diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,3892 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.986425339366516,
+  "eval_steps": 500,
+  "global_step": 550,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00904977375565611,
+      "grad_norm": 6.353778491475079,
+      "learning_rate": 1.4545454545454546e-06,
+      "loss": 0.8197,
+      "step": 1
+    },
+    {
+      "epoch": 0.01809954751131222,
+      "grad_norm": 6.38646932272591,
+      "learning_rate": 2.9090909090909093e-06,
+      "loss": 0.8224,
+      "step": 2
+    },
+    {
+      "epoch": 0.027149321266968326,
+      "grad_norm": 6.053910996051301,
+      "learning_rate": 4.363636363636364e-06,
+      "loss": 0.8034,
+      "step": 3
+    },
+    {
+      "epoch": 0.03619909502262444,
+      "grad_norm": 4.366009246926635,
+      "learning_rate": 5.8181818181818185e-06,
+      "loss": 0.774,
+      "step": 4
+    },
+    {
+      "epoch": 0.04524886877828054,
+      "grad_norm": 2.274884898169855,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 0.7292,
+      "step": 5
+    },
+    {
+      "epoch": 0.05429864253393665,
+      "grad_norm": 1.9029591404903814,
+      "learning_rate": 8.727272727272728e-06,
+      "loss": 0.7138,
+      "step": 6
+    },
+    {
+      "epoch": 0.06334841628959276,
+      "grad_norm": 4.535776001491079,
+      "learning_rate": 1.0181818181818182e-05,
+      "loss": 0.7317,
+      "step": 7
+    },
+    {
+      "epoch": 0.07239819004524888,
+      "grad_norm": 5.273781078723551,
+      "learning_rate": 1.1636363636363637e-05,
+      "loss": 0.7227,
+      "step": 8
+    },
+    {
+      "epoch": 0.08144796380090498,
+      "grad_norm": 6.613983727436227,
+      "learning_rate": 1.3090909090909092e-05,
+      "loss": 0.7225,
+      "step": 9
+    },
+    {
+      "epoch": 0.09049773755656108,
+      "grad_norm": 5.322076805028839,
+      "learning_rate": 1.4545454545454546e-05,
+      "loss": 0.7032,
+      "step": 10
+    },
+    {
+      "epoch": 0.09954751131221719,
+      "grad_norm": 2.6848510839250563,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.6662,
+      "step": 11
+    },
+    {
+      "epoch": 0.1085972850678733,
+      "grad_norm": 2.501883063773779,
+      "learning_rate": 1.7454545454545456e-05,
+      "loss": 0.6361,
+      "step": 12
+    },
+    {
+      "epoch": 0.11764705882352941,
+      "grad_norm": 2.4408014518675407,
+      "learning_rate": 1.8909090909090912e-05,
+      "loss": 0.6328,
+      "step": 13
+    },
+    {
+      "epoch": 0.12669683257918551,
+      "grad_norm": 1.3581412540415263,
+      "learning_rate": 2.0363636363636365e-05,
+      "loss": 0.6038,
+      "step": 14
+    },
+    {
+      "epoch": 0.13574660633484162,
+      "grad_norm": 1.3171074276840706,
+      "learning_rate": 2.1818181818181818e-05,
+      "loss": 0.5949,
+      "step": 15
+    },
+    {
+      "epoch": 0.14479638009049775,
+      "grad_norm": 1.1346220985487154,
+      "learning_rate": 2.3272727272727274e-05,
+      "loss": 0.5812,
+      "step": 16
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.9805713190517853,
+      "learning_rate": 2.4727272727272727e-05,
+      "loss": 0.5777,
+      "step": 17
+    },
+    {
+      "epoch": 0.16289592760180996,
+      "grad_norm": 0.981975629498607,
+      "learning_rate": 2.6181818181818183e-05,
+      "loss": 0.5671,
+      "step": 18
+    },
+    {
+      "epoch": 0.17194570135746606,
+      "grad_norm": 0.8633430106624941,
+      "learning_rate": 2.763636363636364e-05,
+      "loss": 0.5617,
+      "step": 19
+    },
+    {
+      "epoch": 0.18099547511312217,
+      "grad_norm": 0.9927559005340078,
+      "learning_rate": 2.9090909090909093e-05,
+      "loss": 0.5547,
+      "step": 20
+    },
+    {
+      "epoch": 0.19004524886877827,
+      "grad_norm": 0.7908715300138798,
+      "learning_rate": 3.054545454545455e-05,
+      "loss": 0.5392,
+      "step": 21
+    },
+    {
+      "epoch": 0.19909502262443438,
+      "grad_norm": 1.077249549339691,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.5373,
+      "step": 22
+    },
+    {
+      "epoch": 0.2081447963800905,
+      "grad_norm": 1.3602867857746606,
+      "learning_rate": 3.345454545454546e-05,
+      "loss": 0.534,
+      "step": 23
+    },
+    {
+      "epoch": 0.2171945701357466,
+      "grad_norm": 0.687608913929421,
+      "learning_rate": 3.490909090909091e-05,
+      "loss": 0.5286,
+      "step": 24
+    },
+    {
+      "epoch": 0.22624434389140272,
+      "grad_norm": 1.2223237721495865,
+      "learning_rate": 3.6363636363636364e-05,
+      "loss": 0.5301,
+      "step": 25
+    },
+    {
+      "epoch": 0.23529411764705882,
+      "grad_norm": 0.8580084887459775,
+      "learning_rate": 3.7818181818181824e-05,
+      "loss": 0.5214,
+      "step": 26
+    },
+    {
+      "epoch": 0.24434389140271492,
+      "grad_norm": 1.1071155887269823,
+      "learning_rate": 3.927272727272728e-05,
+      "loss": 0.5115,
+      "step": 27
+    },
+    {
+      "epoch": 0.25339366515837103,
+      "grad_norm": 0.7195369247916135,
+      "learning_rate": 4.072727272727273e-05,
+      "loss": 0.5095,
+      "step": 28
+    },
+    {
+      "epoch": 0.26244343891402716,
+      "grad_norm": 1.5961441486786672,
+      "learning_rate": 4.218181818181818e-05,
+      "loss": 0.5087,
+      "step": 29
+    },
+    {
+      "epoch": 0.27149321266968324,
+      "grad_norm": 0.827814164011704,
+      "learning_rate": 4.3636363636363636e-05,
+      "loss": 0.5111,
+      "step": 30
+    },
+    {
+      "epoch": 0.28054298642533937,
+      "grad_norm": 1.1252915824734908,
+      "learning_rate": 4.509090909090909e-05,
+      "loss": 0.5072,
+      "step": 31
+    },
+    {
+      "epoch": 0.2895927601809955,
+      "grad_norm": 1.1219477837842433,
+      "learning_rate": 4.654545454545455e-05,
+      "loss": 0.5046,
+      "step": 32
+    },
+    {
+      "epoch": 0.2986425339366516,
+      "grad_norm": 1.505289650732032,
+      "learning_rate": 4.8e-05,
+      "loss": 0.503,
+      "step": 33
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.8491008498263081,
+      "learning_rate": 4.9454545454545454e-05,
+      "loss": 0.5011,
+      "step": 34
+    },
+    {
+      "epoch": 0.3167420814479638,
+      "grad_norm": 1.4306190204877898,
+      "learning_rate": 5.0909090909090914e-05,
+      "loss": 0.4895,
+      "step": 35
+    },
+    {
+      "epoch": 0.3257918552036199,
+      "grad_norm": 1.0851621930114523,
+      "learning_rate": 5.236363636363637e-05,
+      "loss": 0.4956,
+      "step": 36
+    },
+    {
+      "epoch": 0.334841628959276,
+      "grad_norm": 1.206194741973144,
+      "learning_rate": 5.381818181818182e-05,
+      "loss": 0.4887,
+      "step": 37
+    },
+    {
+      "epoch": 0.3438914027149321,
+      "grad_norm": 1.4311749390727002,
+      "learning_rate": 5.527272727272728e-05,
+      "loss": 0.487,
+      "step": 38
+    },
+    {
+      "epoch": 0.35294117647058826,
+      "grad_norm": 0.738265754189619,
+      "learning_rate": 5.672727272727273e-05,
+      "loss": 0.4854,
+      "step": 39
+    },
+    {
+      "epoch": 0.36199095022624433,
+      "grad_norm": 1.6935271715239177,
+      "learning_rate": 5.8181818181818185e-05,
+      "loss": 0.4943,
+      "step": 40
+    },
+    {
+      "epoch": 0.37104072398190047,
+      "grad_norm": 0.8407873740347103,
+      "learning_rate": 5.9636363636363645e-05,
+      "loss": 0.4846,
+      "step": 41
+    },
+    {
+      "epoch": 0.38009049773755654,
+      "grad_norm": 1.168749658810784,
+      "learning_rate": 6.10909090909091e-05,
+      "loss": 0.4765,
+      "step": 42
+    },
+    {
+      "epoch": 0.3891402714932127,
+      "grad_norm": 0.9365669815926128,
+      "learning_rate": 6.254545454545456e-05,
+      "loss": 0.4865,
+      "step": 43
+    },
+    {
+      "epoch": 0.39819004524886875,
+      "grad_norm": 1.481731917951899,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 0.4858,
+      "step": 44
+    },
+    {
+      "epoch": 0.4072398190045249,
+      "grad_norm": 1.52181493703478,
+      "learning_rate": 6.545454545454546e-05,
+      "loss": 0.4949,
+      "step": 45
+    },
+    {
+      "epoch": 0.416289592760181,
+      "grad_norm": 1.449133028804705,
+      "learning_rate": 6.690909090909092e-05,
+      "loss": 0.4799,
+      "step": 46
+    },
+    {
+      "epoch": 0.4253393665158371,
+      "grad_norm": 1.4529455783010232,
+      "learning_rate": 6.836363636363637e-05,
+      "loss": 0.4779,
+      "step": 47
+    },
+    {
+      "epoch": 0.4343891402714932,
+      "grad_norm": 1.3548957795298309,
+      "learning_rate": 6.981818181818182e-05,
+      "loss": 0.4776,
+      "step": 48
+    },
+    {
+      "epoch": 0.4434389140271493,
+      "grad_norm": 1.3963891315975814,
+      "learning_rate": 7.127272727272728e-05,
+      "loss": 0.4731,
+      "step": 49
+    },
+    {
+      "epoch": 0.45248868778280543,
+      "grad_norm": 1.0657728232300334,
+      "learning_rate": 7.272727272727273e-05,
+      "loss": 0.4702,
+      "step": 50
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 1.2659845812441222,
+      "learning_rate": 7.418181818181818e-05,
+      "loss": 0.4853,
+      "step": 51
+    },
+    {
+      "epoch": 0.47058823529411764,
+      "grad_norm": 1.0364505292451895,
+      "learning_rate": 7.563636363636365e-05,
+      "loss": 0.4663,
+      "step": 52
+    },
+    {
+      "epoch": 0.4796380090497738,
+      "grad_norm": 1.188728427839461,
+      "learning_rate": 7.70909090909091e-05,
+      "loss": 0.4819,
+      "step": 53
+    },
+    {
+      "epoch": 0.48868778280542985,
+      "grad_norm": 1.3147320829735498,
+      "learning_rate": 7.854545454545455e-05,
+      "loss": 0.4765,
+      "step": 54
+    },
+    {
+      "epoch": 0.497737556561086,
+      "grad_norm": 1.1857487638939694,
+      "learning_rate": 8e-05,
+      "loss": 0.4764,
+      "step": 55
+    },
+    {
+      "epoch": 0.5067873303167421,
+      "grad_norm": 1.4561796125999034,
+      "learning_rate": 7.999919440291627e-05,
+      "loss": 0.4853,
+      "step": 56
+    },
+    {
+      "epoch": 0.5158371040723982,
+      "grad_norm": 1.3437982531985815,
+      "learning_rate": 7.999677764411438e-05,
+      "loss": 0.4766,
+      "step": 57
+    },
+    {
+      "epoch": 0.5248868778280543,
+      "grad_norm": 0.7865846274351967,
+      "learning_rate": 7.999274982094104e-05,
+      "loss": 0.4697,
+      "step": 58
+    },
+    {
+      "epoch": 0.5339366515837104,
+      "grad_norm": 1.7177406402851725,
+      "learning_rate": 7.998711109563637e-05,
+      "loss": 0.4753,
+      "step": 59
+    },
+    {
+      "epoch": 0.5429864253393665,
+      "grad_norm": 0.765099828673138,
+      "learning_rate": 7.997986169532741e-05,
+      "loss": 0.4646,
+      "step": 60
+    },
+    {
+      "epoch": 0.5520361990950227,
+      "grad_norm": 1.291319953299644,
+      "learning_rate": 7.997100191201896e-05,
+      "loss": 0.4719,
+      "step": 61
+    },
+    {
+      "epoch": 0.5610859728506787,
+      "grad_norm": 1.0231054776331459,
+      "learning_rate": 7.996053210258176e-05,
+      "loss": 0.4597,
+      "step": 62
+    },
+    {
+      "epoch": 0.5701357466063348,
+      "grad_norm": 1.020678257497354,
+      "learning_rate": 7.994845268873825e-05,
+      "loss": 0.4631,
+      "step": 63
+    },
+    {
+      "epoch": 0.579185520361991,
+      "grad_norm": 1.0920667405954658,
+      "learning_rate": 7.993476415704543e-05,
+      "loss": 0.4557,
+      "step": 64
+    },
+    {
+      "epoch": 0.5882352941176471,
+      "grad_norm": 1.1128002216774169,
+      "learning_rate": 7.991946705887539e-05,
+      "loss": 0.4601,
+      "step": 65
+    },
+    {
+      "epoch": 0.5972850678733032,
+      "grad_norm": 1.0096600697154985,
+      "learning_rate": 7.990256201039297e-05,
+      "loss": 0.4616,
+      "step": 66
+    },
+    {
+      "epoch": 0.6063348416289592,
+      "grad_norm": 1.0630268901035373,
+      "learning_rate": 7.98840496925311e-05,
+      "loss": 0.4536,
+      "step": 67
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.6564221253831782,
+      "learning_rate": 7.986393085096324e-05,
+      "loss": 0.4483,
+      "step": 68
+    },
+    {
+      "epoch": 0.6244343891402715,
+      "grad_norm": 0.6260742925219016,
+      "learning_rate": 7.984220629607336e-05,
+      "loss": 0.4508,
+      "step": 69
+    },
