diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4368 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9951534733441034,
+  "eval_steps": 500,
+  "global_step": 618,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0032310177705977385,
+      "grad_norm": 25.128438265728068,
+      "learning_rate": 8.064516129032259e-08,
+      "loss": 0.1803,
+      "step": 1
+    },
+    {
+      "epoch": 0.006462035541195477,
+      "grad_norm": 20.481937139410615,
+      "learning_rate": 1.6129032258064518e-07,
+      "loss": 0.1841,
+      "step": 2
+    },
+    {
+      "epoch": 0.009693053311793215,
+      "grad_norm": 40.482947214989046,
+      "learning_rate": 2.4193548387096775e-07,
+      "loss": 0.2517,
+      "step": 3
+    },
+    {
+      "epoch": 0.012924071082390954,
+      "grad_norm": 56.10659153508464,
+      "learning_rate": 3.2258064516129035e-07,
+      "loss": 0.289,
+      "step": 4
+    },
+    {
+      "epoch": 0.01615508885298869,
+      "grad_norm": 40.7368526949032,
+      "learning_rate": 4.032258064516129e-07,
+      "loss": 0.2432,
+      "step": 5
+    },
+    {
+      "epoch": 0.01938610662358643,
+      "grad_norm": 36.018819191325036,
+      "learning_rate": 4.838709677419355e-07,
+      "loss": 0.1939,
+      "step": 6
+    },
+    {
+      "epoch": 0.022617124394184167,
+      "grad_norm": 64.22644318288756,
+      "learning_rate": 5.645161290322581e-07,
+      "loss": 0.2186,
+      "step": 7
+    },
+    {
+      "epoch": 0.025848142164781908,
+      "grad_norm": 68.80190588802398,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 0.237,
+      "step": 8
+    },
+    {
+      "epoch": 0.029079159935379646,
+      "grad_norm": 53.87483186618541,
+      "learning_rate": 7.258064516129033e-07,
+      "loss": 0.1904,
+      "step": 9
+    },
+    {
+      "epoch": 0.03231017770597738,
+      "grad_norm": 27.250940737956068,
+      "learning_rate": 8.064516129032258e-07,
+      "loss": 0.1914,
+      "step": 10
+    },
+    {
+      "epoch": 0.035541195476575124,
+      "grad_norm": 19.13516705978285,
+      "learning_rate": 8.870967741935485e-07,
+      "loss": 0.21,
+      "step": 11
+    },
+    {
+      "epoch": 0.03877221324717286,
+      "grad_norm": 5.971043423480582,
+      "learning_rate": 9.67741935483871e-07,
+      "loss": 0.1893,
+      "step": 12
+    },
+    {
+      "epoch": 0.0420032310177706,
+      "grad_norm": 8.917076042584409,
+      "learning_rate": 1.0483870967741936e-06,
+      "loss": 0.1672,
+      "step": 13
+    },
+    {
+      "epoch": 0.045234248788368334,
+      "grad_norm": 6.564634516149644,
+      "learning_rate": 1.1290322580645162e-06,
+      "loss": 0.1639,
+      "step": 14
+    },
+    {
+      "epoch": 0.048465266558966075,
+      "grad_norm": 7.4524362304900995,
+      "learning_rate": 1.2096774193548388e-06,
+      "loss": 0.1904,
+      "step": 15
+    },
+    {
+      "epoch": 0.051696284329563816,
+      "grad_norm": 6.424814341877685,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 0.1539,
+      "step": 16
+    },
+    {
+      "epoch": 0.05492730210016155,
+      "grad_norm": 8.919971119630558,
+      "learning_rate": 1.3709677419354838e-06,
+      "loss": 0.1276,
+      "step": 17
+    },
+    {
+      "epoch": 0.05815831987075929,
+      "grad_norm": 8.858198890334634,
+      "learning_rate": 1.4516129032258066e-06,
+      "loss": 0.148,
+      "step": 18
+    },
+    {
+      "epoch": 0.061389337641357025,
+      "grad_norm": 8.19809300927439,
+      "learning_rate": 1.5322580645161292e-06,
+      "loss": 0.1528,
+      "step": 19
+    },
+    {
+      "epoch": 0.06462035541195477,
+      "grad_norm": 9.263957823023876,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 0.1654,
+      "step": 20
+    },
+    {
+      "epoch": 0.06785137318255251,
+      "grad_norm": 5.211246085563888,
+      "learning_rate": 1.6935483870967742e-06,
+      "loss": 0.1478,
+      "step": 21
+    },
+    {
+      "epoch": 0.07108239095315025,
+      "grad_norm": 4.16568459472329,
+      "learning_rate": 1.774193548387097e-06,
+      "loss": 0.1548,
+      "step": 22
+    },
+    {
+      "epoch": 0.07431340872374798,
+      "grad_norm": 2.1941339248998233,
+      "learning_rate": 1.8548387096774196e-06,
+      "loss": 0.1349,
+      "step": 23
+    },
+    {
+      "epoch": 0.07754442649434572,
+      "grad_norm": 1.5770028567175525,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 0.174,
+      "step": 24
+    },
+    {
+      "epoch": 0.08077544426494346,
+      "grad_norm": 1.606149082877329,
+      "learning_rate": 2.0161290322580646e-06,
+      "loss": 0.1468,
+      "step": 25
+    },
+    {
+      "epoch": 0.0840064620355412,
+      "grad_norm": 1.4435377235758515,
+      "learning_rate": 2.096774193548387e-06,
+      "loss": 0.1482,
+      "step": 26
+    },
+    {
+      "epoch": 0.08723747980613894,
+      "grad_norm": 1.4360703380719761,
+      "learning_rate": 2.17741935483871e-06,
+      "loss": 0.1404,
+      "step": 27
+    },
+    {
+      "epoch": 0.09046849757673667,
+      "grad_norm": 0.8101385337166851,
+      "learning_rate": 2.2580645161290324e-06,
+      "loss": 0.1142,
