+      "loss": 1.527,
+      "step": 16480
+    },
+    {
+      "epoch": 3.840242198416395,
+      "grad_norm": 15.562661170959473,
+      "learning_rate": 0.0,
+      "loss": 1.4266,
+      "step": 16490
+    },
+    {
+      "epoch": 3.8425710293432696,
+      "grad_norm": 8.279319763183594,
+      "learning_rate": 0.0,
+      "loss": 1.4034,
+      "step": 16500
+    },
+    {
+      "epoch": 3.8448998602701443,
+      "grad_norm": 55.923126220703125,
+      "learning_rate": 0.0,
+      "loss": 1.35,
+      "step": 16510
+    },
+    {
+      "epoch": 3.847228691197019,
+      "grad_norm": 15.255816459655762,
+      "learning_rate": 0.0,
+      "loss": 1.5191,
+      "step": 16520
+    },
+    {
+      "epoch": 3.849557522123894,
+      "grad_norm": 15.816597938537598,
+      "learning_rate": 0.0,
+      "loss": 1.0833,
+      "step": 16530
+    },
+    {
+      "epoch": 3.8518863530507685,
+      "grad_norm": 27.234275817871094,
+      "learning_rate": 0.0,
+      "loss": 0.9751,
+      "step": 16540
+    },
+    {
+      "epoch": 3.8542151839776433,
+      "grad_norm": 9.592534065246582,
+      "learning_rate": 0.0,
+      "loss": 1.2378,
+      "step": 16550
+    },
+    {
+      "epoch": 3.856544014904518,
+      "grad_norm": 5.3678388595581055,
+      "learning_rate": 0.0,
+      "loss": 1.6023,
+      "step": 16560
+    },
+    {
+      "epoch": 3.858872845831393,
+      "grad_norm": 16.641244888305664,
+      "learning_rate": 0.0,
+      "loss": 1.348,
+      "step": 16570
+    },
+    {
+      "epoch": 3.8612016767582675,
+      "grad_norm": 8.91340160369873,
+      "learning_rate": 0.0,
+      "loss": 1.0128,
+      "step": 16580
+    },
+    {
+      "epoch": 3.863530507685142,
+      "grad_norm": 27.965681076049805,
+      "learning_rate": 0.0,
+      "loss": 1.3651,
+      "step": 16590
+    },
+    {
+      "epoch": 3.865859338612017,
+      "grad_norm": 24.31036376953125,
+      "learning_rate": 0.0,
+      "loss": 1.1294,
+      "step": 16600
+    },
+    {
+      "epoch": 3.8681881695388913,
+      "grad_norm": 6.568557262420654,
+      "learning_rate": 0.0,
+      "loss": 1.6243,
+      "step": 16610
+    },
+    {
+      "epoch": 3.870517000465766,
+      "grad_norm": 8.548968315124512,
+      "learning_rate": 0.0,
+      "loss": 1.0038,
+      "step": 16620
+    },
+    {
+      "epoch": 3.872845831392641,
+      "grad_norm": 7.129677772521973,
+      "learning_rate": 0.0,
+      "loss": 1.0813,
+      "step": 16630
+    },
+    {
+      "epoch": 3.8751746623195156,
+      "grad_norm": 33.37204360961914,
+      "learning_rate": 0.0,
+      "loss": 0.9409,
+      "step": 16640
+    },
+    {
+      "epoch": 3.8775034932463903,
+      "grad_norm": 43.65248107910156,
+      "learning_rate": 0.0,
+      "loss": 1.1626,
+      "step": 16650
+    },
+    {
+      "epoch": 3.879832324173265,
+      "grad_norm": Infinity,
+      "learning_rate": 0.0,
+      "loss": 1.9948,
+      "step": 16660
+    },
+    {
+      "epoch": 3.88216115510014,
+      "grad_norm": 8.591267585754395,
+      "learning_rate": 0.0,
+      "loss": 1.1729,
+      "step": 16670
+    },
+    {
+      "epoch": 3.8844899860270146,
+      "grad_norm": 17.50629425048828,
+      "learning_rate": 0.0,
+      "loss": 1.2443,
+      "step": 16680
+    },
+    {
+      "epoch": 3.8868188169538893,
+      "grad_norm": 12.087096214294434,
+      "learning_rate": 0.0,
+      "loss": 1.3456,
+      "step": 16690
+    },
+    {
+      "epoch": 3.8891476478807636,
+      "grad_norm": 21.00797462463379,
+      "learning_rate": 0.0,
+      "loss": 1.3057,
+      "step": 16700
+    },
+    {
+      "epoch": 3.891476478807639,
+      "grad_norm": 9.580062866210938,
+      "learning_rate": 0.0,
+      "loss": 0.927,
+      "step": 16710
+    },
+    {
+      "epoch": 3.893805309734513,
+      "grad_norm": 12.416278839111328,
+      "learning_rate": 0.0,
+      "loss": 1.0927,
+      "step": 16720
+    },
+    {
+      "epoch": 3.896134140661388,
+      "grad_norm": 6.420504570007324,
+      "learning_rate": 0.0,
+      "loss": 0.8602,
+      "step": 16730
+    },
+    {
+      "epoch": 3.8984629715882626,
+      "grad_norm": 11.220664978027344,
+      "learning_rate": 0.0,
+      "loss": 1.1429,
+      "step": 16740
+    },
+    {
+      "epoch": 3.9007918025151374,
+      "grad_norm": 5.770464897155762,
+      "learning_rate": 0.0,
+      "loss": 0.6577,
+      "step": 16750
+    },
+    {
+      "epoch": 3.903120633442012,
+      "grad_norm": 17.317913055419922,
+      "learning_rate": 0.0,
+      "loss": 0.9757,
+      "step": 16760
+    },
+    {
+      "epoch": 3.905449464368887,
+      "grad_norm": 252.115234375,
+      "learning_rate": 0.0,
+      "loss": 0.9186,
+      "step": 16770
+    },
+    {
+      "epoch": 3.9077782952957616,
+      "grad_norm": 17.903982162475586,
+      "learning_rate": 0.0,
+      "loss": 1.0096,
+      "step": 16780
+    },
+    {
+      "epoch": 3.9101071262226363,
+      "grad_norm": 9.154571533203125,
+      "learning_rate": 0.0,
+      "loss": 1.1535,
+      "step": 16790
+    },
+    {
+      "epoch": 3.912435957149511,
+      "grad_norm": 5.846363544464111,
+      "learning_rate": 0.0,
+      "loss": 1.0427,
+      "step": 16800
+    },
+    {
+      "epoch": 3.9147647880763854,
+      "grad_norm": 72.26444244384766,
+      "learning_rate": 0.0,
+      "loss": 1.3003,
+      "step": 16810
+    },
+    {
+      "epoch": 3.9170936190032606,
+      "grad_norm": 11.39892578125,
+      "learning_rate": 0.0,
+      "loss": 1.1781,
+      "step": 16820
+    },
+    {
+      "epoch": 3.919422449930135,
+      "grad_norm": 4.515076160430908,
+      "learning_rate": 0.0,
+      "loss": 1.3092,
+      "step": 16830
+    },
+    {
+      "epoch": 3.9217512808570096,
+      "grad_norm": 12.91057300567627,
+      "learning_rate": 0.0,
+      "loss": 1.1982,
+      "step": 16840
+    },
+    {
+      "epoch": 3.9240801117838844,
+      "grad_norm": 11.3361234664917,
+      "learning_rate": 0.0,
+      "loss": 1.3631,
+      "step": 16850
+    },
+    {
+      "epoch": 3.926408942710759,
+      "grad_norm": 9.171483993530273,
+      "learning_rate": 0.0,
+      "loss": 0.8862,
+      "step": 16860
+    },
+    {
+      "epoch": 3.928737773637634,
+      "grad_norm": 14.986222267150879,
+      "learning_rate": 0.0,
+      "loss": 1.1989,
+      "step": 16870
+    },
+    {
+      "epoch": 3.9310666045645086,
+      "grad_norm": 35.58830642700195,
+      "learning_rate": 0.0,
+      "loss": 1.0229,
+      "step": 16880
+    },
+    {
+      "epoch": 3.9333954354913834,
+      "grad_norm": 6.8410234451293945,
+      "learning_rate": 0.0,
+      "loss": 1.095,
+      "step": 16890
+    },
+    {
+      "epoch": 3.935724266418258,
+      "grad_norm": 22.072486877441406,
+      "learning_rate": 0.0,
+      "loss": 1.9416,
+      "step": 16900
+    },
+    {
+      "epoch": 3.938053097345133,
+      "grad_norm": 14.240464210510254,
+      "learning_rate": 0.0,
+      "loss": 0.9298,
+      "step": 16910
+    },
+    {
+      "epoch": 3.9403819282720076,
+      "grad_norm": 8.36672306060791,
+      "learning_rate": 0.0,
+      "loss": 1.1451,
+      "step": 16920
+    },
+    {
+      "epoch": 3.9427107591988824,
+      "grad_norm": 7.652619361877441,
+      "learning_rate": 0.0,
+      "loss": 0.8321,
+      "step": 16930
+    },
+    {
+      "epoch": 3.9450395901257567,
+      "grad_norm": 12.397917747497559,
+      "learning_rate": 0.0,
+      "loss": 0.9605,
+      "step": 16940
+    },
+    {
+      "epoch": 3.9473684210526314,
+      "grad_norm": 5.459069728851318,
+      "learning_rate": 0.0,
+      "loss": 1.0881,
+      "step": 16950
+    },
+    {
+      "epoch": 3.949697251979506,
+      "grad_norm": 29.952774047851562,
+      "learning_rate": 0.0,
+      "loss": 1.303,
+      "step": 16960
+    },
+    {
+      "epoch": 3.952026082906381,
+      "grad_norm": 4.160209655761719,
+      "learning_rate": 0.0,
+      "loss": 1.0394,
+      "step": 16970
+    },
+    {
+      "epoch": 3.9543549138332557,
+      "grad_norm": 8.1553316116333,
+      "learning_rate": 0.0,
+      "loss": 0.9403,
+      "step": 16980
+    },
+    {
+      "epoch": 3.9566837447601304,
+      "grad_norm": 13.145912170410156,
+      "learning_rate": 0.0,
+      "loss": 1.3667,
+      "step": 16990
+    },
+    {
+      "epoch": 3.959012575687005,
+      "grad_norm": 15.345560073852539,
+      "learning_rate": 0.0,
+      "loss": 1.0222,
+      "step": 17000
+    },
+    {
+      "epoch": 3.96134140661388,
+      "grad_norm": 19.302764892578125,
+      "learning_rate": 0.0,
+      "loss": 1.044,
+      "step": 17010
+    },
+    {
+      "epoch": 3.9636702375407546,
+      "grad_norm": 7.825178146362305,
+      "learning_rate": 0.0,
+      "loss": 0.8727,
+      "step": 17020
+    },
+    {
+      "epoch": 3.9659990684676294,
+      "grad_norm": 13.018383026123047,
+      "learning_rate": 0.0,
+      "loss": 1.053,
+      "step": 17030
+    },
+    {
+      "epoch": 3.968327899394504,
+      "grad_norm": 29.119361877441406,
+      "learning_rate": 0.0,
+      "loss": 1.3638,
+      "step": 17040
+    },
+    {
+      "epoch": 3.9706567303213784,
+      "grad_norm": 8.266595840454102,
+      "learning_rate": 0.0,
+      "loss": 1.2011,
+      "step": 17050
+    },
+    {
+      "epoch": 3.9729855612482536,
+      "grad_norm": 8.626734733581543,
+      "learning_rate": 0.0,
+      "loss": 1.249,
+      "step": 17060
+    },
+    {
+      "epoch": 3.975314392175128,
+      "grad_norm": 67.26522064208984,
+      "learning_rate": 0.0,
+      "loss": 1.9316,
+      "step": 17070
+    },
+    {
+      "epoch": 3.9776432231020027,
+      "grad_norm": 8.982076644897461,
+      "learning_rate": 0.0,
+      "loss": 1.4822,
+      "step": 17080
+    },
+    {
+      "epoch": 3.9799720540288774,
+      "grad_norm": 7.744446754455566,
+      "learning_rate": 0.0,
+      "loss": 1.1842,
+      "step": 17090
+    },
+    {
+      "epoch": 3.982300884955752,
+      "grad_norm": 14.065244674682617,
+      "learning_rate": 0.0,
+      "loss": 1.3703,
+      "step": 17100
+    },
+    {
+      "epoch": 3.984629715882627,
+      "grad_norm": 56.45340347290039,
+      "learning_rate": 0.0,
