diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4382 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 620,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0064516129032258064,
+      "grad_norm": 2.3284332752227783,
+      "learning_rate": 4.032258064516129e-07,
+      "loss": 0.1831,
+      "step": 1
+    },
+    {
+      "epoch": 0.012903225806451613,
+      "grad_norm": 3.1032278537750244,
+      "learning_rate": 8.064516129032258e-07,
+      "loss": 0.2496,
+      "step": 2
+    },
+    {
+      "epoch": 0.01935483870967742,
+      "grad_norm": 2.7308666706085205,
+      "learning_rate": 1.2096774193548388e-06,
+      "loss": 0.2497,
+      "step": 3
+    },
+    {
+      "epoch": 0.025806451612903226,
+      "grad_norm": 2.6942598819732666,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 0.2513,
+      "step": 4
+    },
+    {
+      "epoch": 0.03225806451612903,
+      "grad_norm": 2.281903028488159,
+      "learning_rate": 2.0161290322580646e-06,
+      "loss": 0.2021,
+      "step": 5
+    },
+    {
+      "epoch": 0.03870967741935484,
+      "grad_norm": 2.2116780281066895,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.2472,
+      "step": 6
+    },
+    {
+      "epoch": 0.04516129032258064,
+      "grad_norm": 2.3709909915924072,
+      "learning_rate": 2.82258064516129e-06,
+      "loss": 0.2133,
+      "step": 7
+    },
+    {
+      "epoch": 0.05161290322580645,
+      "grad_norm": 1.6222234964370728,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.2072,
+      "step": 8
+    },
+    {
+      "epoch": 0.05806451612903226,
+      "grad_norm": 1.6226286888122559,
+      "learning_rate": 3.6290322580645166e-06,
+      "loss": 0.1732,
+      "step": 9
+    },
+    {
+      "epoch": 0.06451612903225806,
+      "grad_norm": 1.514697551727295,
+      "learning_rate": 4.032258064516129e-06,
+      "loss": 0.2088,
+      "step": 10
+    },
+    {
+      "epoch": 0.07096774193548387,
+      "grad_norm": 1.6407102346420288,
+      "learning_rate": 4.435483870967742e-06,
+      "loss": 0.1733,
+      "step": 11
+    },
+    {
+      "epoch": 0.07741935483870968,
+      "grad_norm": 1.5498034954071045,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.2087,
+      "step": 12
+    },
+    {
+      "epoch": 0.08387096774193549,
+      "grad_norm": 1.9465513229370117,
+      "learning_rate": 5.241935483870968e-06,
+      "loss": 0.2046,
+      "step": 13
+    },
+    {
+      "epoch": 0.09032258064516129,
+      "grad_norm": 1.3483728170394897,
+      "learning_rate": 5.64516129032258e-06,
+      "loss": 0.1783,
+      "step": 14
+    },
+    {
+      "epoch": 0.0967741935483871,
+      "grad_norm": 1.4068559408187866,
+      "learning_rate": 6.048387096774194e-06,
+      "loss": 0.1519,
+      "step": 15
+    },
+    {
+      "epoch": 0.1032258064516129,
+      "grad_norm": 1.3083986043930054,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.1586,
+      "step": 16
+    },
+    {
+      "epoch": 0.10967741935483871,
+      "grad_norm": 1.6559300422668457,
+      "learning_rate": 6.854838709677419e-06,
+      "loss": 0.1476,
+      "step": 17
+    },
+    {
+      "epoch": 0.11612903225806452,
+      "grad_norm": 1.0691255331039429,
+      "learning_rate": 7.258064516129033e-06,
+      "loss": 0.1478,
+      "step": 18
+    },
+    {
+      "epoch": 0.12258064516129032,
+      "grad_norm": 1.1155110597610474,
+      "learning_rate": 7.661290322580646e-06,
+      "loss": 0.1382,
+      "step": 19
+    },
+    {
+      "epoch": 0.12903225806451613,
+      "grad_norm": 0.9816218018531799,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.1204,
+      "step": 20
+    },
+    {
+      "epoch": 0.13548387096774195,
+      "grad_norm": 1.2463096380233765,
+      "learning_rate": 8.46774193548387e-06,
+      "loss": 0.1503,
+      "step": 21
+    },
+    {
+      "epoch": 0.14193548387096774,
+      "grad_norm": 1.3447906970977783,
+      "learning_rate": 8.870967741935484e-06,
+      "loss": 0.1224,
+      "step": 22
+    },
+    {
+      "epoch": 0.14838709677419354,
+      "grad_norm": 1.1465381383895874,
+      "learning_rate": 9.274193548387097e-06,
+      "loss": 0.1374,
+      "step": 23
+    },
+    {
+      "epoch": 0.15483870967741936,
+      "grad_norm": 1.0978549718856812,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.1303,
+      "step": 24
+    },
+    {
+      "epoch": 0.16129032258064516,
+      "grad_norm": 1.1053048372268677,
+      "learning_rate": 1.0080645161290323e-05,
+      "loss": 0.1423,
+      "step": 25
+    },
+    {
+      "epoch": 0.16774193548387098,
+      "grad_norm": 1.0212026834487915,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.1281,
+      "step": 26
+    },
+    {
+      "epoch": 0.17419354838709677,
+      "grad_norm": 0.9742250442504883,
+      "learning_rate": 1.0887096774193549e-05,
+      "loss": 0.1228,
+      "step": 27
+    },
+    {
+      "epoch": 0.18064516129032257,
+      "grad_norm": 1.1676782369613647,
+      "learning_rate": 1.129032258064516e-05,
+      "loss": 0.1429,
+      "step": 28
+    },
+    {
+      "epoch": 0.1870967741935484,
+      "grad_norm": 1.088600516319275,
+      "learning_rate": 1.1693548387096775e-05,
+      "loss": 0.1069,
+      "step": 29
+    },
+    {
+      "epoch": 0.1935483870967742,
+      "grad_norm": 1.3947196006774902,
+      "learning_rate": 1.2096774193548388e-05,
+      "loss": 0.1316,
+      "step": 30
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.9059141874313354,
+      "learning_rate": 1.25e-05,
+      "loss": 0.1121,
+      "step": 31
+    },
+    {
+      "epoch": 0.2064516129032258,
+      "grad_norm": 1.0918734073638916,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.1521,
+      "step": 32
+    },
+    {
+      "epoch": 0.2129032258064516,
+      "grad_norm": 1.1550475358963013,
+      "learning_rate": 1.3306451612903225e-05,
+      "loss": 0.1251,
+      "step": 33
+    },
+    {
+      "epoch": 0.21935483870967742,
+      "grad_norm": 1.2110551595687866,
+      "learning_rate": 1.3709677419354839e-05,
+      "loss": 0.1325,
+      "step": 34
+    },
+    {
+      "epoch": 0.22580645161290322,
+      "grad_norm": 1.0340098142623901,
+      "learning_rate": 1.4112903225806454e-05,
+      "loss": 0.1115,
+      "step": 35
+    },
+    {
+      "epoch": 0.23225806451612904,
+      "grad_norm": 0.9789180159568787,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.1154,
+      "step": 36
+    },
+    {
+      "epoch": 0.23870967741935484,
+      "grad_norm": 0.8362810015678406,
+      "learning_rate": 1.4919354838709679e-05,
+      "loss": 0.1041,
+      "step": 37
+    },
+    {
+      "epoch": 0.24516129032258063,
+      "grad_norm": 1.0805575847625732,
+      "learning_rate": 1.5322580645161292e-05,
+      "loss": 0.1201,
+      "step": 38
+    },
+    {
+      "epoch": 0.25161290322580643,
+      "grad_norm": 1.0794912576675415,
+      "learning_rate": 1.5725806451612903e-05,
+      "loss": 0.1387,
+      "step": 39
+    },
+    {
+      "epoch": 0.25806451612903225,
+      "grad_norm": 1.0303066968917847,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.1381,
+      "step": 40
+    },
+    {
+      "epoch": 0.2645161290322581,
+      "grad_norm": 0.8959848284721375,
+      "learning_rate": 1.653225806451613e-05,
+      "loss": 0.1251,
+      "step": 41
+    },
+    {
+      "epoch": 0.2709677419354839,
+      "grad_norm": 1.0856695175170898,
+      "learning_rate": 1.693548387096774e-05,
+      "loss": 0.1363,
+      "step": 42
+    },
+    {
+      "epoch": 0.27741935483870966,
+      "grad_norm": 0.8375802636146545,
+      "learning_rate": 1.733870967741936e-05,
+      "loss": 0.1009,
+      "step": 43
+    },
+    {
+      "epoch": 0.2838709677419355,
+      "grad_norm": 0.9029824733734131,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.1006,
+      "step": 44
+    },
+    {
+      "epoch": 0.2903225806451613,
+      "grad_norm": 0.8736345767974854,
+      "learning_rate": 1.8145161290322583e-05,
+      "loss": 0.1212,
+      "step": 45
+    },
+    {
+      "epoch": 0.2967741935483871,
+      "grad_norm": 1.165887713432312,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.1241,
+      "step": 46
+    },
+    {
+      "epoch": 0.3032258064516129,
+      "grad_norm": 0.8511247634887695,
+      "learning_rate": 1.8951612903225807e-05,
+      "loss": 0.0942,
+      "step": 47
+    },
+    {
+      "epoch": 0.3096774193548387,
+      "grad_norm": 1.0182602405548096,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.116,
+      "step": 48
+    },
+    {
+      "epoch": 0.3161290322580645,
+      "grad_norm": 0.8452662825584412,
+      "learning_rate": 1.975806451612903e-05,
+      "loss": 0.1023,
+      "step": 49
+    },
+    {
+      "epoch": 0.3225806451612903,
+      "grad_norm": 0.9583229422569275,
+      "learning_rate": 2.0161290322580645e-05,
+      "loss": 0.0956,
+      "step": 50
+    },
+    {
+      "epoch": 0.32903225806451614,
+      "grad_norm": 0.935484766960144,
+      "learning_rate": 2.056451612903226e-05,
+      "loss": 0.1185,
+      "step": 51
+    },
+    {
+      "epoch": 0.33548387096774196,
+      "grad_norm": 0.9844627380371094,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.0998,
+      "step": 52
+    },
+    {
+      "epoch": 0.3419354838709677,
+      "grad_norm": 1.0139315128326416,
+      "learning_rate": 2.1370967741935487e-05,
+      "loss": 0.0901,
+      "step": 53
+    },
+    {
+      "epoch": 0.34838709677419355,
+      "grad_norm": 0.844688892364502,
+      "learning_rate": 2.1774193548387097e-05,
+      "loss": 0.1158,
+      "step": 54
+    },
+    {
+      "epoch": 0.3548387096774194,
+      "grad_norm": 0.778408408164978,
+      "learning_rate": 2.217741935483871e-05,
+      "loss": 0.0884,
+      "step": 55
+    },
+    {
+      "epoch": 0.36129032258064514,
+      "grad_norm": 0.7307286858558655,
+      "learning_rate": 2.258064516129032e-05,
+      "loss": 0.1099,
+      "step": 56
+    },
+    {
+      "epoch": 0.36774193548387096,
+      "grad_norm": 0.681089460849762,
+      "learning_rate": 2.2983870967741935e-05,
+      "loss": 0.0965,
+      "step": 57
+    },
+    {
+      "epoch": 0.3741935483870968,
+      "grad_norm": 0.7206712365150452,
+      "learning_rate": 2.338709677419355e-05,
+      "loss": 0.0898,
+      "step": 58
+    },
+    {
+      "epoch": 0.38064516129032255,
+      "grad_norm": 0.6326794624328613,
+      "learning_rate": 2.3790322580645163e-05,
+      "loss": 0.0896,
+      "step": 59
+    },
+    {
+      "epoch": 0.3870967741935484,
+      "grad_norm": 0.6684013605117798,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.0859,
+      "step": 60
+    },
+    {
+      "epoch": 0.3935483870967742,
+      "grad_norm": 0.7839128971099854,
+      "learning_rate": 2.4596774193548387e-05,
+      "loss": 0.0918,
+      "step": 61
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7025837302207947,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0933,
+      "step": 62
+    },
+    {
+      "epoch": 0.4064516129032258,
+      "grad_norm": 0.7583072185516357,
+      "learning_rate": 2.4999801888257584e-05,
+      "loss": 0.0916,
+      "step": 63
+    },
+    {
+      "epoch": 0.4129032258064516,
+      "grad_norm": 0.8116795420646667,
+      "learning_rate": 2.499920755931005e-05,
+      "loss": 0.0934,
+      "step": 64
+    },
+    {
+      "epoch": 0.41935483870967744,
+      "grad_norm": 0.9053534865379333,
+      "learning_rate": 2.4998217031996375e-05,
+      "loss": 0.1116,
+      "step": 65
+    },
+    {
+      "epoch": 0.4258064516129032,
+      "grad_norm": 0.773985743522644,
+      "learning_rate": 2.4996830337714163e-05,
+      "loss": 0.0874,
+      "step": 66
+    },
+    {
+      "epoch": 0.432258064516129,
+      "grad_norm": 0.8468173146247864,
+      "learning_rate": 2.4995047520418692e-05,
+      "loss": 0.0954,
+      "step": 67
+    },
+    {
+      "epoch": 0.43870967741935485,
+      "grad_norm": 0.7126619815826416,
+      "learning_rate": 2.4992868636621474e-05,
+      "loss": 0.1017,
+      "step": 68
+    },
+    {
+      "epoch": 0.44516129032258067,
+      "grad_norm": 0.7975043654441833,
+      "learning_rate": 2.4990293755388524e-05,
+      "loss": 0.1086,
+      "step": 69
+    },
+    {
+      "epoch": 0.45161290322580644,
+      "grad_norm": 0.8055579662322998,
+      "learning_rate": 2.4987322958338095e-05,
+      "loss": 0.0836,
+      "step": 70
+    },
+    {
+      "epoch": 0.45806451612903226,
+      "grad_norm": 0.6494209170341492,
+      "learning_rate": 2.4983956339638158e-05,
+      "loss": 0.0883,
+      "step": 71
+    },
+    {
+      "epoch": 0.4645161290322581,
+      "grad_norm": 0.6997829675674438,
+      "learning_rate": 2.4980194006003392e-05,
+      "loss": 0.0763,
+      "step": 72
+    },
+    {
+      "epoch": 0.47096774193548385,
+      "grad_norm": 0.596174418926239,
+      "learning_rate": 2.4976036076691787e-05,
+      "loss": 0.0871,
+      "step": 73
+    },
+    {
+      "epoch": 0.4774193548387097,
+      "grad_norm": 0.6535652279853821,
+      "learning_rate": 2.4971482683500884e-05,
+      "loss": 0.0869,
+      "step": 74
+    },
+    {
+      "epoch": 0.4838709677419355,
+      "grad_norm": 0.8003737926483154,
+      "learning_rate": 2.4966533970763586e-05,
+      "loss": 0.1086,
+      "step": 75
+    },
+    {
+      "epoch": 0.49032258064516127,
+      "grad_norm": 0.6992926001548767,
+      "learning_rate": 2.496119009534359e-05,
+      "loss": 0.0822,
+      "step": 76
+    },
+    {
+      "epoch": 0.4967741935483871,
+      "grad_norm": 0.6500689387321472,
+      "learning_rate": 2.4955451226630412e-05,
+      "loss": 0.0876,
+      "step": 77
+    },
+    {
+      "epoch": 0.5032258064516129,
+      "grad_norm": 0.7626132369041443,
+      "learning_rate": 2.4949317546534018e-05,
+      "loss": 0.0911,
+      "step": 78
+    },
+    {
+      "epoch": 0.5096774193548387,