+      "step": 28
+    },
+    {
+      "epoch": 0.09369951534733441,
+      "grad_norm": 1.381962668180297,
+      "learning_rate": 2.338709677419355e-06,
+      "loss": 0.1331,
+      "step": 29
+    },
+    {
+      "epoch": 0.09693053311793215,
+      "grad_norm": 1.2208818326390372,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.1196,
+      "step": 30
+    },
+    {
+      "epoch": 0.10016155088852989,
+      "grad_norm": 1.1209301057498027,
+      "learning_rate": 2.5e-06,
+      "loss": 0.0986,
+      "step": 31
+    },
+    {
+      "epoch": 0.10339256865912763,
+      "grad_norm": 1.9664855281381353,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.133,
+      "step": 32
+    },
+    {
+      "epoch": 0.10662358642972536,
+      "grad_norm": 1.2526567677482312,
+      "learning_rate": 2.6612903225806454e-06,
+      "loss": 0.0969,
+      "step": 33
+    },
+    {
+      "epoch": 0.1098546042003231,
+      "grad_norm": 1.0939478434588246,
+      "learning_rate": 2.7419354838709676e-06,
+      "loss": 0.115,
+      "step": 34
+    },
+    {
+      "epoch": 0.11308562197092084,
+      "grad_norm": 0.8286974625967464,
+      "learning_rate": 2.822580645161291e-06,
+      "loss": 0.0941,
+      "step": 35
+    },
+    {
+      "epoch": 0.11631663974151858,
+      "grad_norm": 1.1449426230450832,
+      "learning_rate": 2.903225806451613e-06,
+      "loss": 0.1182,
+      "step": 36
+    },
+    {
+      "epoch": 0.11954765751211632,
+      "grad_norm": 1.188437444192784,
+      "learning_rate": 2.983870967741936e-06,
+      "loss": 0.1107,
+      "step": 37
+    },
+    {
+      "epoch": 0.12277867528271405,
+      "grad_norm": 1.1499792776405726,
+      "learning_rate": 3.0645161290322584e-06,
+      "loss": 0.1044,
+      "step": 38
+    },
+    {
+      "epoch": 0.1260096930533118,
+      "grad_norm": 1.0118348591126385,
+      "learning_rate": 3.145161290322581e-06,
+      "loss": 0.0926,
+      "step": 39
+    },
+    {
+      "epoch": 0.12924071082390953,
+      "grad_norm": 2.2856482367185813,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.0935,
+      "step": 40
+    },
+    {
+      "epoch": 0.13247172859450726,
+      "grad_norm": 1.3141305507246495,
+      "learning_rate": 3.306451612903226e-06,
+      "loss": 0.1411,
+      "step": 41
+    },
+    {
+      "epoch": 0.13570274636510501,
+      "grad_norm": 1.5200078779321133,
+      "learning_rate": 3.3870967741935484e-06,
+      "loss": 0.1086,
+      "step": 42
+    },
+    {
+      "epoch": 0.13893376413570274,
+      "grad_norm": 1.199141198842882,
+      "learning_rate": 3.4677419354838714e-06,
+      "loss": 0.0963,
+      "step": 43
+    },
+    {
+      "epoch": 0.1421647819063005,
+      "grad_norm": 0.9118534178360143,
+      "learning_rate": 3.548387096774194e-06,
+      "loss": 0.0838,
+      "step": 44
+    },
+    {
+      "epoch": 0.14539579967689822,
+      "grad_norm": 0.8313523608482088,
+      "learning_rate": 3.6290322580645166e-06,
+      "loss": 0.0822,
+      "step": 45
+    },
+    {
+      "epoch": 0.14862681744749595,
+      "grad_norm": 1.0602464666470106,
+      "learning_rate": 3.7096774193548392e-06,
+      "loss": 0.1031,
+      "step": 46
+    },
+    {
+      "epoch": 0.1518578352180937,
+      "grad_norm": 1.4189660331134808,
+      "learning_rate": 3.7903225806451614e-06,
+      "loss": 0.0985,
+      "step": 47
+    },
+    {
+      "epoch": 0.15508885298869143,
+      "grad_norm": 0.7654402241394538,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.0759,
+      "step": 48
+    },
+    {
+      "epoch": 0.1583198707592892,
+      "grad_norm": 1.5213372179407378,
+      "learning_rate": 3.951612903225807e-06,
+      "loss": 0.1288,
+      "step": 49
+    },
+    {
+      "epoch": 0.16155088852988692,
+      "grad_norm": 1.2908078430353152,
+      "learning_rate": 4.032258064516129e-06,
+      "loss": 0.1117,
+      "step": 50
+    },
+    {
+      "epoch": 0.16478190630048464,
+      "grad_norm": 1.0569868777193876,
+      "learning_rate": 4.112903225806452e-06,
+      "loss": 0.0983,
+      "step": 51
+    },
+    {
+      "epoch": 0.1680129240710824,
+      "grad_norm": 0.9711071114660816,
+      "learning_rate": 4.193548387096774e-06,
+      "loss": 0.0911,
+      "step": 52
+    },
+    {
+      "epoch": 0.17124394184168013,
+      "grad_norm": 1.102405320517146,
+      "learning_rate": 4.274193548387097e-06,
+      "loss": 0.1037,
+      "step": 53
+    },
+    {
+      "epoch": 0.17447495961227788,
+      "grad_norm": 0.9937966694938649,
+      "learning_rate": 4.35483870967742e-06,
+      "loss": 0.0815,
+      "step": 54
+    },
+    {
+      "epoch": 0.1777059773828756,
+      "grad_norm": 1.0718680838436345,
+      "learning_rate": 4.435483870967742e-06,
+      "loss": 0.1004,
+      "step": 55
+    },
+    {
+      "epoch": 0.18093699515347333,
+      "grad_norm": 1.1717408640580886,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 0.1278,
+      "step": 56
+    },
+    {
+      "epoch": 0.1841680129240711,
+      "grad_norm": 0.9371050661628969,
+      "learning_rate": 4.596774193548387e-06,
+      "loss": 0.09,
+      "step": 57