+      "loss": 0.9846,
+      "step": 17110
+    },
+    {
+      "epoch": 3.9869585468095017,
+      "grad_norm": 8.13912296295166,
+      "learning_rate": 0.0,
+      "loss": 0.9176,
+      "step": 17120
+    },
+    {
+      "epoch": 3.9892873777363764,
+      "grad_norm": 12.026097297668457,
+      "learning_rate": 0.0,
+      "loss": 1.2079,
+      "step": 17130
+    },
+    {
+      "epoch": 3.991616208663251,
+      "grad_norm": 11.71669864654541,
+      "learning_rate": 0.0,
+      "loss": 1.7823,
+      "step": 17140
+    },
+    {
+      "epoch": 3.993945039590126,
+      "grad_norm": 10.856141090393066,
+      "learning_rate": 0.0,
+      "loss": 1.5076,
+      "step": 17150
+    },
+    {
+      "epoch": 3.9962738705170002,
+      "grad_norm": 111.23175048828125,
+      "learning_rate": 0.0,
+      "loss": 1.7607,
+      "step": 17160
+    },
+    {
+      "epoch": 3.9986027014438754,
+      "grad_norm": 21.408830642700195,
+      "learning_rate": 0.0,
+      "loss": 1.233,
+      "step": 17170
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 0.49088114500045776,
+      "eval_runtime": 15.277,
+      "eval_samples_per_second": 285.134,
+      "eval_steps_per_second": 2.291,
+      "step": 17176
+    },
+    {
+      "epoch": 4.00093153237075,
+      "grad_norm": 42.99901580810547,
+      "learning_rate": 0.0,
+      "loss": 1.3849,
+      "step": 17180
+    },
+    {
+      "epoch": 4.003260363297625,
+      "grad_norm": 15.593567848205566,
+      "learning_rate": 0.0,
+      "loss": 0.9876,
+      "step": 17190
+    },
+    {
+      "epoch": 4.005589194224499,
+      "grad_norm": 9.418315887451172,
+      "learning_rate": 0.0,
+      "loss": 0.775,
+      "step": 17200
+    },
+    {
+      "epoch": 4.007918025151374,
+      "grad_norm": 14.26424503326416,
+      "learning_rate": 0.0,
+      "loss": 0.993,
+      "step": 17210
+    },
+    {
+      "epoch": 4.010246856078249,
+      "grad_norm": 28.059171676635742,
+      "learning_rate": 0.0,
+      "loss": 1.4448,
+      "step": 17220
+    },
+    {
+      "epoch": 4.012575687005123,
+      "grad_norm": 6.072986125946045,
+      "learning_rate": 0.0,
+      "loss": 1.2469,
+      "step": 17230
+    },
+    {
+      "epoch": 4.014904517931998,
+      "grad_norm": 10.683692932128906,
+      "learning_rate": 0.0,
+      "loss": 1.218,
+      "step": 17240
+    },
+    {
+      "epoch": 4.0172333488588725,
+      "grad_norm": 76.68214416503906,
+      "learning_rate": 0.0,
+      "loss": 1.4419,
+      "step": 17250
+    },
+    {
+      "epoch": 4.019562179785748,
+      "grad_norm": 17.647035598754883,
+      "learning_rate": 0.0,
+      "loss": 1.8136,
+      "step": 17260
+    },
+    {
+      "epoch": 4.021891010712622,
+      "grad_norm": 52.65121078491211,
+      "learning_rate": 0.0,
+      "loss": 0.8525,
+      "step": 17270
+    },
+    {
+      "epoch": 4.024219841639497,
+      "grad_norm": 10.610958099365234,
+      "learning_rate": 0.0,
+      "loss": 1.2737,
+      "step": 17280
+    },
+    {
+      "epoch": 4.0265486725663715,
+      "grad_norm": 18.813425064086914,
+      "learning_rate": 0.0,
+      "loss": 0.8997,
+      "step": 17290
+    },
+    {
+      "epoch": 4.028877503493247,
+      "grad_norm": 21.70282554626465,
+      "learning_rate": 0.0,
+      "loss": 0.9929,
+      "step": 17300
+    },
+    {
+      "epoch": 4.031206334420121,
+      "grad_norm": 5.4507036209106445,
+      "learning_rate": 0.0,
+      "loss": 1.9675,
+      "step": 17310
+    },
+    {
+      "epoch": 4.033535165346996,
+      "grad_norm": 7.815976619720459,
+      "learning_rate": 0.0,
+      "loss": 0.6941,
+      "step": 17320
+    },
+    {
+      "epoch": 4.0358639962738705,
+      "grad_norm": 21.757108688354492,
+      "learning_rate": 0.0,
+      "loss": 1.1081,
+      "step": 17330
+    },
+    {
+      "epoch": 4.038192827200745,
+      "grad_norm": 67.65559387207031,
+      "learning_rate": 0.0,
+      "loss": 1.4166,
+      "step": 17340
+    },
+    {
+      "epoch": 4.04052165812762,
+      "grad_norm": 9.288952827453613,
+      "learning_rate": 0.0,
+      "loss": 1.2824,
+      "step": 17350
+    },
+    {
+      "epoch": 4.042850489054494,
+      "grad_norm": 41.26054382324219,
+      "learning_rate": 0.0,
+      "loss": 1.3587,
+      "step": 17360
+    },
+    {
+      "epoch": 4.0451793199813695,
+      "grad_norm": 6.358667850494385,
+      "learning_rate": 0.0,
+      "loss": 0.7649,
+      "step": 17370
+    },
+    {
+      "epoch": 4.047508150908244,
+      "grad_norm": 10.879199981689453,
+      "learning_rate": 0.0,
+      "loss": 1.4232,
+      "step": 17380
+    },
+    {
+      "epoch": 4.049836981835119,
+      "grad_norm": 6.230020523071289,
+      "learning_rate": 0.0,
+      "loss": 1.0463,
+      "step": 17390
+    },
+    {
+      "epoch": 4.052165812761993,
+      "grad_norm": 15.603515625,
+      "learning_rate": 0.0,
+      "loss": 1.223,
+      "step": 17400
+    },
+    {
+      "epoch": 4.0544946436888685,
+      "grad_norm": 14.09308910369873,
+      "learning_rate": 0.0,
+      "loss": 1.0864,
+      "step": 17410
+    },
+    {
+      "epoch": 4.056823474615743,
+      "grad_norm": 65.15744018554688,
+      "learning_rate": 0.0,
+      "loss": 1.2502,
+      "step": 17420
+    },
+    {
+      "epoch": 4.059152305542618,
+      "grad_norm": 53.656063079833984,
+      "learning_rate": 0.0,
+      "loss": 0.9074,
+      "step": 17430
+    },
+    {
+      "epoch": 4.061481136469492,
+      "grad_norm": 26.082855224609375,
+      "learning_rate": 0.0,
+      "loss": 1.3669,
+      "step": 17440
+    },
+    {
+      "epoch": 4.063809967396367,
+      "grad_norm": 37.48140335083008,
+      "learning_rate": 0.0,
+      "loss": 1.2421,
+      "step": 17450
+    },
+    {
+      "epoch": 4.066138798323242,
+      "grad_norm": 8.820276260375977,
+      "learning_rate": 0.0,
+      "loss": 1.2692,
+      "step": 17460
+    },
+    {
+      "epoch": 4.068467629250116,
+      "grad_norm": 11.496400833129883,
+      "learning_rate": 0.0,
+      "loss": 1.0396,
+      "step": 17470
+    },
+    {
+      "epoch": 4.070796460176991,
+      "grad_norm": 18.882984161376953,
+      "learning_rate": 0.0,
+      "loss": 0.9758,
+      "step": 17480
+    },
+    {
+      "epoch": 4.073125291103866,
+      "grad_norm": 8.582090377807617,
+      "learning_rate": 0.0,
+      "loss": 1.1566,
+      "step": 17490
+    },
+    {
+      "epoch": 4.075454122030741,
+      "grad_norm": 13.812816619873047,
+      "learning_rate": 0.0,
+      "loss": 1.6421,
+      "step": 17500
+    },
+    {
+      "epoch": 4.077782952957615,
+      "grad_norm": 52.63260269165039,
+      "learning_rate": 0.0,
+      "loss": 1.0587,
+      "step": 17510
+    },
+    {
+      "epoch": 4.08011178388449,
+      "grad_norm": 10.908785820007324,
+      "learning_rate": 0.0,
+      "loss": 0.855,
+      "step": 17520
+    },
+    {
+      "epoch": 4.0824406148113646,
+      "grad_norm": 8.219367027282715,
+      "learning_rate": 0.0,
+      "loss": 1.009,
+      "step": 17530
+    },
+    {
+      "epoch": 4.08476944573824,
+      "grad_norm": 16.966997146606445,
+      "learning_rate": 0.0,
+      "loss": 1.2723,
+      "step": 17540
+    },
+    {
+      "epoch": 4.087098276665114,
+      "grad_norm": 14.397762298583984,
+      "learning_rate": 0.0,
+      "loss": 1.0312,
+      "step": 17550
+    },
+    {
+      "epoch": 4.089427107591989,
+      "grad_norm": 13.03594970703125,
+      "learning_rate": 0.0,
+      "loss": 1.2744,
+      "step": 17560
+    },
+    {
+      "epoch": 4.0917559385188635,
+      "grad_norm": 6.213973522186279,
+      "learning_rate": 0.0,
+      "loss": 1.0554,
+      "step": 17570
+    },
+    {
+      "epoch": 4.094084769445738,
+      "grad_norm": 42.39612579345703,
+      "learning_rate": 0.0,
+      "loss": 1.4296,
+      "step": 17580
+    },
+    {
+      "epoch": 4.096413600372613,
+      "grad_norm": 6.726439476013184,
+      "learning_rate": 0.0,
+      "loss": 1.0909,
+      "step": 17590
+    },
+    {
+      "epoch": 4.098742431299487,
+      "grad_norm": 12.621766090393066,
+      "learning_rate": 0.0,
+      "loss": 1.542,
+      "step": 17600
+    },
+    {
+      "epoch": 4.1010712622263625,
+      "grad_norm": 11.763903617858887,
+      "learning_rate": 0.0,
+      "loss": 1.6493,
+      "step": 17610
+    },
+    {
+      "epoch": 4.103400093153237,
+      "grad_norm": 6.310747146606445,
+      "learning_rate": 0.0,
+      "loss": 1.0642,
+      "step": 17620
+    },
+    {
+      "epoch": 4.105728924080112,
+      "grad_norm": 13.970367431640625,
+      "learning_rate": 0.0,
+      "loss": 1.2096,
+      "step": 17630
+    },
+    {
+      "epoch": 4.108057755006986,
+      "grad_norm": 18.80079460144043,
+      "learning_rate": 0.0,
+      "loss": 1.1508,
+      "step": 17640
+    },
+    {
+      "epoch": 4.1103865859338615,
+      "grad_norm": 5.596470832824707,
+      "learning_rate": 0.0,
+      "loss": 1.6874,
+      "step": 17650
+    },
+    {
+      "epoch": 4.112715416860736,
+      "grad_norm": 21.15366554260254,
+      "learning_rate": 0.0,
+      "loss": 1.482,
+      "step": 17660
+    },
+    {
+      "epoch": 4.115044247787611,
+      "grad_norm": 14.376870155334473,
+      "learning_rate": 0.0,
+      "loss": 1.2993,
+      "step": 17670
+    },
+    {
+      "epoch": 4.117373078714485,
+      "grad_norm": 3.591233730316162,
+      "learning_rate": 0.0,
+      "loss": 1.1712,
+      "step": 17680
+    },
+    {
+      "epoch": 4.11970190964136,
+      "grad_norm": 85.7278823852539,
+      "learning_rate": 0.0,
+      "loss": 2.0886,
+      "step": 17690
+    },
+    {
+      "epoch": 4.122030740568235,
+      "grad_norm": 24.37750816345215,
+      "learning_rate": 0.0,
+      "loss": 1.1597,
+      "step": 17700
+    },
+    {
+      "epoch": 4.124359571495109,
+      "grad_norm": 9.48483657836914,
+      "learning_rate": 0.0,
+      "loss": 0.9762,
+      "step": 17710
+    },
+    {
+      "epoch": 4.126688402421984,
+      "grad_norm": 16.925823211669922,
+      "learning_rate": 0.0,
+      "loss": 0.8279,
+      "step": 17720
+    },
+    {
+      "epoch": 4.129017233348859,