+      "grad_norm": 0.6485949158668518,
+      "learning_rate": 2.4942789249479054e-05,
+      "loss": 0.0914,
+      "step": 79
+    },
+    {
+      "epoch": 0.5161290322580645,
+      "grad_norm": 0.692364513874054,
+      "learning_rate": 2.493586654239869e-05,
+      "loss": 0.1074,
+      "step": 80
+    },
+    {
+      "epoch": 0.5225806451612903,
+      "grad_norm": 0.7383131980895996,
+      "learning_rate": 2.4928549644728057e-05,
+      "loss": 0.085,
+      "step": 81
+    },
+    {
+      "epoch": 0.5290322580645161,
+      "grad_norm": 0.6585950255393982,
+      "learning_rate": 2.492083878839729e-05,
+      "loss": 0.0795,
+      "step": 82
+    },
+    {
+      "epoch": 0.535483870967742,
+      "grad_norm": 0.7683681845664978,
+      "learning_rate": 2.491273421782417e-05,
+      "loss": 0.073,
+      "step": 83
+    },
+    {
+      "epoch": 0.5419354838709678,
+      "grad_norm": 0.5386450290679932,
+      "learning_rate": 2.4904236189906406e-05,
+      "loss": 0.0814,
+      "step": 84
+    },
+    {
+      "epoch": 0.5483870967741935,
+      "grad_norm": 0.725712239742279,
+      "learning_rate": 2.489534497401345e-05,
+      "loss": 0.0896,
+      "step": 85
+    },
+    {
+      "epoch": 0.5548387096774193,
+      "grad_norm": 0.8596577644348145,
+      "learning_rate": 2.488606085197799e-05,
+      "loss": 0.0816,
+      "step": 86
+    },
+    {
+      "epoch": 0.5612903225806452,
+      "grad_norm": 0.7653164863586426,
+      "learning_rate": 2.4876384118086992e-05,
+      "loss": 0.1078,
+      "step": 87
+    },
+    {
+      "epoch": 0.567741935483871,
+      "grad_norm": 0.713628351688385,
+      "learning_rate": 2.48663150790724e-05,
+      "loss": 0.0887,
+      "step": 88
+    },
+    {
+      "epoch": 0.5741935483870968,
+      "grad_norm": 0.5724640488624573,
+      "learning_rate": 2.4855854054101395e-05,
+      "loss": 0.0849,
+      "step": 89
+    },
+    {
+      "epoch": 0.5806451612903226,
+      "grad_norm": 0.6235289573669434,
+      "learning_rate": 2.484500137476627e-05,
+      "loss": 0.0875,
+      "step": 90
+    },
+    {
+      "epoch": 0.5870967741935483,
+      "grad_norm": 0.785372793674469,
+      "learning_rate": 2.483375738507395e-05,
+      "loss": 0.1225,
+      "step": 91
+    },
+    {
+      "epoch": 0.5935483870967742,
+      "grad_norm": 0.6431748867034912,
+      "learning_rate": 2.4822122441435047e-05,
+      "loss": 0.0913,
+      "step": 92
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.8031719923019409,
+      "learning_rate": 2.4810096912652604e-05,
+      "loss": 0.102,
+      "step": 93
+    },
+    {
+      "epoch": 0.6064516129032258,
+      "grad_norm": 0.5750744938850403,
+      "learning_rate": 2.4797681179910363e-05,
+      "loss": 0.0754,
+      "step": 94
+    },
+    {
+      "epoch": 0.6129032258064516,
+      "grad_norm": 0.7892565727233887,
+      "learning_rate": 2.4784875636760727e-05,
+      "loss": 0.0825,
+      "step": 95
+    },
+    {
+      "epoch": 0.6193548387096774,
+      "grad_norm": 0.7932739853858948,
+      "learning_rate": 2.4771680689112244e-05,
+      "loss": 0.1262,
+      "step": 96
+    },
+    {
+      "epoch": 0.6258064516129033,
+      "grad_norm": 0.7647889852523804,
+      "learning_rate": 2.4758096755216763e-05,
+      "loss": 0.1083,
+      "step": 97
+    },
+    {
+      "epoch": 0.632258064516129,
+      "grad_norm": 0.9550963640213013,
+      "learning_rate": 2.474412426565618e-05,
+      "loss": 0.0828,
+      "step": 98
+    },
+    {
+      "epoch": 0.6387096774193548,
+      "grad_norm": 0.6981013417243958,
+      "learning_rate": 2.4729763663328774e-05,
+      "loss": 0.0943,
+      "step": 99
+    },
+    {
+      "epoch": 0.6451612903225806,
+      "grad_norm": 0.8088532090187073,
+      "learning_rate": 2.4715015403435176e-05,
+      "loss": 0.0954,
+      "step": 100
+    },
+    {
+      "epoch": 0.6516129032258065,
+      "grad_norm": 0.9130911231040955,
+      "learning_rate": 2.4699879953463945e-05,
+      "loss": 0.0973,
+      "step": 101
+    },
+    {
+      "epoch": 0.6580645161290323,
+      "grad_norm": 0.791867196559906,
+      "learning_rate": 2.468435779317673e-05,
+      "loss": 0.0946,
+      "step": 102
+    },
+    {
+      "epoch": 0.6645161290322581,
+      "grad_norm": 0.6049063205718994,
+      "learning_rate": 2.466844941459309e-05,
+      "loss": 0.0797,
+      "step": 103
+    },
+    {
+      "epoch": 0.6709677419354839,
+      "grad_norm": 0.6488558053970337,
+      "learning_rate": 2.4652155321974883e-05,
+      "loss": 0.1004,
+      "step": 104
+    },
+    {
+      "epoch": 0.6774193548387096,
+      "grad_norm": 0.7218672633171082,
+      "learning_rate": 2.4635476031810284e-05,
+      "loss": 0.0943,
+      "step": 105
+    },
+    {
+      "epoch": 0.6838709677419355,
+      "grad_norm": 0.7997153997421265,
+      "learning_rate": 2.4618412072797407e-05,
+      "loss": 0.0831,
+      "step": 106
+    },
+    {
+      "epoch": 0.6903225806451613,
+      "grad_norm": 0.8165119886398315,
+      "learning_rate": 2.4600963985827555e-05,
+      "loss": 0.0919,
+      "step": 107
+    },
+    {
+      "epoch": 0.6967741935483871,
+      "grad_norm": 0.704238772392273,
+      "learning_rate": 2.458313232396808e-05,
+      "loss": 0.0778,
+      "step": 108
+    },
+    {
+      "epoch": 0.7032258064516129,
+      "grad_norm": 0.6857476234436035,
+      "learning_rate": 2.456491765244483e-05,
+      "loss": 0.0914,
+      "step": 109
+    },
+    {
+      "epoch": 0.7096774193548387,
+      "grad_norm": 0.7254015803337097,
+      "learning_rate": 2.4546320548624264e-05,
+      "loss": 0.1102,
+      "step": 110
+    },
+    {
+      "epoch": 0.7161290322580646,
+      "grad_norm": 0.6534197330474854,
+      "learning_rate": 2.4527341601995115e-05,
+      "loss": 0.0841,
+      "step": 111
+    },
+    {
+      "epoch": 0.7225806451612903,
+      "grad_norm": 0.6944810152053833,
+      "learning_rate": 2.450798141414974e-05,
+      "loss": 0.1067,
+      "step": 112
+    },
+    {
+      "epoch": 0.7290322580645161,
+      "grad_norm": 0.7583324909210205,
+      "learning_rate": 2.448824059876503e-05,
+      "loss": 0.0979,
+      "step": 113
+    },
+    {
+      "epoch": 0.7354838709677419,
+      "grad_norm": 0.5010597705841064,
+      "learning_rate": 2.4468119781582948e-05,
+      "loss": 0.069,
+      "step": 114
+    },
+    {
+      "epoch": 0.7419354838709677,
+      "grad_norm": 0.5694583058357239,
+      "learning_rate": 2.444761960039072e-05,
+      "loss": 0.0687,
+      "step": 115
+    },
+    {
+      "epoch": 0.7483870967741936,
+      "grad_norm": 0.803371787071228,
+      "learning_rate": 2.442674070500061e-05,
+      "loss": 0.123,
+      "step": 116
+    },
+    {
+      "epoch": 0.7548387096774194,
+      "grad_norm": 0.6523027420043945,
+      "learning_rate": 2.4405483757229314e-05,
+      "loss": 0.0917,
+      "step": 117
+    },
+    {
+      "epoch": 0.7612903225806451,
+      "grad_norm": 0.6718930006027222,
+      "learning_rate": 2.438384943087698e-05,
+      "loss": 0.0854,
+      "step": 118
+    },
+    {
+      "epoch": 0.7677419354838709,
+      "grad_norm": 0.5987946391105652,
+      "learning_rate": 2.4361838411705865e-05,
+      "loss": 0.0941,
+      "step": 119
+    },
+    {
+      "epoch": 0.7741935483870968,
+      "grad_norm": 0.6336897015571594,
+      "learning_rate": 2.4339451397418584e-05,
+      "loss": 0.0885,
+      "step": 120
+    },
+    {
+      "epoch": 0.7806451612903226,
+      "grad_norm": 0.7484766840934753,
+      "learning_rate": 2.4316689097636008e-05,
+      "loss": 0.0966,
+      "step": 121
+    },
+    {
+      "epoch": 0.7870967741935484,
+      "grad_norm": 0.7096850275993347,
+      "learning_rate": 2.4293552233874754e-05,
+      "loss": 0.0843,
+      "step": 122
+    },
+    {
+      "epoch": 0.7935483870967742,
+      "grad_norm": 0.6953093409538269,
+      "learning_rate": 2.4270041539524322e-05,
+      "loss": 0.079,
+      "step": 123
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.6068540215492249,
+      "learning_rate": 2.4246157759823855e-05,
+      "loss": 0.0846,
+      "step": 124
+    },
+    {
+      "epoch": 0.8064516129032258,
+      "grad_norm": 0.5982446670532227,
+      "learning_rate": 2.4221901651838506e-05,
+      "loss": 0.0864,
+      "step": 125
+    },
+    {
+      "epoch": 0.8129032258064516,
+      "grad_norm": 0.6706437468528748,
+      "learning_rate": 2.419727398443545e-05,
+      "loss": 0.0796,
+      "step": 126
+    },
+    {
+      "epoch": 0.8193548387096774,
+      "grad_norm": 0.6994534730911255,
+      "learning_rate": 2.417227553825949e-05,
+      "loss": 0.0775,
+      "step": 127
+    },
+    {
+      "epoch": 0.8258064516129032,
+      "grad_norm": 0.6935513615608215,
+      "learning_rate": 2.4146907105708357e-05,
+      "loss": 0.1003,
+      "step": 128
+    },
+    {
+      "epoch": 0.832258064516129,
+      "grad_norm": 0.6945312023162842,
+      "learning_rate": 2.4121169490907544e-05,
+      "loss": 0.0901,
+      "step": 129
+    },
+    {
+      "epoch": 0.8387096774193549,
+      "grad_norm": 0.6928992867469788,
+      "learning_rate": 2.409506350968485e-05,
+      "loss": 0.0991,
+      "step": 130
+    },
+    {
+      "epoch": 0.8451612903225807,
+      "grad_norm": 0.6358478665351868,
+      "learning_rate": 2.4068589989544498e-05,
+      "loss": 0.0877,
+      "step": 131
+    },
+    {
+      "epoch": 0.8516129032258064,
+      "grad_norm": 0.6835708022117615,
+      "learning_rate": 2.404174976964092e-05,
+      "loss": 0.1058,
+      "step": 132
+    },
+    {
+      "epoch": 0.8580645161290322,
+      "grad_norm": 0.6372717022895813,
+      "learning_rate": 2.4014543700752156e-05,
+      "loss": 0.0899,
+      "step": 133
+    },
+    {
+      "epoch": 0.864516129032258,
+      "grad_norm": 0.671310544013977,
+      "learning_rate": 2.3986972645252883e-05,
+      "loss": 0.0744,
+      "step": 134
+    },
+    {
+      "epoch": 0.8709677419354839,
+      "grad_norm": 0.5800638794898987,
+      "learning_rate": 2.395903747708707e-05,
+      "loss": 0.0818,
+      "step": 135
+    },
+    {
+      "epoch": 0.8774193548387097,
+      "grad_norm": 0.5142645835876465,
+      "learning_rate": 2.39307390817403e-05,
+      "loss": 0.0811,
+      "step": 136
+    },
+    {
+      "epoch": 0.8838709677419355,
+      "grad_norm": 0.7107434868812561,
+      "learning_rate": 2.390207835621167e-05,
+      "loss": 0.0876,
+      "step": 137
+    },
+    {
+      "epoch": 0.8903225806451613,
+      "grad_norm": 0.6197046637535095,
+      "learning_rate": 2.3873056208985383e-05,
+      "loss": 0.0907,
+      "step": 138
+    },
+    {
+      "epoch": 0.896774193548387,
+      "grad_norm": 0.8946641087532043,
+      "learning_rate": 2.384367356000195e-05,
+      "loss": 0.0867,
+      "step": 139
+    },
+    {
+      "epoch": 0.9032258064516129,
+      "grad_norm": 0.6002138257026672,
+      "learning_rate": 2.3813931340629018e-05,
+      "loss": 0.0766,
+      "step": 140
+    },
+    {
+      "epoch": 0.9096774193548387,
+      "grad_norm": 0.4771173298358917,
+      "learning_rate": 2.378383049363184e-05,
+      "loss": 0.074,
+      "step": 141
+    },
+    {
+      "epoch": 0.9161290322580645,
+      "grad_norm": 0.6188220381736755,
+      "learning_rate": 2.3753371973143433e-05,
+      "loss": 0.0823,
+      "step": 142
+    },
+    {
+      "epoch": 0.9225806451612903,
+      "grad_norm": 0.509564995765686,
+      "learning_rate": 2.3722556744634272e-05,
+      "loss": 0.069,
+      "step": 143
+    },
+    {
+      "epoch": 0.9290322580645162,
+      "grad_norm": 0.5153804421424866,
+      "learning_rate": 2.3691385784881743e-05,
+      "loss": 0.064,
+      "step": 144
+    },
+    {
+      "epoch": 0.9354838709677419,
+      "grad_norm": 0.5935696363449097,
+      "learning_rate": 2.3659860081939146e-05,
+      "loss": 0.0827,
+      "step": 145
+    },
+    {
+      "epoch": 0.9419354838709677,
+      "grad_norm": 0.4910190999507904,
+      "learning_rate": 2.3627980635104396e-05,
+      "loss": 0.0804,
+      "step": 146
+    },
+    {
+      "epoch": 0.9483870967741935,
+      "grad_norm": 0.6524127721786499,
+      "learning_rate": 2.359574845488833e-05,
+      "loss": 0.0956,
+      "step": 147
+    },
+    {
+      "epoch": 0.9548387096774194,
+      "grad_norm": 0.6664571762084961,
+      "learning_rate": 2.356316456298269e-05,
+      "loss": 0.0937,
+      "step": 148
+    },
+    {
+      "epoch": 0.9612903225806452,
+      "grad_norm": 0.579138994216919,
+      "learning_rate": 2.353022999222774e-05,
+      "loss": 0.0936,
+      "step": 149
+    },
+    {
+      "epoch": 0.967741935483871,
+      "grad_norm": 0.6929976940155029,
+      "learning_rate": 2.3496945786579503e-05,
+      "loss": 0.1197,
+      "step": 150
+    },
+    {
+      "epoch": 0.9741935483870968,
+      "grad_norm": 0.6236998438835144,
+      "learning_rate": 2.3463313001076696e-05,
+      "loss": 0.0958,
+      "step": 151
+    },
+    {
+      "epoch": 0.9806451612903225,
+      "grad_norm": 0.4676724672317505,
+      "learning_rate": 2.342933270180728e-05,
+      "loss": 0.0713,
+      "step": 152
+    },
+    {
+      "epoch": 0.9870967741935484,
+      "grad_norm": 0.4456840753555298,
+      "learning_rate": 2.3395005965874657e-05,
+      "loss": 0.0784,
+      "step": 153
+    },
+    {
+      "epoch": 0.9935483870967742,
+      "grad_norm": 0.5712344646453857,
+      "learning_rate": 2.336033388136355e-05,
+      "loss": 0.0935,
+      "step": 154
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4726645350456238,
+      "learning_rate": 2.3325317547305485e-05,
+      "loss": 0.0712,
+      "step": 155
+    },
+    {
+      "epoch": 1.0064516129032257,
+      "grad_norm": 0.48649105429649353,
+      "learning_rate": 2.3289958073643976e-05,
+      "loss": 0.0584,
+      "step": 156
+    },
+    {