+    },
+    {
+      "epoch": 0.18739903069466882,
+      "grad_norm": 0.7008244964012689,
+      "learning_rate": 4.67741935483871e-06,
+      "loss": 0.0704,
+      "step": 58
+    },
+    {
+      "epoch": 0.19063004846526657,
+      "grad_norm": 1.198678087514701,
+      "learning_rate": 4.758064516129033e-06,
+      "loss": 0.0966,
+      "step": 59
+    },
+    {
+      "epoch": 0.1938610662358643,
+      "grad_norm": 1.237062798014623,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.108,
+      "step": 60
+    },
+    {
+      "epoch": 0.19709208400646203,
+      "grad_norm": 0.9791118203985985,
+      "learning_rate": 4.919354838709678e-06,
+      "loss": 0.0991,
+      "step": 61
+    },
+    {
+      "epoch": 0.20032310177705978,
+      "grad_norm": 1.0841634147338435,
+      "learning_rate": 5e-06,
+      "loss": 0.0934,
+      "step": 62
+    },
+    {
+      "epoch": 0.2035541195476575,
+      "grad_norm": 1.258223416727101,
+      "learning_rate": 4.999960092086724e-06,
+      "loss": 0.1219,
+      "step": 63
+    },
+    {
+      "epoch": 0.20678513731825526,
+      "grad_norm": 1.0617428668947297,
+      "learning_rate": 4.999840369621011e-06,
+      "loss": 0.1124,
+      "step": 64
+    },
+    {
+      "epoch": 0.210016155088853,
+      "grad_norm": 1.213736182334779,
+      "learning_rate": 4.999640836425159e-06,
+      "loss": 0.1008,
+      "step": 65
+    },
+    {
+      "epoch": 0.21324717285945072,
+      "grad_norm": 1.100400949090305,
+      "learning_rate": 4.99936149886953e-06,
+      "loss": 0.0732,
+      "step": 66
+    },
+    {
+      "epoch": 0.21647819063004847,
+      "grad_norm": 1.0182260583019065,
+      "learning_rate": 4.999002365872348e-06,
+      "loss": 0.0818,
+      "step": 67
+    },
+    {
+      "epoch": 0.2197092084006462,
+      "grad_norm": 1.0758854772406632,
+      "learning_rate": 4.998563448899413e-06,
+      "loss": 0.1152,
+      "step": 68
+    },
+    {
+      "epoch": 0.22294022617124395,
+      "grad_norm": 0.7655531388319169,
+      "learning_rate": 4.998044761963731e-06,
+      "loss": 0.0629,
+      "step": 69
+    },
+    {
+      "epoch": 0.22617124394184168,
+      "grad_norm": 0.9063698842381128,
+      "learning_rate": 4.9974463216250735e-06,
+      "loss": 0.0947,
+      "step": 70
+    },
+    {
+      "epoch": 0.2294022617124394,
+      "grad_norm": 1.111926907799753,
+      "learning_rate": 4.996768146989446e-06,
+      "loss": 0.0869,
+      "step": 71
+    },
+    {
+      "epoch": 0.23263327948303716,
+      "grad_norm": 1.2171632327405464,
+      "learning_rate": 4.996010259708475e-06,
+      "loss": 0.0906,
+      "step": 72
+    },
+    {
+      "epoch": 0.2358642972536349,
+      "grad_norm": 1.1760501880854652,
+      "learning_rate": 4.99517268397872e-06,
+      "loss": 0.0905,
+      "step": 73
+    },
+    {
+      "epoch": 0.23909531502423265,
+      "grad_norm": 0.9925140924590656,
+      "learning_rate": 4.9942554465409e-06,
+      "loss": 0.0759,
+      "step": 74
+    },
+    {
+      "epoch": 0.24232633279483037,
+      "grad_norm": 1.1900673708039176,
+      "learning_rate": 4.993258576679043e-06,
+      "loss": 0.0959,
+      "step": 75
+    },
+    {
+      "epoch": 0.2455573505654281,
+      "grad_norm": 1.0525284959082881,
+      "learning_rate": 4.9921821062195445e-06,
+      "loss": 0.0854,
+      "step": 76
+    },
+    {
+      "epoch": 0.24878836833602586,
+      "grad_norm": 0.9782484128440981,
+      "learning_rate": 4.991026069530156e-06,
+      "loss": 0.0811,
+      "step": 77
+    },
+    {
+      "epoch": 0.2520193861066236,
+      "grad_norm": 1.099947189666444,
+      "learning_rate": 4.989790503518888e-06,
+      "loss": 0.1213,
+      "step": 78
+    },
+    {
+      "epoch": 0.2552504038772213,
+      "grad_norm": 0.9763034951592139,
+      "learning_rate": 4.988475447632829e-06,
+      "loss": 0.0978,
+      "step": 79
+    },
+    {
+      "epoch": 0.25848142164781907,
+      "grad_norm": 1.1059217088306985,
+      "learning_rate": 4.987080943856887e-06,
+      "loss": 0.1188,
+      "step": 80
+    },
+    {
+      "epoch": 0.2617124394184168,
+      "grad_norm": 0.8811301090737587,
+      "learning_rate": 4.985607036712453e-06,
+      "loss": 0.0959,
+      "step": 81
+    },
+    {
+      "epoch": 0.2649434571890145,
+      "grad_norm": 0.876445346062972,
+      "learning_rate": 4.984053773255971e-06,
+      "loss": 0.0942,
+      "step": 82
+    },
+    {
+      "epoch": 0.2681744749596123,
+      "grad_norm": 0.9532526347121169,
+      "learning_rate": 4.982421203077446e-06,
+      "loss": 0.0877,
+      "step": 83
+    },
+    {
+      "epoch": 0.27140549273021003,
+      "grad_norm": 1.123157927763249,
+      "learning_rate": 4.980709378298851e-06,
+      "loss": 0.1278,
+      "step": 84
+    },
+    {
+      "epoch": 0.27463651050080773,
+      "grad_norm": 1.0388291288831657,
+      "learning_rate": 4.978918353572471e-06,
+      "loss": 0.0711,
+      "step": 85
+    },
+    {
+      "epoch": 0.2778675282714055,
+      "grad_norm": 0.9767051395631638,
+      "learning_rate": 4.977048186079155e-06,
+      "loss": 0.0778,