+      "grad_norm": 5.349981784820557,
+      "learning_rate": 0.0,
+      "loss": 1.1843,
+      "step": 17730
+    },
+    {
+      "epoch": 4.131346064275734,
+      "grad_norm": 6.957655906677246,
+      "learning_rate": 0.0,
+      "loss": 1.16,
+      "step": 17740
+    },
+    {
+      "epoch": 4.133674895202608,
+      "grad_norm": 8.322871208190918,
+      "learning_rate": 0.0,
+      "loss": 0.7904,
+      "step": 17750
+    },
+    {
+      "epoch": 4.136003726129483,
+      "grad_norm": 37.46146774291992,
+      "learning_rate": 0.0,
+      "loss": 1.2393,
+      "step": 17760
+    },
+    {
+      "epoch": 4.138332557056358,
+      "grad_norm": 4.923023700714111,
+      "learning_rate": 0.0,
+      "loss": 2.4185,
+      "step": 17770
+    },
+    {
+      "epoch": 4.140661387983233,
+      "grad_norm": 21.715543746948242,
+      "learning_rate": 0.0,
+      "loss": 1.0933,
+      "step": 17780
+    },
+    {
+      "epoch": 4.142990218910107,
+      "grad_norm": 26.76923179626465,
+      "learning_rate": 0.0,
+      "loss": 1.3966,
+      "step": 17790
+    },
+    {
+      "epoch": 4.145319049836981,
+      "grad_norm": 18.267608642578125,
+      "learning_rate": 0.0,
+      "loss": 1.5447,
+      "step": 17800
+    },
+    {
+      "epoch": 4.147647880763857,
+      "grad_norm": 13.733194351196289,
+      "learning_rate": 0.0,
+      "loss": 1.4679,
+      "step": 17810
+    },
+    {
+      "epoch": 4.149976711690731,
+      "grad_norm": 5.581467628479004,
+      "learning_rate": 0.0,
+      "loss": 1.0773,
+      "step": 17820
+    },
+    {
+      "epoch": 4.152305542617606,
+      "grad_norm": 8.281193733215332,
+      "learning_rate": 0.0,
+      "loss": 1.3519,
+      "step": 17830
+    },
+    {
+      "epoch": 4.15463437354448,
+      "grad_norm": 16.052032470703125,
+      "learning_rate": 0.0,
+      "loss": 1.203,
+      "step": 17840
+    },
+    {
+      "epoch": 4.156963204471356,
+      "grad_norm": 32.61860656738281,
+      "learning_rate": 0.0,
+      "loss": 1.0071,
+      "step": 17850
+    },
+    {
+      "epoch": 4.15929203539823,
+      "grad_norm": 10.298188209533691,
+      "learning_rate": 0.0,
+      "loss": 1.2432,
+      "step": 17860
+    },
+    {
+      "epoch": 4.161620866325105,
+      "grad_norm": 13.705058097839355,
+      "learning_rate": 0.0,
+      "loss": 1.2159,
+      "step": 17870
+    },
+    {
+      "epoch": 4.163949697251979,
+      "grad_norm": 7.2891669273376465,
+      "learning_rate": 0.0,
+      "loss": 1.7573,
+      "step": 17880
+    },
+    {
+      "epoch": 4.166278528178855,
+      "grad_norm": 4.742997169494629,
+      "learning_rate": 0.0,
+      "loss": 1.3245,
+      "step": 17890
+    },
+    {
+      "epoch": 4.168607359105729,
+      "grad_norm": 9.10926342010498,
+      "learning_rate": 0.0,
+      "loss": 1.3587,
+      "step": 17900
+    },
+    {
+      "epoch": 4.170936190032604,
+      "grad_norm": 34.803287506103516,
+      "learning_rate": 0.0,
+      "loss": 0.8937,
+      "step": 17910
+    },
+    {
+      "epoch": 4.173265020959478,
+      "grad_norm": 24.677139282226562,
+      "learning_rate": 0.0,
+      "loss": 1.3338,
+      "step": 17920
+    },
+    {
+      "epoch": 4.175593851886353,
+      "grad_norm": 10.664273262023926,
+      "learning_rate": 0.0,
+      "loss": 1.0591,
+      "step": 17930
+    },
+    {
+      "epoch": 4.177922682813228,
+      "grad_norm": 10.509198188781738,
+      "learning_rate": 0.0,
+      "loss": 1.0005,
+      "step": 17940
+    },
+    {
+      "epoch": 4.180251513740102,
+      "grad_norm": 7.73974609375,
+      "learning_rate": 0.0,
+      "loss": 1.1359,
+      "step": 17950
+    },
+    {
+      "epoch": 4.182580344666977,
+      "grad_norm": 6.254695415496826,
+      "learning_rate": 0.0,
+      "loss": 0.8587,
+      "step": 17960
+    },
+    {
+      "epoch": 4.184909175593852,
+      "grad_norm": 8.58466625213623,
+      "learning_rate": 0.0,
+      "loss": 0.9862,
+      "step": 17970
+    },
+    {
+      "epoch": 4.187238006520727,
+      "grad_norm": 11.516722679138184,
+      "learning_rate": 0.0,
+      "loss": 1.5512,
+      "step": 17980
+    },
+    {
+      "epoch": 4.189566837447601,
+      "grad_norm": 25.665462493896484,
+      "learning_rate": 0.0,
+      "loss": 1.3756,
+      "step": 17990
+    },
+    {
+      "epoch": 4.191895668374476,
+      "grad_norm": 205.93728637695312,
+      "learning_rate": 0.0,
+      "loss": 1.4687,
+      "step": 18000
+    },
+    {
+      "epoch": 4.194224499301351,
+      "grad_norm": 5.3130340576171875,
+      "learning_rate": 0.0,
+      "loss": 1.0205,
+      "step": 18010
+    },
+    {
+      "epoch": 4.196553330228226,
+      "grad_norm": 9.158675193786621,
+      "learning_rate": 0.0,
+      "loss": 0.903,
+      "step": 18020
+    },
+    {
+      "epoch": 4.1988821611551,
+      "grad_norm": 11.879533767700195,
+      "learning_rate": 0.0,
+      "loss": 1.3574,
+      "step": 18030
+    },
+    {
+      "epoch": 4.2012109920819745,
+      "grad_norm": 56.382537841796875,
+      "learning_rate": 0.0,
+      "loss": 1.7197,
+      "step": 18040
+    },
+    {
+      "epoch": 4.20353982300885,
+      "grad_norm": 14.329750061035156,
+      "learning_rate": 0.0,
+      "loss": 1.4055,
+      "step": 18050
+    },
+    {
+      "epoch": 4.205868653935724,
+      "grad_norm": 8.853121757507324,
+      "learning_rate": 0.0,
+      "loss": 1.519,
+      "step": 18060
+    },
+    {
+      "epoch": 4.208197484862599,
+      "grad_norm": 12.149033546447754,
+      "learning_rate": 0.0,
+      "loss": 1.1448,
+      "step": 18070
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 15.463021278381348,
+      "learning_rate": 0.0,
+      "loss": 0.9091,
+      "step": 18080
+    },
+    {
+      "epoch": 4.212855146716349,
+      "grad_norm": 17.922626495361328,
+      "learning_rate": 0.0,
+      "loss": 1.2816,
+      "step": 18090
+    },
+    {
+      "epoch": 4.215183977643223,
+      "grad_norm": 16.048818588256836,
+      "learning_rate": 0.0,
+      "loss": 1.2221,
+      "step": 18100
+    },
+    {
+      "epoch": 4.217512808570098,
+      "grad_norm": 7.998112201690674,
+      "learning_rate": 0.0,
+      "loss": 1.4312,
+      "step": 18110
+    },
+    {
+      "epoch": 4.219841639496972,
+      "grad_norm": 6.729568004608154,
+      "learning_rate": 0.0,
+      "loss": 0.8463,
+      "step": 18120
+    },
+    {
+      "epoch": 4.222170470423848,
+      "grad_norm": 9.382912635803223,
+      "learning_rate": 0.0,
+      "loss": 2.0268,
+      "step": 18130
+    },
+    {
+      "epoch": 4.224499301350722,
+      "grad_norm": 11.232508659362793,
+      "learning_rate": 0.0,
+      "loss": 1.6308,
+      "step": 18140
+    },
+    {
+      "epoch": 4.226828132277596,
+      "grad_norm": 7.6993536949157715,
+      "learning_rate": 0.0,
+      "loss": 0.9748,
+      "step": 18150
+    },
+    {
+      "epoch": 4.229156963204471,
+      "grad_norm": 14.037554740905762,
+      "learning_rate": 0.0,
+      "loss": 1.0376,
+      "step": 18160
+    },
+    {
+      "epoch": 4.231485794131346,
+      "grad_norm": 37.87745666503906,
+      "learning_rate": 0.0,
+      "loss": 2.3579,
+      "step": 18170
+    },
+    {
+      "epoch": 4.233814625058221,
+      "grad_norm": 51.04754638671875,
+      "learning_rate": 0.0,
+      "loss": 1.3791,
+      "step": 18180
+    },
+    {
+      "epoch": 4.236143455985095,
+      "grad_norm": 11.547856330871582,
+      "learning_rate": 0.0,
+      "loss": 0.8157,
+      "step": 18190
+    },
+    {
+      "epoch": 4.23847228691197,
+      "grad_norm": 9.100687980651855,
+      "learning_rate": 0.0,
+      "loss": 0.8207,
+      "step": 18200
+    },
+    {
+      "epoch": 4.240801117838845,
+      "grad_norm": 12.361568450927734,
+      "learning_rate": 0.0,
+      "loss": 1.0341,
+      "step": 18210
+    },
+    {
+      "epoch": 4.24312994876572,
+      "grad_norm": 15.49877643585205,
+      "learning_rate": 0.0,
+      "loss": 1.5983,
+      "step": 18220
+    },
+    {
+      "epoch": 4.245458779692594,
+      "grad_norm": 51.21619415283203,
+      "learning_rate": 0.0,
+      "loss": 1.2015,
+      "step": 18230
+    },
+    {
+      "epoch": 4.247787610619469,
+      "grad_norm": 21.072568893432617,
+      "learning_rate": 0.0,
+      "loss": 0.853,
+      "step": 18240
+    },
+    {
+      "epoch": 4.250116441546344,
+      "grad_norm": 5.333137512207031,
+      "learning_rate": 0.0,
+      "loss": 0.89,
+      "step": 18250
+    },
+    {
+      "epoch": 4.252445272473219,
+      "grad_norm": 5.594559669494629,
+      "learning_rate": 0.0,
+      "loss": 1.3852,
+      "step": 18260
+    },
+    {
+      "epoch": 4.254774103400093,
+      "grad_norm": 9.22107982635498,
+      "learning_rate": 0.0,
+      "loss": 0.9253,
+      "step": 18270
+    },
+    {
+      "epoch": 4.2571029343269675,
+      "grad_norm": 17.17836570739746,
+      "learning_rate": 0.0,
+      "loss": 0.8569,
+      "step": 18280
+    },
+    {
+      "epoch": 4.259431765253843,
+      "grad_norm": 7.7796630859375,
+      "learning_rate": 0.0,
+      "loss": 0.9931,
+      "step": 18290
+    },
+    {
+      "epoch": 4.261760596180717,
+      "grad_norm": 10.182730674743652,
+      "learning_rate": 0.0,
+      "loss": 0.9802,
+      "step": 18300
+    },
+    {
+      "epoch": 4.264089427107592,
+      "grad_norm": 6.655672073364258,
+      "learning_rate": 0.0,
+      "loss": 0.8961,
+      "step": 18310
+    },
+    {
+      "epoch": 4.2664182580344665,
+      "grad_norm": 6.27683162689209,
+      "learning_rate": 0.0,
+      "loss": 1.6628,
+      "step": 18320
+    },
+    {
+      "epoch": 4.268747088961342,
+      "grad_norm": 10.62850570678711,
+      "learning_rate": 0.0,
+      "loss": 1.5432,
+      "step": 18330
+    },
+    {
+      "epoch": 4.271075919888216,
+      "grad_norm": 25.98421287536621,
+      "learning_rate": 0.0,
+      "loss": 1.2067,
+      "step": 18340
+    },
+    {
+      "epoch": 4.273404750815091,
+      "grad_norm": 27.159278869628906,
+      "learning_rate": 0.0,
+      "loss": 1.2735,
+      "step": 18350
+    },
+    {
+      "epoch": 4.2757335817419655,
+      "grad_norm": 23.231164932250977,