+    {
+      "epoch": 0.6334841628959276,
+      "grad_norm": 0.8273325354500644,
+      "learning_rate": 7.981887690292339e-05,
+      "loss": 0.4494,
+      "step": 70
+    },
+    {
+      "epoch": 0.6425339366515838,
+      "grad_norm": 0.884333850023049,
+      "learning_rate": 7.979394361121789e-05,
+      "loss": 0.4519,
+      "step": 71
+    },
+    {
+      "epoch": 0.6515837104072398,
+      "grad_norm": 1.0182608804991282,
+      "learning_rate": 7.97674074252662e-05,
+      "loss": 0.4555,
+      "step": 72
+    },
+    {
+      "epoch": 0.6606334841628959,
+      "grad_norm": 0.8109090124549581,
+      "learning_rate": 7.9739269413942e-05,
+      "loss": 0.4521,
+      "step": 73
+    },
+    {
+      "epoch": 0.669683257918552,
+      "grad_norm": 0.9807135848274302,
+      "learning_rate": 7.970953071064036e-05,
+      "loss": 0.4531,
+      "step": 74
+    },
+    {
+      "epoch": 0.6787330316742082,
+      "grad_norm": 1.504171669224647,
+      "learning_rate": 7.967819251323182e-05,
+      "loss": 0.4705,
+      "step": 75
+    },
+    {
+      "epoch": 0.6877828054298643,
+      "grad_norm": 0.5907918802273645,
+      "learning_rate": 7.964525608401445e-05,
+      "loss": 0.4488,
+      "step": 76
+    },
+    {
+      "epoch": 0.6968325791855203,
+      "grad_norm": 1.198484065683406,
+      "learning_rate": 7.961072274966282e-05,
+      "loss": 0.4551,
+      "step": 77
+    },
+    {
+      "epoch": 0.7058823529411765,
+      "grad_norm": 0.7873070678087877,
+      "learning_rate": 7.957459390117458e-05,
+      "loss": 0.4457,
+      "step": 78
+    },
+    {
+      "epoch": 0.7149321266968326,
+      "grad_norm": 0.8039969398610661,
+      "learning_rate": 7.95368709938145e-05,
+      "loss": 0.454,
+      "step": 79
+    },
+    {
+      "epoch": 0.7239819004524887,
+      "grad_norm": 0.7842737782633313,
+      "learning_rate": 7.949755554705577e-05,
+      "loss": 0.441,
+      "step": 80
+    },
+    {
+      "epoch": 0.7330316742081447,
+      "grad_norm": 0.8912309653170191,
+      "learning_rate": 7.945664914451888e-05,
+      "loss": 0.4442,
+      "step": 81
+    },
+    {
+      "epoch": 0.7420814479638009,
+      "grad_norm": 0.7795917187202375,
+      "learning_rate": 7.941415343390773e-05,
+      "loss": 0.4406,
+      "step": 82
+    },
+    {
+      "epoch": 0.751131221719457,
+      "grad_norm": 0.6205090396508431,
+      "learning_rate": 7.937007012694335e-05,
+      "loss": 0.4507,
+      "step": 83
+    },
+    {
+      "epoch": 0.7601809954751131,
+      "grad_norm": 0.8937106249538119,
+      "learning_rate": 7.932440099929493e-05,
+      "loss": 0.4532,
+      "step": 84
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.6165097160255976,
+      "learning_rate": 7.927714789050826e-05,
+      "loss": 0.4454,
+      "step": 85
+    },
+    {
+      "epoch": 0.7782805429864253,
+      "grad_norm": 0.6003524566440082,
+      "learning_rate": 7.92283127039317e-05,
+      "loss": 0.4367,
+      "step": 86
+    },
+    {
+      "epoch": 0.7873303167420814,
+      "grad_norm": 0.6906315651004887,
+      "learning_rate": 7.917789740663941e-05,
+      "loss": 0.4332,
+      "step": 87
+    },
+    {
+      "epoch": 0.7963800904977375,
+      "grad_norm": 0.515950260791789,
+      "learning_rate": 7.912590402935223e-05,
+      "loss": 0.4345,
+      "step": 88
+    },
+    {
+      "epoch": 0.8054298642533937,
+      "grad_norm": 0.5111081573242718,
+      "learning_rate": 7.907233466635582e-05,
+      "loss": 0.4419,
+      "step": 89
+    },
+    {
+      "epoch": 0.8144796380090498,
+      "grad_norm": 0.4275091010122135,
+      "learning_rate": 7.90171914754163e-05,
+      "loss": 0.4383,
+      "step": 90
+    },
+    {
+      "epoch": 0.8235294117647058,
+      "grad_norm": 0.5029385491118522,
+      "learning_rate": 7.896047667769335e-05,
+      "loss": 0.4348,
+      "step": 91
+    },
+    {
+      "epoch": 0.832579185520362,
+      "grad_norm": 0.4078317539372912,
+      "learning_rate": 7.890219255765077e-05,
+      "loss": 0.4293,
+      "step": 92
+    },
+    {
+      "epoch": 0.8416289592760181,
+      "grad_norm": 0.4768996453352078,
+      "learning_rate": 7.884234146296442e-05,
+      "loss": 0.4375,
+      "step": 93
+    },
+    {
+      "epoch": 0.8506787330316742,
+      "grad_norm": 0.5642712829841973,
+      "learning_rate": 7.878092580442766e-05,
+      "loss": 0.4301,
+      "step": 94
+    },
+    {
+      "epoch": 0.8597285067873304,
+      "grad_norm": 0.7926376614480052,
+      "learning_rate": 7.871794805585427e-05,
+      "loss": 0.4377,
+      "step": 95
+    },
+    {
+      "epoch": 0.8687782805429864,
+      "grad_norm": 0.8704449632643672,
+      "learning_rate": 7.865341075397874e-05,
+      "loss": 0.4303,
+      "step": 96
+    },
+    {
+      "epoch": 0.8778280542986425,
+      "grad_norm": 0.8638439306395513,
+      "learning_rate": 7.858731649835424e-05,
+      "loss": 0.4321,
+      "step": 97
+    },
+    {
+      "epoch": 0.8868778280542986,
+      "grad_norm": 1.1122142269322102,
+      "learning_rate": 7.85196679512477e-05,
+      "loss": 0.4369,
+      "step": 98
+    },
+    {
+      "epoch": 0.8959276018099548,
+      "grad_norm": 0.8695186582366218,
+      "learning_rate": 7.845046783753276e-05,
+      "loss": 0.4229,
+      "step": 99
+    },
+    {
+      "epoch": 0.9049773755656109,
+      "grad_norm": 0.6907557363110941,
+      "learning_rate": 7.837971894457991e-05,
+      "loss": 0.4294,
+      "step": 100
+    },
+    {
+      "epoch": 0.9140271493212669,
+      "grad_norm": 0.5776183515910319,
+      "learning_rate": 7.830742412214422e-05,
+      "loss": 0.4276,
+      "step": 101
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.45053560734923015,
+      "learning_rate": 7.82335862822506e-05,
+      "loss": 0.4284,
+      "step": 102
+    },
+    {
+      "epoch": 0.9321266968325792,
+      "grad_norm": 0.643150601551687,
+      "learning_rate": 7.815820839907651e-05,
+      "loss": 0.4237,
+      "step": 103
+    },
+    {
+      "epoch": 0.9411764705882353,
+      "grad_norm": 0.6867059068503916,
+      "learning_rate": 7.808129350883207e-05,
+      "loss": 0.4325,
+      "step": 104
+    },
+    {
+      "epoch": 0.9502262443438914,
+      "grad_norm": 0.47092962014394424,
+      "learning_rate": 7.800284470963783e-05,
+      "loss": 0.4282,
+      "step": 105
+    },
+    {
+      "epoch": 0.9592760180995475,
+      "grad_norm": 0.5794346362766767,
+      "learning_rate": 7.792286516139999e-05,
+      "loss": 0.426,
+      "step": 106
+    },
+    {
+      "epoch": 0.9683257918552036,
+      "grad_norm": 0.5899947636404669,
+      "learning_rate": 7.784135808568308e-05,
+      "loss": 0.4241,
+      "step": 107
+    },
+    {
+      "epoch": 0.9773755656108597,
+      "grad_norm": 0.5788528664751798,
+      "learning_rate": 7.775832676558028e-05,
+      "loss": 0.4356,
+      "step": 108
+    },
+    {
+      "epoch": 0.9864253393665159,
+      "grad_norm": 0.8755965755346112,
+      "learning_rate": 7.7673774545581e-05,
+      "loss": 0.4366,
+      "step": 109
+    },
+    {
+      "epoch": 0.995475113122172,
+      "grad_norm": 0.9177669118629138,
+      "learning_rate": 7.758770483143634e-05,
+      "loss": 0.4329,
+      "step": 110
+    },
+    {
+      "epoch": 1.006787330316742,
+      "grad_norm": 0.6446624913715124,
+      "learning_rate": 7.750012109002185e-05,
+      "loss": 0.4137,
+      "step": 111
+    },
+    {
+      "epoch": 1.0158371040723981,
+      "grad_norm": 0.5702180405824077,
+      "learning_rate": 7.741102684919787e-05,
+      "loss": 0.414,
+      "step": 112
+    },
+    {
+      "epoch": 1.0248868778280542,
+      "grad_norm": 0.604164406050923,
+      "learning_rate": 7.732042569766741e-05,
+      "loss": 0.4059,
+      "step": 113
+    },
+    {
+      "epoch": 1.0339366515837105,
+      "grad_norm": 0.578300789859933,
+      "learning_rate": 7.722832128483165e-05,
+      "loss": 0.4072,
+      "step": 114
+    },
+    {
+      "epoch": 1.0429864253393666,
+      "grad_norm": 0.7009255889818755,
+      "learning_rate": 7.71347173206429e-05,
+      "loss": 0.4044,
+      "step": 115
+    },
+    {
+      "epoch": 1.0520361990950227,
+      "grad_norm": 0.709442615518018,
+      "learning_rate": 7.703961757545522e-05,
+      "loss": 0.4171,
+      "step": 116
+    },
+    {
+      "epoch": 1.0610859728506787,
+      "grad_norm": 0.6827354691854699,
+      "learning_rate": 7.694302587987245e-05,
+      "loss": 0.4079,
+      "step": 117
+    },
+    {
+      "epoch": 1.0701357466063348,
+      "grad_norm": 0.6818158818500779,
+      "learning_rate": 7.6844946124594e-05,
+      "loss": 0.4126,
+      "step": 118
+    },
+    {
+      "epoch": 1.079185520361991,
+      "grad_norm": 0.863785651181626,
+      "learning_rate": 7.674538226025815e-05,
+      "loss": 0.4109,
+      "step": 119
+    },
+    {
+      "epoch": 1.088235294117647,
+      "grad_norm": 0.5508478361873977,
+      "learning_rate": 7.664433829728279e-05,
+      "loss": 0.4028,
+      "step": 120
+    },
+    {
+      "epoch": 1.0972850678733033,
+      "grad_norm": 0.5954544705810189,
+      "learning_rate": 7.654181830570404e-05,
+      "loss": 0.3969,
+      "step": 121
+    },
+    {
+      "epoch": 1.1063348416289593,
+      "grad_norm": 0.5872119406822112,
+      "learning_rate": 7.64378264150122e-05,
+      "loss": 0.3993,
+      "step": 122
+    },
+    {
+      "epoch": 1.1153846153846154,
+      "grad_norm": 0.5228235141898725,
+      "learning_rate": 7.633236681398549e-05,
+      "loss": 0.4063,
+      "step": 123
+    },
+    {
+      "epoch": 1.1244343891402715,
+      "grad_norm": 0.6996331404201509,
+      "learning_rate": 7.622544375052124e-05,
+      "loss": 0.4076,
+      "step": 124
+    },
+    {
+      "epoch": 1.1334841628959276,
+      "grad_norm": 0.5428162110150009,
+      "learning_rate": 7.611706153146486e-05,
+      "loss": 0.4016,
+      "step": 125
+    },
+    {
+      "epoch": 1.1425339366515836,
+      "grad_norm": 0.4476576669392731,
+      "learning_rate": 7.600722452243632e-05,
+      "loss": 0.4018,
+      "step": 126
+    },
+    {
+      "epoch": 1.1515837104072397,
+      "grad_norm": 0.47501951208322907,
+      "learning_rate": 7.589593714765434e-05,
+      "loss": 0.4003,
+      "step": 127
+    },
+    {
+      "epoch": 1.160633484162896,
+      "grad_norm": 0.39864387167073856,
+      "learning_rate": 7.578320388975816e-05,
+      "loss": 0.4043,
+      "step": 128
+    },
+    {
+      "epoch": 1.169683257918552,
+      "grad_norm": 0.4415298124574496,
+      "learning_rate": 7.566902928962694e-05,
+      "loss": 0.3996,
+      "step": 129
+    },
+    {
+      "epoch": 1.1787330316742082,
+      "grad_norm": 0.3522729758159122,
+      "learning_rate": 7.555341794619695e-05,
+      "loss": 0.402,
+      "step": 130
+    },
+    {
+      "epoch": 1.1877828054298643,
+      "grad_norm": 0.37499820674043755,
+      "learning_rate": 7.543637451627623e-05,
+      "loss": 0.3963,
+      "step": 131
+    },
+    {
+      "epoch": 1.1968325791855203,
+      "grad_norm": 0.3140691864451784,
+      "learning_rate": 7.531790371435709e-05,
+      "loss": 0.3961,
+      "step": 132
+    },
+    {
+      "epoch": 1.2058823529411764,
+      "grad_norm": 0.3383990510694389,
+      "learning_rate": 7.519801031242613e-05,
+      "loss": 0.4026,
+      "step": 133
+    },
+    {
+      "epoch": 1.2149321266968327,
+      "grad_norm": 0.3704940782389753,
+      "learning_rate": 7.507669913977213e-05,
+      "loss": 0.3964,
+      "step": 134
+    },
+    {
+      "epoch": 1.2239819004524888,
+      "grad_norm": 0.3963648582905255,
+      "learning_rate": 7.49539750827914e-05,
+      "loss": 0.3989,
+      "step": 135
+    },
+    {
+      "epoch": 1.2330316742081449,
+      "grad_norm": 0.43661160954834166,
+      "learning_rate": 7.482984308479109e-05,
+      "loss": 0.3992,
+      "step": 136
+    },
+    {
+      "epoch": 1.242081447963801,
+      "grad_norm": 0.4760701506797231,
+      "learning_rate": 7.470430814578997e-05,
+      "loss": 0.4038,
+      "step": 137
+    },
+    {
+      "epoch": 1.251131221719457,
+      "grad_norm": 0.6080541303280323,
+      "learning_rate": 7.457737532231708e-05,
+      "loss": 0.4017,
+      "step": 138
+    },
+    {
+      "epoch": 1.260180995475113,
+      "grad_norm": 0.8992194165468265,
+      "learning_rate": 7.444904972720803e-05,