+      "step": 86
+    },
+    {
+      "epoch": 0.28109854604200324,
+      "grad_norm": 0.796702172980499,
+      "learning_rate": 4.975098935526487e-06,
+      "loss": 0.0737,
+      "step": 87
+    },
+    {
+      "epoch": 0.284329563812601,
+      "grad_norm": 0.930052043825522,
+      "learning_rate": 4.973070664146885e-06,
+      "loss": 0.0797,
+      "step": 88
+    },
+    {
+      "epoch": 0.2875605815831987,
+      "grad_norm": 0.8599629277716169,
+      "learning_rate": 4.970963436695612e-06,
+      "loss": 0.0736,
+      "step": 89
+    },
+    {
+      "epoch": 0.29079159935379645,
+      "grad_norm": 1.2487651297160884,
+      "learning_rate": 4.968777320448707e-06,
+      "loss": 0.1234,
+      "step": 90
+    },
+    {
+      "epoch": 0.2940226171243942,
+      "grad_norm": 1.3265846927372396,
+      "learning_rate": 4.966512385200841e-06,
+      "loss": 0.097,
+      "step": 91
+    },
+    {
+      "epoch": 0.2972536348949919,
+      "grad_norm": 1.2653863711384867,
+      "learning_rate": 4.964168703263086e-06,
+      "loss": 0.0851,
+      "step": 92
+    },
+    {
+      "epoch": 0.30048465266558966,
+      "grad_norm": 1.05226163192955,
+      "learning_rate": 4.961746349460607e-06,
+      "loss": 0.0901,
+      "step": 93
+    },
+    {
+      "epoch": 0.3037156704361874,
+      "grad_norm": 0.7053264391935079,
+      "learning_rate": 4.959245401130269e-06,
+      "loss": 0.0472,
+      "step": 94
+    },
+    {
+      "epoch": 0.3069466882067851,
+      "grad_norm": 1.1693965687082317,
+      "learning_rate": 4.956665938118179e-06,
+      "loss": 0.0909,
+      "step": 95
+    },
+    {
+      "epoch": 0.31017770597738287,
+      "grad_norm": 1.0877301086902282,
+      "learning_rate": 4.954008042777125e-06,
+      "loss": 0.0897,
+      "step": 96
+    },
+    {
+      "epoch": 0.3134087237479806,
+      "grad_norm": 0.9145033111059097,
+      "learning_rate": 4.951271799963952e-06,
+      "loss": 0.0871,
+      "step": 97
+    },
+    {
+      "epoch": 0.3166397415185784,
+      "grad_norm": 0.8706912416502465,
+      "learning_rate": 4.9484572970368516e-06,
+      "loss": 0.0656,
+      "step": 98
+    },
+    {
+      "epoch": 0.3198707592891761,
+      "grad_norm": 1.1252067732287794,
+      "learning_rate": 4.945564623852577e-06,
+      "loss": 0.0916,
+      "step": 99
+    },
+    {
+      "epoch": 0.32310177705977383,
+      "grad_norm": 0.9809013164181603,
+      "learning_rate": 4.942593872763566e-06,
+      "loss": 0.0671,
+      "step": 100
+    },
+    {
+      "epoch": 0.3263327948303716,
+      "grad_norm": 0.9939061476745602,
+      "learning_rate": 4.939545138615003e-06,
+      "loss": 0.0796,
+      "step": 101
+    },
+    {
+      "epoch": 0.3295638126009693,
+      "grad_norm": 0.8729446605586859,
+      "learning_rate": 4.93641851874178e-06,
+      "loss": 0.1031,
+      "step": 102
+    },
+    {
+      "epoch": 0.33279483037156704,
+      "grad_norm": 0.8275137143927102,
+      "learning_rate": 4.933214112965399e-06,
+      "loss": 0.098,
+      "step": 103
+    },
+    {
+      "epoch": 0.3360258481421648,
+      "grad_norm": 1.1250877000414508,
+      "learning_rate": 4.929932023590776e-06,
+      "loss": 0.0818,
+      "step": 104
+    },
+    {
+      "epoch": 0.3392568659127625,
+      "grad_norm": 0.6635247677362335,
+      "learning_rate": 4.926572355402983e-06,
+      "loss": 0.069,
+      "step": 105
+    },
+    {
+      "epoch": 0.34248788368336025,
+      "grad_norm": 0.7585575652472656,
+      "learning_rate": 4.923135215663897e-06,
+      "loss": 0.0574,
+      "step": 106
+    },
+    {
+      "epoch": 0.345718901453958,
+      "grad_norm": 0.8301240151962479,
+      "learning_rate": 4.919620714108777e-06,
+      "loss": 0.0894,
+      "step": 107
+    },
+    {
+      "epoch": 0.34894991922455576,
+      "grad_norm": 0.7584921353525548,
+      "learning_rate": 4.916028962942763e-06,
+      "loss": 0.0693,
+      "step": 108
+    },
+    {
+      "epoch": 0.35218093699515346,
+      "grad_norm": 0.8546005601207664,
+      "learning_rate": 4.912360076837289e-06,
+      "loss": 0.0853,
+      "step": 109
+    },
+    {
+      "epoch": 0.3554119547657512,
+      "grad_norm": 0.7665244261956478,
+      "learning_rate": 4.908614172926426e-06,
+      "loss": 0.0555,
+      "step": 110
+    },
+    {
+      "epoch": 0.35864297253634897,
+      "grad_norm": 0.7090721453253053,
+      "learning_rate": 4.904791370803141e-06,
+      "loss": 0.0809,
+      "step": 111
+    },
+    {
+      "epoch": 0.36187399030694667,
+      "grad_norm": 0.9004480711031451,
+      "learning_rate": 4.9008917925154795e-06,
+      "loss": 0.0798,
+      "step": 112
+    },
+    {
+      "epoch": 0.3651050080775444,
+      "grad_norm": 0.6681566874499523,
+      "learning_rate": 4.896915562562665e-06,
+      "loss": 0.0567,
+      "step": 113
+    },
+    {
+      "epoch": 0.3683360258481422,
+      "grad_norm": 0.7831071753829433,
+      "learning_rate": 4.892862807891131e-06,
+      "loss": 0.0886,
+      "step": 114
+    },
+    {
+      "epoch": 0.3715670436187399,
+      "grad_norm": 0.8572105799904943,
+      "learning_rate": 4.888733657890463e-06,