+      "epoch": 1.0129032258064516,
+      "grad_norm": 0.5159472823143005,
+      "learning_rate": 2.3254256581199336e-05,
+      "loss": 0.0579,
+      "step": 157
+    },
+    {
+      "epoch": 1.0193548387096774,
+      "grad_norm": 0.5775710344314575,
+      "learning_rate": 2.3218214201633136e-05,
+      "loss": 0.0676,
+      "step": 158
+    },
+    {
+      "epoch": 1.0258064516129033,
+      "grad_norm": 0.5070593357086182,
+      "learning_rate": 2.318183207741237e-05,
+      "loss": 0.0794,
+      "step": 159
+    },
+    {
+      "epoch": 1.032258064516129,
+      "grad_norm": 0.38065212965011597,
+      "learning_rate": 2.3145111361773186e-05,
+      "loss": 0.051,
+      "step": 160
+    },
+    {
+      "epoch": 1.038709677419355,
+      "grad_norm": 0.562282383441925,
+      "learning_rate": 2.310805321868439e-05,
+      "loss": 0.0753,
+      "step": 161
+    },
+    {
+      "epoch": 1.0451612903225806,
+      "grad_norm": 0.49883219599723816,
+      "learning_rate": 2.30706588228105e-05,
+      "loss": 0.0554,
+      "step": 162
+    },
+    {
+      "epoch": 1.0516129032258064,
+      "grad_norm": 0.5298740863800049,
+      "learning_rate": 2.303292935947455e-05,
+      "loss": 0.0602,
+      "step": 163
+    },
+    {
+      "epoch": 1.0580645161290323,
+      "grad_norm": 0.5768100619316101,
+      "learning_rate": 2.2994866024620486e-05,
+      "loss": 0.0585,
+      "step": 164
+    },
+    {
+      "epoch": 1.064516129032258,
+      "grad_norm": 0.5079744458198547,
+      "learning_rate": 2.2956470024775294e-05,
+      "loss": 0.0459,
+      "step": 165
+    },
+    {
+      "epoch": 1.070967741935484,
+      "grad_norm": 0.5212790966033936,
+      "learning_rate": 2.291774257701072e-05,
+      "loss": 0.0619,
+      "step": 166
+    },
+    {
+      "epoch": 1.0774193548387097,
+      "grad_norm": 0.5063428282737732,
+      "learning_rate": 2.2878684908904707e-05,
+      "loss": 0.0609,
+      "step": 167
+    },
+    {
+      "epoch": 1.0838709677419356,
+      "grad_norm": 0.6523650288581848,
+      "learning_rate": 2.2839298258502483e-05,
+      "loss": 0.067,
+      "step": 168
+    },
+    {
+      "epoch": 1.0903225806451613,
+      "grad_norm": 0.57984459400177,
+      "learning_rate": 2.279958387427732e-05,
+      "loss": 0.0703,
+      "step": 169
+    },
+    {
+      "epoch": 1.096774193548387,
+      "grad_norm": 0.6002654433250427,
+      "learning_rate": 2.2759543015090955e-05,
+      "loss": 0.074,
+      "step": 170
+    },
+    {
+      "epoch": 1.103225806451613,
+      "grad_norm": 0.3899862766265869,
+      "learning_rate": 2.2719176950153688e-05,
+      "loss": 0.0461,
+      "step": 171
+    },
+    {
+      "epoch": 1.1096774193548387,
+      "grad_norm": 0.5003259778022766,
+      "learning_rate": 2.267848695898416e-05,
+      "loss": 0.0613,
+      "step": 172
+    },
+    {
+      "epoch": 1.1161290322580646,
+      "grad_norm": 0.558653712272644,
+      "learning_rate": 2.2637474331368766e-05,
+      "loss": 0.0658,
+      "step": 173
+    },
+    {
+      "epoch": 1.1225806451612903,
+      "grad_norm": 0.5032625794410706,
+      "learning_rate": 2.2596140367320813e-05,
+      "loss": 0.0564,
+      "step": 174
+    },
+    {
+      "epoch": 1.129032258064516,
+      "grad_norm": 0.5199857950210571,
+      "learning_rate": 2.2554486377039282e-05,
+      "loss": 0.0587,
+      "step": 175
+    },
+    {
+      "epoch": 1.135483870967742,
+      "grad_norm": 0.6159687042236328,
+      "learning_rate": 2.251251368086731e-05,
+      "loss": 0.0585,
+      "step": 176
+    },
+    {
+      "epoch": 1.1419354838709677,
+      "grad_norm": 0.5216447114944458,
+      "learning_rate": 2.2470223609250328e-05,
+      "loss": 0.0501,
+      "step": 177
+    },
+    {
+      "epoch": 1.1483870967741936,
+      "grad_norm": 0.49131321907043457,
+      "learning_rate": 2.24276175026939e-05,
+      "loss": 0.053,
+      "step": 178
+    },
+    {
+      "epoch": 1.1548387096774193,
+      "grad_norm": 0.8894760608673096,
+      "learning_rate": 2.238469671172123e-05,
+      "loss": 0.0854,
+      "step": 179
+    },
+    {
+      "epoch": 1.1612903225806452,
+      "grad_norm": 0.6628456711769104,
+      "learning_rate": 2.2341462596830354e-05,
+      "loss": 0.064,
+      "step": 180
+    },
+    {
+      "epoch": 1.167741935483871,
+      "grad_norm": 0.4577731788158417,
+      "learning_rate": 2.229791652845099e-05,
+      "loss": 0.0543,
+      "step": 181
+    },
+    {
+      "epoch": 1.1741935483870969,
+      "grad_norm": 0.49301421642303467,
+      "learning_rate": 2.225405988690115e-05,
+      "loss": 0.0598,
+      "step": 182
+    },
+    {
+      "epoch": 1.1806451612903226,
+      "grad_norm": 0.523009717464447,
+      "learning_rate": 2.220989406234333e-05,
+      "loss": 0.0752,
+      "step": 183
+    },
+    {
+      "epoch": 1.1870967741935483,
+      "grad_norm": 0.7591210007667542,
+      "learning_rate": 2.2165420454740494e-05,
+      "loss": 0.0643,
+      "step": 184
+    },
+    {
+      "epoch": 1.1935483870967742,
+      "grad_norm": 0.532319962978363,
+      "learning_rate": 2.2120640473811656e-05,
+      "loss": 0.0464,
+      "step": 185
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.4334715008735657,
+      "learning_rate": 2.2075555538987227e-05,
+      "loss": 0.0669,
+      "step": 186
+    },
+    {
+      "epoch": 1.206451612903226,
+      "grad_norm": 0.4052492380142212,
+      "learning_rate": 2.2030167079364007e-05,
+      "loss": 0.0506,
+      "step": 187
+    },
+    {
+      "epoch": 1.2129032258064516,
+      "grad_norm": 0.763782799243927,
+      "learning_rate": 2.1984476533659888e-05,
+      "loss": 0.0477,
+      "step": 188
+    },
+    {
+      "epoch": 1.2193548387096773,
+      "grad_norm": 0.46810972690582275,
+      "learning_rate": 2.1938485350168248e-05,
+      "loss": 0.055,
+      "step": 189
+    },
+    {
+      "epoch": 1.2258064516129032,
+      "grad_norm": 0.4722144901752472,
+      "learning_rate": 2.1892194986712045e-05,
+      "loss": 0.053,
+      "step": 190
+    },
+    {
+      "epoch": 1.232258064516129,
+      "grad_norm": 0.5537333488464355,
+      "learning_rate": 2.1845606910597616e-05,
+      "loss": 0.0686,
+      "step": 191
+    },
+    {
+      "epoch": 1.238709677419355,
+      "grad_norm": 0.5123704671859741,
+      "learning_rate": 2.179872259856814e-05,
+      "loss": 0.0627,
+      "step": 192
+    },
+    {
+      "epoch": 1.2451612903225806,
+      "grad_norm": 0.5691571831703186,
+      "learning_rate": 2.175154353675686e-05,
+      "loss": 0.0601,
+      "step": 193
+    },
+    {
+      "epoch": 1.2516129032258063,
+      "grad_norm": 0.4747653007507324,
+      "learning_rate": 2.1704071220639965e-05,
+      "loss": 0.0551,
+      "step": 194
+    },
+    {
+      "epoch": 1.2580645161290323,
+      "grad_norm": 0.5692989826202393,
+      "learning_rate": 2.1656307154989174e-05,
+      "loss": 0.0482,
+      "step": 195
+    },
+    {
+      "epoch": 1.2645161290322582,
+      "grad_norm": 0.7472412586212158,
+      "learning_rate": 2.1608252853824047e-05,
+      "loss": 0.0609,
+      "step": 196
+    },
+    {
+      "epoch": 1.270967741935484,
+      "grad_norm": 0.568708062171936,
+      "learning_rate": 2.1559909840364e-05,
+      "loss": 0.0572,
+      "step": 197
+    },
+    {
+      "epoch": 1.2774193548387096,
+      "grad_norm": 0.6601235866546631,
+      "learning_rate": 2.1511279646980016e-05,
+      "loss": 0.0777,
+      "step": 198
+    },
+    {
+      "epoch": 1.2838709677419355,
+      "grad_norm": 0.429850697517395,
+      "learning_rate": 2.1462363815146065e-05,
+      "loss": 0.0454,
+      "step": 199
+    },
+    {
+      "epoch": 1.2903225806451613,
+      "grad_norm": 0.7702894806861877,
+      "learning_rate": 2.1413163895390254e-05,
+      "loss": 0.0655,
+      "step": 200
+    },
+    {
+      "epoch": 1.2967741935483872,
+      "grad_norm": 0.4497153162956238,
+      "learning_rate": 2.1363681447245686e-05,
+      "loss": 0.0512,
+      "step": 201
+    },
+    {
+      "epoch": 1.303225806451613,
+      "grad_norm": 0.5631290674209595,
+      "learning_rate": 2.1313918039200995e-05,
+      "loss": 0.0645,
+      "step": 202
+    },
+    {
+      "epoch": 1.3096774193548386,
+      "grad_norm": 0.7414901852607727,
+      "learning_rate": 2.1263875248650662e-05,
+      "loss": 0.0561,
+      "step": 203
+    },
+    {
+      "epoch": 1.3161290322580645,
+      "grad_norm": 0.5053102970123291,
+      "learning_rate": 2.121355466184499e-05,
+      "loss": 0.0608,
+      "step": 204
+    },
+    {
+      "epoch": 1.3225806451612903,
+      "grad_norm": 10.00545597076416,
+      "learning_rate": 2.116295787383985e-05,
+      "loss": 0.0826,
+      "step": 205
+    },
+    {
+      "epoch": 1.3290322580645162,
+      "grad_norm": 0.6418637037277222,
+      "learning_rate": 2.1112086488446085e-05,
+      "loss": 0.0743,
+      "step": 206
+    },
+    {
+      "epoch": 1.335483870967742,
+      "grad_norm": 0.4627211391925812,
+      "learning_rate": 2.1060942118178706e-05,
+      "loss": 0.0476,
+      "step": 207
+    },
+    {
+      "epoch": 1.3419354838709676,
+      "grad_norm": 0.5375849604606628,
+      "learning_rate": 2.1009526384205767e-05,
+      "loss": 0.048,
+      "step": 208
+    },
+    {
+      "epoch": 1.3483870967741935,
+      "grad_norm": 0.606073796749115,
+      "learning_rate": 2.095784091629697e-05,
+      "loss": 0.0704,
+      "step": 209
+    },
+    {
+      "epoch": 1.3548387096774195,
+      "grad_norm": 0.44339242577552795,
+      "learning_rate": 2.0905887352772004e-05,
+      "loss": 0.0516,
+      "step": 210
+    },
+    {
+      "epoch": 1.3612903225806452,
+      "grad_norm": 0.6248610019683838,
+      "learning_rate": 2.085366734044864e-05,
+      "loss": 0.066,
+      "step": 211
+    },
+    {
+      "epoch": 1.367741935483871,
+      "grad_norm": 0.5914815664291382,
+      "learning_rate": 2.080118253459049e-05,
+      "loss": 0.0611,
+      "step": 212
+    },
+    {
+      "epoch": 1.3741935483870968,
+      "grad_norm": 0.45894381403923035,
+      "learning_rate": 2.0748434598854573e-05,
+      "loss": 0.0501,
+      "step": 213
+    },
+    {
+      "epoch": 1.3806451612903226,
+      "grad_norm": 0.49100032448768616,
+      "learning_rate": 2.0695425205238557e-05,
+      "loss": 0.0552,
+      "step": 214
+    },
+    {
+      "epoch": 1.3870967741935485,
+      "grad_norm": 0.528611958026886,
+      "learning_rate": 2.0642156034027783e-05,
+      "loss": 0.0639,
+      "step": 215
+    },
+    {
+      "epoch": 1.3935483870967742,
+      "grad_norm": 0.4487656056880951,
+      "learning_rate": 2.0588628773741973e-05,
+      "loss": 0.0435,
+      "step": 216
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.547444224357605,
+      "learning_rate": 2.0534845121081742e-05,
+      "loss": 0.0547,
+      "step": 217
+    },
+    {
+      "epoch": 1.4064516129032258,
+      "grad_norm": 0.5207445621490479,
+      "learning_rate": 2.0480806780874794e-05,
+      "loss": 0.0574,
+      "step": 218
+    },
+    {
+      "epoch": 1.4129032258064516,
+      "grad_norm": 0.5784499049186707,
+      "learning_rate": 2.0426515466021887e-05,
+      "loss": 0.0608,
+      "step": 219
+    },
+    {
+      "epoch": 1.4193548387096775,
+      "grad_norm": 0.7198527455329895,
+      "learning_rate": 2.0371972897442532e-05,
+      "loss": 0.0639,
+      "step": 220
+    },
+    {
+      "epoch": 1.4258064516129032,
+      "grad_norm": 0.4550151228904724,
+      "learning_rate": 2.031718080402046e-05,
+      "loss": 0.0547,
+      "step": 221
+    },
+    {
+      "epoch": 1.432258064516129,
+      "grad_norm": 0.48588842153549194,
+      "learning_rate": 2.026214092254881e-05,
+      "loss": 0.0603,
+      "step": 222
+    },
+    {
+      "epoch": 1.4387096774193548,
+      "grad_norm": 0.5426737666130066,
+      "learning_rate": 2.0206854997675072e-05,
+      "loss": 0.0616,
+      "step": 223
+    },
+    {
+      "epoch": 1.4451612903225808,
+      "grad_norm": 0.5034387707710266,
+      "learning_rate": 2.0151324781845787e-05,
+      "loss": 0.0644,
+      "step": 224
+    },
+    {
+      "epoch": 1.4516129032258065,
+      "grad_norm": 0.5200063586235046,
+      "learning_rate": 2.0095552035251007e-05,
+      "loss": 0.0596,
+      "step": 225
+    },
+    {
+      "epoch": 1.4580645161290322,
+      "grad_norm": 0.4462428390979767,
+      "learning_rate": 2.0039538525768496e-05,
+      "loss": 0.0523,
+      "step": 226
+    },
+    {
+      "epoch": 1.4645161290322581,
+      "grad_norm": 0.5513397455215454,
+      "learning_rate": 1.9983286028907687e-05,
+      "loss": 0.0528,
+      "step": 227
+    },
+    {
+      "epoch": 1.4709677419354839,
+      "grad_norm": 0.44743800163269043,
+      "learning_rate": 1.992679632775341e-05,
+      "loss": 0.0649,
+      "step": 228
+    },
+    {
+      "epoch": 1.4774193548387098,
+      "grad_norm": 0.4505648910999298,
+      "learning_rate": 1.9870071212909357e-05,
+      "loss": 0.0453,
+      "step": 229
+    },
+    {
+      "epoch": 1.4838709677419355,
+      "grad_norm": 0.48718520998954773,
+      "learning_rate": 1.9813112482441345e-05,
+      "loss": 0.0664,
+      "step": 230
+    },
+    {
+      "epoch": 1.4903225806451612,
+      "grad_norm": 0.4392196834087372,
+      "learning_rate": 1.9755921941820314e-05,
+      "loss": 0.0504,
+      "step": 231
+    },
+    {
+      "epoch": 1.4967741935483871,
+      "grad_norm": 0.5312716364860535,
+      "learning_rate": 1.9698501403865083e-05,
+      "loss": 0.0699,
+      "step": 232
+    },
+    {
+      "epoch": 1.5032258064516129,
+      "grad_norm": 0.5387852787971497,
+      "learning_rate": 1.9640852688684904e-05,
+      "loss": 0.071,
+      "step": 233
+    },
+    {
+      "epoch": 1.5096774193548388,
+      "grad_norm": 0.4734801650047302,
+      "learning_rate": 1.9582977623621766e-05,