+      "learning_rate": 0.0,
+      "loss": 1.3331,
+      "step": 18360
+    },
+    {
+      "epoch": 4.27806241266884,
+      "grad_norm": 96.513427734375,
+      "learning_rate": 0.0,
+      "loss": 1.5835,
+      "step": 18370
+    },
+    {
+      "epoch": 4.280391243595715,
+      "grad_norm": 6.849877834320068,
+      "learning_rate": 0.0,
+      "loss": 1.0837,
+      "step": 18380
+    },
+    {
+      "epoch": 4.282720074522589,
+      "grad_norm": 7.442630767822266,
+      "learning_rate": 0.0,
+      "loss": 1.4346,
+      "step": 18390
+    },
+    {
+      "epoch": 4.2850489054494645,
+      "grad_norm": 62.38019561767578,
+      "learning_rate": 0.0,
+      "loss": 1.2822,
+      "step": 18400
+    },
+    {
+      "epoch": 4.287377736376339,
+      "grad_norm": 7.65008020401001,
+      "learning_rate": 0.0,
+      "loss": 1.3873,
+      "step": 18410
+    },
+    {
+      "epoch": 4.289706567303214,
+      "grad_norm": 31.235200881958008,
+      "learning_rate": 0.0,
+      "loss": 1.4172,
+      "step": 18420
+    },
+    {
+      "epoch": 4.292035398230088,
+      "grad_norm": 5.858349800109863,
+      "learning_rate": 0.0,
+      "loss": 0.7914,
+      "step": 18430
+    },
+    {
+      "epoch": 4.2943642291569635,
+      "grad_norm": 8.754595756530762,
+      "learning_rate": 0.0,
+      "loss": 1.2774,
+      "step": 18440
+    },
+    {
+      "epoch": 4.296693060083838,
+      "grad_norm": 64.94791412353516,
+      "learning_rate": 0.0,
+      "loss": 1.2843,
+      "step": 18450
+    },
+    {
+      "epoch": 4.299021891010713,
+      "grad_norm": 3.240407943725586,
+      "learning_rate": 0.0,
+      "loss": 0.8924,
+      "step": 18460
+    },
+    {
+      "epoch": 4.301350721937587,
+      "grad_norm": 15.686180114746094,
+      "learning_rate": 0.0,
+      "loss": 1.1514,
+      "step": 18470
+    },
+    {
+      "epoch": 4.3036795528644625,
+      "grad_norm": 10.717156410217285,
+      "learning_rate": 0.0,
+      "loss": 1.4731,
+      "step": 18480
+    },
+    {
+      "epoch": 4.306008383791337,
+      "grad_norm": 8.399251937866211,
+      "learning_rate": 0.0,
+      "loss": 1.0628,
+      "step": 18490
+    },
+    {
+      "epoch": 4.308337214718211,
+      "grad_norm": 5.603919506072998,
+      "learning_rate": 0.0,
+      "loss": 1.1527,
+      "step": 18500
+    },
+    {
+      "epoch": 4.310666045645086,
+      "grad_norm": 30.59088706970215,
+      "learning_rate": 0.0,
+      "loss": 0.9832,
+      "step": 18510
+    },
+    {
+      "epoch": 4.312994876571961,
+      "grad_norm": 9.03952407836914,
+      "learning_rate": 0.0,
+      "loss": 1.0315,
+      "step": 18520
+    },
+    {
+      "epoch": 4.315323707498836,
+      "grad_norm": 59.89068603515625,
+      "learning_rate": 0.0,
+      "loss": 1.0991,
+      "step": 18530
+    },
+    {
+      "epoch": 4.31765253842571,
+      "grad_norm": 105.38227844238281,
+      "learning_rate": 0.0,
+      "loss": 1.2814,
+      "step": 18540
+    },
+    {
+      "epoch": 4.319981369352585,
+      "grad_norm": 15.269956588745117,
+      "learning_rate": 0.0,
+      "loss": 1.2359,
+      "step": 18550
+    },
+    {
+      "epoch": 4.32231020027946,
+      "grad_norm": 57.37482833862305,
+      "learning_rate": 0.0,
+      "loss": 1.163,
+      "step": 18560
+    },
+    {
+      "epoch": 4.324639031206335,
+      "grad_norm": 23.360641479492188,
+      "learning_rate": 0.0,
+      "loss": 0.8576,
+      "step": 18570
+    },
+    {
+      "epoch": 4.326967862133209,
+      "grad_norm": 14.31608772277832,
+      "learning_rate": 0.0,
+      "loss": 1.4788,
+      "step": 18580
+    },
+    {
+      "epoch": 4.329296693060084,
+      "grad_norm": 40.62135314941406,
+      "learning_rate": 0.0,
+      "loss": 1.2003,
+      "step": 18590
+    },
+    {
+      "epoch": 4.3316255239869585,
+      "grad_norm": 39.43674850463867,
+      "learning_rate": 0.0,
+      "loss": 1.2967,
+      "step": 18600
+    },
+    {
+      "epoch": 4.333954354913833,
+      "grad_norm": 25.50300407409668,
+      "learning_rate": 0.0,
+      "loss": 1.8007,
+      "step": 18610
+    },
+    {
+      "epoch": 4.336283185840708,
+      "grad_norm": 56.91150665283203,
+      "learning_rate": 0.0,
+      "loss": 1.2731,
+      "step": 18620
+    },
+    {
+      "epoch": 4.338612016767582,
+      "grad_norm": 13.99095344543457,
+      "learning_rate": 0.0,
+      "loss": 1.4154,
+      "step": 18630
+    },
+    {
+      "epoch": 4.3409408476944575,
+      "grad_norm": 15.199361801147461,
+      "learning_rate": 0.0,
+      "loss": 1.5806,
+      "step": 18640
+    },
+    {
+      "epoch": 4.343269678621332,
+      "grad_norm": 45.43046188354492,
+      "learning_rate": 0.0,
+      "loss": 1.4277,
+      "step": 18650
+    },
+    {
+      "epoch": 4.345598509548207,
+      "grad_norm": 4.511590480804443,
+      "learning_rate": 0.0,
+      "loss": 0.894,
+      "step": 18660
+    },
+    {
+      "epoch": 4.347927340475081,
+      "grad_norm": 51.11252212524414,
+      "learning_rate": 0.0,
+      "loss": 1.3697,
+      "step": 18670
+    },
+    {
+      "epoch": 4.3502561714019565,
+      "grad_norm": 5.312637805938721,
+      "learning_rate": 0.0,
+      "loss": 1.1775,
+      "step": 18680
+    },
+    {
+      "epoch": 4.352585002328831,
+      "grad_norm": 23.01420021057129,
+      "learning_rate": 0.0,
+      "loss": 1.209,
+      "step": 18690
+    },
+    {
+      "epoch": 4.354913833255706,
+      "grad_norm": 7.891007423400879,
+      "learning_rate": 0.0,
+      "loss": 0.8061,
+      "step": 18700
+    },
+    {
+      "epoch": 4.35724266418258,
+      "grad_norm": 11.448382377624512,
+      "learning_rate": 0.0,
+      "loss": 1.6594,
+      "step": 18710
+    },
+    {
+      "epoch": 4.359571495109455,
+      "grad_norm": 18.895126342773438,
+      "learning_rate": 0.0,
+      "loss": 1.0112,
+      "step": 18720
+    },
+    {
+      "epoch": 4.36190032603633,
+      "grad_norm": 15.76264762878418,
+      "learning_rate": 0.0,
+      "loss": 0.9219,
+      "step": 18730
+    },
+    {
+      "epoch": 4.364229156963204,
+      "grad_norm": 13.89644718170166,
+      "learning_rate": 0.0,
+      "loss": 1.4194,
+      "step": 18740
+    },
+    {
+      "epoch": 4.366557987890079,
+      "grad_norm": 7.580521106719971,
+      "learning_rate": 0.0,
+      "loss": 1.0652,
+      "step": 18750
+    },
+    {
+      "epoch": 4.368886818816954,
+      "grad_norm": 4.65724515914917,
+      "learning_rate": 0.0,
+      "loss": 1.1159,
+      "step": 18760
+    },
+    {
+      "epoch": 4.371215649743829,
+      "grad_norm": 7.993378162384033,
+      "learning_rate": 0.0,
+      "loss": 0.9724,
+      "step": 18770
+    },
+    {
+      "epoch": 4.373544480670703,
+      "grad_norm": 5.239194393157959,
+      "learning_rate": 0.0,
+      "loss": 0.9784,
+      "step": 18780
+    },
+    {
+      "epoch": 4.375873311597578,
+      "grad_norm": 7.044046878814697,
+      "learning_rate": 0.0,
+      "loss": 0.8269,
+      "step": 18790
+    },
+    {
+      "epoch": 4.378202142524453,
+      "grad_norm": 13.0368070602417,
+      "learning_rate": 0.0,
+      "loss": 2.584,
+      "step": 18800
+    },
+    {
+      "epoch": 4.380530973451328,
+      "grad_norm": 9.664212226867676,
+      "learning_rate": 0.0,
+      "loss": 1.1443,
+      "step": 18810
+    },
+    {
+      "epoch": 4.382859804378202,
+      "grad_norm": 13.87582015991211,
+      "learning_rate": 0.0,
+      "loss": 0.9475,
+      "step": 18820
+    },
+    {
+      "epoch": 4.385188635305077,
+      "grad_norm": 7.444732666015625,
+      "learning_rate": 0.0,
+      "loss": 1.1998,
+      "step": 18830
+    },
+    {
+      "epoch": 4.387517466231952,
+      "grad_norm": 17.309480667114258,
+      "learning_rate": 0.0,
+      "loss": 0.9564,
+      "step": 18840
+    },
+    {
+      "epoch": 4.389846297158826,
+      "grad_norm": 16.705272674560547,
+      "learning_rate": 0.0,
+      "loss": 1.2803,
+      "step": 18850
+    },
+    {
+      "epoch": 4.392175128085701,
+      "grad_norm": 12.118639945983887,
+      "learning_rate": 0.0,
+      "loss": 1.1282,
+      "step": 18860
+    },
+    {
+      "epoch": 4.394503959012575,
+      "grad_norm": 15.16536808013916,
+      "learning_rate": 0.0,
+      "loss": 1.1275,
+      "step": 18870
+    },
+    {
+      "epoch": 4.396832789939451,
+      "grad_norm": 4.360487461090088,
+      "learning_rate": 0.0,
+      "loss": 1.4061,
+      "step": 18880
+    },
+    {
+      "epoch": 4.399161620866325,
+      "grad_norm": 22.97736930847168,
+      "learning_rate": 0.0,
+      "loss": 1.6106,
+      "step": 18890
+    },
+    {
+      "epoch": 4.4014904517932,
+      "grad_norm": 62.73188018798828,
+      "learning_rate": 0.0,
+      "loss": 1.2969,
+      "step": 18900
+    },
+    {
+      "epoch": 4.403819282720074,
+      "grad_norm": 15.173748970031738,
+      "learning_rate": 0.0,
+      "loss": 1.6601,
+      "step": 18910
+    },
+    {
+      "epoch": 4.40614811364695,
+      "grad_norm": 12.659820556640625,
+      "learning_rate": 0.0,
+      "loss": 1.4902,
+      "step": 18920
+    },
+    {
+      "epoch": 4.408476944573824,
+      "grad_norm": 9.84178352355957,
+      "learning_rate": 0.0,
+      "loss": 1.0183,
+      "step": 18930
+    },
+    {
+      "epoch": 4.410805775500698,
+      "grad_norm": 16.406572341918945,
+      "learning_rate": 0.0,
+      "loss": 0.9501,
+      "step": 18940
+    },
+    {
+      "epoch": 4.413134606427573,
+      "grad_norm": 8.493030548095703,
+      "learning_rate": 0.0,
+      "loss": 0.9876,
+      "step": 18950
+    },
+    {
+      "epoch": 4.415463437354448,
+      "grad_norm": 8.988524436950684,
+      "learning_rate": 0.0,
+      "loss": 1.1732,
+      "step": 18960
+    },
+    {
+      "epoch": 4.417792268281323,
+      "grad_norm": 7.769420146942139,
+      "learning_rate": 0.0,
+      "loss": 1.3663,
+      "step": 18970
+    },
+    {
+      "epoch": 4.420121099208197,
+      "grad_norm": 10.500604629516602,
+      "learning_rate": 0.0,
+      "loss": 1.9919,
+      "step": 18980
+    },
+    {
+      "epoch": 4.422449930135072,
+      "grad_norm": 13.152361869812012,