+      "loss": 0.4021,
+      "step": 139
+    },
+    {
+      "epoch": 1.2692307692307692,
+      "grad_norm": 0.9388790657173226,
+      "learning_rate": 7.431933652939909e-05,
+      "loss": 0.4012,
+      "step": 140
+    },
+    {
+      "epoch": 1.2782805429864252,
+      "grad_norm": 0.7115639008601973,
+      "learning_rate": 7.418824095371895e-05,
+      "loss": 0.3981,
+      "step": 141
+    },
+    {
+      "epoch": 1.2873303167420813,
+      "grad_norm": 0.482309574495901,
+      "learning_rate": 7.405576828067829e-05,
+      "loss": 0.4016,
+      "step": 142
+    },
+    {
+      "epoch": 1.2963800904977376,
+      "grad_norm": 0.4429825222163161,
+      "learning_rate": 7.392192384625704e-05,
+      "loss": 0.3992,
+      "step": 143
+    },
+    {
+      "epoch": 1.3054298642533937,
+      "grad_norm": 0.46955938360600935,
+      "learning_rate": 7.378671304168955e-05,
+      "loss": 0.3963,
+      "step": 144
+    },
+    {
+      "epoch": 1.3144796380090498,
+      "grad_norm": 0.41143211547851066,
+      "learning_rate": 7.365014131324725e-05,
+      "loss": 0.3989,
+      "step": 145
+    },
+    {
+      "epoch": 1.3235294117647058,
+      "grad_norm": 0.3655361428229988,
+      "learning_rate": 7.35122141620195e-05,
+      "loss": 0.3981,
+      "step": 146
+    },
+    {
+      "epoch": 1.332579185520362,
+      "grad_norm": 0.3675624921847117,
+      "learning_rate": 7.337293714369182e-05,
+      "loss": 0.3888,
+      "step": 147
+    },
+    {
+      "epoch": 1.3416289592760182,
+      "grad_norm": 0.38410932117144336,
+      "learning_rate": 7.323231586832219e-05,
+      "loss": 0.3983,
+      "step": 148
+    },
+    {
+      "epoch": 1.3506787330316743,
+      "grad_norm": 0.35565241834369704,
+      "learning_rate": 7.30903560001151e-05,
+      "loss": 0.3927,
+      "step": 149
+    },
+    {
+      "epoch": 1.3597285067873304,
+      "grad_norm": 0.3699524089900378,
+      "learning_rate": 7.294706325719331e-05,
+      "loss": 0.4009,
+      "step": 150
+    },
+    {
+      "epoch": 1.3687782805429864,
+      "grad_norm": 0.37565752487407766,
+      "learning_rate": 7.280244341136765e-05,
+      "loss": 0.3968,
+      "step": 151
+    },
+    {
+      "epoch": 1.3778280542986425,
+      "grad_norm": 0.41101581863985015,
+      "learning_rate": 7.26565022879044e-05,
+      "loss": 0.396,
+      "step": 152
+    },
+    {
+      "epoch": 1.3868778280542986,
+      "grad_norm": 0.4755720566371037,
+      "learning_rate": 7.250924576529072e-05,
+      "loss": 0.3941,
+      "step": 153
+    },
+    {
+      "epoch": 1.3959276018099547,
+      "grad_norm": 0.5478106344364355,
+      "learning_rate": 7.236067977499791e-05,
+      "loss": 0.396,
+      "step": 154
+    },
+    {
+      "epoch": 1.4049773755656108,
+      "grad_norm": 0.6612146719144061,
+      "learning_rate": 7.221081030124235e-05,
+      "loss": 0.3927,
+      "step": 155
+    },
+    {
+      "epoch": 1.4140271493212668,
+      "grad_norm": 0.7051474533138959,
+      "learning_rate": 7.205964338074462e-05,
+      "loss": 0.3991,
+      "step": 156
+    },
+    {
+      "epoch": 1.4230769230769231,
+      "grad_norm": 0.6630123949633135,
+      "learning_rate": 7.190718510248622e-05,
+      "loss": 0.3969,
+      "step": 157
+    },
+    {
+      "epoch": 1.4321266968325792,
+      "grad_norm": 0.5841525031103902,
+      "learning_rate": 7.175344160746438e-05,
+      "loss": 0.4009,
+      "step": 158
+    },
+    {
+      "epoch": 1.4411764705882353,
+      "grad_norm": 0.5791330941522758,
+      "learning_rate": 7.159841908844465e-05,
+      "loss": 0.3904,
+      "step": 159
+    },
+    {
+      "epoch": 1.4502262443438914,
+      "grad_norm": 0.6974262225802738,
+      "learning_rate": 7.144212378971151e-05,
+      "loss": 0.3965,
+      "step": 160
+    },
+    {
+      "epoch": 1.4592760180995474,
+      "grad_norm": 0.7970932314626533,
+      "learning_rate": 7.128456200681678e-05,
+      "loss": 0.3913,
+      "step": 161
+    },
+    {
+      "epoch": 1.4683257918552037,
+      "grad_norm": 0.7048456834043483,
+      "learning_rate": 7.11257400863261e-05,
+      "loss": 0.3901,
+      "step": 162
+    },
+    {
+      "epoch": 1.4773755656108598,
+      "grad_norm": 0.4096523489923981,
+      "learning_rate": 7.096566442556331e-05,
+      "loss": 0.3937,
+      "step": 163
+    },
+    {
+      "epoch": 1.4864253393665159,
+      "grad_norm": 0.3554945327024847,
+      "learning_rate": 7.080434147235263e-05,
+      "loss": 0.3892,
+      "step": 164
+    },
+    {
+      "epoch": 1.495475113122172,
+      "grad_norm": 0.5448243159096916,
+      "learning_rate": 7.064177772475912e-05,
+      "loss": 0.3985,
+      "step": 165
+    },
+    {
+      "epoch": 1.504524886877828,
+      "grad_norm": 0.4087602063081316,
+      "learning_rate": 7.047797973082684e-05,
+      "loss": 0.3953,
+      "step": 166
+    },
+    {
+      "epoch": 1.5135746606334841,
+      "grad_norm": 0.27977995475090656,
+      "learning_rate": 7.031295408831508e-05,
+      "loss": 0.3859,
+      "step": 167
+    },
+    {
+      "epoch": 1.5226244343891402,
+      "grad_norm": 0.4128081941990591,
+      "learning_rate": 7.014670744443267e-05,
+      "loss": 0.3993,
+      "step": 168
+    },
+    {
+      "epoch": 1.5316742081447963,
+      "grad_norm": 0.3728102691478845,
+      "learning_rate": 6.997924649557017e-05,
+      "loss": 0.3924,
+      "step": 169
+    },
+    {
+      "epoch": 1.5407239819004523,
+      "grad_norm": 0.2814741945781907,
+      "learning_rate": 6.98105779870302e-05,
+      "loss": 0.396,
+      "step": 170
+    },
+    {
+      "epoch": 1.5497737556561086,
+      "grad_norm": 0.3864339523003812,
+      "learning_rate": 6.964070871275567e-05,
+      "loss": 0.3958,
+      "step": 171
+    },
+    {
+      "epoch": 1.5588235294117647,
+      "grad_norm": 0.32053481390391325,
+      "learning_rate": 6.94696455150562e-05,
+      "loss": 0.3927,
+      "step": 172
+    },
+    {
+      "epoch": 1.5678733031674208,
+      "grad_norm": 0.25239085998303645,
+      "learning_rate": 6.929739528433244e-05,
+      "loss": 0.3931,
+      "step": 173
+    },
+    {
+      "epoch": 1.5769230769230769,
+      "grad_norm": 0.3797864909903561,
+      "learning_rate": 6.912396495879857e-05,
+      "loss": 0.3947,
+      "step": 174
+    },
+    {
+      "epoch": 1.5859728506787332,
+      "grad_norm": 0.3778896067077122,
+      "learning_rate": 6.89493615242028e-05,
+      "loss": 0.3958,
+      "step": 175
+    },
+    {
+      "epoch": 1.5950226244343892,
+      "grad_norm": 0.34652871604736857,
+      "learning_rate": 6.877359201354606e-05,
+      "loss": 0.3875,
+      "step": 176
+    },
+    {
+      "epoch": 1.6040723981900453,
+      "grad_norm": 0.42674803575084275,
+      "learning_rate": 6.859666350679854e-05,
+      "loss": 0.3909,
+      "step": 177
+    },
+    {
+      "epoch": 1.6131221719457014,
+      "grad_norm": 0.29475704922050416,
+      "learning_rate": 6.841858313061477e-05,
+      "loss": 0.3861,
+      "step": 178
+    },
+    {
+      "epoch": 1.6221719457013575,
+      "grad_norm": 0.3605164686190752,
+      "learning_rate": 6.823935805804626e-05,
+      "loss": 0.3946,
+      "step": 179
+    },
+    {
+      "epoch": 1.6312217194570136,
+      "grad_norm": 0.3793403076875403,
+      "learning_rate": 6.805899550825285e-05,
+      "loss": 0.3849,
+      "step": 180
+    },
+    {
+      "epoch": 1.6402714932126696,
+      "grad_norm": 0.38647041207728616,
+      "learning_rate": 6.787750274621175e-05,
+      "loss": 0.3841,
+      "step": 181
+    },
+    {
+      "epoch": 1.6493212669683257,
+      "grad_norm": 0.41289104580852704,
+      "learning_rate": 6.769488708242492e-05,
+      "loss": 0.3901,
+      "step": 182
+    },
+    {
+      "epoch": 1.6583710407239818,
+      "grad_norm": 0.4086406307183557,
+      "learning_rate": 6.751115587262469e-05,
+      "loss": 0.3886,
+      "step": 183
+    },
+    {
+      "epoch": 1.6674208144796379,
+      "grad_norm": 0.35780257992696674,
+      "learning_rate": 6.732631651747739e-05,
+      "loss": 0.3832,
+      "step": 184
+    },
+    {
+      "epoch": 1.6764705882352942,
+      "grad_norm": 0.30719607252305003,
+      "learning_rate": 6.714037646228529e-05,
+      "loss": 0.3904,
+      "step": 185
+    },
+    {
+      "epoch": 1.6855203619909502,
+      "grad_norm": 0.38674289134885453,
+      "learning_rate": 6.695334319668672e-05,
+      "loss": 0.3886,
+      "step": 186
+    },
+    {
+      "epoch": 1.6945701357466063,
+      "grad_norm": 0.4005595923099328,
+      "learning_rate": 6.676522425435433e-05,
+      "loss": 0.3919,
+      "step": 187
+    },
+    {
+      "epoch": 1.7036199095022626,
+      "grad_norm": 0.2349267690302485,
+      "learning_rate": 6.65760272126917e-05,
+      "loss": 0.3883,
+      "step": 188
+    },
+    {
+      "epoch": 1.7126696832579187,
+      "grad_norm": 0.3574524568496641,
+      "learning_rate": 6.638575969252806e-05,
+      "loss": 0.3865,
+      "step": 189
+    },
+    {
+      "epoch": 1.7217194570135748,
+      "grad_norm": 0.42670317548076125,
+      "learning_rate": 6.619442935781141e-05,
+      "loss": 0.3914,
+      "step": 190
+    },
+    {
+      "epoch": 1.7307692307692308,
+      "grad_norm": 0.30707518042373566,
+      "learning_rate": 6.600204391529971e-05,
+      "loss": 0.3865,
+      "step": 191
+    },
+    {
+      "epoch": 1.739819004524887,
+      "grad_norm": 0.2622083300455114,
+      "learning_rate": 6.580861111425053e-05,
+      "loss": 0.391,
+      "step": 192
+    },
+    {
+      "epoch": 1.748868778280543,
+      "grad_norm": 0.38093733410787994,
+      "learning_rate": 6.56141387461089e-05,
+      "loss": 0.3872,
+      "step": 193
+    },
+    {
+      "epoch": 1.757918552036199,
+      "grad_norm": 0.43742318682713477,
+      "learning_rate": 6.541863464419346e-05,
+      "loss": 0.3938,
+      "step": 194
+    },
+    {
+      "epoch": 1.7669683257918551,
+      "grad_norm": 0.4418280959306971,
+      "learning_rate": 6.52221066833809e-05,
+      "loss": 0.3848,
+      "step": 195
+    },
+    {
+      "epoch": 1.7760180995475112,
+      "grad_norm": 0.41049856835634324,
+      "learning_rate": 6.502456277978887e-05,
+      "loss": 0.3852,
+      "step": 196
+    },
+    {
+      "epoch": 1.7850678733031673,
+      "grad_norm": 0.4150478314670342,
+      "learning_rate": 6.482601089045696e-05,
+      "loss": 0.3844,
+      "step": 197
+    },
+    {
+      "epoch": 1.7941176470588234,
+      "grad_norm": 0.43894520954893845,
+      "learning_rate": 6.462645901302633e-05,
+      "loss": 0.3814,
+      "step": 198
+    },
+    {
+      "epoch": 1.8031674208144797,
+      "grad_norm": 0.41995958530789795,
+      "learning_rate": 6.442591518541753e-05,
+      "loss": 0.381,
+      "step": 199
+    },
+    {
+      "epoch": 1.8122171945701357,
+      "grad_norm": 0.3201288359968715,
+      "learning_rate": 6.422438748550667e-05,
+      "loss": 0.3852,
+      "step": 200
+    },
+    {
+      "epoch": 1.8212669683257918,
+      "grad_norm": 0.2931969021729461,
+      "learning_rate": 6.402188403080013e-05,
+      "loss": 0.3878,
+      "step": 201
+    },
+    {
+      "epoch": 1.8303167420814481,
+      "grad_norm": 0.4066483749649814,
+      "learning_rate": 6.381841297810753e-05,
+      "loss": 0.3834,
+      "step": 202
+    },
+    {
+      "epoch": 1.8393665158371042,
+      "grad_norm": 0.40957398047436605,
+      "learning_rate": 6.361398252321321e-05,
+      "loss": 0.3886,
+      "step": 203
+    },
+    {
+      "epoch": 1.8484162895927603,
+      "grad_norm": 0.32160755061403157,
+      "learning_rate": 6.340860090054608e-05,
+      "loss": 0.3841,
+      "step": 204
+    },
+    {
+      "epoch": 1.8574660633484164,
+      "grad_norm": 0.23090971242916025,
+      "learning_rate": 6.320227638284793e-05,
+      "loss": 0.3842,
+      "step": 205
+    },
+    {
+      "epoch": 1.8665158371040724,
+      "grad_norm": 0.32946206927022925,
+      "learning_rate": 6.29950172808403e-05,
+      "loss": 0.3901,
+      "step": 206
+    },
+    {
+      "epoch": 1.8755656108597285,
+      "grad_norm": 0.3530001057183439,
+      "learning_rate": 6.278683194288956e-05,
+      "loss": 0.3852,
+      "step": 207
+    },
+    {
+      "epoch": 1.8846153846153846,
+      "grad_norm": 0.28218082313802967,