+      "loss": 0.0941,
+      "step": 115
+    },
+    {
+      "epoch": 0.37479806138933763,
+      "grad_norm": 0.6839139362191016,
+      "learning_rate": 4.884528244389269e-06,
+      "loss": 0.0499,
+      "step": 116
+    },
+    {
+      "epoch": 0.3780290791599354,
+      "grad_norm": 0.7535937955610514,
+      "learning_rate": 4.8802467016509704e-06,
+      "loss": 0.0648,
+      "step": 117
+    },
+    {
+      "epoch": 0.38126009693053314,
+      "grad_norm": 0.7278973767612748,
+      "learning_rate": 4.8758891663695165e-06,
+      "loss": 0.0647,
+      "step": 118
+    },
+    {
+      "epoch": 0.38449111470113084,
+      "grad_norm": 0.8658527730630694,
+      "learning_rate": 4.87145577766502e-06,
+      "loss": 0.0694,
+      "step": 119
+    },
+    {
+      "epoch": 0.3877221324717286,
+      "grad_norm": 0.5993174371068741,
+      "learning_rate": 4.866946677079314e-06,
+      "loss": 0.0532,
+      "step": 120
+    },
+    {
+      "epoch": 0.39095315024232635,
+      "grad_norm": 0.9585898899909292,
+      "learning_rate": 4.862362008571434e-06,
+      "loss": 0.081,
+      "step": 121
+    },
+    {
+      "epoch": 0.39418416801292405,
+      "grad_norm": 0.8402658697480521,
+      "learning_rate": 4.857701918513023e-06,
+      "loss": 0.0661,
+      "step": 122
+    },
+    {
+      "epoch": 0.3974151857835218,
+      "grad_norm": 0.8695072904889961,
+      "learning_rate": 4.852966555683657e-06,
+      "loss": 0.0777,
+      "step": 123
+    },
+    {
+      "epoch": 0.40064620355411956,
+      "grad_norm": 0.8700878343519065,
+      "learning_rate": 4.848156071266095e-06,
+      "loss": 0.0669,
+      "step": 124
+    },
+    {
+      "epoch": 0.40387722132471726,
+      "grad_norm": 0.7573866683757483,
+      "learning_rate": 4.843270618841455e-06,
+      "loss": 0.0659,
+      "step": 125
+    },
+    {
+      "epoch": 0.407108239095315,
+      "grad_norm": 0.7144861602580213,
+      "learning_rate": 4.838310354384304e-06,
+      "loss": 0.0673,
+      "step": 126
+    },
+    {
+      "epoch": 0.41033925686591277,
+      "grad_norm": 0.6521772244924818,
+      "learning_rate": 4.833275436257684e-06,
+      "loss": 0.0487,
+      "step": 127
+    },
+    {
+      "epoch": 0.4135702746365105,
+      "grad_norm": 0.9937642189435811,
+      "learning_rate": 4.828166025208059e-06,
+      "loss": 0.0748,
+      "step": 128
+    },
+    {
+      "epoch": 0.4168012924071082,
+      "grad_norm": 0.838066231893655,
+      "learning_rate": 4.822982284360173e-06,
+      "loss": 0.0732,
+      "step": 129
+    },
+    {
+      "epoch": 0.420032310177706,
+      "grad_norm": 0.9685536935346779,
+      "learning_rate": 4.8177243792118515e-06,
+      "loss": 0.1053,
+      "step": 130
+    },
+    {
+      "epoch": 0.42326332794830374,
+      "grad_norm": 0.8362566096259404,
+      "learning_rate": 4.8123924776287115e-06,
+      "loss": 0.0767,
+      "step": 131
+    },
+    {
+      "epoch": 0.42649434571890144,
+      "grad_norm": 0.7707131560838676,
+      "learning_rate": 4.8069867498388066e-06,
+      "loss": 0.059,
+      "step": 132
+    },
+    {
+      "epoch": 0.4297253634894992,
+      "grad_norm": 0.8446766172300785,
+      "learning_rate": 4.80150736842719e-06,
+      "loss": 0.0715,
+      "step": 133
+    },
+    {
+      "epoch": 0.43295638126009695,
+      "grad_norm": 0.9888301615866294,
+      "learning_rate": 4.795954508330403e-06,
+      "loss": 0.0816,
+      "step": 134
+    },
+    {
+      "epoch": 0.43618739903069464,
+      "grad_norm": 0.5818611202414509,
+      "learning_rate": 4.790328346830893e-06,
+      "loss": 0.0579,
+      "step": 135
+    },
+    {
+      "epoch": 0.4394184168012924,
+      "grad_norm": 0.7358503512846166,
+      "learning_rate": 4.784629063551354e-06,
+      "loss": 0.0729,
+      "step": 136
+    },
+    {
+      "epoch": 0.44264943457189015,
+      "grad_norm": 0.6475363027245936,
+      "learning_rate": 4.778856840448985e-06,
+      "loss": 0.0595,
+      "step": 137
+    },
+    {
+      "epoch": 0.4458804523424879,
+      "grad_norm": 0.8615033443900738,
+      "learning_rate": 4.773011861809694e-06,
+      "loss": 0.0951,
+      "step": 138
+    },
+    {
+      "epoch": 0.4491114701130856,
+      "grad_norm": 0.6927048507197057,
+      "learning_rate": 4.7670943142421955e-06,
+      "loss": 0.0524,
+      "step": 139
+    },
+    {
+      "epoch": 0.45234248788368336,
+      "grad_norm": 1.051050914153071,
+      "learning_rate": 4.761104386672074e-06,
+      "loss": 0.0766,
+      "step": 140
+    },
+    {
+      "epoch": 0.4555735056542811,
+      "grad_norm": 0.5989675474056907,
+      "learning_rate": 4.7550422703357355e-06,
+      "loss": 0.0701,
+      "step": 141
+    },
+    {
+      "epoch": 0.4588045234248788,
+      "grad_norm": 0.7775916205990767,
+      "learning_rate": 4.748908158774312e-06,
+      "loss": 0.0587,
+      "step": 142
+    },
+    {
+      "epoch": 0.4620355411954766,
+      "grad_norm": 0.6567814970993029,
+      "learning_rate": 4.742702247827476e-06,
+      "loss": 0.0558,
+      "step": 143
+    },
+    {
+      "epoch": 0.46526655896607433,