+      "learning_rate": 0.0,
+      "loss": 1.5332,
+      "step": 18990
+    },
+    {
+      "epoch": 4.424778761061947,
+      "grad_norm": 29.10773468017578,
+      "learning_rate": 0.0,
+      "loss": 1.6014,
+      "step": 19000
+    },
+    {
+      "epoch": 4.427107591988822,
+      "grad_norm": 9.866641998291016,
+      "learning_rate": 0.0,
+      "loss": 1.3321,
+      "step": 19010
+    },
+    {
+      "epoch": 4.429436422915696,
+      "grad_norm": 7.406885623931885,
+      "learning_rate": 0.0,
+      "loss": 0.9355,
+      "step": 19020
+    },
+    {
+      "epoch": 4.431765253842571,
+      "grad_norm": 64.86759948730469,
+      "learning_rate": 0.0,
+      "loss": 1.4649,
+      "step": 19030
+    },
+    {
+      "epoch": 4.434094084769446,
+      "grad_norm": 24.239585876464844,
+      "learning_rate": 0.0,
+      "loss": 1.3549,
+      "step": 19040
+    },
+    {
+      "epoch": 4.436422915696321,
+      "grad_norm": 12.164912223815918,
+      "learning_rate": 0.0,
+      "loss": 0.9648,
+      "step": 19050
+    },
+    {
+      "epoch": 4.438751746623195,
+      "grad_norm": 7.17809534072876,
+      "learning_rate": 0.0,
+      "loss": 1.0822,
+      "step": 19060
+    },
+    {
+      "epoch": 4.4410805775500695,
+      "grad_norm": 15.311285972595215,
+      "learning_rate": 0.0,
+      "loss": 0.918,
+      "step": 19070
+    },
+    {
+      "epoch": 4.443409408476945,
+      "grad_norm": 6.92308235168457,
+      "learning_rate": 0.0,
+      "loss": 1.1334,
+      "step": 19080
+    },
+    {
+      "epoch": 4.445738239403819,
+      "grad_norm": 21.64579963684082,
+      "learning_rate": 0.0,
+      "loss": 1.011,
+      "step": 19090
+    },
+    {
+      "epoch": 4.448067070330694,
+      "grad_norm": 17.369667053222656,
+      "learning_rate": 0.0,
+      "loss": 1.0185,
+      "step": 19100
+    },
+    {
+      "epoch": 4.4503959012575685,
+      "grad_norm": 6.284876823425293,
+      "learning_rate": 0.0,
+      "loss": 1.3311,
+      "step": 19110
+    },
+    {
+      "epoch": 4.452724732184444,
+      "grad_norm": 4.605877876281738,
+      "learning_rate": 0.0,
+      "loss": 0.9037,
+      "step": 19120
+    },
+    {
+      "epoch": 4.455053563111318,
+      "grad_norm": 8.776147842407227,
+      "learning_rate": 0.0,
+      "loss": 1.2707,
+      "step": 19130
+    },
+    {
+      "epoch": 4.457382394038193,
+      "grad_norm": 8.23446273803711,
+      "learning_rate": 0.0,
+      "loss": 1.194,
+      "step": 19140
+    },
+    {
+      "epoch": 4.4597112249650674,
+      "grad_norm": 6.68794059753418,
+      "learning_rate": 0.0,
+      "loss": 1.1309,
+      "step": 19150
+    },
+    {
+      "epoch": 4.462040055891943,
+      "grad_norm": 5.445720672607422,
+      "learning_rate": 0.0,
+      "loss": 1.2681,
+      "step": 19160
+    },
+    {
+      "epoch": 4.464368886818817,
+      "grad_norm": 47.82493591308594,
+      "learning_rate": 0.0,
+      "loss": 1.0086,
+      "step": 19170
+    },
+    {
+      "epoch": 4.466697717745692,
+      "grad_norm": 12.19786548614502,
+      "learning_rate": 0.0,
+      "loss": 1.3866,
+      "step": 19180
+    },
+    {
+      "epoch": 4.469026548672566,
+      "grad_norm": 27.628253936767578,
+      "learning_rate": 0.0,
+      "loss": 1.8903,
+      "step": 19190
+    },
+    {
+      "epoch": 4.471355379599441,
+      "grad_norm": 6.904911041259766,
+      "learning_rate": 0.0,
+      "loss": 1.2476,
+      "step": 19200
+    },
+    {
+      "epoch": 4.473684210526316,
+      "grad_norm": 13.863846778869629,
+      "learning_rate": 0.0,
+      "loss": 0.859,
+      "step": 19210
+    },
+    {
+      "epoch": 4.47601304145319,
+      "grad_norm": 124.06563568115234,
+      "learning_rate": 0.0,
+      "loss": 1.8265,
+      "step": 19220
+    },
+    {
+      "epoch": 4.478341872380065,
+      "grad_norm": 104.48321533203125,
+      "learning_rate": 0.0,
+      "loss": 1.2933,
+      "step": 19230
+    },
+    {
+      "epoch": 4.48067070330694,
+      "grad_norm": 29.74410629272461,
+      "learning_rate": 0.0,
+      "loss": 1.3518,
+      "step": 19240
+    },
+    {
+      "epoch": 4.482999534233815,
+      "grad_norm": 9.21904468536377,
+      "learning_rate": 0.0,
+      "loss": 1.3145,
+      "step": 19250
+    },
+    {
+      "epoch": 4.485328365160689,
+      "grad_norm": 17.82377815246582,
+      "learning_rate": 0.0,
+      "loss": 1.1376,
+      "step": 19260
+    },
+    {
+      "epoch": 4.487657196087564,
+      "grad_norm": 7.17800235748291,
+      "learning_rate": 0.0,
+      "loss": 1.6021,
+      "step": 19270
+    },
+    {
+      "epoch": 4.489986027014439,
+      "grad_norm": 8.264793395996094,
+      "learning_rate": 0.0,
+      "loss": 1.0782,
+      "step": 19280
+    },
+    {
+      "epoch": 4.492314857941313,
+      "grad_norm": 10.051447868347168,
+      "learning_rate": 0.0,
+      "loss": 1.046,
+      "step": 19290
+    },
+    {
+      "epoch": 4.494643688868188,
+      "grad_norm": 9.666632652282715,
+      "learning_rate": 0.0,
+      "loss": 0.8611,
+      "step": 19300
+    },
+    {
+      "epoch": 4.4969725197950625,
+      "grad_norm": 5.8455986976623535,
+      "learning_rate": 0.0,
+      "loss": 1.1852,
+      "step": 19310
+    },
+    {
+      "epoch": 4.499301350721938,
+      "grad_norm": 40.72843933105469,
+      "learning_rate": 0.0,
+      "loss": 1.4481,
+      "step": 19320
+    },
+    {
+      "epoch": 4.501630181648812,
+      "grad_norm": 14.480032920837402,
+      "learning_rate": 0.0,
+      "loss": 1.086,
+      "step": 19330
+    },
+    {
+      "epoch": 4.503959012575687,
+      "grad_norm": 8.125821113586426,
+      "learning_rate": 0.0,
+      "loss": 1.1448,
+      "step": 19340
+    },
+    {
+      "epoch": 4.5062878435025615,
+      "grad_norm": 8.457114219665527,
+      "learning_rate": 0.0,
+      "loss": 0.9219,
+      "step": 19350
+    },
+    {
+      "epoch": 4.508616674429437,
+      "grad_norm": 46.61256790161133,
+      "learning_rate": 0.0,
+      "loss": 1.4796,
+      "step": 19360
+    },
+    {
+      "epoch": 4.510945505356311,
+      "grad_norm": 12.012411117553711,
+      "learning_rate": 0.0,
+      "loss": 0.9379,
+      "step": 19370
+    },
+    {
+      "epoch": 4.513274336283186,
+      "grad_norm": 10.32507038116455,
+      "learning_rate": 0.0,
+      "loss": 1.0856,
+      "step": 19380
+    },
+    {
+      "epoch": 4.5156031672100605,
+      "grad_norm": 18.621065139770508,
+      "learning_rate": 0.0,
+      "loss": 1.2173,
+      "step": 19390
+    },
+    {
+      "epoch": 4.517931998136936,
+      "grad_norm": 5.736082553863525,
+      "learning_rate": 0.0,
+      "loss": 0.866,
+      "step": 19400
+    },
+    {
+      "epoch": 4.52026082906381,
+      "grad_norm": 8.20267391204834,
+      "learning_rate": 0.0,
+      "loss": 1.3217,
+      "step": 19410
+    },
+    {
+      "epoch": 4.522589659990684,
+      "grad_norm": 7.2390947341918945,
+      "learning_rate": 0.0,
+      "loss": 1.6424,
+      "step": 19420
+    },
+    {
+      "epoch": 4.5249184909175595,
+      "grad_norm": 13.071420669555664,
+      "learning_rate": 0.0,
+      "loss": 0.9959,
+      "step": 19430
+    },
+    {
+      "epoch": 4.527247321844434,
+      "grad_norm": 7.594562530517578,
+      "learning_rate": 0.0,
+      "loss": 1.2662,
+      "step": 19440
+    },
+    {
+      "epoch": 4.529576152771309,
+      "grad_norm": 31.20307159423828,
+      "learning_rate": 0.0,
+      "loss": 1.7857,
+      "step": 19450
+    },
+    {
+      "epoch": 4.531904983698183,
+      "grad_norm": 11.719870567321777,
+      "learning_rate": 0.0,
+      "loss": 1.0293,
+      "step": 19460
+    },
+    {
+      "epoch": 4.5342338146250585,
+      "grad_norm": 16.678586959838867,
+      "learning_rate": 0.0,
+      "loss": 1.2137,
+      "step": 19470
+    },
+    {
+      "epoch": 4.536562645551933,
+      "grad_norm": 6.821103096008301,
+      "learning_rate": 0.0,
+      "loss": 0.7388,
+      "step": 19480
+    },
+    {
+      "epoch": 4.538891476478808,
+      "grad_norm": 22.45350456237793,
+      "learning_rate": 0.0,
+      "loss": 1.3895,
+      "step": 19490
+    },
+    {
+      "epoch": 4.541220307405682,
+      "grad_norm": 3.4364640712738037,
+      "learning_rate": 0.0,
+      "loss": 1.3045,
+      "step": 19500
+    },
+    {
+      "epoch": 4.543549138332557,
+      "grad_norm": 12.992962837219238,
+      "learning_rate": 0.0,
+      "loss": 0.9766,
+      "step": 19510
+    },
+    {
+      "epoch": 4.545877969259432,
+      "grad_norm": 8.960369110107422,
+      "learning_rate": 0.0,
+      "loss": 1.4743,
+      "step": 19520
+    },
+    {
+      "epoch": 4.548206800186307,
+      "grad_norm": 9.852701187133789,
+      "learning_rate": 0.0,
+      "loss": 1.0763,
+      "step": 19530
+    },
+    {
+      "epoch": 4.550535631113181,
+      "grad_norm": 5.533620357513428,
+      "learning_rate": 0.0,
+      "loss": 1.6823,
+      "step": 19540
+    },
+    {
+      "epoch": 4.552864462040056,
+      "grad_norm": 11.03526496887207,
+      "learning_rate": 0.0,
+      "loss": 1.3972,
+      "step": 19550
+    },
+    {
+      "epoch": 4.555193292966931,
+      "grad_norm": 11.094636917114258,
+      "learning_rate": 0.0,
+      "loss": 0.9835,
+      "step": 19560
+    },
+    {
+      "epoch": 4.557522123893805,
+      "grad_norm": 9.783246994018555,
+      "learning_rate": 0.0,
+      "loss": 1.2277,
+      "step": 19570
+    },
+    {
+      "epoch": 4.55985095482068,
+      "grad_norm": 32.54227828979492,
+      "learning_rate": 0.0,
+      "loss": 1.4958,
+      "step": 19580
+    },
+    {
+      "epoch": 4.562179785747555,
+      "grad_norm": 15.014932632446289,
+      "learning_rate": 0.0,
+      "loss": 1.0942,
+      "step": 19590
+    },
+    {
+      "epoch": 4.56450861667443,
+      "grad_norm": 10.05160140991211,
+      "learning_rate": 0.0,
+      "loss": 1.4059,
+      "step": 19600
+    },
+    {
+      "epoch": 4.566837447601304,
+      "grad_norm": 7.201435089111328,
+      "learning_rate": 0.0,
+      "loss": 1.5083,
+      "step": 19610
+    },
+    {
+      "epoch": 4.569166278528179,
+      "grad_norm": 9.957510948181152,