+      "learning_rate": 6.257772875467078e-05,
+      "loss": 0.387,
+      "step": 208
+    },
+    {
+      "epoch": 1.8936651583710407,
+      "grad_norm": 0.3059885996215407,
+      "learning_rate": 6.236771613882987e-05,
+      "loss": 0.3883,
+      "step": 209
+    },
+    {
+      "epoch": 1.9027149321266967,
+      "grad_norm": 0.34171931460456756,
+      "learning_rate": 6.215680255464442e-05,
+      "loss": 0.3824,
+      "step": 210
+    },
+    {
+      "epoch": 1.9117647058823528,
+      "grad_norm": 0.29374439911760764,
+      "learning_rate": 6.194499649768281e-05,
+      "loss": 0.3864,
+      "step": 211
+    },
+    {
+      "epoch": 1.9208144796380089,
+      "grad_norm": 0.29144021228496453,
+      "learning_rate": 6.173230649946213e-05,
+      "loss": 0.3836,
+      "step": 212
+    },
+    {
+      "epoch": 1.9298642533936652,
+      "grad_norm": 0.29148303948179005,
+      "learning_rate": 6.15187411271045e-05,
+      "loss": 0.3796,
+      "step": 213
+    },
+    {
+      "epoch": 1.9389140271493213,
+      "grad_norm": 0.2754963794966634,
+      "learning_rate": 6.130430898299199e-05,
+      "loss": 0.3848,
+      "step": 214
+    },
+    {
+      "epoch": 1.9479638009049773,
+      "grad_norm": 0.26672266015023416,
+      "learning_rate": 6.10890187044201e-05,
+      "loss": 0.3836,
+      "step": 215
+    },
+    {
+      "epoch": 1.9570135746606336,
+      "grad_norm": 0.3413044280319467,
+      "learning_rate": 6.087287896324984e-05,
+      "loss": 0.385,
+      "step": 216
+    },
+    {
+      "epoch": 1.9660633484162897,
+      "grad_norm": 0.4653279793260335,
+      "learning_rate": 6.0655898465558484e-05,
+      "loss": 0.3871,
+      "step": 217
+    },
+    {
+      "epoch": 1.9751131221719458,
+      "grad_norm": 0.4607547435763325,
+      "learning_rate": 6.043808595128883e-05,
+      "loss": 0.3827,
+      "step": 218
+    },
+    {
+      "epoch": 1.9841628959276019,
+      "grad_norm": 0.3114177233194163,
+      "learning_rate": 6.021945019389719e-05,
+      "loss": 0.391,
+      "step": 219
+    },
+    {
+      "epoch": 1.993212669683258,
+      "grad_norm": 0.263615834735127,
+      "learning_rate": 6.000000000000001e-05,
+      "loss": 0.381,
+      "step": 220
+    },
+    {
+      "epoch": 2.004524886877828,
+      "grad_norm": 0.3080883305611625,
+      "learning_rate": 5.977974420901908e-05,
+      "loss": 0.3732,
+      "step": 221
+    },
+    {
+      "epoch": 2.013574660633484,
+      "grad_norm": 0.2983316325982498,
+      "learning_rate": 5.955869169282556e-05,
+      "loss": 0.3539,
+      "step": 222
+    },
+    {
+      "epoch": 2.02262443438914,
+      "grad_norm": 0.35442755044439017,
+      "learning_rate": 5.9336851355382557e-05,
+      "loss": 0.3625,
+      "step": 223
+    },
+    {
+      "epoch": 2.0316742081447963,
+      "grad_norm": 0.5038394603992613,
+      "learning_rate": 5.911423213238653e-05,
+      "loss": 0.3565,
+      "step": 224
+    },
+    {
+      "epoch": 2.0407239819004523,
+      "grad_norm": 0.6565318273183188,
+      "learning_rate": 5.889084299090732e-05,
+      "loss": 0.3562,
+      "step": 225
+    },
+    {
+      "epoch": 2.0497737556561084,
+      "grad_norm": 0.8028771505743133,
+      "learning_rate": 5.866669292902695e-05,
+      "loss": 0.36,
+      "step": 226
+    },
+    {
+      "epoch": 2.0588235294117645,
+      "grad_norm": 0.8257109533011188,
+      "learning_rate": 5.844179097547725e-05,
+      "loss": 0.3602,
+      "step": 227
+    },
+    {
+      "epoch": 2.067873303167421,
+      "grad_norm": 0.6350182381420444,
+      "learning_rate": 5.821614618927613e-05,
+      "loss": 0.3571,
+      "step": 228
+    },
+    {
+      "epoch": 2.076923076923077,
+      "grad_norm": 0.38135287559376224,
+      "learning_rate": 5.798976765936264e-05,
+      "loss": 0.3559,
+      "step": 229
+    },
+    {
+      "epoch": 2.085972850678733,
+      "grad_norm": 0.5026969066810225,
+      "learning_rate": 5.776266450423097e-05,
+      "loss": 0.355,
+      "step": 230
+    },
+    {
+      "epoch": 2.0950226244343892,
+      "grad_norm": 0.6298538131496503,
+      "learning_rate": 5.75348458715631e-05,
+      "loss": 0.3595,
+      "step": 231
+    },
+    {
+      "epoch": 2.1040723981900453,
+      "grad_norm": 0.3867446080865823,
+      "learning_rate": 5.7306320937860336e-05,
+      "loss": 0.352,
+      "step": 232
+    },
+    {
+      "epoch": 2.1131221719457014,
+      "grad_norm": 0.3115968628085643,
+      "learning_rate": 5.7077098908073676e-05,
+      "loss": 0.3582,
+      "step": 233
+    },
+    {
+      "epoch": 2.1221719457013575,
+      "grad_norm": 0.3939923894024485,
+      "learning_rate": 5.684718901523307e-05,
+      "loss": 0.3533,
+      "step": 234
+    },
+    {
+      "epoch": 2.1312217194570136,
+      "grad_norm": 0.25112142036368945,
+      "learning_rate": 5.661660052007547e-05,
+      "loss": 0.3564,
+      "step": 235
+    },
+    {
+      "epoch": 2.1402714932126696,
+      "grad_norm": 0.3261992149324723,
+      "learning_rate": 5.6385342710671815e-05,
+      "loss": 0.3518,
+      "step": 236
+    },
+    {
+      "epoch": 2.1493212669683257,
+      "grad_norm": 0.32489012611444446,
+      "learning_rate": 5.6153424902053e-05,
+      "loss": 0.3535,
+      "step": 237
+    },
+    {
+      "epoch": 2.158371040723982,
+      "grad_norm": 0.24162335661441722,
+      "learning_rate": 5.59208564358345e-05,
+      "loss": 0.3548,
+      "step": 238
+    },
+    {
+      "epoch": 2.167420814479638,
+      "grad_norm": 0.3778776749277093,
+      "learning_rate": 5.568764667984022e-05,
+      "loss": 0.3605,
+      "step": 239
+    },
+    {
+      "epoch": 2.176470588235294,
+      "grad_norm": 0.33672093596701086,
+      "learning_rate": 5.5453805027725145e-05,
+      "loss": 0.3538,
+      "step": 240
+    },
+    {
+      "epoch": 2.1855203619909505,
+      "grad_norm": 0.24527198542362685,
+      "learning_rate": 5.521934089859692e-05,
+      "loss": 0.3497,
+      "step": 241
+    },
+    {
+      "epoch": 2.1945701357466065,
+      "grad_norm": 0.3958849461770139,
+      "learning_rate": 5.4984263736636494e-05,
+      "loss": 0.355,
+      "step": 242
+    },
+    {
+      "epoch": 2.2036199095022626,
+      "grad_norm": 0.38916418294477323,
+      "learning_rate": 5.4748583010717636e-05,
+      "loss": 0.3557,
+      "step": 243
+    },
+    {
+      "epoch": 2.2126696832579187,
+      "grad_norm": 0.36390563994328023,
+      "learning_rate": 5.451230821402564e-05,
+      "loss": 0.362,
+      "step": 244
+    },
+    {
+      "epoch": 2.2217194570135748,
+      "grad_norm": 0.3301056539987755,
+      "learning_rate": 5.427544886367488e-05,
+      "loss": 0.3519,
+      "step": 245
+    },
+    {
+      "epoch": 2.230769230769231,
+      "grad_norm": 0.215969613599219,
+      "learning_rate": 5.403801450032544e-05,
+      "loss": 0.3551,
+      "step": 246
+    },
+    {
+      "epoch": 2.239819004524887,
+      "grad_norm": 0.2378223811885189,
+      "learning_rate": 5.380001468779883e-05,
+      "loss": 0.359,
+      "step": 247
+    },
+    {
+      "epoch": 2.248868778280543,
+      "grad_norm": 0.3256287018733439,
+      "learning_rate": 5.356145901269282e-05,
+      "loss": 0.3569,
+      "step": 248
+    },
+    {
+      "epoch": 2.257918552036199,
+      "grad_norm": 0.2802481742472565,
+      "learning_rate": 5.3322357083995235e-05,
+      "loss": 0.3584,
+      "step": 249
+    },
+    {
+      "epoch": 2.266968325791855,
+      "grad_norm": 0.30050063645860137,
+      "learning_rate": 5.3082718532696874e-05,
+      "loss": 0.3527,
+      "step": 250
+    },
+    {
+      "epoch": 2.276018099547511,
+      "grad_norm": 0.337462061586617,
+      "learning_rate": 5.284255301140364e-05,
+      "loss": 0.3508,
+      "step": 251
+    },
+    {
+      "epoch": 2.2850678733031673,
+      "grad_norm": 0.22314780810636892,
+      "learning_rate": 5.2601870193947716e-05,
+      "loss": 0.3486,
+      "step": 252
+    },
+    {
+      "epoch": 2.2941176470588234,
+      "grad_norm": 0.2251099676952424,
+      "learning_rate": 5.23606797749979e-05,
+      "loss": 0.3521,
+      "step": 253
+    },
+    {
+      "epoch": 2.3031674208144794,
+      "grad_norm": 0.2657181581176672,
+      "learning_rate": 5.21189914696691e-05,
+      "loss": 0.3506,
+      "step": 254
+    },
+    {
+      "epoch": 2.3122171945701355,
+      "grad_norm": 0.23800570673769336,
+      "learning_rate": 5.1876815013131e-05,
+      "loss": 0.3511,
+      "step": 255
+    },
+    {
+      "epoch": 2.321266968325792,
+      "grad_norm": 0.2673484662572871,
+      "learning_rate": 5.163416016021597e-05,
+      "loss": 0.3537,
+      "step": 256
+    },
+    {
+      "epoch": 2.330316742081448,
+      "grad_norm": 0.25378700284589706,
+      "learning_rate": 5.1391036685026093e-05,
+      "loss": 0.3492,
+      "step": 257
+    },
+    {
+      "epoch": 2.339366515837104,
+      "grad_norm": 0.2198551274983534,
+      "learning_rate": 5.114745438053952e-05,
+      "loss": 0.3522,
+      "step": 258
+    },
+    {
+      "epoch": 2.3484162895927603,
+      "grad_norm": 0.22757974892064678,
+      "learning_rate": 5.0903423058215925e-05,
+      "loss": 0.3531,
+      "step": 259
+    },
+    {
+      "epoch": 2.3574660633484164,
+      "grad_norm": 0.2162305337190899,
+      "learning_rate": 5.06589525476014e-05,
+      "loss": 0.3509,
+      "step": 260
+    },
+    {
+      "epoch": 2.3665158371040724,
+      "grad_norm": 0.22176152775329794,
+      "learning_rate": 5.0414052695932486e-05,
+      "loss": 0.3539,
+      "step": 261
+    },
+    {
+      "epoch": 2.3755656108597285,
+      "grad_norm": 0.20465996494402094,
+      "learning_rate": 5.016873336773949e-05,
+      "loss": 0.3513,
+      "step": 262
+    },
+    {
+      "epoch": 2.3846153846153846,
+      "grad_norm": 0.19749781398286828,
+      "learning_rate": 4.992300444444916e-05,
+      "loss": 0.3569,
+      "step": 263
+    },
+    {
+      "epoch": 2.3936651583710407,
+      "grad_norm": 0.21862557545293346,
+      "learning_rate": 4.967687582398671e-05,
+      "loss": 0.3532,
+      "step": 264
+    },
+    {
+      "epoch": 2.4027149321266967,
+      "grad_norm": 0.22129803132569947,
+      "learning_rate": 4.94303574203771e-05,
+      "loss": 0.3546,
+      "step": 265
+    },
+    {
+      "epoch": 2.411764705882353,
+      "grad_norm": 0.22177138969097251,
+      "learning_rate": 4.9183459163345644e-05,
+      "loss": 0.3502,
+      "step": 266
+    },
+    {
+      "epoch": 2.420814479638009,
+      "grad_norm": 0.19327131140015494,
+      "learning_rate": 4.893619099791817e-05,
+      "loss": 0.3522,
+      "step": 267
+    },
+    {
+      "epoch": 2.4298642533936654,
+      "grad_norm": 0.20052109770619267,
+      "learning_rate": 4.868856288402032e-05,
+      "loss": 0.356,
+      "step": 268
+    },
+    {
+      "epoch": 2.4389140271493215,
+      "grad_norm": 0.19644855379050266,
+      "learning_rate": 4.8440584796076395e-05,
+      "loss": 0.3534,
+      "step": 269
+    },
+    {
+      "epoch": 2.4479638009049776,
+      "grad_norm": 0.20192604708935719,
+      "learning_rate": 4.819226672260763e-05,
+      "loss": 0.3592,
+      "step": 270
+    },
+    {
+      "epoch": 2.4570135746606336,
+      "grad_norm": 0.1922532533153157,
+      "learning_rate": 4.794361866582982e-05,
+      "loss": 0.3552,
+      "step": 271
+    },
+    {
+      "epoch": 2.4660633484162897,
+      "grad_norm": 0.22962467692658226,
+      "learning_rate": 4.7694650641250446e-05,
+      "loss": 0.3531,
+      "step": 272
+    },
+    {
+      "epoch": 2.475113122171946,
+      "grad_norm": 0.21255494327094565,
+      "learning_rate": 4.7445372677265205e-05,
+      "loss": 0.3554,
+      "step": 273
+    },
+    {
+      "epoch": 2.484162895927602,
+      "grad_norm": 0.19556932541533853,
+      "learning_rate": 4.719579481475416e-05,
+      "loss": 0.3558,
+      "step": 274
+    },
+    {
+      "epoch": 2.493212669683258,
+      "grad_norm": 0.2127502030224185,
+      "learning_rate": 4.694592710667723e-05,
+      "loss": 0.3558,
+      "step": 275
+    },
+    {
+      "epoch": 2.502262443438914,
+      "grad_norm": 0.2051192723904285,
+      "learning_rate": 4.6695779617669236e-05,
+      "loss": 0.355,
+      "step": 276
+    },
+    {
+      "epoch": 2.51131221719457,
+      "grad_norm": 0.21459679896962008,
+      "learning_rate": 4.6445362423634574e-05,