+      "grad_norm": 0.8201443957683763,
+      "learning_rate": 4.736424735627193e-06,
+      "loss": 0.066,
+      "step": 144
+    },
+    {
+      "epoch": 0.46849757673667203,
+      "grad_norm": 0.767995936704651,
+      "learning_rate": 4.730075822591392e-06,
+      "loss": 0.0698,
+      "step": 145
+    },
+    {
+      "epoch": 0.4717285945072698,
+      "grad_norm": 0.8765270563094147,
+      "learning_rate": 4.7236557114175705e-06,
+      "loss": 0.0752,
+      "step": 146
+    },
+    {
+      "epoch": 0.47495961227786754,
+      "grad_norm": 0.8060072499383857,
+      "learning_rate": 4.71716460707632e-06,
+      "loss": 0.0776,
+      "step": 147
+    },
+    {
+      "epoch": 0.4781906300484653,
+      "grad_norm": 0.7249857982210036,
+      "learning_rate": 4.710602716804784e-06,
+      "loss": 0.0645,
+      "step": 148
+    },
+    {
+      "epoch": 0.481421647819063,
+      "grad_norm": 0.9856585082866961,
+      "learning_rate": 4.703970250100041e-06,
+      "loss": 0.0813,
+      "step": 149
+    },
+    {
+      "epoch": 0.48465266558966075,
+      "grad_norm": 0.7832993384259341,
+      "learning_rate": 4.697267418712415e-06,
+      "loss": 0.0855,
+      "step": 150
+    },
+    {
+      "epoch": 0.4878836833602585,
+      "grad_norm": 0.7537716514815709,
+      "learning_rate": 4.690494436638718e-06,
+      "loss": 0.0662,
+      "step": 151
+    },
+    {
+      "epoch": 0.4911147011308562,
+      "grad_norm": 0.6970576661817561,
+      "learning_rate": 4.683651520115414e-06,
+      "loss": 0.0526,
+      "step": 152
+    },
+    {
+      "epoch": 0.49434571890145396,
+      "grad_norm": 0.6025061116050546,
+      "learning_rate": 4.67673888761172e-06,
+      "loss": 0.0423,
+      "step": 153
+    },
+    {
+      "epoch": 0.4975767366720517,
+      "grad_norm": 0.8666593448997953,
+      "learning_rate": 4.669756759822625e-06,
+      "loss": 0.0769,
+      "step": 154
+    },
+    {
+      "epoch": 0.5008077544426495,
+      "grad_norm": 0.8822574390448034,
+      "learning_rate": 4.66270535966185e-06,
+      "loss": 0.0835,
+      "step": 155
+    },
+    {
+      "epoch": 0.5040387722132472,
+      "grad_norm": 0.7966361049316173,
+      "learning_rate": 4.655584912254727e-06,
+      "loss": 0.0702,
+      "step": 156
+    },
+    {
+      "epoch": 0.5072697899838449,
+      "grad_norm": 0.975295266397441,
+      "learning_rate": 4.6483956449310155e-06,
+      "loss": 0.0868,
+      "step": 157
+    },
+    {
+      "epoch": 0.5105008077544426,
+      "grad_norm": 0.6419609425416163,
+      "learning_rate": 4.64113778721764e-06,
+      "loss": 0.0563,
+      "step": 158
+    },
+    {
+      "epoch": 0.5137318255250404,
+      "grad_norm": 1.074461428386011,
+      "learning_rate": 4.633811570831367e-06,
+      "loss": 0.1019,
+      "step": 159
+    },
+    {
+      "epoch": 0.5169628432956381,
+      "grad_norm": 0.7470516638155429,
+      "learning_rate": 4.626417229671401e-06,
+      "loss": 0.0699,
+      "step": 160
+    },
+    {
+      "epoch": 0.5201938610662359,
+      "grad_norm": 0.6593218212471074,
+      "learning_rate": 4.6189549998119235e-06,
+      "loss": 0.0489,
+      "step": 161
+    },
+    {
+      "epoch": 0.5234248788368336,
+      "grad_norm": 0.9595803117243349,
+      "learning_rate": 4.611425119494552e-06,
+      "loss": 0.0784,
+      "step": 162
+    },
+    {
+      "epoch": 0.5266558966074314,
+      "grad_norm": 0.6236918909086313,
+      "learning_rate": 4.603827829120734e-06,
+      "loss": 0.0581,
+      "step": 163
+    },
+    {
+      "epoch": 0.529886914378029,
+      "grad_norm": 0.94480008993302,
+      "learning_rate": 4.596163371244076e-06,
+      "loss": 0.0672,
+      "step": 164
+    },
+    {
+      "epoch": 0.5331179321486268,
+      "grad_norm": 0.8140229354538353,
+      "learning_rate": 4.588431990562593e-06,
+      "loss": 0.0557,
+      "step": 165
+    },
+    {
+      "epoch": 0.5363489499192245,
+      "grad_norm": 0.6881752667822972,
+      "learning_rate": 4.580633933910901e-06,
+      "loss": 0.0499,
+      "step": 166
+    },
+    {
+      "epoch": 0.5395799676898223,
+      "grad_norm": 0.6835421331522927,
+      "learning_rate": 4.572769450252335e-06,
+      "loss": 0.0726,
+      "step": 167
+    },
+    {
+      "epoch": 0.5428109854604201,
+      "grad_norm": 0.6690598961735207,
+      "learning_rate": 4.564838790671e-06,
+      "loss": 0.0553,
+      "step": 168
+    },
+    {
+      "epoch": 0.5460420032310178,
+      "grad_norm": 1.0189532436328723,
+      "learning_rate": 4.556842208363756e-06,
+      "loss": 0.0865,
+      "step": 169
+    },
+    {
+      "epoch": 0.5492730210016155,
+      "grad_norm": 0.6275318178066402,
+      "learning_rate": 4.548779958632134e-06,
+      "loss": 0.0549,
+      "step": 170
+    },
+    {
+      "epoch": 0.5525040387722132,
+      "grad_norm": 0.6758827324437686,
+      "learning_rate": 4.540652298874183e-06,
+      "loss": 0.055,
+      "step": 171
+    },
+    {
+      "epoch": 0.555735056542811,
+      "grad_norm": 0.7969599802065607,
+      "learning_rate": 4.532459488576258e-06,
+      "loss": 0.0667,
+      "step": 172