+      "loss": 0.0561,
+      "step": 234
+    },
+    {
+      "epoch": 1.5161290322580645,
+      "grad_norm": 0.4738084375858307,
+      "learning_rate": 1.9524878043192463e-05,
+      "loss": 0.0545,
+      "step": 235
+    },
+    {
+      "epoch": 1.5225806451612902,
+      "grad_norm": 0.5166822671890259,
+      "learning_rate": 1.9466555789030456e-05,
+      "loss": 0.0708,
+      "step": 236
+    },
+    {
+      "epoch": 1.5290322580645161,
+      "grad_norm": 0.5719185471534729,
+      "learning_rate": 1.9408012709827485e-05,
+      "loss": 0.073,
+      "step": 237
+    },
+    {
+      "epoch": 1.535483870967742,
+      "grad_norm": 0.5363075733184814,
+      "learning_rate": 1.934925066127498e-05,
+      "loss": 0.0581,
+      "step": 238
+    },
+    {
+      "epoch": 1.5419354838709678,
+      "grad_norm": 0.551699697971344,
+      "learning_rate": 1.9290271506005236e-05,
+      "loss": 0.0598,
+      "step": 239
+    },
+    {
+      "epoch": 1.5483870967741935,
+      "grad_norm": 0.5568850636482239,
+      "learning_rate": 1.9231077113532363e-05,
+      "loss": 0.0471,
+      "step": 240
+    },
+    {
+      "epoch": 1.5548387096774192,
+      "grad_norm": 0.5048776268959045,
+      "learning_rate": 1.917166936019304e-05,
+      "loss": 0.0613,
+      "step": 241
+    },
+    {
+      "epoch": 1.5612903225806452,
+      "grad_norm": 0.516986608505249,
+      "learning_rate": 1.911205012908703e-05,
+      "loss": 0.0678,
+      "step": 242
+    },
+    {
+      "epoch": 1.567741935483871,
+      "grad_norm": 0.48142287135124207,
+      "learning_rate": 1.90522213100175e-05,
+      "loss": 0.0557,
+      "step": 243
+    },
+    {
+      "epoch": 1.5741935483870968,
+      "grad_norm": 0.4997798800468445,
+      "learning_rate": 1.8992184799431095e-05,
+      "loss": 0.042,
+      "step": 244
+    },
+    {
+      "epoch": 1.5806451612903225,
+      "grad_norm": 0.5074776411056519,
+      "learning_rate": 1.893194250035786e-05,
+      "loss": 0.073,
+      "step": 245
+    },
+    {
+      "epoch": 1.5870967741935482,
+      "grad_norm": 0.5136696696281433,
+      "learning_rate": 1.8871496322350883e-05,
+      "loss": 0.0547,
+      "step": 246
+    },
+    {
+      "epoch": 1.5935483870967742,
+      "grad_norm": 0.6183574795722961,
+      "learning_rate": 1.881084818142579e-05,
+      "loss": 0.0708,
+      "step": 247
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.5576770901679993,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 0.0629,
+      "step": 248
+    },
+    {
+      "epoch": 1.6064516129032258,
+      "grad_norm": 0.4211249351501465,
+      "learning_rate": 1.868895370683179e-05,
+      "loss": 0.0544,
+      "step": 249
+    },
+    {
+      "epoch": 1.6129032258064515,
+      "grad_norm": 0.43533676862716675,
+      "learning_rate": 1.862771123695917e-05,
+      "loss": 0.0568,
+      "step": 250
+    },
+    {
+      "epoch": 1.6193548387096774,
+      "grad_norm": 0.48171842098236084,
+      "learning_rate": 1.8566274531638516e-05,
+      "loss": 0.0524,
+      "step": 251
+    },
+    {
+      "epoch": 1.6258064516129034,
+      "grad_norm": 0.459471195936203,
+      "learning_rate": 1.850464553828307e-05,
+      "loss": 0.0471,
+      "step": 252
+    },
+    {
+      "epoch": 1.632258064516129,
+      "grad_norm": 0.5311537384986877,
+      "learning_rate": 1.844282621040119e-05,
+      "loss": 0.0766,
+      "step": 253
+    },
+    {
+      "epoch": 1.6387096774193548,
+      "grad_norm": 0.5022658109664917,
+      "learning_rate": 1.838081850753445e-05,
+      "loss": 0.0579,
+      "step": 254
+    },
+    {
+      "epoch": 1.6451612903225805,
+      "grad_norm": 0.5516560077667236,
+      "learning_rate": 1.8318624395195483e-05,
+      "loss": 0.0616,
+      "step": 255
+    },
+    {
+      "epoch": 1.6516129032258065,
+      "grad_norm": 0.4552045166492462,
+      "learning_rate": 1.825624584480573e-05,
+      "loss": 0.0512,
+      "step": 256
+    },
+    {
+      "epoch": 1.6580645161290324,
+      "grad_norm": 0.5871717929840088,
+      "learning_rate": 1.8193684833632925e-05,
+      "loss": 0.0641,
+      "step": 257
+    },
+    {
+      "epoch": 1.664516129032258,
+      "grad_norm": 0.46038615703582764,
+      "learning_rate": 1.8130943344728414e-05,
+      "loss": 0.0459,
+      "step": 258
+    },
+    {
+      "epoch": 1.6709677419354838,
+      "grad_norm": 0.5282014608383179,
+      "learning_rate": 1.8068023366864305e-05,
+      "loss": 0.0569,
+      "step": 259
+    },
+    {
+      "epoch": 1.6774193548387095,
+      "grad_norm": 0.3797317147254944,
+      "learning_rate": 1.800492689447043e-05,
+      "loss": 0.0459,
+      "step": 260
+    },
+    {
+      "epoch": 1.6838709677419355,
+      "grad_norm": 0.5863360166549683,
+      "learning_rate": 1.7941655927571125e-05,
+      "loss": 0.0695,
+      "step": 261
+    },
+    {
+      "epoch": 1.6903225806451614,
+      "grad_norm": 0.5562090873718262,
+      "learning_rate": 1.7878212471721824e-05,
+      "loss": 0.0554,
+      "step": 262
+    },
+    {
+      "epoch": 1.696774193548387,
+      "grad_norm": 0.5164937973022461,
+      "learning_rate": 1.781459853794551e-05,
+      "loss": 0.0542,
+      "step": 263
+    },
+    {
+      "epoch": 1.7032258064516128,
+      "grad_norm": 0.5710752010345459,
+      "learning_rate": 1.7750816142668937e-05,
+      "loss": 0.0641,
+      "step": 264
+    },
+    {
+      "epoch": 1.7096774193548387,
+      "grad_norm": 0.43633976578712463,
+      "learning_rate": 1.7686867307658743e-05,
+      "loss": 0.0498,
+      "step": 265
+    },
+    {
+      "epoch": 1.7161290322580647,
+      "grad_norm": 0.5214335322380066,
+      "learning_rate": 1.7622754059957343e-05,
+      "loss": 0.054,
+      "step": 266
+    },
+    {
+      "epoch": 1.7225806451612904,
+      "grad_norm": 0.6009476780891418,
+      "learning_rate": 1.7558478431818702e-05,
+      "loss": 0.0538,
+      "step": 267
+    },
+    {
+      "epoch": 1.729032258064516,
+      "grad_norm": 0.5809276700019836,
+      "learning_rate": 1.749404246064388e-05,
+      "loss": 0.0751,
+      "step": 268
+    },
+    {
+      "epoch": 1.7354838709677418,
+      "grad_norm": 0.5733875632286072,
+      "learning_rate": 1.7429448188916483e-05,
+      "loss": 0.0685,
+      "step": 269
+    },
+    {
+      "epoch": 1.7419354838709677,
+      "grad_norm": 0.3861143887042999,
+      "learning_rate": 1.7364697664137912e-05,
+      "loss": 0.044,
+      "step": 270
+    },
+    {
+      "epoch": 1.7483870967741937,
+      "grad_norm": 0.8718386292457581,
+      "learning_rate": 1.7299792938762443e-05,
+      "loss": 0.0807,
+      "step": 271
+    },
+    {
+      "epoch": 1.7548387096774194,
+      "grad_norm": 0.6809967160224915,
+      "learning_rate": 1.72347360701322e-05,
+      "loss": 0.0698,
+      "step": 272
+    },
+    {
+      "epoch": 1.761290322580645,
+      "grad_norm": 0.45045140385627747,
+      "learning_rate": 1.7169529120411922e-05,
+      "loss": 0.0552,
+      "step": 273
+    },
+    {
+      "epoch": 1.7677419354838708,
+      "grad_norm": 0.46889108419418335,
+      "learning_rate": 1.710417415652359e-05,
+      "loss": 0.0576,
+      "step": 274
+    },
+    {
+      "epoch": 1.7741935483870968,
+      "grad_norm": 0.5053566098213196,
+      "learning_rate": 1.7038673250080934e-05,
+      "loss": 0.0535,
+      "step": 275
+    },
+    {
+      "epoch": 1.7806451612903227,
+      "grad_norm": 0.4330599904060364,
+      "learning_rate": 1.6973028477323742e-05,
+      "loss": 0.0518,
+      "step": 276
+    },
+    {
+      "epoch": 1.7870967741935484,
+      "grad_norm": 0.4866834580898285,
+      "learning_rate": 1.6907241919052068e-05,
+      "loss": 0.051,
+      "step": 277
+    },
+    {
+      "epoch": 1.793548387096774,
+      "grad_norm": 0.6048698425292969,
+      "learning_rate": 1.6841315660560252e-05,
+      "loss": 0.0683,
+      "step": 278
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.39490604400634766,
+      "learning_rate": 1.677525179157086e-05,
+      "loss": 0.0515,
+      "step": 279
+    },
+    {
+      "epoch": 1.8064516129032258,
+      "grad_norm": 0.4544709324836731,
+      "learning_rate": 1.6709052406168393e-05,
+      "loss": 0.0624,
+      "step": 280
+    },
+    {
+      "epoch": 1.8129032258064517,
+      "grad_norm": 0.5158767700195312,
+      "learning_rate": 1.664271960273295e-05,
+      "loss": 0.0575,
+      "step": 281
+    },
+    {
+      "epoch": 1.8193548387096774,
+      "grad_norm": 0.5172263979911804,
+      "learning_rate": 1.6576255483873686e-05,
+      "loss": 0.0578,
+      "step": 282
+    },
+    {
+      "epoch": 1.8258064516129031,
+      "grad_norm": 0.4233238995075226,
+      "learning_rate": 1.6509662156362196e-05,
+      "loss": 0.0547,
+      "step": 283
+    },
+    {
+      "epoch": 1.832258064516129,
+      "grad_norm": 0.45361143350601196,
+      "learning_rate": 1.6442941731065697e-05,
+      "loss": 0.0512,
+      "step": 284
+    },
+    {
+      "epoch": 1.838709677419355,
+      "grad_norm": 0.5802233219146729,
+      "learning_rate": 1.637609632288014e-05,
+      "loss": 0.0596,
+      "step": 285
+    },
+    {
+      "epoch": 1.8451612903225807,
+      "grad_norm": 0.5369323492050171,
+      "learning_rate": 1.630912805066317e-05,
+      "loss": 0.0646,
+      "step": 286
+    },
+    {
+      "epoch": 1.8516129032258064,
+      "grad_norm": 0.45122525095939636,
+      "learning_rate": 1.6242039037166977e-05,
+      "loss": 0.0517,
+      "step": 287
+    },
+    {
+      "epoch": 1.8580645161290321,
+      "grad_norm": 0.39205196499824524,
+      "learning_rate": 1.6174831408970964e-05,
+      "loss": 0.0491,
+      "step": 288
+    },
+    {
+      "epoch": 1.864516129032258,
+      "grad_norm": 0.4472959637641907,
+      "learning_rate": 1.6107507296414383e-05,
+      "loss": 0.049,
+      "step": 289
+    },
+    {
+      "epoch": 1.870967741935484,
+      "grad_norm": 0.41624531149864197,
+      "learning_rate": 1.6040068833528797e-05,
+      "loss": 0.0483,
+      "step": 290
+    },
+    {
+      "epoch": 1.8774193548387097,
+      "grad_norm": 0.43875551223754883,
+      "learning_rate": 1.597251815797044e-05,
+      "loss": 0.0487,
+      "step": 291
+    },
+    {
+      "epoch": 1.8838709677419354,
+      "grad_norm": 0.4978736937046051,
+      "learning_rate": 1.5904857410952417e-05,
+      "loss": 0.0573,
+      "step": 292
+    },
+    {
+      "epoch": 1.8903225806451613,
+      "grad_norm": 0.5798497796058655,
+      "learning_rate": 1.5837088737176896e-05,
+      "loss": 0.0683,
+      "step": 293
+    },
+    {
+      "epoch": 1.896774193548387,
+      "grad_norm": 0.7377052903175354,
+      "learning_rate": 1.5769214284767086e-05,
+      "loss": 0.0583,
+      "step": 294
+    },
+    {
+      "epoch": 1.903225806451613,
+      "grad_norm": 0.4153827428817749,
+      "learning_rate": 1.570123620519915e-05,
+      "loss": 0.0543,
+      "step": 295
+    },
+    {
+      "epoch": 1.9096774193548387,
+      "grad_norm": 0.4852810800075531,
+      "learning_rate": 1.563315665323401e-05,
+      "loss": 0.0636,
+      "step": 296
+    },
+    {
+      "epoch": 1.9161290322580644,
+      "grad_norm": 0.5545767545700073,
+      "learning_rate": 1.5564977786849055e-05,
+      "loss": 0.062,
+      "step": 297
+    },
+    {
+      "epoch": 1.9225806451612903,
+      "grad_norm": 0.4363822937011719,
+      "learning_rate": 1.549670176716973e-05,
+      "loss": 0.0516,
+      "step": 298
+    },
+    {
+      "epoch": 1.9290322580645163,
+      "grad_norm": 0.5309383273124695,
+      "learning_rate": 1.5428330758401027e-05,
+      "loss": 0.0647,
+      "step": 299
+    },
+    {
+      "epoch": 1.935483870967742,
+      "grad_norm": 0.6617056131362915,
+      "learning_rate": 1.53598669277589e-05,
+      "loss": 0.0641,
+      "step": 300
+    },
+    {
+      "epoch": 1.9419354838709677,
+      "grad_norm": 0.49968254566192627,
+      "learning_rate": 1.529131244540155e-05,
+      "loss": 0.0585,
+      "step": 301
+    },
+    {
+      "epoch": 1.9483870967741934,
+      "grad_norm": 0.40158751606941223,
+      "learning_rate": 1.5222669484360644e-05,
+      "loss": 0.0537,
+      "step": 302
+    },
+    {
+      "epoch": 1.9548387096774194,
+      "grad_norm": 0.4537198543548584,
+      "learning_rate": 1.5153940220472451e-05,
+      "loss": 0.0511,
+      "step": 303
+    },
+    {
+      "epoch": 1.9612903225806453,
+      "grad_norm": 0.47163766622543335,
+      "learning_rate": 1.5085126832308843e-05,
+      "loss": 0.0532,
+      "step": 304
+    },
+    {
+      "epoch": 1.967741935483871,
+      "grad_norm": 0.4738634526729584,
+      "learning_rate": 1.5016231501108253e-05,
+      "loss": 0.0615,
+      "step": 305
+    },
+    {
+      "epoch": 1.9741935483870967,
+      "grad_norm": 0.38560569286346436,
+      "learning_rate": 1.494725641070654e-05,
+      "loss": 0.0522,
+      "step": 306
+    },
+    {
+      "epoch": 1.9806451612903224,
+      "grad_norm": 0.5569445490837097,
+      "learning_rate": 1.4878203747467764e-05,
+      "loss": 0.0731,
+      "step": 307
+    },
+    {
+      "epoch": 1.9870967741935484,
+      "grad_norm": 0.38958773016929626,
+      "learning_rate": 1.480907570021487e-05,
+      "loss": 0.0461,
+      "step": 308
+    },
+    {
+      "epoch": 1.9935483870967743,
+      "grad_norm": 0.4473820924758911,
+      "learning_rate": 1.4739874460160316e-05,
+      "loss": 0.0555,
+      "step": 309
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.39891934394836426,
+      "learning_rate": 1.4670602220836633e-05,
+      "loss": 0.051,
+      "step": 310
+    },
+    {
+      "epoch": 2.0064516129032257,
+      "grad_norm": 0.474127858877182,
+      "learning_rate": 1.4601261178026854e-05,
+      "loss": 0.0401,
+      "step": 311
+    },
+    {
+      "epoch": 2.0129032258064514,
+      "grad_norm": 0.3391839563846588,