+      "loss": 0.3547,
+      "step": 277
+    },
+    {
+      "epoch": 2.520361990950226,
+      "grad_norm": 0.21249030888822018,
+      "learning_rate": 4.61946856113413e-05,
+      "loss": 0.3524,
+      "step": 278
+    },
+    {
+      "epoch": 2.5294117647058822,
+      "grad_norm": 0.18973130323510315,
+      "learning_rate": 4.594375927801487e-05,
+      "loss": 0.3518,
+      "step": 279
+    },
+    {
+      "epoch": 2.5384615384615383,
+      "grad_norm": 0.20099116045455737,
+      "learning_rate": 4.5692593530931416e-05,
+      "loss": 0.3551,
+      "step": 280
+    },
+    {
+      "epoch": 2.5475113122171944,
+      "grad_norm": 0.2404974178503931,
+      "learning_rate": 4.5441198487010574e-05,
+      "loss": 0.3546,
+      "step": 281
+    },
+    {
+      "epoch": 2.5565610859728505,
+      "grad_norm": 0.21416667678544446,
+      "learning_rate": 4.5189584272408074e-05,
+      "loss": 0.3521,
+      "step": 282
+    },
+    {
+      "epoch": 2.5656108597285066,
+      "grad_norm": 0.1821577098280573,
+      "learning_rate": 4.493776102210779e-05,
+      "loss": 0.351,
+      "step": 283
+    },
+    {
+      "epoch": 2.5746606334841626,
+      "grad_norm": 0.1940123893538777,
+      "learning_rate": 4.468573887951354e-05,
+      "loss": 0.352,
+      "step": 284
+    },
+    {
+      "epoch": 2.583710407239819,
+      "grad_norm": 0.21908628404417796,
+      "learning_rate": 4.4433527996040443e-05,
+      "loss": 0.3555,
+      "step": 285
+    },
+    {
+      "epoch": 2.5927601809954752,
+      "grad_norm": 0.16351576977199667,
+      "learning_rate": 4.418113853070614e-05,
+      "loss": 0.3534,
+      "step": 286
+    },
+    {
+      "epoch": 2.6018099547511313,
+      "grad_norm": 0.18499158266871718,
+      "learning_rate": 4.392858064972149e-05,
+      "loss": 0.3555,
+      "step": 287
+    },
+    {
+      "epoch": 2.6108597285067874,
+      "grad_norm": 0.17222994187080978,
+      "learning_rate": 4.3675864526081106e-05,
+      "loss": 0.3522,
+      "step": 288
+    },
+    {
+      "epoch": 2.6199095022624435,
+      "grad_norm": 0.17596242704110496,
+      "learning_rate": 4.34230003391536e-05,
+      "loss": 0.3529,
+      "step": 289
+    },
+    {
+      "epoch": 2.6289592760180995,
+      "grad_norm": 0.1788336399501724,
+      "learning_rate": 4.316999827427154e-05,
+      "loss": 0.3501,
+      "step": 290
+    },
+    {
+      "epoch": 2.6380090497737556,
+      "grad_norm": 0.15674058262172438,
+      "learning_rate": 4.2916868522321235e-05,
+      "loss": 0.3593,
+      "step": 291
+    },
+    {
+      "epoch": 2.6470588235294117,
+      "grad_norm": 0.19461132000856615,
+      "learning_rate": 4.266362127933216e-05,
+      "loss": 0.3542,
+      "step": 292
+    },
+    {
+      "epoch": 2.6561085972850678,
+      "grad_norm": 0.18265415777078592,
+      "learning_rate": 4.2410266746066345e-05,
+      "loss": 0.3529,
+      "step": 293
+    },
+    {
+      "epoch": 2.665158371040724,
+      "grad_norm": 0.19366748651198787,
+      "learning_rate": 4.215681512760744e-05,
+      "loss": 0.3545,
+      "step": 294
+    },
+    {
+      "epoch": 2.6742081447963804,
+      "grad_norm": 0.19701470985076372,
+      "learning_rate": 4.19032766329497e-05,
+      "loss": 0.3533,
+      "step": 295
+    },
+    {
+      "epoch": 2.6832579185520364,
+      "grad_norm": 0.2051473001306171,
+      "learning_rate": 4.1649661474586694e-05,
+      "loss": 0.3543,
+      "step": 296
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 0.18550660724506887,
+      "learning_rate": 4.139597986810005e-05,
+      "loss": 0.3569,
+      "step": 297
+    },
+    {
+      "epoch": 2.7013574660633486,
+      "grad_norm": 0.20456848012158177,
+      "learning_rate": 4.114224203174785e-05,
+      "loss": 0.3496,
+      "step": 298
+    },
+    {
+      "epoch": 2.7104072398190047,
+      "grad_norm": 0.1780622844651236,
+      "learning_rate": 4.0888458186053184e-05,
+      "loss": 0.354,
+      "step": 299
+    },
+    {
+      "epoch": 2.7194570135746607,
+      "grad_norm": 0.1640758315975294,
+      "learning_rate": 4.063463855339232e-05,
+      "loss": 0.3496,
+      "step": 300
+    },
+    {
+      "epoch": 2.728506787330317,
+      "grad_norm": 0.19228477606561048,
+      "learning_rate": 4.0380793357583076e-05,
+      "loss": 0.3491,
+      "step": 301
+    },
+    {
+      "epoch": 2.737556561085973,
+      "grad_norm": 0.1755323101374011,
+      "learning_rate": 4.012693282347289e-05,
+      "loss": 0.3564,
+      "step": 302
+    },
+    {
+      "epoch": 2.746606334841629,
+      "grad_norm": 0.1809635624979587,
+      "learning_rate": 3.9873067176527114e-05,
+      "loss": 0.3544,
+      "step": 303
+    },
+    {
+      "epoch": 2.755656108597285,
+      "grad_norm": 0.18201509342269132,
+      "learning_rate": 3.961920664241694e-05,
+      "loss": 0.3546,
+      "step": 304
+    },
+    {
+      "epoch": 2.764705882352941,
+      "grad_norm": 0.17873879422611214,
+      "learning_rate": 3.9365361446607684e-05,
+      "loss": 0.355,
+      "step": 305
+    },
+    {
+      "epoch": 2.773755656108597,
+      "grad_norm": 0.21315929069191436,
+      "learning_rate": 3.911154181394682e-05,
+      "loss": 0.3564,
+      "step": 306
+    },
+    {
+      "epoch": 2.7828054298642533,
+      "grad_norm": 0.19819053615429819,
+      "learning_rate": 3.885775796825216e-05,
+      "loss": 0.3548,
+      "step": 307
+    },
+    {
+      "epoch": 2.7918552036199094,
+      "grad_norm": 0.13490710454922517,
+      "learning_rate": 3.860402013189998e-05,
+      "loss": 0.3485,
+      "step": 308
+    },
+    {
+      "epoch": 2.8009049773755654,
+      "grad_norm": 0.20244714474335393,
+      "learning_rate": 3.835033852541332e-05,
+      "loss": 0.3562,
+      "step": 309
+    },
+    {
+      "epoch": 2.8099547511312215,
+      "grad_norm": 0.1724672504964484,
+      "learning_rate": 3.809672336705031e-05,
+      "loss": 0.348,
+      "step": 310
+    },
+    {
+      "epoch": 2.8190045248868776,
+      "grad_norm": 0.15081349665835933,
+      "learning_rate": 3.784318487239257e-05,
+      "loss": 0.3546,
+      "step": 311
+    },
+    {
+      "epoch": 2.8280542986425337,
+      "grad_norm": 0.1783285611622551,
+      "learning_rate": 3.758973325393367e-05,
+      "loss": 0.35,
+      "step": 312
+    },
+    {
+      "epoch": 2.83710407239819,
+      "grad_norm": 0.16120590818140387,
+      "learning_rate": 3.7336378720667846e-05,
+      "loss": 0.3537,
+      "step": 313
+    },
+    {
+      "epoch": 2.8461538461538463,
+      "grad_norm": 0.20165982856627884,
+      "learning_rate": 3.708313147767878e-05,
+      "loss": 0.3543,
+      "step": 314
+    },
+    {
+      "epoch": 2.8552036199095023,
+      "grad_norm": 0.21095988182947403,
+      "learning_rate": 3.683000172572846e-05,
+      "loss": 0.3544,
+      "step": 315
+    },
+    {
+      "epoch": 2.8642533936651584,
+      "grad_norm": 0.18625234007151376,
+      "learning_rate": 3.657699966084642e-05,
+      "loss": 0.3531,
+      "step": 316
+    },
+    {
+      "epoch": 2.8733031674208145,
+      "grad_norm": 0.22479750462747256,
+      "learning_rate": 3.632413547391891e-05,
+      "loss": 0.3556,
+      "step": 317
+    },
+    {
+      "epoch": 2.8823529411764706,
+      "grad_norm": 0.20909188100700477,
+      "learning_rate": 3.6071419350278515e-05,
+      "loss": 0.3471,
+      "step": 318
+    },
+    {
+      "epoch": 2.8914027149321266,
+      "grad_norm": 0.18778607505398776,
+      "learning_rate": 3.581886146929387e-05,
+      "loss": 0.3451,
+      "step": 319
+    },
+    {
+      "epoch": 2.9004524886877827,
+      "grad_norm": 0.21561728314983256,
+      "learning_rate": 3.556647200395956e-05,
+      "loss": 0.352,
+      "step": 320
+    },
+    {
+      "epoch": 2.909502262443439,
+      "grad_norm": 0.20718371330348473,
+      "learning_rate": 3.5314261120486474e-05,
+      "loss": 0.3493,
+      "step": 321
+    },
+    {
+      "epoch": 2.918552036199095,
+      "grad_norm": 0.16443205391902374,
+      "learning_rate": 3.5062238977892214e-05,
+      "loss": 0.3508,
+      "step": 322
+    },
+    {
+      "epoch": 2.9276018099547514,
+      "grad_norm": 0.19396119036177084,
+      "learning_rate": 3.481041572759193e-05,
+      "loss": 0.355,
+      "step": 323
+    },
+    {
+      "epoch": 2.9366515837104075,
+      "grad_norm": 0.18132782101939904,
+      "learning_rate": 3.4558801512989446e-05,
+      "loss": 0.354,
+      "step": 324
+    },
+    {
+      "epoch": 2.9457013574660635,
+      "grad_norm": 0.17332578387246808,
+      "learning_rate": 3.4307406469068604e-05,
+      "loss": 0.3542,
+      "step": 325
+    },
+    {
+      "epoch": 2.9547511312217196,
+      "grad_norm": 0.15354450735940553,
+      "learning_rate": 3.405624072198514e-05,
+      "loss": 0.3553,
+      "step": 326
+    },
+    {
+      "epoch": 2.9638009049773757,
+      "grad_norm": 0.17733336095930732,
+      "learning_rate": 3.3805314388658714e-05,
+      "loss": 0.3507,
+      "step": 327
+    },
+    {
+      "epoch": 2.9728506787330318,
+      "grad_norm": 0.16922858743324629,
+      "learning_rate": 3.3554637576365446e-05,
+      "loss": 0.3469,
+      "step": 328
+    },
+    {
+      "epoch": 2.981900452488688,
+      "grad_norm": 0.1635466514191799,
+      "learning_rate": 3.330422038233078e-05,
+      "loss": 0.3499,
+      "step": 329
+    },
+    {
+      "epoch": 2.990950226244344,
+      "grad_norm": 0.181396866738228,
+      "learning_rate": 3.305407289332279e-05,
+      "loss": 0.3475,
+      "step": 330
+    },
+    {
+      "epoch": 3.002262443438914,
+      "grad_norm": 0.17232098683504665,
+      "learning_rate": 3.280420518524585e-05,
+      "loss": 0.3429,
+      "step": 331
+    },
+    {
+      "epoch": 3.01131221719457,
+      "grad_norm": 0.2085779039684738,
+      "learning_rate": 3.25546273227348e-05,
+      "loss": 0.3276,
+      "step": 332
+    },
+    {
+      "epoch": 3.020361990950226,
+      "grad_norm": 0.20454372196591433,
+      "learning_rate": 3.230534935874958e-05,
+      "loss": 0.322,
+      "step": 333
+    },
+    {
+      "epoch": 3.0294117647058822,
+      "grad_norm": 0.21400683996161396,
+      "learning_rate": 3.205638133417019e-05,
+      "loss": 0.3247,
+      "step": 334
+    },
+    {
+      "epoch": 3.0384615384615383,
+      "grad_norm": 0.20994330105610337,
+      "learning_rate": 3.180773327739238e-05,
+      "loss": 0.328,
+      "step": 335
+    },
+    {
+      "epoch": 3.0475113122171944,
+      "grad_norm": 0.2107968970213985,
+      "learning_rate": 3.155941520392362e-05,
+      "loss": 0.3235,
+      "step": 336
+    },
+    {
+      "epoch": 3.0565610859728505,
+      "grad_norm": 0.20164240350868537,
+      "learning_rate": 3.1311437115979696e-05,
+      "loss": 0.3203,
+      "step": 337
+    },
+    {
+      "epoch": 3.065610859728507,
+      "grad_norm": 0.20071440677649943,
+      "learning_rate": 3.1063809002081834e-05,
+      "loss": 0.3221,
+      "step": 338
+    },
+    {
+      "epoch": 3.074660633484163,
+      "grad_norm": 0.17663930655160845,
+      "learning_rate": 3.0816540836654356e-05,
+      "loss": 0.3226,
+      "step": 339
+    },
+    {
+      "epoch": 3.083710407239819,
+      "grad_norm": 0.18106709441769947,
+      "learning_rate": 3.0569642579622905e-05,
+      "loss": 0.3214,
+      "step": 340
+    },
+    {
+      "epoch": 3.0927601809954752,
+      "grad_norm": 0.17780441082604515,
+      "learning_rate": 3.0323124176013297e-05,
+      "loss": 0.3225,
+      "step": 341
+    },
+    {
+      "epoch": 3.1018099547511313,
+      "grad_norm": 0.17074344590150428,
+      "learning_rate": 3.007699555555086e-05,
+      "loss": 0.3183,
+      "step": 342
+    },
+    {
+      "epoch": 3.1108597285067874,
+      "grad_norm": 0.22118329908847872,
+      "learning_rate": 2.9831266632260534e-05,
+      "loss": 0.3221,
+      "step": 343
+    },
+    {
+      "epoch": 3.1199095022624435,
+      "grad_norm": 0.1862795261167787,
+      "learning_rate": 2.958594730406752e-05,
+      "loss": 0.3233,
+      "step": 344
+    },
+    {
+      "epoch": 3.1289592760180995,
+      "grad_norm": 0.18624339502853474,
+      "learning_rate": 2.9341047452398607e-05,
+      "loss": 0.3232,
+      "step": 345
+    },
+    {
+      "epoch": 3.1380090497737556,
+      "grad_norm": 0.19906920288843233,