+    },
+    {
+      "epoch": 0.5589660743134087,
+      "grad_norm": 0.8835274962394156,
+      "learning_rate": 4.524201789304727e-06,
+      "loss": 0.1042,
+      "step": 173
+    },
+    {
+      "epoch": 0.5621970920840065,
+      "grad_norm": 1.0052690224759626,
+      "learning_rate": 4.515879464697629e-06,
+      "loss": 0.0676,
+      "step": 174
+    },
+    {
+      "epoch": 0.5654281098546042,
+      "grad_norm": 1.2550940752148396,
+      "learning_rate": 4.507492780456249e-06,
+      "loss": 0.0873,
+      "step": 175
+    },
+    {
+      "epoch": 0.568659127625202,
+      "grad_norm": 0.7510769288869017,
+      "learning_rate": 4.499042004336642e-06,
+      "loss": 0.0593,
+      "step": 176
+    },
+    {
+      "epoch": 0.5718901453957996,
+      "grad_norm": 1.0704861891317858,
+      "learning_rate": 4.490527406141081e-06,
+      "loss": 0.1084,
+      "step": 177
+    },
+    {
+      "epoch": 0.5751211631663974,
+      "grad_norm": 0.6284718527814975,
+      "learning_rate": 4.481949257709442e-06,
+      "loss": 0.0508,
+      "step": 178
+    },
+    {
+      "epoch": 0.5783521809369951,
+      "grad_norm": 0.7735314263403176,
+      "learning_rate": 4.4733078329105296e-06,
+      "loss": 0.071,
+      "step": 179
+    },
+    {
+      "epoch": 0.5815831987075929,
+      "grad_norm": 0.7328207313228111,
+      "learning_rate": 4.464603407633326e-06,
+      "loss": 0.0616,
+      "step": 180
+    },
+    {
+      "epoch": 0.5848142164781907,
+      "grad_norm": 0.8860553425416037,
+      "learning_rate": 4.455836259778193e-06,
+      "loss": 0.0933,
+      "step": 181
+    },
+    {
+      "epoch": 0.5880452342487884,
+      "grad_norm": 0.7335813522711138,
+      "learning_rate": 4.44700666924799e-06,
+      "loss": 0.0778,
+      "step": 182
+    },
+    {
+      "epoch": 0.5912762520193862,
+      "grad_norm": 1.0224460571070495,
+      "learning_rate": 4.438114917939145e-06,
+      "loss": 0.0816,
+      "step": 183
+    },
+    {
+      "epoch": 0.5945072697899838,
+      "grad_norm": 0.5929045336796159,
+      "learning_rate": 4.42916128973265e-06,
+      "loss": 0.0544,
+      "step": 184
+    },
+    {
+      "epoch": 0.5977382875605816,
+      "grad_norm": 0.9356825212983167,
+      "learning_rate": 4.420146070484997e-06,
+      "loss": 0.0697,
+      "step": 185
+    },
+    {
+      "epoch": 0.6009693053311793,
+      "grad_norm": 0.9940780034167175,
+      "learning_rate": 4.41106954801906e-06,
+      "loss": 0.0849,
+      "step": 186
+    },
+    {
+      "epoch": 0.6042003231017771,
+      "grad_norm": 1.1206234673942042,
+      "learning_rate": 4.401932012114893e-06,
+      "loss": 0.0822,
+      "step": 187
+    },
+    {
+      "epoch": 0.6074313408723748,
+      "grad_norm": 0.5583759826731935,
+      "learning_rate": 4.39273375450049e-06,
+      "loss": 0.0516,
+      "step": 188
+    },
+    {
+      "epoch": 0.6106623586429726,
+      "grad_norm": 0.6454315459498958,
+      "learning_rate": 4.383475068842464e-06,
+      "loss": 0.0577,
+      "step": 189
+    },
+    {
+      "epoch": 0.6138933764135702,
+      "grad_norm": 1.1332732216792498,
+      "learning_rate": 4.3741562507366754e-06,
+      "loss": 0.0655,
+      "step": 190
+    },
+    {
+      "epoch": 0.617124394184168,
+      "grad_norm": 0.8318517770779469,
+      "learning_rate": 4.36477759769879e-06,
+      "loss": 0.0954,
+      "step": 191
+    },
+    {
+      "epoch": 0.6203554119547657,
+      "grad_norm": 1.3985194163380206,
+      "learning_rate": 4.355339409154788e-06,
+      "loss": 0.119,
+      "step": 192
+    },
+    {
+      "epoch": 0.6235864297253635,
+      "grad_norm": 0.8270239497445794,
+      "learning_rate": 4.345841986431396e-06,
+      "loss": 0.0637,
+      "step": 193
+    },
+    {
+      "epoch": 0.6268174474959612,
+      "grad_norm": 0.7760464412137208,
+      "learning_rate": 4.336285632746472e-06,
+      "loss": 0.0884,
+      "step": 194
+    },
+    {
+      "epoch": 0.630048465266559,
+      "grad_norm": 0.7692351681728313,
+      "learning_rate": 4.326670653199323e-06,
+      "loss": 0.0585,
+      "step": 195
+    },
+    {
+      "epoch": 0.6332794830371568,
+      "grad_norm": 0.66023769139009,
+      "learning_rate": 4.316997354760965e-06,
+      "loss": 0.0678,
+      "step": 196
+    },
+    {
+      "epoch": 0.6365105008077544,
+      "grad_norm": 0.6890195746804293,
+      "learning_rate": 4.307266046264323e-06,
+      "loss": 0.0664,
+      "step": 197
+    },
+    {
+      "epoch": 0.6397415185783522,
+      "grad_norm": 0.847865514008143,
+      "learning_rate": 4.297477038394368e-06,
+      "loss": 0.0825,
+      "step": 198
+    },
+    {
+      "epoch": 0.6429725363489499,
+      "grad_norm": 0.9550408154272941,
+      "learning_rate": 4.287630643678204e-06,
+      "loss": 0.0744,
+      "step": 199
+    },
+    {
+      "epoch": 0.6462035541195477,
+      "grad_norm": 0.857502218920431,
+      "learning_rate": 4.2777271764750805e-06,
+      "loss": 0.0671,
+      "step": 200
+    },
+    {
+      "epoch": 0.6494345718901454,
+      "grad_norm": 0.8052709789557946,
+      "learning_rate": 4.267766952966369e-06,