+      "learning_rate": 0.0,
+      "loss": 1.0267,
+      "step": 19620
+    },
+    {
+      "epoch": 4.5714951094550536,
+      "grad_norm": 38.4839973449707,
+      "learning_rate": 0.0,
+      "loss": 1.0244,
+      "step": 19630
+    },
+    {
+      "epoch": 4.573823940381928,
+      "grad_norm": 15.924169540405273,
+      "learning_rate": 0.0,
+      "loss": 1.2288,
+      "step": 19640
+    },
+    {
+      "epoch": 4.576152771308803,
+      "grad_norm": 5.762319087982178,
+      "learning_rate": 0.0,
+      "loss": 1.0638,
+      "step": 19650
+    },
+    {
+      "epoch": 4.578481602235677,
+      "grad_norm": 14.740281105041504,
+      "learning_rate": 0.0,
+      "loss": 1.151,
+      "step": 19660
+    },
+    {
+      "epoch": 4.5808104331625525,
+      "grad_norm": 11.056538581848145,
+      "learning_rate": 0.0,
+      "loss": 0.9192,
+      "step": 19670
+    },
+    {
+      "epoch": 4.583139264089427,
+      "grad_norm": 20.958890914916992,
+      "learning_rate": 0.0,
+      "loss": 1.0669,
+      "step": 19680
+    },
+    {
+      "epoch": 4.585468095016302,
+      "grad_norm": 29.393579483032227,
+      "learning_rate": 0.0,
+      "loss": 1.5113,
+      "step": 19690
+    },
+    {
+      "epoch": 4.587796925943176,
+      "grad_norm": 45.95991134643555,
+      "learning_rate": 0.0,
+      "loss": 1.4022,
+      "step": 19700
+    },
+    {
+      "epoch": 4.5901257568700515,
+      "grad_norm": 6.977542400360107,
+      "learning_rate": 0.0,
+      "loss": 0.9212,
+      "step": 19710
+    },
+    {
+      "epoch": 4.592454587796926,
+      "grad_norm": 7.12802791595459,
+      "learning_rate": 0.0,
+      "loss": 1.0174,
+      "step": 19720
+    },
+    {
+      "epoch": 4.594783418723801,
+      "grad_norm": 12.21475601196289,
+      "learning_rate": 0.0,
+      "loss": 1.7109,
+      "step": 19730
+    },
+    {
+      "epoch": 4.597112249650675,
+      "grad_norm": 21.435813903808594,
+      "learning_rate": 0.0,
+      "loss": 1.817,
+      "step": 19740
+    },
+    {
+      "epoch": 4.5994410805775505,
+      "grad_norm": 7.1112446784973145,
+      "learning_rate": 0.0,
+      "loss": 1.0705,
+      "step": 19750
+    },
+    {
+      "epoch": 4.601769911504425,
+      "grad_norm": 3.9490299224853516,
+      "learning_rate": 0.0,
+      "loss": 1.2692,
+      "step": 19760
+    },
+    {
+      "epoch": 4.604098742431299,
+      "grad_norm": 50.142601013183594,
+      "learning_rate": 0.0,
+      "loss": 1.5913,
+      "step": 19770
+    },
+    {
+      "epoch": 4.606427573358174,
+      "grad_norm": 7.299134254455566,
+      "learning_rate": 0.0,
+      "loss": 1.096,
+      "step": 19780
+    },
+    {
+      "epoch": 4.608756404285049,
+      "grad_norm": 69.69446563720703,
+      "learning_rate": 0.0,
+      "loss": 1.1911,
+      "step": 19790
+    },
+    {
+      "epoch": 4.611085235211924,
+      "grad_norm": 6.633697509765625,
+      "learning_rate": 0.0,
+      "loss": 1.1309,
+      "step": 19800
+    },
+    {
+      "epoch": 4.613414066138798,
+      "grad_norm": 10.545255661010742,
+      "learning_rate": 0.0,
+      "loss": 1.0185,
+      "step": 19810
+    },
+    {
+      "epoch": 4.615742897065673,
+      "grad_norm": 14.515172004699707,
+      "learning_rate": 0.0,
+      "loss": 1.7769,
+      "step": 19820
+    },
+    {
+      "epoch": 4.618071727992548,
+      "grad_norm": 12.217021942138672,
+      "learning_rate": 0.0,
+      "loss": 1.0632,
+      "step": 19830
+    },
+    {
+      "epoch": 4.620400558919423,
+      "grad_norm": 29.141551971435547,
+      "learning_rate": 0.0,
+      "loss": 1.2936,
+      "step": 19840
+    },
+    {
+      "epoch": 4.622729389846297,
+      "grad_norm": 18.290485382080078,
+      "learning_rate": 0.0,
+      "loss": 1.0439,
+      "step": 19850
+    },
+    {
+      "epoch": 4.625058220773171,
+      "grad_norm": 9.301356315612793,
+      "learning_rate": 0.0,
+      "loss": 1.2435,
+      "step": 19860
+    },
+    {
+      "epoch": 4.627387051700047,
+      "grad_norm": 8.590415954589844,
+      "learning_rate": 0.0,
+      "loss": 0.8793,
+      "step": 19870
+    },
+    {
+      "epoch": 4.629715882626921,
+      "grad_norm": 15.62922477722168,
+      "learning_rate": 0.0,
+      "loss": 1.1785,
+      "step": 19880
+    },
+    {
+      "epoch": 4.632044713553796,
+      "grad_norm": 6.69285249710083,
+      "learning_rate": 0.0,
+      "loss": 0.9444,
+      "step": 19890
+    },
+    {
+      "epoch": 4.63437354448067,
+      "grad_norm": 62.42060852050781,
+      "learning_rate": 0.0,
+      "loss": 1.3757,
+      "step": 19900
+    },
+    {
+      "epoch": 4.636702375407546,
+      "grad_norm": 13.940834045410156,
+      "learning_rate": 0.0,
+      "loss": 1.0621,
+      "step": 19910
+    },
+    {
+      "epoch": 4.63903120633442,
+      "grad_norm": 12.126023292541504,
+      "learning_rate": 0.0,
+      "loss": 1.0629,
+      "step": 19920
+    },
+    {
+      "epoch": 4.641360037261295,
+      "grad_norm": 19.56361198425293,
+      "learning_rate": 0.0,
+      "loss": 0.8859,
+      "step": 19930
+    },
+    {
+      "epoch": 4.643688868188169,
+      "grad_norm": 13.945938110351562,
+      "learning_rate": 0.0,
+      "loss": 1.4923,
+      "step": 19940
+    },
+    {
+      "epoch": 4.646017699115045,
+      "grad_norm": 6.583313941955566,
+      "learning_rate": 0.0,
+      "loss": 1.1695,
+      "step": 19950
+    },
+    {
+      "epoch": 4.648346530041919,
+      "grad_norm": 11.310225486755371,
+      "learning_rate": 0.0,
+      "loss": 1.7277,
+      "step": 19960
+    },
+    {
+      "epoch": 4.650675360968794,
+      "grad_norm": 104.10961151123047,
+      "learning_rate": 0.0,
+      "loss": 1.4273,
+      "step": 19970
+    },
+    {
+      "epoch": 4.653004191895668,
+      "grad_norm": 19.168676376342773,
+      "learning_rate": 0.0,
+      "loss": 1.5528,
+      "step": 19980
+    },
+    {
+      "epoch": 4.655333022822543,
+      "grad_norm": 6.544228553771973,
+      "learning_rate": 0.0,
+      "loss": 1.234,
+      "step": 19990
+    },
+    {
+      "epoch": 4.657661853749418,
+      "grad_norm": 5.276122093200684,
+      "learning_rate": 0.0,
+      "loss": 1.2581,
+      "step": 20000
+    },
+    {
+      "epoch": 4.659990684676292,
+      "grad_norm": 94.88931274414062,
+      "learning_rate": 0.0,
+      "loss": 1.3947,
+      "step": 20010
+    },
+    {
+      "epoch": 4.662319515603167,
+      "grad_norm": 8.365758895874023,
+      "learning_rate": 0.0,
+      "loss": 1.5685,
+      "step": 20020
+    },
+    {
+      "epoch": 4.664648346530042,
+      "grad_norm": 23.34221839904785,
+      "learning_rate": 0.0,
+      "loss": 1.0813,
+      "step": 20030
+    },
+    {
+      "epoch": 4.666977177456917,
+      "grad_norm": 19.669198989868164,
+      "learning_rate": 0.0,
+      "loss": 1.0667,
+      "step": 20040
+    },
+    {
+      "epoch": 4.669306008383791,
+      "grad_norm": 11.661108016967773,
+      "learning_rate": 0.0,
+      "loss": 1.1768,
+      "step": 20050
+    },
+    {
+      "epoch": 4.671634839310666,
+      "grad_norm": 10.606729507446289,
+      "learning_rate": 0.0,
+      "loss": 0.9829,
+      "step": 20060
+    },
+    {
+      "epoch": 4.673963670237541,
+      "grad_norm": 48.090267181396484,
+      "learning_rate": 0.0,
+      "loss": 1.153,
+      "step": 20070
+    },
+    {
+      "epoch": 4.676292501164415,
+      "grad_norm": 5.69580602645874,
+      "learning_rate": 0.0,
+      "loss": 1.1018,
+      "step": 20080
+    },
+    {
+      "epoch": 4.67862133209129,
+      "grad_norm": 11.6792631149292,
+      "learning_rate": 0.0,
+      "loss": 1.0087,
+      "step": 20090
+    },
+    {
+      "epoch": 4.680950163018165,
+      "grad_norm": 10.280180931091309,
+      "learning_rate": 0.0,
+      "loss": 1.0452,
+      "step": 20100
+    },
+    {
+      "epoch": 4.68327899394504,
+      "grad_norm": 25.463672637939453,
+      "learning_rate": 0.0,
+      "loss": 1.6864,
+      "step": 20110
+    },
+    {
+      "epoch": 4.685607824871914,
+      "grad_norm": 25.444168090820312,
+      "learning_rate": 0.0,
+      "loss": 1.4118,
+      "step": 20120
+    },
+    {
+      "epoch": 4.687936655798789,
+      "grad_norm": 3.998115062713623,
+      "learning_rate": 0.0,
+      "loss": 1.1572,
+      "step": 20130
+    },
+    {
+      "epoch": 4.6902654867256635,
+      "grad_norm": 54.24504852294922,
+      "learning_rate": 0.0,
+      "loss": 1.1301,
+      "step": 20140
+    },
+    {
+      "epoch": 4.692594317652539,
+      "grad_norm": 8.593291282653809,
+      "learning_rate": 0.0,
+      "loss": 1.0843,
+      "step": 20150
+    },
+    {
+      "epoch": 4.694923148579413,
+      "grad_norm": 20.388153076171875,
+      "learning_rate": 0.0,
+      "loss": 1.1582,
+      "step": 20160
+    },
+    {
+      "epoch": 4.697251979506288,
+      "grad_norm": 7.956397533416748,
+      "learning_rate": 0.0,
+      "loss": 1.1456,
+      "step": 20170
+    },
+    {
+      "epoch": 4.6995808104331624,
+      "grad_norm": 13.123111724853516,
+      "learning_rate": 0.0,
+      "loss": 1.4565,
+      "step": 20180
+    },
+    {
+      "epoch": 4.701909641360038,
+      "grad_norm": 4.361729145050049,
+      "learning_rate": 0.0,
+      "loss": 0.9043,
+      "step": 20190
+    },
+    {
+      "epoch": 4.704238472286912,
+      "grad_norm": 16.26031494140625,
+      "learning_rate": 0.0,
+      "loss": 1.3562,
+      "step": 20200
+    },
+    {
+      "epoch": 4.706567303213786,
+      "grad_norm": 39.821197509765625,
+      "learning_rate": 0.0,
+      "loss": 1.3656,
+      "step": 20210
+    },
+    {
+      "epoch": 4.708896134140661,
+      "grad_norm": 5.075674057006836,
+      "learning_rate": 0.0,
+      "loss": 1.0091,
+      "step": 20220
+    },
+    {
+      "epoch": 4.711224965067536,
+      "grad_norm": 10.941627502441406,
+      "learning_rate": 0.0,
+      "loss": 1.6002,
+      "step": 20230
+    },
+    {
+      "epoch": 4.713553795994411,
+      "grad_norm": 15.39237117767334,
+      "learning_rate": 0.0,
+      "loss": 0.986,
+      "step": 20240
+    },
+    {
+      "epoch": 4.715882626921285,
+      "grad_norm": 33.28059005737305,