+      "learning_rate": 2.9096576941784095e-05,
+      "loss": 0.3217,
+      "step": 346
+    },
+    {
+      "epoch": 3.1470588235294117,
+      "grad_norm": 0.1759509265715728,
+      "learning_rate": 2.8852545619460495e-05,
+      "loss": 0.3294,
+      "step": 347
+    },
+    {
+      "epoch": 3.1561085972850678,
+      "grad_norm": 0.18215420840842134,
+      "learning_rate": 2.860896331497391e-05,
+      "loss": 0.3178,
+      "step": 348
+    },
+    {
+      "epoch": 3.165158371040724,
+      "grad_norm": 0.16928396196250736,
+      "learning_rate": 2.8365839839784038e-05,
+      "loss": 0.3208,
+      "step": 349
+    },
+    {
+      "epoch": 3.17420814479638,
+      "grad_norm": 0.18758285635894378,
+      "learning_rate": 2.8123184986869022e-05,
+      "loss": 0.3252,
+      "step": 350
+    },
+    {
+      "epoch": 3.183257918552036,
+      "grad_norm": 0.1727196365880379,
+      "learning_rate": 2.7881008530330914e-05,
+      "loss": 0.3219,
+      "step": 351
+    },
+    {
+      "epoch": 3.1923076923076925,
+      "grad_norm": 0.19928771398379677,
+      "learning_rate": 2.7639320225002108e-05,
+      "loss": 0.3279,
+      "step": 352
+    },
+    {
+      "epoch": 3.2013574660633486,
+      "grad_norm": 0.15934892115019736,
+      "learning_rate": 2.7398129806052298e-05,
+      "loss": 0.3211,
+      "step": 353
+    },
+    {
+      "epoch": 3.2104072398190047,
+      "grad_norm": 0.1620987722032314,
+      "learning_rate": 2.715744698859637e-05,
+      "loss": 0.3264,
+      "step": 354
+    },
+    {
+      "epoch": 3.2194570135746607,
+      "grad_norm": 0.14580042456343553,
+      "learning_rate": 2.691728146730314e-05,
+      "loss": 0.3247,
+      "step": 355
+    },
+    {
+      "epoch": 3.228506787330317,
+      "grad_norm": 0.1669926797929679,
+      "learning_rate": 2.6677642916004772e-05,
+      "loss": 0.3219,
+      "step": 356
+    },
+    {
+      "epoch": 3.237556561085973,
+      "grad_norm": 0.13912969057523117,
+      "learning_rate": 2.6438540987307174e-05,
+      "loss": 0.3224,
+      "step": 357
+    },
+    {
+      "epoch": 3.246606334841629,
+      "grad_norm": 0.17393961594448798,
+      "learning_rate": 2.6199985312201185e-05,
+      "loss": 0.3287,
+      "step": 358
+    },
+    {
+      "epoch": 3.255656108597285,
+      "grad_norm": 0.13336325385823727,
+      "learning_rate": 2.5961985499674587e-05,
+      "loss": 0.3195,
+      "step": 359
+    },
+    {
+      "epoch": 3.264705882352941,
+      "grad_norm": 0.16625718933722608,
+      "learning_rate": 2.5724551136325132e-05,
+      "loss": 0.3155,
+      "step": 360
+    },
+    {
+      "epoch": 3.273755656108597,
+      "grad_norm": 0.1313925578933839,
+      "learning_rate": 2.5487691785974366e-05,
+      "loss": 0.3245,
+      "step": 361
+    },
+    {
+      "epoch": 3.2828054298642533,
+      "grad_norm": 0.1580151751002634,
+      "learning_rate": 2.5251416989282377e-05,
+      "loss": 0.3244,
+      "step": 362
+    },
+    {
+      "epoch": 3.2918552036199094,
+      "grad_norm": 0.13919373737905993,
+      "learning_rate": 2.501573626336352e-05,
+      "loss": 0.3204,
+      "step": 363
+    },
+    {
+      "epoch": 3.3009049773755654,
+      "grad_norm": 0.1394478161996878,
+      "learning_rate": 2.478065910140308e-05,
+      "loss": 0.319,
+      "step": 364
+    },
+    {
+      "epoch": 3.3099547511312215,
+      "grad_norm": 0.1422678230130045,
+      "learning_rate": 2.4546194972274852e-05,
+      "loss": 0.3234,
+      "step": 365
+    },
+    {
+      "epoch": 3.3190045248868776,
+      "grad_norm": 0.13767311772002824,
+      "learning_rate": 2.431235332015978e-05,
+      "loss": 0.3222,
+      "step": 366
+    },
+    {
+      "epoch": 3.328054298642534,
+      "grad_norm": 0.15146521934073406,
+      "learning_rate": 2.4079143564165524e-05,
+      "loss": 0.3196,
+      "step": 367
+    },
+    {
+      "epoch": 3.33710407239819,
+      "grad_norm": 0.14554574710226792,
+      "learning_rate": 2.3846575097947015e-05,
+      "loss": 0.3238,
+      "step": 368
+    },
+    {
+      "epoch": 3.3461538461538463,
+      "grad_norm": 0.12428809362324093,
+      "learning_rate": 2.361465728932819e-05,
+      "loss": 0.3281,
+      "step": 369
+    },
+    {
+      "epoch": 3.3552036199095023,
+      "grad_norm": 0.1359067551518604,
+      "learning_rate": 2.338339947992455e-05,
+      "loss": 0.3197,
+      "step": 370
+    },
+    {
+      "epoch": 3.3642533936651584,
+      "grad_norm": 0.12263338859613652,
+      "learning_rate": 2.315281098476694e-05,
+      "loss": 0.3217,
+      "step": 371
+    },
+    {
+      "epoch": 3.3733031674208145,
+      "grad_norm": 0.13047474507331128,
+      "learning_rate": 2.2922901091926334e-05,
+      "loss": 0.3244,
+      "step": 372
+    },
+    {
+      "epoch": 3.3823529411764706,
+      "grad_norm": 0.12872903204410682,
+      "learning_rate": 2.269367906213966e-05,
+      "loss": 0.3229,
+      "step": 373
+    },
+    {
+      "epoch": 3.3914027149321266,
+      "grad_norm": 0.11879478583728734,
+      "learning_rate": 2.24651541284369e-05,
+      "loss": 0.3246,
+      "step": 374
+    },
+    {
+      "epoch": 3.4004524886877827,
+      "grad_norm": 0.13996450916284633,
+      "learning_rate": 2.2237335495769035e-05,
+      "loss": 0.3255,
+      "step": 375
+    },
+    {
+      "epoch": 3.409502262443439,
+      "grad_norm": 0.12258279755344627,
+      "learning_rate": 2.2010232340637375e-05,
+      "loss": 0.3222,
+      "step": 376
+    },
+    {
+      "epoch": 3.418552036199095,
+      "grad_norm": 0.13396529243580008,
+      "learning_rate": 2.1783853810723895e-05,
+      "loss": 0.3188,
+      "step": 377
+    },
+    {
+      "epoch": 3.427601809954751,
+      "grad_norm": 0.12827246370155101,
+      "learning_rate": 2.155820902452276e-05,
+      "loss": 0.3233,
+      "step": 378
+    },
+    {
+      "epoch": 3.4366515837104075,
+      "grad_norm": 0.13361073513914423,
+      "learning_rate": 2.1333307070973054e-05,
+      "loss": 0.32,
+      "step": 379
+    },
+    {
+      "epoch": 3.4457013574660635,
+      "grad_norm": 0.1294537429532072,
+      "learning_rate": 2.11091570090927e-05,
+      "loss": 0.3234,
+      "step": 380
+    },
+    {
+      "epoch": 3.4547511312217196,
+      "grad_norm": 0.12477628801726541,
+      "learning_rate": 2.0885767867613485e-05,
+      "loss": 0.3236,
+      "step": 381
+    },
+    {
+      "epoch": 3.4638009049773757,
+      "grad_norm": 0.12556744064400813,
+      "learning_rate": 2.0663148644617443e-05,
+      "loss": 0.3193,
+      "step": 382
+    },
+    {
+      "epoch": 3.4728506787330318,
+      "grad_norm": 0.12838442646777212,
+      "learning_rate": 2.044130830717445e-05,
+      "loss": 0.3197,
+      "step": 383
+    },
+    {
+      "epoch": 3.481900452488688,
+      "grad_norm": 0.11666989370139778,
+      "learning_rate": 2.0220255790980935e-05,
+      "loss": 0.3222,
+      "step": 384
+    },
+    {
+      "epoch": 3.490950226244344,
+      "grad_norm": 0.1201654171036639,
+      "learning_rate": 2.0000000000000012e-05,
+      "loss": 0.323,
+      "step": 385
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 0.12406834396457092,
+      "learning_rate": 1.9780549806102827e-05,
+      "loss": 0.3233,
+      "step": 386
+    },
+    {
+      "epoch": 3.509049773755656,
+      "grad_norm": 0.13211561659340326,
+      "learning_rate": 1.9561914048711182e-05,
+      "loss": 0.3328,
+      "step": 387
+    },
+    {
+      "epoch": 3.518099547511312,
+      "grad_norm": 0.11794332998469956,
+      "learning_rate": 1.934410153444153e-05,
+      "loss": 0.3206,
+      "step": 388
+    },
+    {
+      "epoch": 3.5271493212669682,
+      "grad_norm": 0.11751517121074768,
+      "learning_rate": 1.912712103675017e-05,
+      "loss": 0.3234,
+      "step": 389
+    },
+    {
+      "epoch": 3.5361990950226243,
+      "grad_norm": 0.11384664165774953,
+      "learning_rate": 1.8910981295579903e-05,
+      "loss": 0.326,
+      "step": 390
+    },
+    {
+      "epoch": 3.5452488687782804,
+      "grad_norm": 0.11619434472524914,
+      "learning_rate": 1.8695691017008005e-05,
+      "loss": 0.328,
+      "step": 391
+    },
+    {
+      "epoch": 3.5542986425339365,
+      "grad_norm": 0.11461255200579494,
+      "learning_rate": 1.8481258872895503e-05,
+      "loss": 0.3249,
+      "step": 392
+    },
+    {
+      "epoch": 3.5633484162895925,
+      "grad_norm": 0.1269351876097148,
+      "learning_rate": 1.826769350053789e-05,
+      "loss": 0.325,
+      "step": 393
+    },
+    {
+      "epoch": 3.5723981900452486,
+      "grad_norm": 0.1262020845365235,
+      "learning_rate": 1.8055003502317213e-05,
+      "loss": 0.3245,
+      "step": 394
+    },
+    {
+      "epoch": 3.581447963800905,
+      "grad_norm": 0.1244584292444969,
+      "learning_rate": 1.7843197445355593e-05,
+      "loss": 0.3257,
+      "step": 395
+    },
+    {
+      "epoch": 3.590497737556561,
+      "grad_norm": 0.121141018365639,
+      "learning_rate": 1.7632283861170135e-05,
+      "loss": 0.3267,
+      "step": 396
+    },
+    {
+      "epoch": 3.5995475113122173,
+      "grad_norm": 0.1199129930925027,
+      "learning_rate": 1.7422271245329244e-05,
+      "loss": 0.3257,
+      "step": 397
+    },
+    {
+      "epoch": 3.6085972850678734,
+      "grad_norm": 0.11856999524364402,
+      "learning_rate": 1.721316805711045e-05,
+      "loss": 0.3211,
+      "step": 398
+    },
+    {
+      "epoch": 3.6176470588235294,
+      "grad_norm": 0.11909188120229579,
+      "learning_rate": 1.7004982719159712e-05,
+      "loss": 0.3201,
+      "step": 399
+    },
+    {
+      "epoch": 3.6266968325791855,
+      "grad_norm": 0.11144474240232284,
+      "learning_rate": 1.679772361715208e-05,
+      "loss": 0.3223,
+      "step": 400
+    },
+    {
+      "epoch": 3.6357466063348416,
+      "grad_norm": 0.12526137904233012,
+      "learning_rate": 1.6591399099453952e-05,
+      "loss": 0.3251,
+      "step": 401
+    },
+    {
+      "epoch": 3.6447963800904977,
+      "grad_norm": 0.1129996909861758,
+      "learning_rate": 1.6386017476786818e-05,
+      "loss": 0.3264,
+      "step": 402
+    },
+    {
+      "epoch": 3.6538461538461537,
+      "grad_norm": 0.11339940885061946,
+      "learning_rate": 1.6181587021892484e-05,
+      "loss": 0.3252,
+      "step": 403
+    },
+    {
+      "epoch": 3.66289592760181,
+      "grad_norm": 0.11557131743659978,
+      "learning_rate": 1.5978115969199882e-05,
+      "loss": 0.3241,
+      "step": 404
+    },
+    {
+      "epoch": 3.6719457013574663,
+      "grad_norm": 0.12094104342387753,
+      "learning_rate": 1.5775612514493343e-05,
+      "loss": 0.3224,
+      "step": 405
+    },
+    {
+      "epoch": 3.6809954751131224,
+      "grad_norm": 0.12996591315042658,
+      "learning_rate": 1.557408481458247e-05,
+      "loss": 0.3214,
+      "step": 406
+    },
+    {
+      "epoch": 3.6900452488687785,
+      "grad_norm": 0.10932053351798611,
+      "learning_rate": 1.537354098697367e-05,
+      "loss": 0.3243,
+      "step": 407
+    },
+    {
+      "epoch": 3.6990950226244346,
+      "grad_norm": 0.119643634861931,
+      "learning_rate": 1.5173989109543055e-05,
+      "loss": 0.3262,
+      "step": 408
+    },
+    {
+      "epoch": 3.7081447963800906,
+      "grad_norm": 0.11650012137085076,
+      "learning_rate": 1.497543722021114e-05,
+      "loss": 0.3248,
+      "step": 409
+    },
+    {
+      "epoch": 3.7171945701357467,
+      "grad_norm": 0.11291091689614972,
+      "learning_rate": 1.4777893316619114e-05,
+      "loss": 0.3203,
+      "step": 410
+    },
+    {
+      "epoch": 3.726244343891403,
+      "grad_norm": 0.1162086489522257,
+      "learning_rate": 1.4581365355806555e-05,
+      "loss": 0.3243,
+      "step": 411
+    },
+    {
+      "epoch": 3.735294117647059,
+      "grad_norm": 0.10927589657003309,
+      "learning_rate": 1.4385861253891111e-05,
+      "loss": 0.327,
+      "step": 412
+    },
+    {
+      "epoch": 3.744343891402715,
+      "grad_norm": 0.10801027792504748,
+      "learning_rate": 1.4191388885749487e-05,
+      "loss": 0.3237,
+      "step": 413
+    },
+    {
+      "epoch": 3.753393665158371,
+      "grad_norm": 0.11625407101384182,
+      "learning_rate": 1.3997956084700301e-05,
+      "loss": 0.3223,
+      "step": 414
+    },
+    {
+      "epoch": 3.762443438914027,