+      "learning_rate": 1.4531853529694956e-05,
+      "loss": 0.0333,
+      "step": 312
+    },
+    {
+      "epoch": 2.0193548387096776,
+      "grad_norm": 0.3230273723602295,
+      "learning_rate": 1.446238147591616e-05,
+      "loss": 0.0282,
+      "step": 313
+    },
+    {
+      "epoch": 2.0258064516129033,
+      "grad_norm": 0.3246399462223053,
+      "learning_rate": 1.439284721880721e-05,
+      "loss": 0.0345,
+      "step": 314
+    },
+    {
+      "epoch": 2.032258064516129,
+      "grad_norm": 0.41817039251327515,
+      "learning_rate": 1.4323252962456554e-05,
+      "loss": 0.0288,
+      "step": 315
+    },
+    {
+      "epoch": 2.0387096774193547,
+      "grad_norm": 0.48674166202545166,
+      "learning_rate": 1.4253600912854497e-05,
+      "loss": 0.0354,
+      "step": 316
+    },
+    {
+      "epoch": 2.0451612903225804,
+      "grad_norm": 0.42214757204055786,
+      "learning_rate": 1.4183893277823265e-05,
+      "loss": 0.0388,
+      "step": 317
+    },
+    {
+      "epoch": 2.0516129032258066,
+      "grad_norm": 0.5475701093673706,
+      "learning_rate": 1.411413226694702e-05,
+      "loss": 0.0294,
+      "step": 318
+    },
+    {
+      "epoch": 2.0580645161290323,
+      "grad_norm": 0.5432962775230408,
+      "learning_rate": 1.4044320091501834e-05,
+      "loss": 0.0372,
+      "step": 319
+    },
+    {
+      "epoch": 2.064516129032258,
+      "grad_norm": 0.49539855122566223,
+      "learning_rate": 1.3974458964385579e-05,
+      "loss": 0.0425,
+      "step": 320
+    },
+    {
+      "epoch": 2.0709677419354837,
+      "grad_norm": 0.340425044298172,
+      "learning_rate": 1.3904551100047791e-05,
+      "loss": 0.026,
+      "step": 321
+    },
+    {
+      "epoch": 2.07741935483871,
+      "grad_norm": 0.4815217852592468,
+      "learning_rate": 1.3834598714419486e-05,
+      "loss": 0.0352,
+      "step": 322
+    },
+    {
+      "epoch": 2.0838709677419356,
+      "grad_norm": 0.4457317888736725,
+      "learning_rate": 1.3764604024842903e-05,
+      "loss": 0.028,
+      "step": 323
+    },
+    {
+      "epoch": 2.0903225806451613,
+      "grad_norm": 0.45776546001434326,
+      "learning_rate": 1.369456925000123e-05,
+      "loss": 0.0287,
+      "step": 324
+    },
+    {
+      "epoch": 2.096774193548387,
+      "grad_norm": 0.3825792968273163,
+      "learning_rate": 1.362449660984826e-05,
+      "loss": 0.0257,
+      "step": 325
+    },
+    {
+      "epoch": 2.1032258064516127,
+      "grad_norm": 0.44209763407707214,
+      "learning_rate": 1.3554388325538059e-05,
+      "loss": 0.0274,
+      "step": 326
+    },
+    {
+      "epoch": 2.109677419354839,
+      "grad_norm": 0.37732046842575073,
+      "learning_rate": 1.3484246619354524e-05,
+      "loss": 0.0263,
+      "step": 327
+    },
+    {
+      "epoch": 2.1161290322580646,
+      "grad_norm": 0.4975365698337555,
+      "learning_rate": 1.3414073714640951e-05,
+      "loss": 0.0294,
+      "step": 328
+    },
+    {
+      "epoch": 2.1225806451612903,
+      "grad_norm": 0.37548768520355225,
+      "learning_rate": 1.3343871835729565e-05,
+      "loss": 0.0261,
+      "step": 329
+    },
+    {
+      "epoch": 2.129032258064516,
+      "grad_norm": 0.722154438495636,
+      "learning_rate": 1.3273643207871025e-05,
+      "loss": 0.0296,
+      "step": 330
+    },
+    {
+      "epoch": 2.135483870967742,
+      "grad_norm": 0.513611912727356,
+      "learning_rate": 1.3203390057163855e-05,
+      "loss": 0.0326,
+      "step": 331
+    },
+    {
+      "epoch": 2.141935483870968,
+      "grad_norm": 0.43579375743865967,
+      "learning_rate": 1.3133114610483909e-05,
+      "loss": 0.035,
+      "step": 332
+    },
+    {
+      "epoch": 2.1483870967741936,
+      "grad_norm": 0.4927336275577545,
+      "learning_rate": 1.3062819095413786e-05,
+      "loss": 0.0358,
+      "step": 333
+    },
+    {
+      "epoch": 2.1548387096774193,
+      "grad_norm": 0.43542489409446716,
+      "learning_rate": 1.2992505740172196e-05,
+      "loss": 0.035,
+      "step": 334
+    },
+    {
+      "epoch": 2.161290322580645,
+      "grad_norm": 0.34009236097335815,
+      "learning_rate": 1.2922176773543355e-05,
+      "loss": 0.0264,
+      "step": 335
+    },
+    {
+      "epoch": 2.167741935483871,
+      "grad_norm": 0.4710192084312439,
+      "learning_rate": 1.2851834424806314e-05,
+      "loss": 0.0403,
+      "step": 336
+    },
+    {
+      "epoch": 2.174193548387097,
+      "grad_norm": 0.8653304576873779,
+      "learning_rate": 1.2781480923664326e-05,
+      "loss": 0.0839,
+      "step": 337
+    },
+    {
+      "epoch": 2.1806451612903226,
+      "grad_norm": 0.7528795599937439,
+      "learning_rate": 1.2711118500174138e-05,
+      "loss": 0.0488,
+      "step": 338
+    },
+    {
+      "epoch": 2.1870967741935483,
+      "grad_norm": 0.5551451444625854,
+      "learning_rate": 1.2640749384675324e-05,
+      "loss": 0.0223,
+      "step": 339
+    },
+    {
+      "epoch": 2.193548387096774,
+      "grad_norm": 0.42200708389282227,
+      "learning_rate": 1.2570375807719576e-05,
+      "loss": 0.0305,
+      "step": 340
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.5258976817131042,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0455,
+      "step": 341
+    },
+    {
+      "epoch": 2.206451612903226,
+      "grad_norm": 0.495807945728302,
+      "learning_rate": 1.242962419228043e-05,
+      "loss": 0.0323,
+      "step": 342
+    },
+    {
+      "epoch": 2.2129032258064516,
+      "grad_norm": 0.5464356541633606,
+      "learning_rate": 1.2359250615324678e-05,
+      "loss": 0.0325,
+      "step": 343
+    },
+    {
+      "epoch": 2.2193548387096773,
+      "grad_norm": 0.5555934906005859,
+      "learning_rate": 1.2288881499825863e-05,
+      "loss": 0.0504,
+      "step": 344
+    },
+    {
+      "epoch": 2.225806451612903,
+      "grad_norm": 0.41927701234817505,
+      "learning_rate": 1.2218519076335677e-05,
+      "loss": 0.0288,
+      "step": 345
+    },
+    {
+      "epoch": 2.232258064516129,
+      "grad_norm": 0.5449569821357727,
+      "learning_rate": 1.2148165575193685e-05,
+      "loss": 0.0328,
+      "step": 346
+    },
+    {
+      "epoch": 2.238709677419355,
+      "grad_norm": 0.4198172688484192,
+      "learning_rate": 1.2077823226456648e-05,
+      "loss": 0.0284,
+      "step": 347
+    },
+    {
+      "epoch": 2.2451612903225806,
+      "grad_norm": 0.5396814346313477,
+      "learning_rate": 1.2007494259827809e-05,
+      "loss": 0.0379,
+      "step": 348
+    },
+    {
+      "epoch": 2.2516129032258063,
+      "grad_norm": 0.4842919409275055,
+      "learning_rate": 1.1937180904586215e-05,
+      "loss": 0.0316,
+      "step": 349
+    },
+    {
+      "epoch": 2.258064516129032,
+      "grad_norm": 0.5152572989463806,
+      "learning_rate": 1.1866885389516092e-05,
+      "loss": 0.0321,
+      "step": 350
+    },
+    {
+      "epoch": 2.264516129032258,
+      "grad_norm": 0.556614875793457,
+      "learning_rate": 1.179660994283615e-05,
+      "loss": 0.0372,
+      "step": 351
+    },
+    {
+      "epoch": 2.270967741935484,
+      "grad_norm": 0.5159235000610352,
+      "learning_rate": 1.1726356792128978e-05,
+      "loss": 0.0328,
+      "step": 352
+    },
+    {
+      "epoch": 2.2774193548387096,
+      "grad_norm": 0.5564429759979248,
+      "learning_rate": 1.1656128164270436e-05,
+      "loss": 0.0304,
+      "step": 353
+    },
+    {
+      "epoch": 2.2838709677419353,
+      "grad_norm": 0.6227903366088867,
+      "learning_rate": 1.1585926285359049e-05,
+      "loss": 0.0321,
+      "step": 354
+    },
+    {
+      "epoch": 2.2903225806451615,
+      "grad_norm": 0.5218878388404846,
+      "learning_rate": 1.1515753380645479e-05,
+      "loss": 0.0358,
+      "step": 355
+    },
+    {
+      "epoch": 2.296774193548387,
+      "grad_norm": 0.49731266498565674,
+      "learning_rate": 1.1445611674461942e-05,
+      "loss": 0.0331,
+      "step": 356
+    },
+    {
+      "epoch": 2.303225806451613,
+      "grad_norm": 0.5095941424369812,
+      "learning_rate": 1.1375503390151737e-05,
+      "loss": 0.0315,
+      "step": 357
+    },
+    {
+      "epoch": 2.3096774193548386,
+      "grad_norm": 0.4576358199119568,
+      "learning_rate": 1.1305430749998775e-05,
+      "loss": 0.0304,
+      "step": 358
+    },
+    {
+      "epoch": 2.3161290322580643,
+      "grad_norm": 0.5103798508644104,
+      "learning_rate": 1.12353959751571e-05,
+      "loss": 0.0281,
+      "step": 359
+    },
+    {
+      "epoch": 2.3225806451612905,
+      "grad_norm": 0.5072308778762817,
+      "learning_rate": 1.1165401285580515e-05,
+      "loss": 0.0267,
+      "step": 360
+    },
+    {
+      "epoch": 2.329032258064516,
+      "grad_norm": 0.45558691024780273,
+      "learning_rate": 1.1095448899952212e-05,
+      "loss": 0.0302,
+      "step": 361
+    },
+    {
+      "epoch": 2.335483870967742,
+      "grad_norm": 0.4773171842098236,
+      "learning_rate": 1.1025541035614427e-05,
+      "loss": 0.0307,
+      "step": 362
+    },
+    {
+      "epoch": 2.3419354838709676,
+      "grad_norm": 0.4630301892757416,
+      "learning_rate": 1.0955679908498171e-05,
+      "loss": 0.0292,
+      "step": 363
+    },
+    {
+      "epoch": 2.3483870967741938,
+      "grad_norm": 0.5814460515975952,
+      "learning_rate": 1.0885867733052985e-05,
+      "loss": 0.034,
+      "step": 364
+    },
+    {
+      "epoch": 2.3548387096774195,
+      "grad_norm": 0.3135308623313904,
+      "learning_rate": 1.0816106722176741e-05,
+      "loss": 0.0264,
+      "step": 365
+    },
+    {
+      "epoch": 2.361290322580645,
+      "grad_norm": 0.4219888150691986,
+      "learning_rate": 1.0746399087145504e-05,
+      "loss": 0.0304,
+      "step": 366
+    },
+    {
+      "epoch": 2.367741935483871,
+      "grad_norm": 0.4246158003807068,
+      "learning_rate": 1.0676747037543447e-05,
+      "loss": 0.032,
+      "step": 367
+    },
+    {
+      "epoch": 2.3741935483870966,
+      "grad_norm": 0.4565359950065613,
+      "learning_rate": 1.0607152781192796e-05,
+      "loss": 0.0326,
+      "step": 368
+    },
+    {
+      "epoch": 2.3806451612903228,
+      "grad_norm": 0.4495943486690521,
+      "learning_rate": 1.053761852408384e-05,
+      "loss": 0.0307,
+      "step": 369
+    },
+    {
+      "epoch": 2.3870967741935485,
+      "grad_norm": 0.47505924105644226,
+      "learning_rate": 1.0468146470305047e-05,
+      "loss": 0.0366,
+      "step": 370
+    },
+    {
+      "epoch": 2.393548387096774,
+      "grad_norm": 0.41802337765693665,
+      "learning_rate": 1.039873882197315e-05,
+      "loss": 0.0242,
+      "step": 371
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.4308302104473114,
+      "learning_rate": 1.0329397779163372e-05,
+      "loss": 0.0303,
+      "step": 372
+    },
+    {
+      "epoch": 2.4064516129032256,
+      "grad_norm": 0.3776704967021942,
+      "learning_rate": 1.0260125539839686e-05,
+      "loss": 0.0224,
+      "step": 373
+    },
+    {
+      "epoch": 2.412903225806452,
+      "grad_norm": 0.3952430188655853,
+      "learning_rate": 1.0190924299785138e-05,
+      "loss": 0.0236,
+      "step": 374
+    },
+    {
+      "epoch": 2.4193548387096775,
+      "grad_norm": 0.5212628841400146,
+      "learning_rate": 1.0121796252532237e-05,
+      "loss": 0.0352,
+      "step": 375
+    },
+    {
+      "epoch": 2.425806451612903,
+      "grad_norm": 0.5264010429382324,
+      "learning_rate": 1.0052743589293463e-05,
+      "loss": 0.0366,
+      "step": 376
+    },
+    {
+      "epoch": 2.432258064516129,
+      "grad_norm": 0.42148974537849426,
+      "learning_rate": 9.983768498891747e-06,
+      "loss": 0.0281,
+      "step": 377
+    },
+    {
+      "epoch": 2.4387096774193546,
+      "grad_norm": 0.4387865960597992,
+      "learning_rate": 9.91487316769116e-06,
+      "loss": 0.0321,
+      "step": 378
+    },
+    {
+      "epoch": 2.445161290322581,
+      "grad_norm": 0.4530801475048065,
+      "learning_rate": 9.846059779527552e-06,
+      "loss": 0.03,
+      "step": 379
+    },
+    {
+      "epoch": 2.4516129032258065,
+      "grad_norm": 0.44786474108695984,
+      "learning_rate": 9.777330515639356e-06,
+      "loss": 0.0312,
+      "step": 380
+    },
+    {
+      "epoch": 2.458064516129032,
+      "grad_norm": 0.42808324098587036,
+      "learning_rate": 9.708687554598454e-06,
+      "loss": 0.0321,
+      "step": 381
+    },
+    {
+      "epoch": 2.464516129032258,
+      "grad_norm": 0.4658293128013611,
+      "learning_rate": 9.640133072241105e-06,
+      "loss": 0.0335,
+      "step": 382
+    },
+    {
+      "epoch": 2.4709677419354836,
+      "grad_norm": 0.45854416489601135,
+      "learning_rate": 9.571669241598974e-06,
+      "loss": 0.0306,
+      "step": 383
+    },
+    {
+      "epoch": 2.47741935483871,
+      "grad_norm": 0.5602400302886963,
+      "learning_rate": 9.503298232830274e-06,
+      "loss": 0.0425,
+      "step": 384
+    },
+    {
+      "epoch": 2.4838709677419355,
+      "grad_norm": 0.4135296940803528,
+      "learning_rate": 9.43502221315095e-06,
+      "loss": 0.0317,
+      "step": 385
+    },
+    {
+      "epoch": 2.490322580645161,
+      "grad_norm": 0.6756112575531006,
+      "learning_rate": 9.366843346765992e-06,
+      "loss": 0.0492,
+      "step": 386
+    },
+    {
+      "epoch": 2.496774193548387,
+      "grad_norm": 0.6048617362976074,
+      "learning_rate": 9.298763794800856e-06,
+      "loss": 0.0326,
+      "step": 387
+    },
+    {
+      "epoch": 2.5032258064516126,
+      "grad_norm": 0.3737858831882477,
+      "learning_rate": 9.230785715232917e-06,
+      "loss": 0.0226,
+      "step": 388
+    },
+    {
+      "epoch": 2.509677419354839,
+      "grad_norm": 0.49958306550979614,
+      "learning_rate": 9.162911262823104e-06,
+      "loss": 0.0293,
+      "step": 389
+    },
+    {
+      "epoch": 2.5161290322580645,