+      "learning_rate": 0.0,
+      "loss": 1.1865,
+      "step": 20250
+    },
+    {
+      "epoch": 4.71821145784816,
+      "grad_norm": 20.488203048706055,
+      "learning_rate": 0.0,
+      "loss": 1.221,
+      "step": 20260
+    },
+    {
+      "epoch": 4.720540288775035,
+      "grad_norm": 31.694875717163086,
+      "learning_rate": 0.0,
+      "loss": 1.441,
+      "step": 20270
+    },
+    {
+      "epoch": 4.72286911970191,
+      "grad_norm": 11.912035942077637,
+      "learning_rate": 0.0,
+      "loss": 1.2058,
+      "step": 20280
+    },
+    {
+      "epoch": 4.725197950628784,
+      "grad_norm": 95.51344299316406,
+      "learning_rate": 0.0,
+      "loss": 1.3537,
+      "step": 20290
+    },
+    {
+      "epoch": 4.727526781555659,
+      "grad_norm": 7.430610656738281,
+      "learning_rate": 0.0,
+      "loss": 1.6614,
+      "step": 20300
+    },
+    {
+      "epoch": 4.729855612482534,
+      "grad_norm": 16.465394973754883,
+      "learning_rate": 0.0,
+      "loss": 1.3616,
+      "step": 20310
+    },
+    {
+      "epoch": 4.732184443409409,
+      "grad_norm": 8.750808715820312,
+      "learning_rate": 0.0,
+      "loss": 1.4499,
+      "step": 20320
+    },
+    {
+      "epoch": 4.734513274336283,
+      "grad_norm": 15.627165794372559,
+      "learning_rate": 0.0,
+      "loss": 1.854,
+      "step": 20330
+    },
+    {
+      "epoch": 4.7368421052631575,
+      "grad_norm": 10.590481758117676,
+      "learning_rate": 0.0,
+      "loss": 1.2535,
+      "step": 20340
+    },
+    {
+      "epoch": 4.739170936190033,
+      "grad_norm": 11.492149353027344,
+      "learning_rate": 0.0,
+      "loss": 1.2967,
+      "step": 20350
+    },
+    {
+      "epoch": 4.741499767116907,
+      "grad_norm": 21.636804580688477,
+      "learning_rate": 0.0,
+      "loss": 1.6113,
+      "step": 20360
+    },
+    {
+      "epoch": 4.743828598043782,
+      "grad_norm": 5.577965259552002,
+      "learning_rate": 0.0,
+      "loss": 0.8558,
+      "step": 20370
+    },
+    {
+      "epoch": 4.7461574289706565,
+      "grad_norm": 7.697096824645996,
+      "learning_rate": 0.0,
+      "loss": 1.211,
+      "step": 20380
+    },
+    {
+      "epoch": 4.748486259897532,
+      "grad_norm": 35.868289947509766,
+      "learning_rate": 0.0,
+      "loss": 0.9857,
+      "step": 20390
+    },
+    {
+      "epoch": 4.750815090824406,
+      "grad_norm": 5.876249313354492,
+      "learning_rate": 0.0,
+      "loss": 0.9203,
+      "step": 20400
+    },
+    {
+      "epoch": 4.753143921751281,
+      "grad_norm": 12.119768142700195,
+      "learning_rate": 0.0,
+      "loss": 1.9162,
+      "step": 20410
+    },
+    {
+      "epoch": 4.7554727526781555,
+      "grad_norm": 10.388832092285156,
+      "learning_rate": 0.0,
+      "loss": 1.0832,
+      "step": 20420
+    },
+    {
+      "epoch": 4.75780158360503,
+      "grad_norm": 10.35787296295166,
+      "learning_rate": 0.0,
+      "loss": 1.0746,
+      "step": 20430
+    },
+    {
+      "epoch": 4.760130414531905,
+      "grad_norm": 8.285577774047852,
+      "learning_rate": 0.0,
+      "loss": 0.9798,
+      "step": 20440
+    },
+    {
+      "epoch": 4.76245924545878,
+      "grad_norm": 8.039005279541016,
+      "learning_rate": 0.0,
+      "loss": 1.0026,
+      "step": 20450
+    },
+    {
+      "epoch": 4.7647880763856545,
+      "grad_norm": 14.241586685180664,
+      "learning_rate": 0.0,
+      "loss": 1.4068,
+      "step": 20460
+    },
+    {
+      "epoch": 4.767116907312529,
+      "grad_norm": 14.098332405090332,
+      "learning_rate": 0.0,
+      "loss": 2.0716,
+      "step": 20470
+    },
+    {
+      "epoch": 4.769445738239404,
+      "grad_norm": 13.193544387817383,
+      "learning_rate": 0.0,
+      "loss": 1.0225,
+      "step": 20480
+    },
+    {
+      "epoch": 4.771774569166278,
+      "grad_norm": 12.964882850646973,
+      "learning_rate": 0.0,
+      "loss": 0.9621,
+      "step": 20490
+    },
+    {
+      "epoch": 4.7741034000931535,
+      "grad_norm": 66.96244049072266,
+      "learning_rate": 0.0,
+      "loss": 1.0828,
+      "step": 20500
+    },
+    {
+      "epoch": 4.776432231020028,
+      "grad_norm": 21.851341247558594,
+      "learning_rate": 0.0,
+      "loss": 1.2226,
+      "step": 20510
+    },
+    {
+      "epoch": 4.778761061946903,
+      "grad_norm": 11.199970245361328,
+      "learning_rate": 0.0,
+      "loss": 0.9426,
+      "step": 20520
+    },
+    {
+      "epoch": 4.781089892873777,
+      "grad_norm": 40.490753173828125,
+      "learning_rate": 0.0,
+      "loss": 1.5914,
+      "step": 20530
+    },
+    {
+      "epoch": 4.7834187238006525,
+      "grad_norm": 17.287166595458984,
+      "learning_rate": 0.0,
+      "loss": 1.4681,
+      "step": 20540
+    },
+    {
+      "epoch": 4.785747554727527,
+      "grad_norm": 13.948269844055176,
+      "learning_rate": 0.0,
+      "loss": 1.2498,
+      "step": 20550
+    },
+    {
+      "epoch": 4.788076385654401,
+      "grad_norm": 57.58226776123047,
+      "learning_rate": 0.0,
+      "loss": 1.7445,
+      "step": 20560
+    },
+    {
+      "epoch": 4.790405216581276,
+      "grad_norm": 5.487739086151123,
+      "learning_rate": 0.0,
+      "loss": 0.898,
+      "step": 20570
+    },
+    {
+      "epoch": 4.792734047508151,
+      "grad_norm": 15.564270973205566,
+      "learning_rate": 0.0,
+      "loss": 1.1383,
+      "step": 20580
+    },
+    {
+      "epoch": 4.795062878435026,
+      "grad_norm": 7.663552761077881,
+      "learning_rate": 0.0,
+      "loss": 1.2161,
+      "step": 20590
+    },
+    {
+      "epoch": 4.7973917093619,
+      "grad_norm": 6.975662708282471,
+      "learning_rate": 0.0,
+      "loss": 1.0381,
+      "step": 20600
+    },
+    {
+      "epoch": 4.799720540288775,
+      "grad_norm": 68.53016662597656,
+      "learning_rate": 0.0,
+      "loss": 1.3573,
+      "step": 20610
+    },
+    {
+      "epoch": 4.80204937121565,
+      "grad_norm": 51.50092315673828,
+      "learning_rate": 0.0,
+      "loss": 1.2739,
+      "step": 20620
+    },
+    {
+      "epoch": 4.804378202142525,
+      "grad_norm": 41.12253189086914,
+      "learning_rate": 0.0,
+      "loss": 1.0314,
+      "step": 20630
+    },
+    {
+      "epoch": 4.806707033069399,
+      "grad_norm": 8.585794448852539,
+      "learning_rate": 0.0,
+      "loss": 1.1131,
+      "step": 20640
+    },
+    {
+      "epoch": 4.809035863996274,
+      "grad_norm": 15.281457901000977,
+      "learning_rate": 0.0,
+      "loss": 1.1092,
+      "step": 20650
+    },
+    {
+      "epoch": 4.8113646949231486,
+      "grad_norm": 9.958845138549805,
+      "learning_rate": 0.0,
+      "loss": 1.1242,
+      "step": 20660
+    },
+    {
+      "epoch": 4.813693525850024,
+      "grad_norm": 9.2029390335083,
+      "learning_rate": 0.0,
+      "loss": 0.8147,
+      "step": 20670
+    },
+    {
+      "epoch": 4.816022356776898,
+      "grad_norm": 8.422203063964844,
+      "learning_rate": 0.0,
+      "loss": 1.2394,
+      "step": 20680
+    },
+    {
+      "epoch": 4.818351187703772,
+      "grad_norm": 6.612034320831299,
+      "learning_rate": 0.0,
+      "loss": 0.9382,
+      "step": 20690
+    },
+    {
+      "epoch": 4.8206800186306475,
+      "grad_norm": 9.461136817932129,
+      "learning_rate": 0.0,
+      "loss": 0.8587,
+      "step": 20700
+    },
+    {
+      "epoch": 4.823008849557522,
+      "grad_norm": 10.280345916748047,
+      "learning_rate": 0.0,
+      "loss": 0.8139,
+      "step": 20710
+    },
+    {
+      "epoch": 4.825337680484397,
+      "grad_norm": 6.057394504547119,
+      "learning_rate": 0.0,
+      "loss": 1.129,
+      "step": 20720
+    },
+    {
+      "epoch": 4.827666511411271,
+      "grad_norm": 9.460865020751953,
+      "learning_rate": 0.0,
+      "loss": 0.8686,
+      "step": 20730
+    },
+    {
+      "epoch": 4.8299953423381465,
+      "grad_norm": 31.428016662597656,
+      "learning_rate": 0.0,
+      "loss": 1.4765,
+      "step": 20740
+    },
+    {
+      "epoch": 4.832324173265021,
+      "grad_norm": 6.0293426513671875,
+      "learning_rate": 0.0,
+      "loss": 1.4691,
+      "step": 20750
+    },
+    {
+      "epoch": 4.834653004191896,
+      "grad_norm": 9.366103172302246,
+      "learning_rate": 0.0,
+      "loss": 0.9476,
+      "step": 20760
+    },
+    {
+      "epoch": 4.83698183511877,
+      "grad_norm": 7.880916595458984,
+      "learning_rate": 0.0,
+      "loss": 1.6768,
+      "step": 20770
+    },
+    {
+      "epoch": 4.839310666045645,
+      "grad_norm": 5.079976558685303,
+      "learning_rate": 0.0,
+      "loss": 0.8185,
+      "step": 20780
+    },
+    {
+      "epoch": 4.84163949697252,
+      "grad_norm": 5.588425159454346,
+      "learning_rate": 0.0,
+      "loss": 0.8626,
+      "step": 20790
+    },
+    {
+      "epoch": 4.843968327899394,
+      "grad_norm": 2.570796489715576,
+      "learning_rate": 0.0,
+      "loss": 1.2061,
+      "step": 20800
+    },
+    {
+      "epoch": 4.846297158826269,
+      "grad_norm": 21.089445114135742,
+      "learning_rate": 0.0,
+      "loss": 1.4568,
+      "step": 20810
+    },
+    {
+      "epoch": 4.848625989753144,
+      "grad_norm": 8.1835355758667,
+      "learning_rate": 0.0,
+      "loss": 0.8973,
+      "step": 20820
+    },
+    {
+      "epoch": 4.850954820680019,
+      "grad_norm": 12.01059341430664,
+      "learning_rate": 0.0,
+      "loss": 0.9296,
+      "step": 20830
+    },
+    {
+      "epoch": 4.853283651606893,
+      "grad_norm": 25.412508010864258,
+      "learning_rate": 0.0,
+      "loss": 1.787,
+      "step": 20840
+    },
+    {
+      "epoch": 4.855612482533768,
+      "grad_norm": 6.831169605255127,
+      "learning_rate": 0.0,
+      "loss": 1.0895,
+      "step": 20850
+    },
+    {
+      "epoch": 4.857941313460643,
+      "grad_norm": 8.809972763061523,
+      "learning_rate": 0.0,
+      "loss": 1.2377,
+      "step": 20860
+    },
+    {
+      "epoch": 4.860270144387518,
+      "grad_norm": 6.069893836975098,
+      "learning_rate": 0.0,
+      "loss": 1.2144,
+      "step": 20870
+    },
+    {
+      "epoch": 4.862598975314392,
+      "grad_norm": 11.856980323791504,