+      "grad_norm": 0.10784892069818107,
+      "learning_rate": 1.3805570642188602e-05,
+      "loss": 0.3258,
+      "step": 415
+    },
+    {
+      "epoch": 3.771493212669683,
+      "grad_norm": 0.11300531775753117,
+      "learning_rate": 1.3614240307471942e-05,
+      "loss": 0.3248,
+      "step": 416
+    },
+    {
+      "epoch": 3.7805429864253393,
+      "grad_norm": 0.10835333011179236,
+      "learning_rate": 1.34239727873083e-05,
+      "loss": 0.3223,
+      "step": 417
+    },
+    {
+      "epoch": 3.7895927601809953,
+      "grad_norm": 0.11188577965419376,
+      "learning_rate": 1.3234775745645684e-05,
+      "loss": 0.321,
+      "step": 418
+    },
+    {
+      "epoch": 3.7986425339366514,
+      "grad_norm": 0.12447909695863361,
+      "learning_rate": 1.3046656803313287e-05,
+      "loss": 0.324,
+      "step": 419
+    },
+    {
+      "epoch": 3.8076923076923075,
+      "grad_norm": 0.10209002045981093,
+      "learning_rate": 1.2859623537714719e-05,
+      "loss": 0.3227,
+      "step": 420
+    },
+    {
+      "epoch": 3.8167420814479636,
+      "grad_norm": 0.11319252778270614,
+      "learning_rate": 1.2673683482522629e-05,
+      "loss": 0.3289,
+      "step": 421
+    },
+    {
+      "epoch": 3.8257918552036196,
+      "grad_norm": 0.10456931028449813,
+      "learning_rate": 1.248884412737532e-05,
+      "loss": 0.3236,
+      "step": 422
+    },
+    {
+      "epoch": 3.834841628959276,
+      "grad_norm": 0.10976135305835821,
+      "learning_rate": 1.2305112917575092e-05,
+      "loss": 0.3273,
+      "step": 423
+    },
+    {
+      "epoch": 3.8438914027149322,
+      "grad_norm": 0.10710488304052018,
+      "learning_rate": 1.2122497253788267e-05,
+      "loss": 0.3224,
+      "step": 424
+    },
+    {
+      "epoch": 3.8529411764705883,
+      "grad_norm": 0.10332457574753726,
+      "learning_rate": 1.1941004491747145e-05,
+      "loss": 0.3174,
+      "step": 425
+    },
+    {
+      "epoch": 3.8619909502262444,
+      "grad_norm": 0.10300034100616205,
+      "learning_rate": 1.1760641941953744e-05,
+      "loss": 0.3252,
+      "step": 426
+    },
+    {
+      "epoch": 3.8710407239819005,
+      "grad_norm": 0.09955598289499694,
+      "learning_rate": 1.1581416869385253e-05,
+      "loss": 0.3223,
+      "step": 427
+    },
+    {
+      "epoch": 3.8800904977375565,
+      "grad_norm": 0.10841097410633975,
+      "learning_rate": 1.1403336493201462e-05,
+      "loss": 0.3215,
+      "step": 428
+    },
+    {
+      "epoch": 3.8891402714932126,
+      "grad_norm": 0.10174185620822972,
+      "learning_rate": 1.1226407986453963e-05,
+      "loss": 0.3238,
+      "step": 429
+    },
+    {
+      "epoch": 3.8981900452488687,
+      "grad_norm": 0.10861737022921748,
+      "learning_rate": 1.1050638475797193e-05,
+      "loss": 0.3256,
+      "step": 430
+    },
+    {
+      "epoch": 3.9072398190045248,
+      "grad_norm": 0.10297039570713754,
+      "learning_rate": 1.0876035041201436e-05,
+      "loss": 0.3221,
+      "step": 431
+    },
+    {
+      "epoch": 3.916289592760181,
+      "grad_norm": 0.1076446648237558,
+      "learning_rate": 1.070260471566757e-05,
+      "loss": 0.3252,
+      "step": 432
+    },
+    {
+      "epoch": 3.9253393665158374,
+      "grad_norm": 0.10925197562208173,
+      "learning_rate": 1.0530354484943798e-05,
+      "loss": 0.3185,
+      "step": 433
+    },
+    {
+      "epoch": 3.9343891402714934,
+      "grad_norm": 0.09823521379039317,
+      "learning_rate": 1.0359291287244334e-05,
+      "loss": 0.3241,
+      "step": 434
+    },
+    {
+      "epoch": 3.9434389140271495,
+      "grad_norm": 0.11042331281486452,
+      "learning_rate": 1.0189422012969814e-05,
+      "loss": 0.3264,
+      "step": 435
+    },
+    {
+      "epoch": 3.9524886877828056,
+      "grad_norm": 0.0998774378258076,
+      "learning_rate": 1.0020753504429845e-05,
+      "loss": 0.3184,
+      "step": 436
+    },
+    {
+      "epoch": 3.9615384615384617,
+      "grad_norm": 0.10587358010063602,
+      "learning_rate": 9.85329255556735e-06,
+      "loss": 0.3259,
+      "step": 437
+    },
+    {
+      "epoch": 3.9705882352941178,
+      "grad_norm": 0.10052895526823628,
+      "learning_rate": 9.687045911684928e-06,
+      "loss": 0.3239,
+      "step": 438
+    },
+    {
+      "epoch": 3.979638009049774,
+      "grad_norm": 0.09541866120111013,
+      "learning_rate": 9.522020269173172e-06,
+      "loss": 0.323,
+      "step": 439
+    },
+    {
+      "epoch": 3.98868778280543,
+      "grad_norm": 0.09557886438970163,
+      "learning_rate": 9.358222275240884e-06,
+      "loss": 0.324,
+      "step": 440
+    },
+    {
+      "epoch": 3.997737556561086,
+      "grad_norm": 0.10151624518760664,
+      "learning_rate": 9.195658527647371e-06,
+      "loss": 0.3264,
+      "step": 441
+    },
+    {
+      "epoch": 4.009049773755656,
+      "grad_norm": 0.16524757813062563,
+      "learning_rate": 9.034335574436701e-06,
+      "loss": 0.3062,
+      "step": 442
+    },
+    {
+      "epoch": 4.018099547511312,
+      "grad_norm": 0.12618495103542215,
+      "learning_rate": 8.8742599136739e-06,
+      "loss": 0.3037,
+      "step": 443
+    },
+    {
+      "epoch": 4.027149321266968,
+      "grad_norm": 0.12794305399302355,
+      "learning_rate": 8.715437993183235e-06,
+      "loss": 0.3071,
+      "step": 444
+    },
+    {
+      "epoch": 4.036199095022624,
+      "grad_norm": 0.1752311726452474,
+      "learning_rate": 8.55787621028851e-06,
+      "loss": 0.3049,
+      "step": 445
+    },
+    {
+      "epoch": 4.04524886877828,
+      "grad_norm": 0.13744941353125903,
+      "learning_rate": 8.401580911555353e-06,
+      "loss": 0.3022,
+      "step": 446
+    },
+    {
+      "epoch": 4.0542986425339365,
+      "grad_norm": 0.1442214999179444,
+      "learning_rate": 8.24655839253563e-06,
+      "loss": 0.304,
+      "step": 447
+    },
+    {
+      "epoch": 4.0633484162895925,
+      "grad_norm": 0.14721326705532867,
+      "learning_rate": 8.09281489751379e-06,
+      "loss": 0.3014,
+      "step": 448
+    },
+    {
+      "epoch": 4.072398190045249,
+      "grad_norm": 0.13164956472111614,
+      "learning_rate": 7.940356619255385e-06,
+      "loss": 0.3,
+      "step": 449
+    },
+    {
+      "epoch": 4.081447963800905,
+      "grad_norm": 0.14645232900110186,
+      "learning_rate": 7.789189698757656e-06,
+      "loss": 0.3051,
+      "step": 450
+    },
+    {
+      "epoch": 4.090497737556561,
+      "grad_norm": 0.13594041957864836,
+      "learning_rate": 7.639320225002106e-06,
+      "loss": 0.3008,
+      "step": 451
+    },
+    {
+      "epoch": 4.099547511312217,
+      "grad_norm": 0.12959173947863253,
+      "learning_rate": 7.490754234709285e-06,
+      "loss": 0.3029,
+      "step": 452
+    },
+    {
+      "epoch": 4.108597285067873,
+      "grad_norm": 0.1312102043451062,
+      "learning_rate": 7.343497712095619e-06,
+      "loss": 0.3018,
+      "step": 453
+    },
+    {
+      "epoch": 4.117647058823529,
+      "grad_norm": 0.12721628165894824,
+      "learning_rate": 7.1975565886323575e-06,
+      "loss": 0.3021,
+      "step": 454
+    },
+    {
+      "epoch": 4.126696832579185,
+      "grad_norm": 0.12135804195023794,
+      "learning_rate": 7.052936742806693e-06,
+      "loss": 0.3029,
+      "step": 455
+    },
+    {
+      "epoch": 4.135746606334842,
+      "grad_norm": 0.11021717009086096,
+      "learning_rate": 6.909643999884918e-06,
+      "loss": 0.3036,
+      "step": 456
+    },
+    {
+      "epoch": 4.144796380090498,
+      "grad_norm": 0.1171461454928136,
+      "learning_rate": 6.767684131677814e-06,
+      "loss": 0.3019,
+      "step": 457
+    },
+    {
+      "epoch": 4.153846153846154,
+      "grad_norm": 0.12398145068200955,
+      "learning_rate": 6.627062856308191e-06,
+      "loss": 0.3057,
+      "step": 458
+    },
+    {
+      "epoch": 4.16289592760181,
+      "grad_norm": 0.10813946756475247,
+      "learning_rate": 6.487785837980509e-06,
+      "loss": 0.3055,
+      "step": 459
+    },
+    {
+      "epoch": 4.171945701357466,
+      "grad_norm": 0.10512317676255085,
+      "learning_rate": 6.349858686752748e-06,
+      "loss": 0.3028,
+      "step": 460
+    },
+    {
+      "epoch": 4.180995475113122,
+      "grad_norm": 0.11457252381271353,
+      "learning_rate": 6.213286958310476e-06,
+      "loss": 0.3029,
+      "step": 461
+    },
+    {
+      "epoch": 4.1900452488687785,
+      "grad_norm": 0.11257398302051942,
+      "learning_rate": 6.078076153742962e-06,
+      "loss": 0.3042,
+      "step": 462
+    },
+    {
+      "epoch": 4.199095022624435,
+      "grad_norm": 0.1068866308205338,
+      "learning_rate": 5.9442317193217245e-06,
+      "loss": 0.3051,
+      "step": 463
+    },
+    {
+      "epoch": 4.208144796380091,
+      "grad_norm": 0.09665417595653125,
+      "learning_rate": 5.811759046281062e-06,
+      "loss": 0.3044,
+      "step": 464
+    },
+    {
+      "epoch": 4.217194570135747,
+      "grad_norm": 0.10132756213863214,
+      "learning_rate": 5.680663470600918e-06,
+      "loss": 0.3047,
+      "step": 465
+    },
+    {
+      "epoch": 4.226244343891403,
+      "grad_norm": 0.10197900247318212,
+      "learning_rate": 5.550950272791977e-06,
+      "loss": 0.3036,
+      "step": 466
+    },
+    {
+      "epoch": 4.235294117647059,
+      "grad_norm": 0.10011788827590404,
+      "learning_rate": 5.422624677682935e-06,
+      "loss": 0.3042,
+      "step": 467
+    },
+    {
+      "epoch": 4.244343891402715,
+      "grad_norm": 0.09704681985529298,
+      "learning_rate": 5.29569185421003e-06,
+      "loss": 0.2997,
+      "step": 468
+    },
+    {
+      "epoch": 4.253393665158371,
+      "grad_norm": 0.0889335395278227,
+      "learning_rate": 5.1701569152089196e-06,
+      "loss": 0.3033,
+      "step": 469
+    },
+    {
+      "epoch": 4.262443438914027,
+      "grad_norm": 0.09701179356263233,
+      "learning_rate": 5.046024917208603e-06,
+      "loss": 0.3074,
+      "step": 470
+    },
+    {
+      "epoch": 4.271493212669683,
+      "grad_norm": 0.09991090929407241,
+      "learning_rate": 4.923300860227884e-06,
+      "loss": 0.3074,
+      "step": 471
+    },
+    {
+      "epoch": 4.280542986425339,
+      "grad_norm": 0.09090875982688397,
+      "learning_rate": 4.801989687573878e-06,
+      "loss": 0.3049,
+      "step": 472
+    },
+    {
+      "epoch": 4.289592760180995,
+      "grad_norm": 0.08947832790676682,
+      "learning_rate": 4.6820962856429205e-06,
+      "loss": 0.3039,
+      "step": 473
+    },
+    {
+      "epoch": 4.298642533936651,
+      "grad_norm": 0.08534002305718665,
+      "learning_rate": 4.563625483723777e-06,
+      "loss": 0.3079,
+      "step": 474
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.08940524333139004,
+      "learning_rate": 4.446582053803066e-06,
+      "loss": 0.2974,
+      "step": 475
+    },
+    {
+      "epoch": 4.316742081447964,
+      "grad_norm": 0.09171172616230228,
+      "learning_rate": 4.330970710373063e-06,
+      "loss": 0.3076,
+      "step": 476
+    },
+    {
+      "epoch": 4.32579185520362,
+      "grad_norm": 0.08806621206621702,
+      "learning_rate": 4.216796110241852e-06,
+      "loss": 0.3048,
+      "step": 477
+    },
+    {
+      "epoch": 4.334841628959276,
+      "grad_norm": 0.08409826711122455,
+      "learning_rate": 4.104062852345671e-06,
+      "loss": 0.3013,
+      "step": 478
+    },
+    {
+      "epoch": 4.343891402714932,
+      "grad_norm": 0.0893222760374292,
+      "learning_rate": 3.9927754775636886e-06,
+      "loss": 0.3052,
+      "step": 479
+    },
+    {
+      "epoch": 4.352941176470588,
+      "grad_norm": 0.08399072448662723,
+      "learning_rate": 3.882938468535158e-06,
+      "loss": 0.3024,
+      "step": 480
+    },
+    {
+      "epoch": 4.361990950226244,
+      "grad_norm": 0.08499472493876709,
+      "learning_rate": 3.7745562494787645e-06,
+      "loss": 0.3022,
+      "step": 481
+    },
+    {
+      "epoch": 4.371040723981901,
+      "grad_norm": 0.08145638475145033,
+      "learning_rate": 3.667633186014512e-06,
+      "loss": 0.3054,
+      "step": 482
+    },
+    {
+      "epoch": 4.380090497737557,
+      "grad_norm": 0.08565742656060206,
+      "learning_rate": 3.5621735849877957e-06,
+      "loss": 0.3026,
+      "step": 483
+    },
+    {
+      "epoch": 4.389140271493213,