+      "grad_norm": 0.4132345914840698,
+      "learning_rate": 9.095142589047586e-06,
+      "loss": 0.0268,
+      "step": 390
+    },
+    {
+      "epoch": 2.52258064516129,
+      "grad_norm": 0.5339500308036804,
+      "learning_rate": 9.027481842029567e-06,
+      "loss": 0.0308,
+      "step": 391
+    },
+    {
+      "epoch": 2.5290322580645164,
+      "grad_norm": 0.5680338740348816,
+      "learning_rate": 8.9599311664712e-06,
+      "loss": 0.026,
+      "step": 392
+    },
+    {
+      "epoch": 2.535483870967742,
+      "grad_norm": 0.4945621192455292,
+      "learning_rate": 8.89249270358562e-06,
+      "loss": 0.0414,
+      "step": 393
+    },
+    {
+      "epoch": 2.541935483870968,
+      "grad_norm": 0.478188157081604,
+      "learning_rate": 8.825168591029042e-06,
+      "loss": 0.0325,
+      "step": 394
+    },
+    {
+      "epoch": 2.5483870967741935,
+      "grad_norm": 0.41539856791496277,
+      "learning_rate": 8.757960962833026e-06,
+      "loss": 0.0276,
+      "step": 395
+    },
+    {
+      "epoch": 2.554838709677419,
+      "grad_norm": 0.41548025608062744,
+      "learning_rate": 8.69087194933683e-06,
+      "loss": 0.0258,
+      "step": 396
+    },
+    {
+      "epoch": 2.5612903225806454,
+      "grad_norm": 0.7209835052490234,
+      "learning_rate": 8.623903677119866e-06,
+      "loss": 0.0275,
+      "step": 397
+    },
+    {
+      "epoch": 2.567741935483871,
+      "grad_norm": 0.45113834738731384,
+      "learning_rate": 8.557058268934306e-06,
+      "loss": 0.0276,
+      "step": 398
+    },
+    {
+      "epoch": 2.574193548387097,
+      "grad_norm": 0.4919924736022949,
+      "learning_rate": 8.490337843637807e-06,
+      "loss": 0.0352,
+      "step": 399
+    },
+    {
+      "epoch": 2.5806451612903225,
+      "grad_norm": 0.4441167414188385,
+      "learning_rate": 8.423744516126313e-06,
+      "loss": 0.0312,
+      "step": 400
+    },
+    {
+      "epoch": 2.587096774193548,
+      "grad_norm": 0.3870048522949219,
+      "learning_rate": 8.357280397267054e-06,
+      "loss": 0.0273,
+      "step": 401
+    },
+    {
+      "epoch": 2.5935483870967744,
+      "grad_norm": 0.4747593104839325,
+      "learning_rate": 8.29094759383161e-06,
+      "loss": 0.0428,
+      "step": 402
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.3494237959384918,
+      "learning_rate": 8.224748208429142e-06,
+      "loss": 0.0249,
+      "step": 403
+    },
+    {
+      "epoch": 2.606451612903226,
+      "grad_norm": 0.3618505001068115,
+      "learning_rate": 8.158684339439748e-06,
+      "loss": 0.0221,
+      "step": 404
+    },
+    {
+      "epoch": 2.6129032258064515,
+      "grad_norm": 0.45744070410728455,
+      "learning_rate": 8.092758080947939e-06,
+      "loss": 0.0303,
+      "step": 405
+    },
+    {
+      "epoch": 2.6193548387096772,
+      "grad_norm": 0.3921363055706024,
+      "learning_rate": 8.02697152267626e-06,
+      "loss": 0.0267,
+      "step": 406
+    },
+    {
+      "epoch": 2.6258064516129034,
+      "grad_norm": 0.5149343013763428,
+      "learning_rate": 7.961326749919069e-06,
+      "loss": 0.0347,
+      "step": 407
+    },
+    {
+      "epoch": 2.632258064516129,
+      "grad_norm": 0.5246243476867676,
+      "learning_rate": 7.895825843476412e-06,
+      "loss": 0.0318,
+      "step": 408
+    },
+    {
+      "epoch": 2.638709677419355,
+      "grad_norm": 0.5338672995567322,
+      "learning_rate": 7.83047087958808e-06,
+      "loss": 0.0331,
+      "step": 409
+    },
+    {
+      "epoch": 2.6451612903225805,
+      "grad_norm": 0.4028920531272888,
+      "learning_rate": 7.7652639298678e-06,
+      "loss": 0.0251,
+      "step": 410
+    },
+    {
+      "epoch": 2.6516129032258062,
+      "grad_norm": 0.3391985297203064,
+      "learning_rate": 7.70020706123756e-06,
+      "loss": 0.0206,
+      "step": 411
+    },
+    {
+      "epoch": 2.6580645161290324,
+      "grad_norm": 0.4651046097278595,
+      "learning_rate": 7.635302335862094e-06,
+      "loss": 0.0242,
+      "step": 412
+    },
+    {
+      "epoch": 2.664516129032258,
+      "grad_norm": 0.4581477642059326,
+      "learning_rate": 7.570551811083521e-06,
+      "loss": 0.0334,
+      "step": 413
+    },
+    {
+      "epoch": 2.670967741935484,
+      "grad_norm": 0.629748523235321,
+      "learning_rate": 7.505957539356126e-06,
+      "loss": 0.0426,
+      "step": 414
+    },
+    {
+      "epoch": 2.6774193548387095,
+      "grad_norm": 0.44972798228263855,
+      "learning_rate": 7.441521568181299e-06,
+      "loss": 0.0279,
+      "step": 415
+    },
+    {
+      "epoch": 2.6838709677419352,
+      "grad_norm": 0.49497148394584656,
+      "learning_rate": 7.37724594004266e-06,
+      "loss": 0.0331,
+      "step": 416
+    },
+    {
+      "epoch": 2.6903225806451614,
+      "grad_norm": 0.4186260998249054,
+      "learning_rate": 7.313132692341263e-06,
+      "loss": 0.0294,
+      "step": 417
+    },
+    {
+      "epoch": 2.696774193548387,
+      "grad_norm": 0.4715961813926697,
+      "learning_rate": 7.249183857331064e-06,
+      "loss": 0.0293,
+      "step": 418
+    },
+    {
+      "epoch": 2.703225806451613,
+      "grad_norm": 0.48064178228378296,
+      "learning_rate": 7.185401462054495e-06,
+      "loss": 0.0312,
+      "step": 419
+    },
+    {
+      "epoch": 2.709677419354839,
+      "grad_norm": 0.4826470613479614,
+      "learning_rate": 7.121787528278177e-06,
+      "loss": 0.0303,
+      "step": 420
+    },
+    {
+      "epoch": 2.7161290322580647,
+      "grad_norm": 0.39333951473236084,
+      "learning_rate": 7.058344072428877e-06,
+      "loss": 0.0211,
+      "step": 421
+    },
+    {
+      "epoch": 2.7225806451612904,
+      "grad_norm": 0.3964556157588959,
+      "learning_rate": 6.99507310552957e-06,
+      "loss": 0.031,
+      "step": 422
+    },
+    {
+      "epoch": 2.729032258064516,
+      "grad_norm": 0.5450259447097778,
+      "learning_rate": 6.931976633135695e-06,
+      "loss": 0.0344,
+      "step": 423
+    },
+    {
+      "epoch": 2.735483870967742,
+      "grad_norm": 0.4331640601158142,
+      "learning_rate": 6.869056655271588e-06,
+      "loss": 0.0261,
+      "step": 424
+    },
+    {
+      "epoch": 2.741935483870968,
+      "grad_norm": 0.46446603536605835,
+      "learning_rate": 6.806315166367075e-06,
+      "loss": 0.0311,
+      "step": 425
+    },
+    {
+      "epoch": 2.7483870967741937,
+      "grad_norm": 0.5200790166854858,
+      "learning_rate": 6.743754155194268e-06,
+      "loss": 0.0292,
+      "step": 426
+    },
+    {
+      "epoch": 2.7548387096774194,
+      "grad_norm": 0.6154363751411438,
+      "learning_rate": 6.681375604804521e-06,
+      "loss": 0.0252,
+      "step": 427
+    },
+    {
+      "epoch": 2.761290322580645,
+      "grad_norm": 0.43054288625717163,
+      "learning_rate": 6.619181492465557e-06,
+      "loss": 0.0225,
+      "step": 428
+    },
+    {
+      "epoch": 2.767741935483871,
+      "grad_norm": 0.5042747259140015,
+      "learning_rate": 6.55717378959881e-06,
+      "loss": 0.0266,
+      "step": 429
+    },
+    {
+      "epoch": 2.774193548387097,
+      "grad_norm": 0.5118414759635925,
+      "learning_rate": 6.4953544617169376e-06,
+      "loss": 0.0324,
+      "step": 430
+    },
+    {
+      "epoch": 2.7806451612903227,
+      "grad_norm": 0.4071415364742279,
+      "learning_rate": 6.43372546836149e-06,
+      "loss": 0.0306,
+      "step": 431
+    },
+    {
+      "epoch": 2.7870967741935484,
+      "grad_norm": 0.43498843908309937,
+      "learning_rate": 6.372288763040833e-06,
+      "loss": 0.0267,
+      "step": 432
+    },
+    {
+      "epoch": 2.793548387096774,
+      "grad_norm": 0.6407294273376465,
+      "learning_rate": 6.3110462931682075e-06,
+      "loss": 0.0529,
+      "step": 433
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.4058496356010437,
+      "learning_rate": 6.250000000000003e-06,
+      "loss": 0.0289,
+      "step": 434
+    },
+    {
+      "epoch": 2.806451612903226,
+      "grad_norm": 0.34818795323371887,
+      "learning_rate": 6.1891518185742116e-06,
+      "loss": 0.0236,
+      "step": 435
+    },
+    {
+      "epoch": 2.8129032258064517,
+      "grad_norm": 0.4517665505409241,
+      "learning_rate": 6.1285036776491165e-06,
+      "loss": 0.0341,
+      "step": 436
+    },
+    {
+      "epoch": 2.8193548387096774,
+      "grad_norm": 0.5423181056976318,
+      "learning_rate": 6.068057499642144e-06,
+      "loss": 0.0406,
+      "step": 437
+    },
+    {
+      "epoch": 2.825806451612903,
+      "grad_norm": 0.4574117362499237,
+      "learning_rate": 6.007815200568906e-06,
+      "loss": 0.0344,
+      "step": 438
+    },
+    {
+      "epoch": 2.832258064516129,
+      "grad_norm": 0.4028095006942749,
+      "learning_rate": 5.9477786899825024e-06,
+      "loss": 0.026,
+      "step": 439
+    },
+    {
+      "epoch": 2.838709677419355,
+      "grad_norm": 0.4277281165122986,
+      "learning_rate": 5.8879498709129735e-06,
+      "loss": 0.032,
+      "step": 440
+    },
+    {
+      "epoch": 2.8451612903225807,
+      "grad_norm": 0.4217607080936432,
+      "learning_rate": 5.82833063980696e-06,
+      "loss": 0.0275,
+      "step": 441
+    },
+    {
+      "epoch": 2.8516129032258064,
+      "grad_norm": 0.4865557849407196,
+      "learning_rate": 5.7689228864676394e-06,
+      "loss": 0.0344,
+      "step": 442
+    },
+    {
+      "epoch": 2.858064516129032,
+      "grad_norm": 0.44111689925193787,
+      "learning_rate": 5.70972849399477e-06,
+      "loss": 0.028,
+      "step": 443
+    },
+    {
+      "epoch": 2.864516129032258,
+      "grad_norm": 0.5612359046936035,
+      "learning_rate": 5.650749338725019e-06,
+      "loss": 0.0433,
+      "step": 444
+    },
+    {
+      "epoch": 2.870967741935484,
+      "grad_norm": 0.42652663588523865,
+      "learning_rate": 5.591987290172518e-06,
+      "loss": 0.0266,
+      "step": 445
+    },
+    {
+      "epoch": 2.8774193548387097,
+      "grad_norm": 0.43139341473579407,
+      "learning_rate": 5.533444210969546e-06,
+      "loss": 0.0228,
+      "step": 446
+    },
+    {
+      "epoch": 2.8838709677419354,
+      "grad_norm": 0.3348155915737152,
+      "learning_rate": 5.475121956807537e-06,
+      "loss": 0.0236,
+      "step": 447
+    },
+    {
+      "epoch": 2.8903225806451616,
+      "grad_norm": 0.41803082823753357,
+      "learning_rate": 5.417022376378239e-06,
+      "loss": 0.0275,
+      "step": 448
+    },
+    {
+      "epoch": 2.896774193548387,
+      "grad_norm": 0.4131038784980774,
+      "learning_rate": 5.359147311315094e-06,
+      "loss": 0.0265,
+      "step": 449
+    },
+    {
+      "epoch": 2.903225806451613,
+      "grad_norm": 0.5227479934692383,
+      "learning_rate": 5.30149859613492e-06,
+      "loss": 0.025,
+      "step": 450
+    },
+    {
+      "epoch": 2.9096774193548387,
+      "grad_norm": 0.43142953515052795,
+      "learning_rate": 5.244078058179691e-06,
+      "loss": 0.0249,
+      "step": 451
+    },
+    {
+      "epoch": 2.9161290322580644,
+      "grad_norm": 0.4158158600330353,
+      "learning_rate": 5.186887517558653e-06,
+      "loss": 0.0297,
+      "step": 452
+    },
+    {
+      "epoch": 2.9225806451612906,
+      "grad_norm": 0.3552153706550598,
+      "learning_rate": 5.129928787090646e-06,
+      "loss": 0.0234,
+      "step": 453
+    },
+    {
+      "epoch": 2.9290322580645163,
+      "grad_norm": 0.49204781651496887,
+      "learning_rate": 5.073203672246593e-06,
+      "loss": 0.0379,
+      "step": 454
+    },
+    {
+      "epoch": 2.935483870967742,
+      "grad_norm": 0.38140571117401123,
+      "learning_rate": 5.016713971092311e-06,
+      "loss": 0.0294,
+      "step": 455
+    },
+    {
+      "epoch": 2.9419354838709677,
+      "grad_norm": 0.5261517763137817,
+      "learning_rate": 4.960461474231505e-06,
+      "loss": 0.0305,
+      "step": 456
+    },
+    {
+      "epoch": 2.9483870967741934,
+      "grad_norm": 0.6391315460205078,
+      "learning_rate": 4.904447964748993e-06,
+      "loss": 0.038,
+      "step": 457
+    },
+    {
+      "epoch": 2.9548387096774196,
+      "grad_norm": 0.3812016546726227,
+      "learning_rate": 4.848675218154214e-06,
+      "loss": 0.0259,
+      "step": 458
+    },
+    {
+      "epoch": 2.9612903225806453,
+      "grad_norm": 0.4748527407646179,
+      "learning_rate": 4.793145002324933e-06,
+      "loss": 0.0329,
+      "step": 459
+    },
+    {
+      "epoch": 2.967741935483871,
+      "grad_norm": 0.4919755458831787,
+      "learning_rate": 4.737859077451191e-06,
+      "loss": 0.0253,
+      "step": 460
+    },
+    {
+      "epoch": 2.9741935483870967,
+      "grad_norm": 0.4986102879047394,
+      "learning_rate": 4.68281919597954e-06,
+      "loss": 0.0293,
+      "step": 461
+    },
+    {
+      "epoch": 2.9806451612903224,
+      "grad_norm": 0.48589223623275757,
+      "learning_rate": 4.6280271025574695e-06,
+      "loss": 0.0287,
+      "step": 462
+    },
+    {
+      "epoch": 2.9870967741935486,
+      "grad_norm": 0.4930824935436249,
+      "learning_rate": 4.573484533978119e-06,
+      "loss": 0.0258,
+      "step": 463
+    },
+    {
+      "epoch": 2.9935483870967743,
+      "grad_norm": 0.38358667492866516,
+      "learning_rate": 4.5191932191252075e-06,
+      "loss": 0.0235,
+      "step": 464
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.5300599336624146,
+      "learning_rate": 4.465154878918258e-06,
+      "loss": 0.0309,
+      "step": 465
+    },
+    {
+      "epoch": 3.0064516129032257,
+      "grad_norm": 0.24884271621704102,
+      "learning_rate": 4.411371226258032e-06,
+      "loss": 0.0182,
+      "step": 466
+    },
+    {
+      "epoch": 3.0129032258064514,
+      "grad_norm": 0.3061859607696533,
+      "learning_rate": 4.3578439659722246e-06,
+      "loss": 0.0161,
+      "step": 467
+    },
+    {
+      "epoch": 3.0193548387096776,