+      "learning_rate": 0.0,
+      "loss": 1.1085,
+      "step": 20880
+    },
+    {
+      "epoch": 4.864927806241267,
+      "grad_norm": 41.11824035644531,
+      "learning_rate": 0.0,
+      "loss": 1.5651,
+      "step": 20890
+    },
+    {
+      "epoch": 4.867256637168142,
+      "grad_norm": 13.00899887084961,
+      "learning_rate": 0.0,
+      "loss": 1.0189,
+      "step": 20900
+    },
+    {
+      "epoch": 4.869585468095016,
+      "grad_norm": 11.95178508758545,
+      "learning_rate": 0.0,
+      "loss": 0.8974,
+      "step": 20910
+    },
+    {
+      "epoch": 4.871914299021891,
+      "grad_norm": 31.843204498291016,
+      "learning_rate": 0.0,
+      "loss": 1.059,
+      "step": 20920
+    },
+    {
+      "epoch": 4.874243129948765,
+      "grad_norm": 18.231882095336914,
+      "learning_rate": 0.0,
+      "loss": 1.4373,
+      "step": 20930
+    },
+    {
+      "epoch": 4.876571960875641,
+      "grad_norm": 14.467658042907715,
+      "learning_rate": 0.0,
+      "loss": 0.9063,
+      "step": 20940
+    },
+    {
+      "epoch": 4.878900791802515,
+      "grad_norm": 47.42462158203125,
+      "learning_rate": 0.0,
+      "loss": 1.6727,
+      "step": 20950
+    },
+    {
+      "epoch": 4.88122962272939,
+      "grad_norm": 12.333345413208008,
+      "learning_rate": 0.0,
+      "loss": 1.077,
+      "step": 20960
+    },
+    {
+      "epoch": 4.883558453656264,
+      "grad_norm": 25.662342071533203,
+      "learning_rate": 0.0,
+      "loss": 0.9945,
+      "step": 20970
+    },
+    {
+      "epoch": 4.88588728458314,
+      "grad_norm": 141.52272033691406,
+      "learning_rate": 0.0,
+      "loss": 0.9511,
+      "step": 20980
+    },
+    {
+      "epoch": 4.888216115510014,
+      "grad_norm": 68.67302703857422,
+      "learning_rate": 0.0,
+      "loss": 1.3785,
+      "step": 20990
+    },
+    {
+      "epoch": 4.890544946436888,
+      "grad_norm": 4.360086441040039,
+      "learning_rate": 0.0,
+      "loss": 0.931,
+      "step": 21000
+    },
+    {
+      "epoch": 4.892873777363763,
+      "grad_norm": 9.454261779785156,
+      "learning_rate": 0.0,
+      "loss": 1.0979,
+      "step": 21010
+    },
+    {
+      "epoch": 4.895202608290639,
+      "grad_norm": 15.307931900024414,
+      "learning_rate": 0.0,
+      "loss": 1.0663,
+      "step": 21020
+    },
+    {
+      "epoch": 4.897531439217513,
+      "grad_norm": 6.253856182098389,
+      "learning_rate": 0.0,
+      "loss": 1.1837,
+      "step": 21030
+    },
+    {
+      "epoch": 4.899860270144387,
+      "grad_norm": 10.178426742553711,
+      "learning_rate": 0.0,
+      "loss": 0.9836,
+      "step": 21040
+    },
+    {
+      "epoch": 4.902189101071262,
+      "grad_norm": 43.5223274230957,
+      "learning_rate": 0.0,
+      "loss": 0.9516,
+      "step": 21050
+    },
+    {
+      "epoch": 4.904517931998137,
+      "grad_norm": 63.62842559814453,
+      "learning_rate": 0.0,
+      "loss": 0.9483,
+      "step": 21060
+    },
+    {
+      "epoch": 4.906846762925012,
+      "grad_norm": 11.0003662109375,
+      "learning_rate": 0.0,
+      "loss": 0.6298,
+      "step": 21070
+    },
+    {
+      "epoch": 4.909175593851886,
+      "grad_norm": 45.8305778503418,
+      "learning_rate": 0.0,
+      "loss": 1.1087,
+      "step": 21080
+    },
+    {
+      "epoch": 4.911504424778761,
+      "grad_norm": 18.97077178955078,
+      "learning_rate": 0.0,
+      "loss": 1.3575,
+      "step": 21090
+    },
+    {
+      "epoch": 4.913833255705636,
+      "grad_norm": 12.255200386047363,
+      "learning_rate": 0.0,
+      "loss": 1.0349,
+      "step": 21100
+    },
+    {
+      "epoch": 4.916162086632511,
+      "grad_norm": 7.784099102020264,
+      "learning_rate": 0.0,
+      "loss": 1.0109,
+      "step": 21110
+    },
+    {
+      "epoch": 4.918490917559385,
+      "grad_norm": 10.180665016174316,
+      "learning_rate": 0.0,
+      "loss": 1.1474,
+      "step": 21120
+    },
+    {
+      "epoch": 4.9208197484862595,
+      "grad_norm": 8.419008255004883,
+      "learning_rate": 0.0,
+      "loss": 1.1899,
+      "step": 21130
+    },
+    {
+      "epoch": 4.923148579413135,
+      "grad_norm": 105.6771469116211,
+      "learning_rate": 0.0,
+      "loss": 1.0455,
+      "step": 21140
+    },
+    {
+      "epoch": 4.925477410340009,
+      "grad_norm": 8.955032348632812,
+      "learning_rate": 0.0,
+      "loss": 1.3199,
+      "step": 21150
+    },
+    {
+      "epoch": 4.927806241266884,
+      "grad_norm": 4.457620620727539,
+      "learning_rate": 0.0,
+      "loss": 1.0334,
+      "step": 21160
+    },
+    {
+      "epoch": 4.9301350721937585,
+      "grad_norm": 21.67259979248047,
+      "learning_rate": 0.0,
+      "loss": 1.1492,
+      "step": 21170
+    },
+    {
+      "epoch": 4.932463903120634,
+      "grad_norm": 19.54806900024414,
+      "learning_rate": 0.0,
+      "loss": 1.7704,
+      "step": 21180
+    },
+    {
+      "epoch": 4.934792734047508,
+      "grad_norm": 56.51215744018555,
+      "learning_rate": 0.0,
+      "loss": 1.379,
+      "step": 21190
+    },
+    {
+      "epoch": 4.937121564974383,
+      "grad_norm": 11.147592544555664,
+      "learning_rate": 0.0,
+      "loss": 1.6373,
+      "step": 21200
+    },
+    {
+      "epoch": 4.9394503959012575,
+      "grad_norm": 7.846542835235596,
+      "learning_rate": 0.0,
+      "loss": 1.4208,
+      "step": 21210
+    },
+    {
+      "epoch": 4.941779226828133,
+      "grad_norm": 22.17070960998535,
+      "learning_rate": 0.0,
+      "loss": 1.2205,
+      "step": 21220
+    },
+    {
+      "epoch": 4.944108057755007,
+      "grad_norm": 12.682791709899902,
+      "learning_rate": 0.0,
+      "loss": 0.9231,
+      "step": 21230
+    },
+    {
+      "epoch": 4.946436888681882,
+      "grad_norm": 10.368877410888672,
+      "learning_rate": 0.0,
+      "loss": 1.1445,
+      "step": 21240
+    },
+    {
+      "epoch": 4.948765719608756,
+      "grad_norm": 14.382050514221191,
+      "learning_rate": 0.0,
+      "loss": 2.1009,
+      "step": 21250
+    },
+    {
+      "epoch": 4.951094550535631,
+      "grad_norm": 12.668513298034668,
+      "learning_rate": 0.0,
+      "loss": 1.2432,
+      "step": 21260
+    },
+    {
+      "epoch": 4.953423381462506,
+      "grad_norm": 5.586184024810791,
+      "learning_rate": 0.0,
+      "loss": 1.1811,
+      "step": 21270
+    },
+    {
+      "epoch": 4.95575221238938,
+      "grad_norm": 7.907278060913086,
+      "learning_rate": 0.0,
+      "loss": 1.0929,
+      "step": 21280
+    },
+    {
+      "epoch": 4.958081043316255,
+      "grad_norm": 45.115882873535156,
+      "learning_rate": 0.0,
+      "loss": 1.2946,
+      "step": 21290
+    },
+    {
+      "epoch": 4.96040987424313,
+      "grad_norm": 59.10978317260742,
+      "learning_rate": 0.0,
+      "loss": 1.472,
+      "step": 21300
+    },
+    {
+      "epoch": 4.962738705170005,
+      "grad_norm": 14.696610450744629,
+      "learning_rate": 0.0,
+      "loss": 1.6996,
+      "step": 21310
+    },
+    {
+      "epoch": 4.965067536096879,
+      "grad_norm": 17.75439453125,
+      "learning_rate": 0.0,
+      "loss": 1.1951,
+      "step": 21320
+    },
+    {
+      "epoch": 4.967396367023754,
+      "grad_norm": 58.72871398925781,
+      "learning_rate": 0.0,
+      "loss": 1.2667,
+      "step": 21330
+    },
+    {
+      "epoch": 4.969725197950629,
+      "grad_norm": 19.66732406616211,
+      "learning_rate": 0.0,
+      "loss": 1.4438,
+      "step": 21340
+    },
+    {
+      "epoch": 4.972054028877503,
+      "grad_norm": 9.646601676940918,
+      "learning_rate": 0.0,
+      "loss": 1.081,
+      "step": 21350
+    },
+    {
+      "epoch": 4.974382859804378,
+      "grad_norm": 7.841832160949707,
+      "learning_rate": 0.0,
+      "loss": 0.8552,
+      "step": 21360
+    },
+    {
+      "epoch": 4.9767116907312525,
+      "grad_norm": 11.626233100891113,
+      "learning_rate": 0.0,
+      "loss": 1.0067,
+      "step": 21370
+    },
+    {
+      "epoch": 4.979040521658128,
+      "grad_norm": 39.45273971557617,
+      "learning_rate": 0.0,
+      "loss": 1.0806,
+      "step": 21380
+    },
+    {
+      "epoch": 4.981369352585002,
+      "grad_norm": 11.194560050964355,
+      "learning_rate": 0.0,
+      "loss": 0.908,
+      "step": 21390
+    },
+    {
+      "epoch": 4.983698183511877,
+      "grad_norm": 158.674072265625,
+      "learning_rate": 0.0,
+      "loss": 1.4795,
+      "step": 21400
+    },
+    {
+      "epoch": 4.9860270144387515,
+      "grad_norm": 32.56852722167969,
+      "learning_rate": 0.0,
+      "loss": 1.6169,
+      "step": 21410
+    },
+    {
+      "epoch": 4.988355845365627,
+      "grad_norm": 7.956556797027588,
+      "learning_rate": 0.0,
+      "loss": 1.3402,
+      "step": 21420
+    },
+    {
+      "epoch": 4.990684676292501,
+      "grad_norm": 29.174211502075195,
+      "learning_rate": 0.0,
+      "loss": 1.0919,
+      "step": 21430
+    },
+    {
+      "epoch": 4.993013507219376,
+      "grad_norm": 88.53599548339844,
+      "learning_rate": 0.0,
+      "loss": 1.2261,
+      "step": 21440
+    },
+    {
+      "epoch": 4.9953423381462505,
+      "grad_norm": 15.843962669372559,
+      "learning_rate": 0.0,
+      "loss": 1.2406,
+      "step": 21450
+    },
+    {
+      "epoch": 4.997671169073126,
+      "grad_norm": 38.2321662902832,
+      "learning_rate": 0.0,
+      "loss": 1.4553,
+      "step": 21460
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 12.097282409667969,
+      "learning_rate": 0.0,
+      "loss": 1.2166,
+      "step": 21470
+    },
+    {
+      "epoch": 5.0,
+      "eval_loss": 0.49088114500045776,
+      "eval_runtime": 15.246,
+      "eval_samples_per_second": 285.715,
+      "eval_steps_per_second": 2.296,
+      "step": 21470
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 21470,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "EarlyStoppingCallback": {
+      "args": {
+        "early_stopping_patience": 2,
+        "early_stopping_threshold": 0.0
+      },
+      "attributes": {
+        "early_stopping_patience_counter": 2
+      }
+    },
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}