+      "grad_norm": 0.08492763740421684,
+      "learning_rate": 3.458181694295961e-06,
+      "loss": 0.3055,
+      "step": 484
+    },
+    {
+      "epoch": 4.398190045248869,
+      "grad_norm": 0.08463274662713963,
+      "learning_rate": 3.3556617027172168e-06,
+      "loss": 0.3024,
+      "step": 485
+    },
+    {
+      "epoch": 4.407239819004525,
+      "grad_norm": 0.08462399748218868,
+      "learning_rate": 3.2546177397418677e-06,
+      "loss": 0.3066,
+      "step": 486
+    },
+    {
+      "epoch": 4.416289592760181,
+      "grad_norm": 0.08376453881287438,
+      "learning_rate": 3.155053875406e-06,
+      "loss": 0.3042,
+      "step": 487
+    },
+    {
+      "epoch": 4.425339366515837,
+      "grad_norm": 0.08830270926768068,
+      "learning_rate": 3.0569741201275626e-06,
+      "loss": 0.2987,
+      "step": 488
+    },
+    {
+      "epoch": 4.4343891402714934,
+      "grad_norm": 0.08240870254024944,
+      "learning_rate": 2.96038242454479e-06,
+      "loss": 0.3003,
+      "step": 489
+    },
+    {
+      "epoch": 4.4434389140271495,
+      "grad_norm": 0.09052455529057644,
+      "learning_rate": 2.8652826793570975e-06,
+      "loss": 0.3028,
+      "step": 490
+    },
+    {
+      "epoch": 4.452488687782806,
+      "grad_norm": 0.08455563785960757,
+      "learning_rate": 2.77167871516836e-06,
+      "loss": 0.3073,
+      "step": 491
+    },
+    {
+      "epoch": 4.461538461538462,
+      "grad_norm": 0.08127731938352123,
+      "learning_rate": 2.679574302332597e-06,
+      "loss": 0.2999,
+      "step": 492
+    },
+    {
+      "epoch": 4.470588235294118,
+      "grad_norm": 0.08359245964188998,
+      "learning_rate": 2.5889731508021363e-06,
+      "loss": 0.3049,
+      "step": 493
+    },
+    {
+      "epoch": 4.479638009049774,
+      "grad_norm": 0.08004277394855593,
+      "learning_rate": 2.4998789099781507e-06,
+      "loss": 0.2995,
+      "step": 494
+    },
+    {
+      "epoch": 4.48868778280543,
+      "grad_norm": 0.09293016471118876,
+      "learning_rate": 2.4122951685636674e-06,
+      "loss": 0.3084,
+      "step": 495
+    },
+    {
+      "epoch": 4.497737556561086,
+      "grad_norm": 0.08857129082558779,
+      "learning_rate": 2.3262254544190154e-06,
+      "loss": 0.3028,
+      "step": 496
+    },
+    {
+      "epoch": 4.506787330316742,
+      "grad_norm": 0.0851690989816207,
+      "learning_rate": 2.2416732344197368e-06,
+      "loss": 0.3052,
+      "step": 497
+    },
+    {
+      "epoch": 4.515837104072398,
+      "grad_norm": 0.08380605677642579,
+      "learning_rate": 2.1586419143169125e-06,
+      "loss": 0.3071,
+      "step": 498
+    },
+    {
+      "epoch": 4.524886877828054,
+      "grad_norm": 0.07847574450950273,
+      "learning_rate": 2.0771348386000233e-06,
+      "loss": 0.3034,
+      "step": 499
+    },
+    {
+      "epoch": 4.53393665158371,
+      "grad_norm": 0.08166491411892216,
+      "learning_rate": 1.997155290362187e-06,
+      "loss": 0.3065,
+      "step": 500
+    },
+    {
+      "epoch": 4.542986425339366,
+      "grad_norm": 0.07778971565605453,
+      "learning_rate": 1.9187064911679432e-06,
+      "loss": 0.3024,
+      "step": 501
+    },
+    {
+      "epoch": 4.552036199095022,
+      "grad_norm": 0.07933626982513363,
+      "learning_rate": 1.841791600923495e-06,
+      "loss": 0.3023,
+      "step": 502
+    },
+    {
+      "epoch": 4.5610859728506785,
+      "grad_norm": 0.08136425063541126,
+      "learning_rate": 1.766413717749389e-06,
+      "loss": 0.304,
+      "step": 503
+    },
+    {
+      "epoch": 4.570135746606335,
+      "grad_norm": 0.08138859986883137,
+      "learning_rate": 1.6925758778557933e-06,
+      "loss": 0.3055,
+      "step": 504
+    },
+    {
+      "epoch": 4.579185520361991,
+      "grad_norm": 0.0787210528579053,
+      "learning_rate": 1.6202810554201099e-06,
+      "loss": 0.3049,
+      "step": 505
+    },
+    {
+      "epoch": 4.588235294117647,
+      "grad_norm": 0.08051236562441969,
+      "learning_rate": 1.5495321624672443e-06,
+      "loss": 0.3044,
+      "step": 506
+    },
+    {
+      "epoch": 4.597285067873303,
+      "grad_norm": 0.0788996281741959,
+      "learning_rate": 1.4803320487523087e-06,
+      "loss": 0.3057,
+      "step": 507
+    },
+    {
+      "epoch": 4.606334841628959,
+      "grad_norm": 0.07764118519403024,
+      "learning_rate": 1.4126835016457752e-06,
+      "loss": 0.3038,
+      "step": 508
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.08001544469825624,
+      "learning_rate": 1.3465892460212594e-06,
+      "loss": 0.3064,
+      "step": 509
+    },
+    {
+      "epoch": 4.624434389140271,
+      "grad_norm": 0.09063120326499433,
+      "learning_rate": 1.2820519441457502e-06,
+      "loss": 0.3047,
+      "step": 510
+    },
+    {
+      "epoch": 4.633484162895927,
+      "grad_norm": 0.07794623861050551,
+      "learning_rate": 1.2190741955723495e-06,
+      "loss": 0.3045,
+      "step": 511
+    },
+    {
+      "epoch": 4.642533936651584,
+      "grad_norm": 0.07595212127483163,
+      "learning_rate": 1.1576585370355908e-06,
+      "loss": 0.3065,
+      "step": 512
+    },
+    {
+      "epoch": 4.65158371040724,
+      "grad_norm": 0.07739705488047151,
+      "learning_rate": 1.0978074423492368e-06,
+      "loss": 0.3014,
+      "step": 513
+    },
+    {
+      "epoch": 4.660633484162896,
+      "grad_norm": 0.07685164772052698,
+      "learning_rate": 1.0395233223066614e-06,
+      "loss": 0.2996,
+      "step": 514
+    },
+    {
+      "epoch": 4.669683257918552,
+      "grad_norm": 0.08015550295148605,
+      "learning_rate": 9.828085245837183e-07,
+      "loss": 0.307,
+      "step": 515
+    },
+    {
+      "epoch": 4.678733031674208,
+      "grad_norm": 0.08113161848829428,
+      "learning_rate": 9.276653336441943e-07,
+      "loss": 0.3091,
+      "step": 516
+    },
+    {
+      "epoch": 4.6877828054298645,
+      "grad_norm": 0.08234301018528185,
+      "learning_rate": 8.740959706477725e-07,
+      "loss": 0.3022,
+      "step": 517
+    },
+    {
+      "epoch": 4.6968325791855206,
+      "grad_norm": 0.07811510319887344,
+      "learning_rate": 8.221025933605919e-07,
+      "loss": 0.3027,
+      "step": 518
+    },
+    {
+      "epoch": 4.705882352941177,
+      "grad_norm": 0.07966253839252949,
+      "learning_rate": 7.716872960683041e-07,
+      "loss": 0.304,
+      "step": 519
+    },
+    {
+      "epoch": 4.714932126696833,
+      "grad_norm": 0.07556300199073022,
+      "learning_rate": 7.228521094917318e-07,
+      "loss": 0.303,
+      "step": 520
+    },
+    {
+      "epoch": 4.723981900452489,
+      "grad_norm": 0.07949886614317854,
+      "learning_rate": 6.755990007050761e-07,
+      "loss": 0.3061,
+      "step": 521
+    },
+    {
+      "epoch": 4.733031674208145,
+      "grad_norm": 0.07841434742669152,
+      "learning_rate": 6.299298730566516e-07,
+      "loss": 0.3068,
+      "step": 522
+    },
+    {
+      "epoch": 4.742081447963801,
+      "grad_norm": 0.07541915667043782,
+      "learning_rate": 5.858465660922808e-07,
+      "loss": 0.3066,
+      "step": 523
+    },
+    {
+      "epoch": 4.751131221719457,
+      "grad_norm": 0.07733742513960733,
+      "learning_rate": 5.433508554811307e-07,
+      "loss": 0.3057,
+      "step": 524
+    },
+    {
+      "epoch": 4.760180995475113,
+      "grad_norm": 0.07836214413259006,
+      "learning_rate": 5.024444529442285e-07,
+      "loss": 0.3031,
+      "step": 525
+    },
+    {
+      "epoch": 4.769230769230769,
+      "grad_norm": 0.07953356517050592,
+      "learning_rate": 4.6312900618550317e-07,
+      "loss": 0.3016,
+      "step": 526
+    },
+    {
+      "epoch": 4.778280542986425,
+      "grad_norm": 0.07617521054682652,
+      "learning_rate": 4.254060988254205e-07,
+      "loss": 0.2999,
+      "step": 527
+    },
+    {
+      "epoch": 4.787330316742081,
+      "grad_norm": 0.07846308080577853,
+      "learning_rate": 3.8927725033718553e-07,
+      "loss": 0.3051,
+      "step": 528
+    },
+    {
+      "epoch": 4.796380090497737,
+      "grad_norm": 0.07896885874732473,
+      "learning_rate": 3.547439159855559e-07,
+      "loss": 0.3063,
+      "step": 529
+    },
+    {
+      "epoch": 4.8054298642533935,
+      "grad_norm": 0.07878929850813644,
+      "learning_rate": 3.218074867681864e-07,
+      "loss": 0.302,
+      "step": 530
+    },
+    {
+      "epoch": 4.8144796380090495,
+      "grad_norm": 0.07613359977307133,
+      "learning_rate": 2.9046928935966056e-07,
+      "loss": 0.2972,
+      "step": 531
+    },
+    {
+      "epoch": 4.823529411764706,
+      "grad_norm": 0.07688492198921053,
+      "learning_rate": 2.607305860579912e-07,
+      "loss": 0.3067,
+      "step": 532
+    },
+    {
+      "epoch": 4.832579185520362,
+      "grad_norm": 0.07476518062895497,
+      "learning_rate": 2.3259257473381647e-07,
+      "loss": 0.2991,
+      "step": 533
+    },
+    {
+      "epoch": 4.841628959276018,
+      "grad_norm": 0.07552831124187401,
+      "learning_rate": 2.060563887821232e-07,
+      "loss": 0.3005,
+      "step": 534
+    },
+    {
+      "epoch": 4.850678733031674,
+      "grad_norm": 0.07631228481012646,
+      "learning_rate": 1.8112309707661647e-07,
+      "loss": 0.3035,
+      "step": 535
+    },
+    {
+      "epoch": 4.859728506787331,
+      "grad_norm": 0.07467074977475399,
+      "learning_rate": 1.577937039266475e-07,
+      "loss": 0.3056,
+      "step": 536
+    },
+    {
+      "epoch": 4.868778280542987,
+      "grad_norm": 0.07431003159382249,
+      "learning_rate": 1.3606914903677492e-07,
+      "loss": 0.2996,
+      "step": 537
+    },
+    {
+      "epoch": 4.877828054298643,
+      "grad_norm": 0.07468768960052453,
+      "learning_rate": 1.1595030746890168e-07,
+      "loss": 0.3077,
+      "step": 538
+    },
+    {
+      "epoch": 4.886877828054299,
+      "grad_norm": 0.07691379852821048,
+      "learning_rate": 9.74379896070321e-08,
+      "loss": 0.303,
+      "step": 539
+    },
+    {
+      "epoch": 4.895927601809955,
+      "grad_norm": 0.07405617978413666,
+      "learning_rate": 8.053294112462696e-08,
+      "loss": 0.3029,
+      "step": 540
+    },
+    {
+      "epoch": 4.904977375565611,
+      "grad_norm": 0.07237727103301038,
+      "learning_rate": 6.523584295457408e-08,
+      "loss": 0.3045,
+      "step": 541
+    },
+    {
+      "epoch": 4.914027149321267,
+      "grad_norm": 0.07736995275320246,
+      "learning_rate": 5.154731126176149e-08,
+      "loss": 0.3072,
+      "step": 542
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.07450947266190626,
+      "learning_rate": 3.946789741824386e-08,
+      "loss": 0.2999,
+      "step": 543
+    },
+    {
+      "epoch": 4.932126696832579,
+      "grad_norm": 0.07656920155960474,
+      "learning_rate": 2.8998087981055854e-08,
+      "loss": 0.3001,
+      "step": 544
+    },
+    {
+      "epoch": 4.9411764705882355,
+      "grad_norm": 0.07427882247951784,
+      "learning_rate": 2.01383046725967e-08,
+      "loss": 0.3011,
+      "step": 545
+    },
+    {
+      "epoch": 4.950226244343892,
+      "grad_norm": 0.07493198557289557,
+      "learning_rate": 1.288890436363488e-08,
+      "loss": 0.3073,
+      "step": 546
+    },
+    {
+      "epoch": 4.959276018099548,
+      "grad_norm": 0.07595703866942234,
+      "learning_rate": 7.250179058968521e-09,
+      "loss": 0.302,
+      "step": 547
+    },
+    {
+      "epoch": 4.968325791855204,
+      "grad_norm": 0.07667669947739308,
+      "learning_rate": 3.222355885625916e-09,
+      "loss": 0.3058,
+      "step": 548
+    },
+    {
+      "epoch": 4.97737556561086,
+      "grad_norm": 0.0778877092524117,
+      "learning_rate": 8.055970837395066e-10,
+      "loss": 0.3071,
+      "step": 549
+    },
+    {
+      "epoch": 4.986425339366516,
+      "grad_norm": 0.07598122336675614,
+      "learning_rate": 0.0,
+      "loss": 0.3021,
+      "step": 550
+    },
+    {
+      "epoch": 4.986425339366516,
+      "step": 550,
+      "total_flos": 1.3210208775492862e+19,
+      "train_loss": 0.37584147680889474,
+      "train_runtime": 27424.4359,
+      "train_samples_per_second": 10.307,
+      "train_steps_per_second": 0.02
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 550,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.3210208775492862e+19,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}