+      "grad_norm": 0.3014248311519623,
+      "learning_rate": 4.304574794761447e-06,
+      "loss": 0.0135,
+      "step": 468
+    },
+    {
+      "epoch": 3.0258064516129033,
+      "grad_norm": 0.31640511751174927,
+      "learning_rate": 4.251565401145432e-06,
+      "loss": 0.0168,
+      "step": 469
+    },
+    {
+      "epoch": 3.032258064516129,
+      "grad_norm": 0.25447705388069153,
+      "learning_rate": 4.1988174654095104e-06,
+      "loss": 0.0117,
+      "step": 470
+    },
+    {
+      "epoch": 3.0387096774193547,
+      "grad_norm": 0.31153520941734314,
+      "learning_rate": 4.146332659551364e-06,
+      "loss": 0.0147,
+      "step": 471
+    },
+    {
+      "epoch": 3.0451612903225804,
+      "grad_norm": 0.32956749200820923,
+      "learning_rate": 4.094112647227996e-06,
+      "loss": 0.015,
+      "step": 472
+    },
+    {
+      "epoch": 3.0516129032258066,
+      "grad_norm": 0.3105918765068054,
+      "learning_rate": 4.042159083703031e-06,
+      "loss": 0.0136,
+      "step": 473
+    },
+    {
+      "epoch": 3.0580645161290323,
+      "grad_norm": 0.3890332281589508,
+      "learning_rate": 3.9904736157942355e-06,
+      "loss": 0.0128,
+      "step": 474
+    },
+    {
+      "epoch": 3.064516129032258,
+      "grad_norm": 0.2500901222229004,
+      "learning_rate": 3.939057881821295e-06,
+      "loss": 0.0109,
+      "step": 475
+    },
+    {
+      "epoch": 3.0709677419354837,
+      "grad_norm": 0.33506497740745544,
+      "learning_rate": 3.887913511553917e-06,
+      "loss": 0.0138,
+      "step": 476
+    },
+    {
+      "epoch": 3.07741935483871,
+      "grad_norm": 0.35200193524360657,
+      "learning_rate": 3.837042126160157e-06,
+      "loss": 0.0163,
+      "step": 477
+    },
+    {
+      "epoch": 3.0838709677419356,
+      "grad_norm": 0.33882763981819153,
+      "learning_rate": 3.786445338155013e-06,
+      "loss": 0.0125,
+      "step": 478
+    },
+    {
+      "epoch": 3.0903225806451613,
+      "grad_norm": 0.4239828586578369,
+      "learning_rate": 3.736124751349343e-06,
+      "loss": 0.0146,
+      "step": 479
+    },
+    {
+      "epoch": 3.096774193548387,
+      "grad_norm": 0.39569729566574097,
+      "learning_rate": 3.6860819607990108e-06,
+      "loss": 0.0149,
+      "step": 480
+    },
+    {
+      "epoch": 3.1032258064516127,
+      "grad_norm": 0.4424724578857422,
+      "learning_rate": 3.6363185527543156e-06,
+      "loss": 0.0147,
+      "step": 481
+    },
+    {
+      "epoch": 3.109677419354839,
+      "grad_norm": 0.54300457239151,
+      "learning_rate": 3.5868361046097475e-06,
+      "loss": 0.0166,
+      "step": 482
+    },
+    {
+      "epoch": 3.1161290322580646,
+      "grad_norm": 0.41813674569129944,
+      "learning_rate": 3.537636184853939e-06,
+      "loss": 0.0131,
+      "step": 483
+    },
+    {
+      "epoch": 3.1225806451612903,
+      "grad_norm": 0.4122736155986786,
+      "learning_rate": 3.4887203530199864e-06,
+      "loss": 0.0141,
+      "step": 484
+    },
+    {
+      "epoch": 3.129032258064516,
+      "grad_norm": 0.39058443903923035,
+      "learning_rate": 3.440090159636003e-06,
+      "loss": 0.0115,
+      "step": 485
+    },
+    {
+      "epoch": 3.135483870967742,
+      "grad_norm": 0.42365285754203796,
+      "learning_rate": 3.391747146175954e-06,
+      "loss": 0.0097,
+      "step": 486
+    },
+    {
+      "epoch": 3.141935483870968,
+      "grad_norm": 0.5885961055755615,
+      "learning_rate": 3.3436928450108264e-06,
+      "loss": 0.0219,
+      "step": 487
+    },
+    {
+      "epoch": 3.1483870967741936,
+      "grad_norm": 0.39843595027923584,
+      "learning_rate": 3.2959287793600356e-06,
+      "loss": 0.0098,
+      "step": 488
+    },
+    {
+      "epoch": 3.1548387096774193,
+      "grad_norm": 0.27729499340057373,
+      "learning_rate": 3.2484564632431396e-06,
+      "loss": 0.0068,
+      "step": 489
+    },
+    {
+      "epoch": 3.161290322580645,
+      "grad_norm": 0.4415301978588104,
+      "learning_rate": 3.2012774014318625e-06,
+      "loss": 0.0151,
+      "step": 490
+    },
+    {
+      "epoch": 3.167741935483871,
+      "grad_norm": 0.40353745222091675,
+      "learning_rate": 3.154393089402391e-06,
+      "loss": 0.016,
+      "step": 491
+    },
+    {
+      "epoch": 3.174193548387097,
+      "grad_norm": 0.4263345003128052,
+      "learning_rate": 3.107805013287958e-06,
+      "loss": 0.0112,
+      "step": 492
+    },
+    {
+      "epoch": 3.1806451612903226,
+      "grad_norm": 0.37084028124809265,
+      "learning_rate": 3.061514649831755e-06,
+      "loss": 0.014,
+      "step": 493
+    },
+    {
+      "epoch": 3.1870967741935483,
+      "grad_norm": 0.5308308005332947,
+      "learning_rate": 3.0155234663401146e-06,
+      "loss": 0.0146,
+      "step": 494
+    },
+    {
+      "epoch": 3.193548387096774,
+      "grad_norm": 0.47034651041030884,
+      "learning_rate": 2.9698329206359925e-06,
+      "loss": 0.0124,
+      "step": 495
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 0.3541916012763977,
+      "learning_rate": 2.9244444610127764e-06,
+      "loss": 0.0121,
+      "step": 496
+    },
+    {
+      "epoch": 3.206451612903226,
+      "grad_norm": 0.4616714417934418,
+      "learning_rate": 2.8793595261883465e-06,
+      "loss": 0.0181,
+      "step": 497
+    },
+    {
+      "epoch": 3.2129032258064516,
+      "grad_norm": 0.36847707629203796,
+      "learning_rate": 2.8345795452595095e-06,
+      "loss": 0.0165,
+      "step": 498
+    },
+    {
+      "epoch": 3.2193548387096773,
+      "grad_norm": 0.3731675148010254,
+      "learning_rate": 2.790105937656673e-06,
+      "loss": 0.013,
+      "step": 499
+    },
+    {
+      "epoch": 3.225806451612903,
+      "grad_norm": 0.42840346693992615,
+      "learning_rate": 2.7459401130988534e-06,
+      "loss": 0.0109,
+      "step": 500
+    },
+    {
+      "epoch": 3.232258064516129,
+      "grad_norm": 0.33602991700172424,
+      "learning_rate": 2.7020834715490093e-06,
+      "loss": 0.0106,
+      "step": 501
+    },
+    {
+      "epoch": 3.238709677419355,
+      "grad_norm": 0.47426825761795044,
+      "learning_rate": 2.6585374031696474e-06,
+      "loss": 0.0133,
+      "step": 502
+    },
+    {
+      "epoch": 3.2451612903225806,
+      "grad_norm": 0.41371604800224304,
+      "learning_rate": 2.61530328827877e-06,
+      "loss": 0.0094,
+      "step": 503
+    },
+    {
+      "epoch": 3.2516129032258063,
+      "grad_norm": 0.4533410668373108,
+      "learning_rate": 2.5723824973061e-06,
+      "loss": 0.0123,
+      "step": 504
+    },
+    {
+      "epoch": 3.258064516129032,
+      "grad_norm": 0.2636722922325134,
+      "learning_rate": 2.5297763907496746e-06,
+      "loss": 0.0086,
+      "step": 505
+    },
+    {
+      "epoch": 3.264516129032258,
+      "grad_norm": 0.48444676399230957,
+      "learning_rate": 2.4874863191326953e-06,
+      "loss": 0.0169,
+      "step": 506
+    },
+    {
+      "epoch": 3.270967741935484,
+      "grad_norm": 0.5979859828948975,
+      "learning_rate": 2.44551362296072e-06,
+      "loss": 0.0112,
+      "step": 507
+    },
+    {
+      "epoch": 3.2774193548387096,
+      "grad_norm": 0.43151959776878357,
+      "learning_rate": 2.4038596326791884e-06,
+      "loss": 0.0109,
+      "step": 508
+    },
+    {
+      "epoch": 3.2838709677419353,
+      "grad_norm": 0.4825892746448517,
+      "learning_rate": 2.362525668631238e-06,
+      "loss": 0.013,
+      "step": 509
+    },
+    {
+      "epoch": 3.2903225806451615,
+      "grad_norm": 0.3168151080608368,
+      "learning_rate": 2.3215130410158424e-06,
+      "loss": 0.0106,
+      "step": 510
+    },
+    {
+      "epoch": 3.296774193548387,
+      "grad_norm": 0.4605632722377777,
+      "learning_rate": 2.2808230498463116e-06,
+      "loss": 0.0189,
+      "step": 511
+    },
+    {
+      "epoch": 3.303225806451613,
+      "grad_norm": 0.47640544176101685,
+      "learning_rate": 2.240456984909049e-06,
+      "loss": 0.015,
+      "step": 512
+    },
+    {
+      "epoch": 3.3096774193548386,
+      "grad_norm": 0.5328596234321594,
+      "learning_rate": 2.2004161257226805e-06,
+      "loss": 0.0201,
+      "step": 513
+    },
+    {
+      "epoch": 3.3161290322580643,
+      "grad_norm": 0.5342445969581604,
+      "learning_rate": 2.16070174149752e-06,
+      "loss": 0.0109,
+      "step": 514
+    },
+    {
+      "epoch": 3.3225806451612905,
+      "grad_norm": 0.5308839082717896,
+      "learning_rate": 2.121315091095297e-06,
+      "loss": 0.014,
+      "step": 515
+    },
+    {
+      "epoch": 3.329032258064516,
+      "grad_norm": 0.4669474959373474,
+      "learning_rate": 2.082257422989281e-06,
+      "loss": 0.0105,
+      "step": 516
+    },
+    {
+      "epoch": 3.335483870967742,
+      "grad_norm": 0.37382492423057556,
+      "learning_rate": 2.0435299752247077e-06,
+      "loss": 0.0132,
+      "step": 517
+    },
+    {
+      "epoch": 3.3419354838709676,
+      "grad_norm": 0.4566926062107086,
+      "learning_rate": 2.0051339753795125e-06,
+      "loss": 0.0159,
+      "step": 518
+    },
+    {
+      "epoch": 3.3483870967741938,
+      "grad_norm": 0.4399929642677307,
+      "learning_rate": 1.9670706405254548e-06,
+      "loss": 0.0149,
+      "step": 519
+    },
+    {
+      "epoch": 3.3548387096774195,
+      "grad_norm": 0.3071390986442566,
+      "learning_rate": 1.929341177189506e-06,
+      "loss": 0.0085,
+      "step": 520
+    },
+    {
+      "epoch": 3.361290322580645,
+      "grad_norm": 0.39541929960250854,
+      "learning_rate": 1.8919467813156121e-06,
+      "loss": 0.0088,
+      "step": 521
+    },
+    {
+      "epoch": 3.367741935483871,
+      "grad_norm": 0.49959710240364075,
+      "learning_rate": 1.854888638226815e-06,
+      "loss": 0.0147,
+      "step": 522
+    },
+    {
+      "epoch": 3.3741935483870966,
+      "grad_norm": 0.3740963339805603,
+      "learning_rate": 1.8181679225876324e-06,
+      "loss": 0.0099,
+      "step": 523
+    },
+    {
+      "epoch": 3.3806451612903228,
+      "grad_norm": 0.27066710591316223,
+      "learning_rate": 1.7817857983668612e-06,
+      "loss": 0.0071,
+      "step": 524
+    },
+    {
+      "epoch": 3.3870967741935485,
+      "grad_norm": 0.47990116477012634,
+      "learning_rate": 1.745743418800669e-06,
+      "loss": 0.012,
+      "step": 525
+    },
+    {
+      "epoch": 3.393548387096774,
+      "grad_norm": 0.40311211347579956,
+      "learning_rate": 1.7100419263560263e-06,
+      "loss": 0.0102,
+      "step": 526
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 0.4134123623371124,
+      "learning_rate": 1.6746824526945163e-06,
+      "loss": 0.0123,
+      "step": 527
+    },
+    {
+      "epoch": 3.4064516129032256,
+      "grad_norm": 0.44432902336120605,
+      "learning_rate": 1.6396661186364543e-06,
+      "loss": 0.0106,
+      "step": 528
+    },
+    {
+      "epoch": 3.412903225806452,
+      "grad_norm": 0.4438421428203583,
+      "learning_rate": 1.6049940341253442e-06,
+      "loss": 0.0172,
+      "step": 529
+    },
+    {
+      "epoch": 3.4193548387096775,
+      "grad_norm": 0.5001305341720581,
+      "learning_rate": 1.570667298192724e-06,
+      "loss": 0.0115,
+      "step": 530
+    },
+    {
+      "epoch": 3.425806451612903,
+      "grad_norm": 0.36009618639945984,
+      "learning_rate": 1.5366869989233062e-06,
+      "loss": 0.0111,
+      "step": 531
+    },
+    {
+      "epoch": 3.432258064516129,
+      "grad_norm": 0.35556045174598694,
+      "learning_rate": 1.5030542134205003e-06,
+      "loss": 0.0129,
+      "step": 532
+    },
+    {
+      "epoch": 3.4387096774193546,
+      "grad_norm": 0.38800477981567383,
+      "learning_rate": 1.4697700077722616e-06,
+      "loss": 0.0111,
+      "step": 533
+    },
+    {
+      "epoch": 3.445161290322581,
+      "grad_norm": 0.35132497549057007,
+      "learning_rate": 1.4368354370173073e-06,
+      "loss": 0.0133,
+      "step": 534
+    },
+    {
+      "epoch": 3.4516129032258065,
+      "grad_norm": 0.49624019861221313,
+      "learning_rate": 1.404251545111672e-06,
+      "loss": 0.0152,
+      "step": 535
+    },
+    {
+      "epoch": 3.458064516129032,
+      "grad_norm": 0.34581199288368225,
+      "learning_rate": 1.3720193648956062e-06,
+      "loss": 0.0093,
+      "step": 536
+    },
+    {
+      "epoch": 3.464516129032258,
+      "grad_norm": 0.4548514783382416,
+      "learning_rate": 1.3401399180608551e-06,
+      "loss": 0.0174,
+      "step": 537
+    },
+    {
+      "epoch": 3.4709677419354836,
+      "grad_norm": 0.33973830938339233,
+      "learning_rate": 1.3086142151182605e-06,
+      "loss": 0.0143,
+      "step": 538
+    },
+    {
+      "epoch": 3.47741935483871,
+      "grad_norm": 0.3562283515930176,
+      "learning_rate": 1.2774432553657303e-06,
+      "loss": 0.0129,
+      "step": 539
+    },
+    {
+      "epoch": 3.4838709677419355,
+      "grad_norm": 0.42894405126571655,
+      "learning_rate": 1.2466280268565708e-06,
+      "loss": 0.0136,
+      "step": 540
+    },
+    {
+      "epoch": 3.490322580645161,
+      "grad_norm": 0.36266642808914185,
+      "learning_rate": 1.2161695063681589e-06,
+      "loss": 0.0152,
+      "step": 541
+    },
+    {
+      "epoch": 3.496774193548387,
+      "grad_norm": 0.41463732719421387,
+      "learning_rate": 1.186068659370984e-06,
+      "loss": 0.0126,
+      "step": 542
+    },
+    {
+      "epoch": 3.5032258064516126,
+      "grad_norm": 0.3517482876777649,
+      "learning_rate": 1.1563264399980512e-06,
+      "loss": 0.0106,
+      "step": 543
+    },
+    {
+      "epoch": 3.509677419354839,
+      "grad_norm": 0.3592299520969391,
+      "learning_rate": 1.1269437910146173e-06,
+      "loss": 0.01,
+      "step": 544
+    },
+    {
+      "epoch": 3.5161290322580645,
+      "grad_norm": 0.3486897051334381,
+      "learning_rate": 1.0979216437883327e-06,
+      "loss": 0.0132,
+      "step": 545
+    },
+    {
+      "epoch": 3.52258064516129,
+      "grad_norm": 0.3669939935207367,
+      "learning_rate": 1.069260918259704e-06,
+      "loss": 0.0108,
+      "step": 546
+    },
+    {
+      "epoch": 3.5290322580645164,
+      "grad_norm": 0.345688134431839,
+      "learning_rate": 1.0409625229129292e-06,
+      "loss": 0.0112,
+      "step": 547
+    },
+    {
+      "epoch": 3.535483870967742,
+      "grad_norm": 0.40567103028297424,
+      "learning_rate": 1.0130273547471176e-06,
+      "loss": 0.017,
+      "step": 548
+    },
+    {
+      "epoch": 3.541935483870968,
+      "grad_norm": 0.45447733998298645,
+      "learning_rate": 9.854562992478445e-07,
+      "loss": 0.0293,
+      "step": 549
+    },
+    {
+      "epoch": 3.5483870967741935,
+      "grad_norm": 0.40127208828926086,
+      "learning_rate": 9.582502303590798e-07,
+      "loss": 0.0151,
+      "step": 550
+    },
+    {
+      "epoch": 3.554838709677419,
+      "grad_norm": 0.32802486419677734,
+      "learning_rate": 9.314100104555066e-07,
+      "loss": 0.0101,
+      "step": 551
+    },
+    {
+      "epoch": 3.5612903225806454,
+      "grad_norm": 0.35557428002357483,
+      "learning_rate": 9.049364903151558e-07,
+      "loss": 0.0108,
+      "step": 552
+    },
+    {
+      "epoch": 3.567741935483871,
+      "grad_norm": 0.5105459094047546,
+      "learning_rate": 8.788305090924556e-07,
+      "loss": 0.0167,
+      "step": 553
+    },
+    {
+      "epoch": 3.574193548387097,
+      "grad_norm": 0.3577045202255249,
+      "learning_rate": 8.530928942916447e-07,
+      "loss": 0.0076,
+      "step": 554
+    },
+    {
+      "epoch": 3.5806451612903225,
+      "grad_norm": 0.2892685532569885,
+      "learning_rate": 8.277244617405102e-07,
+      "loss": 0.0077,
+      "step": 555
+    },
+    {
+      "epoch": 3.587096774193548,
+      "grad_norm": 0.47886940836906433,
+      "learning_rate": 8.027260155645546e-07,
+      "loss": 0.0109,
+      "step": 556
+    },
+    {
+      "epoch": 3.5935483870967744,
+      "grad_norm": 0.3236874043941498,
+      "learning_rate": 7.780983481614962e-07,
+      "loss": 0.0074,
+      "step": 557
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.46806618571281433,
+      "learning_rate": 7.538422401761461e-07,
+      "loss": 0.0174,
+      "step": 558
+    },
+    {
+      "epoch": 3.606451612903226,
+      "grad_norm": 0.4186045229434967,
+      "learning_rate": 7.299584604756784e-07,
+      "loss": 0.0111,
+      "step": 559
+    },
+    {
+      "epoch": 3.6129032258064515,
+      "grad_norm": 0.4132605791091919,
+      "learning_rate": 7.064477661252483e-07,
+      "loss": 0.0132,
+      "step": 560
+    },
+    {
+      "epoch": 3.6193548387096772,
+      "grad_norm": 0.5827385783195496,
+      "learning_rate": 6.833109023639928e-07,
+      "loss": 0.017,
+      "step": 561
+    },
+    {
+      "epoch": 3.6258064516129034,
+      "grad_norm": 0.3105774521827698,
+      "learning_rate": 6.605486025814164e-07,
+      "loss": 0.0091,
+      "step": 562
+    },
+    {
+      "epoch": 3.632258064516129,
+      "grad_norm": 0.34796178340911865,
+      "learning_rate": 6.381615882941366e-07,
+      "loss": 0.0083,
+      "step": 563
+    },
+    {
+      "epoch": 3.638709677419355,
+      "grad_norm": 0.3462621867656708,
+      "learning_rate": 6.16150569123021e-07,
+      "loss": 0.0143,
+      "step": 564
+    },
+    {
+      "epoch": 3.6451612903225805,
+      "grad_norm": 0.4699903428554535,
+      "learning_rate": 5.945162427706888e-07,
+      "loss": 0.0145,
+      "step": 565
+    },
+    {
+      "epoch": 3.6516129032258062,
+      "grad_norm": 0.42084646224975586,
+      "learning_rate": 5.732592949993898e-07,
+      "loss": 0.015,
+      "step": 566
+    },
+    {
+      "epoch": 3.6580645161290324,
+      "grad_norm": 0.4539680778980255,
+      "learning_rate": 5.5238039960928e-07,
+      "loss": 0.0154,
+      "step": 567
+    },
+    {
+      "epoch": 3.664516129032258,
+      "grad_norm": 0.3853324353694916,
+      "learning_rate": 5.318802184170565e-07,
+      "loss": 0.0126,
+      "step": 568
+    },
+    {
+      "epoch": 3.670967741935484,
+      "grad_norm": 0.409679651260376,
+      "learning_rate": 5.117594012349735e-07,
+      "loss": 0.0143,
+      "step": 569
+    },
+    {
+      "epoch": 3.6774193548387095,
+      "grad_norm": 0.42005378007888794,
+      "learning_rate": 4.920185858502596e-07,
+      "loss": 0.0129,
+      "step": 570
+    },
+    {
+      "epoch": 3.6838709677419352,
+      "grad_norm": 0.34078460931777954,
+      "learning_rate": 4.7265839800488543e-07,
+      "loss": 0.0132,
+      "step": 571
+    },
+    {
+      "epoch": 3.6903225806451614,
+      "grad_norm": 0.5289260149002075,
+      "learning_rate": 4.5367945137573946e-07,
+      "loss": 0.0114,
+      "step": 572
+    },
+    {
+      "epoch": 3.696774193548387,
+      "grad_norm": 0.43742361664772034,
+      "learning_rate": 4.350823475551713e-07,
+      "loss": 0.0099,
+      "step": 573
+    },
+    {
+      "epoch": 3.703225806451613,
+      "grad_norm": 0.3581911623477936,
+      "learning_rate": 4.1686767603192344e-07,
+      "loss": 0.0153,
+      "step": 574
+    },
+    {
+      "epoch": 3.709677419354839,
+      "grad_norm": 0.3517850637435913,
+      "learning_rate": 3.990360141724478e-07,
+      "loss": 0.0123,
+      "step": 575
+    },
+    {
+      "epoch": 3.7161290322580647,
+      "grad_norm": 0.44253072142601013,
+      "learning_rate": 3.815879272025966e-07,
+      "loss": 0.0068,
+      "step": 576
+    },
+    {
+      "epoch": 3.7225806451612904,
+      "grad_norm": 0.3429562449455261,
+      "learning_rate": 3.6452396818971863e-07,
+      "loss": 0.0121,
+      "step": 577
+    },
+    {
+      "epoch": 3.729032258064516,
+      "grad_norm": 0.7084751129150391,
+      "learning_rate": 3.4784467802511797e-07,
+      "loss": 0.0145,
+      "step": 578
+    },
+    {
+      "epoch": 3.735483870967742,
+      "grad_norm": 0.388698011636734,
+      "learning_rate": 3.3155058540691037e-07,
+      "loss": 0.0107,
+      "step": 579
+    },
+    {
+      "epoch": 3.741935483870968,
+      "grad_norm": 0.41982078552246094,
+      "learning_rate": 3.1564220682327314e-07,
+      "loss": 0.014,
+      "step": 580
+    },
+    {
+      "epoch": 3.7483870967741937,
+      "grad_norm": 0.645720899105072,
+      "learning_rate": 3.001200465360593e-07,
+      "loss": 0.015,
+      "step": 581
+    },
+    {
+      "epoch": 3.7548387096774194,
+      "grad_norm": 0.5690763592720032,
+      "learning_rate": 2.8498459656482317e-07,
+      "loss": 0.0189,
+      "step": 582
+    },
+    {
+      "epoch": 3.761290322580645,
+      "grad_norm": 0.4491289556026459,
+      "learning_rate": 2.702363366712257e-07,
+      "loss": 0.0127,
+      "step": 583
+    },
+    {
+      "epoch": 3.767741935483871,
+      "grad_norm": 0.4925324618816376,
+      "learning_rate": 2.5587573434381895e-07,
+      "loss": 0.0138,
+      "step": 584
+    },
+    {
+      "epoch": 3.774193548387097,
+      "grad_norm": 0.6388445496559143,
+      "learning_rate": 2.41903244783237e-07,
+      "loss": 0.0145,
+      "step": 585
+    },
+    {
+      "epoch": 3.7806451612903227,
+      "grad_norm": 0.31376415491104126,
+      "learning_rate": 2.2831931088775904e-07,
+      "loss": 0.0135,
+      "step": 586
+    },
+    {
+      "epoch": 3.7870967741935484,
+      "grad_norm": 0.3549552261829376,
+      "learning_rate": 2.1512436323927604e-07,
+      "loss": 0.014,
+      "step": 587
+    },
+    {
+      "epoch": 3.793548387096774,
+      "grad_norm": 0.446304053068161,
+      "learning_rate": 2.0231882008963783e-07,
+      "loss": 0.0144,
+      "step": 588
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.45615440607070923,
+      "learning_rate": 1.8990308734739976e-07,
+      "loss": 0.0143,
+      "step": 589
+    },
+    {
+      "epoch": 3.806451612903226,
+      "grad_norm": 0.3915248513221741,
+      "learning_rate": 1.7787755856495254e-07,
+      "loss": 0.0131,
+      "step": 590
+    },
+    {
+      "epoch": 3.8129032258064517,
+      "grad_norm": 0.2556948959827423,
+      "learning_rate": 1.6624261492605153e-07,
+      "loss": 0.0061,
+      "step": 591
+    },
+    {
+      "epoch": 3.8193548387096774,
+      "grad_norm": 0.5648970603942871,
+      "learning_rate": 1.5499862523372933e-07,
+      "loss": 0.011,
+      "step": 592
+    },
+    {
+      "epoch": 3.825806451612903,
+      "grad_norm": 0.30211833119392395,
+      "learning_rate": 1.4414594589860774e-07,
+      "loss": 0.0087,
+      "step": 593
+    },
+    {
+      "epoch": 3.832258064516129,
+      "grad_norm": 0.21295692026615143,
+      "learning_rate": 1.3368492092760142e-07,
+      "loss": 0.0059,
+      "step": 594
+    },
+    {
+      "epoch": 3.838709677419355,
+      "grad_norm": 0.49749764800071716,
+      "learning_rate": 1.2361588191300983e-07,
+      "loss": 0.0112,
+      "step": 595
+    },
+    {
+      "epoch": 3.8451612903225807,
+      "grad_norm": 0.3051888048648834,
+      "learning_rate": 1.139391480220145e-07,
+      "loss": 0.0077,
+      "step": 596
+    },
+    {
+      "epoch": 3.8516129032258064,
+      "grad_norm": 0.3583107590675354,
+      "learning_rate": 1.0465502598655114e-07,
+      "loss": 0.0115,
+      "step": 597
+    },
+    {
+      "epoch": 3.858064516129032,
+      "grad_norm": 0.4116378426551819,
+      "learning_rate": 9.576381009359508e-08,
+      "loss": 0.0127,
+      "step": 598
+    },
+    {
+      "epoch": 3.864516129032258,
+      "grad_norm": 0.3633911907672882,
+      "learning_rate": 8.726578217582993e-08,
+      "loss": 0.0109,
+      "step": 599
+    },
+    {
+      "epoch": 3.870967741935484,
+      "grad_norm": 0.35112428665161133,
+      "learning_rate": 7.916121160271572e-08,
+      "loss": 0.0112,
+      "step": 600
+    },
+    {
+      "epoch": 3.8774193548387097,
+      "grad_norm": 0.4247336685657501,
+      "learning_rate": 7.145035527194588e-08,
+      "loss": 0.0103,
+      "step": 601
+    },
+    {
+      "epoch": 3.8838709677419354,
+      "grad_norm": 0.36888250708580017,
+      "learning_rate": 6.413345760131057e-08,
+      "loss": 0.0145,
+      "step": 602
+    },
+    {
+      "epoch": 3.8903225806451616,
+      "grad_norm": 0.3833377957344055,
+      "learning_rate": 5.721075052094599e-08,
+      "loss": 0.0094,
+      "step": 603
+    },
+    {
+      "epoch": 3.896774193548387,
+      "grad_norm": 0.45575153827667236,
+      "learning_rate": 5.068245346598332e-08,
+      "loss": 0.0111,
+      "step": 604
+    },
+    {
+      "epoch": 3.903225806451613,
+      "grad_norm": 0.3035842776298523,
+      "learning_rate": 4.454877336958763e-08,
+      "loss": 0.0071,
+      "step": 605
+    },
+    {
+      "epoch": 3.9096774193548387,
+      "grad_norm": 0.3646605908870697,
+      "learning_rate": 3.8809904656410264e-08,
+      "loss": 0.0111,
+      "step": 606
+    },
+    {
+      "epoch": 3.9161290322580644,
+      "grad_norm": 0.3893778920173645,
+      "learning_rate": 3.346602923641473e-08,
+      "loss": 0.0084,
+      "step": 607
+    },
+    {
+      "epoch": 3.9225806451612906,
+      "grad_norm": 0.29794201254844666,
+      "learning_rate": 2.8517316499115932e-08,
+      "loss": 0.0069,
+      "step": 608
+    },
+    {
+      "epoch": 3.9290322580645163,
+      "grad_norm": 0.38667142391204834,
+      "learning_rate": 2.3963923308212288e-08,
+      "loss": 0.0144,
+      "step": 609
+    },
+    {
+      "epoch": 3.935483870967742,
+      "grad_norm": 0.39387401938438416,
+      "learning_rate": 1.9805993996606376e-08,
+      "loss": 0.0087,
+      "step": 610
+    },
+    {
+      "epoch": 3.9419354838709677,
+      "grad_norm": 0.3552229106426239,
+      "learning_rate": 1.604366036184052e-08,
+      "loss": 0.0093,
+      "step": 611
+    },
+    {
+      "epoch": 3.9483870967741934,
+      "grad_norm": 0.5617074966430664,
+      "learning_rate": 1.2677041661907085e-08,
+      "loss": 0.008,
+      "step": 612
+    },
+    {
+      "epoch": 3.9548387096774196,
+      "grad_norm": 0.3588564693927765,
+      "learning_rate": 9.706244611480674e-09,
+      "loss": 0.016,
+      "step": 613
+    },
+    {
+      "epoch": 3.9612903225806453,
+      "grad_norm": 0.36114564538002014,
+      "learning_rate": 7.131363378524991e-09,
+      "loss": 0.0113,
+      "step": 614
+    },
+    {
+      "epoch": 3.967741935483871,
+      "grad_norm": 0.5072866678237915,
+      "learning_rate": 4.952479581311897e-09,
+      "loss": 0.0128,
+      "step": 615
+    },
+    {
+      "epoch": 3.9741935483870967,
+      "grad_norm": 0.3469507396221161,
+      "learning_rate": 3.1696622858373716e-09,
+      "loss": 0.01,
+      "step": 616
+    },
+    {
+      "epoch": 3.9806451612903224,
+      "grad_norm": 0.6267412900924683,
+      "learning_rate": 1.7829680036274276e-09,
+      "loss": 0.0201,
+      "step": 617
+    },
+    {
+      "epoch": 3.9870967741935486,
+      "grad_norm": 0.4204852879047394,
+      "learning_rate": 7.924406899492698e-10,
+      "loss": 0.0108,
+      "step": 618
+    },
+    {
+      "epoch": 3.9935483870967743,
+      "grad_norm": 0.2933880090713501,
+      "learning_rate": 1.9811174241796127e-10,
+      "loss": 0.0113,
+      "step": 619
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.3794356882572174,
+      "learning_rate": 0.0,
+      "loss": 0.0109,
+      "step": 620
+    },
+    {
+      "epoch": 4.0,
+      "step": 620,
+      "total_flos": 2.582134138035241e+17,
+      "train_loss": 0.053207253262911355,
+      "train_runtime": 778.335,
+      "train_samples_per_second": 25.449,
+      "train_steps_per_second": 0.797
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 620,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "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": 2.582134138035241e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}