+      "loss": 0.0612,
+      "step": 201
+    },
+    {
+      "epoch": 0.6526655896607432,
+      "grad_norm": 1.083089645941659,
+      "learning_rate": 4.257750291145457e-06,
+      "loss": 0.0822,
+      "step": 202
+    },
+    {
+      "epoch": 0.6558966074313409,
+      "grad_norm": 0.984740295179814,
+      "learning_rate": 4.247677510807602e-06,
+      "loss": 0.056,
+      "step": 203
+    },
+    {
+      "epoch": 0.6591276252019386,
+      "grad_norm": 0.8835020576624913,
+      "learning_rate": 4.237548933539718e-06,
+      "loss": 0.0758,
+      "step": 204
+    },
+    {
+      "epoch": 0.6623586429725363,
+      "grad_norm": 0.6564577266758135,
+      "learning_rate": 4.227364882710114e-06,
+      "loss": 0.0563,
+      "step": 205
+    },
+    {
+      "epoch": 0.6655896607431341,
+      "grad_norm": 1.1066641198832807,
+      "learning_rate": 4.217125683458162e-06,
+      "loss": 0.083,
+      "step": 206
+    },
+    {
+      "epoch": 0.6688206785137318,
+      "grad_norm": 0.6596522236176221,
+      "learning_rate": 4.206831662683922e-06,
+      "loss": 0.0661,
+      "step": 207
+    },
+    {
+      "epoch": 0.6720516962843296,
+      "grad_norm": 1.3814266436856864,
+      "learning_rate": 4.196483149037707e-06,
+      "loss": 0.1065,
+      "step": 208
+    },
+    {
+      "epoch": 0.6752827140549273,
+      "grad_norm": 0.8566187355631336,
+      "learning_rate": 4.186080472909582e-06,
+      "loss": 0.0798,
+      "step": 209
+    },
+    {
+      "epoch": 0.678513731825525,
+      "grad_norm": 0.7295355541151259,
+      "learning_rate": 4.1756239664188275e-06,
+      "loss": 0.0575,
+      "step": 210
+    },
+    {
+      "epoch": 0.6817447495961227,
+      "grad_norm": 1.1292799421470592,
+      "learning_rate": 4.165113963403326e-06,
+      "loss": 0.0721,
+      "step": 211
+    },
+    {
+      "epoch": 0.6849757673667205,
+      "grad_norm": 0.6756065514670196,
+      "learning_rate": 4.154550799408906e-06,
+      "loss": 0.0627,
+      "step": 212
+    },
+    {
+      "epoch": 0.6882067851373183,
+      "grad_norm": 0.7404052802160156,
+      "learning_rate": 4.143934811678637e-06,
+      "loss": 0.0535,
+      "step": 213
+    },
+    {
+      "epoch": 0.691437802907916,
+      "grad_norm": 1.022807263526455,
+      "learning_rate": 4.1332663391420515e-06,
+      "loss": 0.0854,
+      "step": 214
+    },
+    {
+      "epoch": 0.6946688206785138,
+      "grad_norm": 0.8472576700436739,
+      "learning_rate": 4.1225457224043316e-06,
+      "loss": 0.0623,
+      "step": 215
+    },
+    {
+      "epoch": 0.6978998384491115,
+      "grad_norm": 0.682205907382696,
+      "learning_rate": 4.111773303735432e-06,
+      "loss": 0.0523,
+      "step": 216
+    },
+    {
+      "epoch": 0.7011308562197092,
+      "grad_norm": 0.8113418772511839,
+      "learning_rate": 4.100949427059151e-06,
+      "loss": 0.0789,
+      "step": 217
+    },
+    {
+      "epoch": 0.7043618739903069,
+      "grad_norm": 0.9832925631785678,
+      "learning_rate": 4.090074437942155e-06,
+      "loss": 0.0723,
+      "step": 218
+    },
+    {
+      "epoch": 0.7075928917609047,
+      "grad_norm": 0.8721900501399756,
+      "learning_rate": 4.079148683582943e-06,
+      "loss": 0.0882,
+      "step": 219
+    },
+    {
+      "epoch": 0.7108239095315024,
+      "grad_norm": 0.6650598159524921,
+      "learning_rate": 4.06817251280076e-06,
+      "loss": 0.0645,
+      "step": 220
+    },
+    {
+      "epoch": 0.7140549273021002,
+      "grad_norm": 0.5794028281825343,
+      "learning_rate": 4.0571462760244626e-06,
+      "loss": 0.0505,
+      "step": 221
+    },
+    {
+      "epoch": 0.7172859450726979,
+      "grad_norm": 0.766211455982521,
+      "learning_rate": 4.046070325281333e-06,
+      "loss": 0.0716,
+      "step": 222
+    },
+    {
+      "epoch": 0.7205169628432956,
+      "grad_norm": 0.6405405389062213,
+      "learning_rate": 4.034945014185836e-06,
+      "loss": 0.0682,
+      "step": 223
+    },
+    {
+      "epoch": 0.7237479806138933,
+      "grad_norm": 0.7302480630044077,
+      "learning_rate": 4.0237706979283306e-06,
+      "loss": 0.0813,
+      "step": 224
+    },
+    {
+      "epoch": 0.7269789983844911,
+      "grad_norm": 0.8456954793821594,
+      "learning_rate": 4.012547733263734e-06,
+      "loss": 0.0672,
+      "step": 225
+    },
+    {
+      "epoch": 0.7302100161550888,
+      "grad_norm": 0.6987533622613603,
+      "learning_rate": 4.001276478500127e-06,
+      "loss": 0.0606,
+      "step": 226
+    },
+    {
+      "epoch": 0.7334410339256866,
+      "grad_norm": 0.7583800796325042,
+      "learning_rate": 3.989957293487314e-06,
+      "loss": 0.048,
+      "step": 227
+    },
+    {
+      "epoch": 0.7366720516962844,
+      "grad_norm": 0.8822810810943023,
+      "learning_rate": 3.978590539605338e-06,
+      "loss": 0.0604,
+      "step": 228
+    },
+    {
+      "epoch": 0.7399030694668821,
+      "grad_norm": 0.8557818511714256,
+      "learning_rate": 3.967176579752943e-06,
+      "loss": 0.0589,
+      "step": 229
+    },
+    {
+      "epoch": 0.7431340872374798,
+      "grad_norm": 0.5503873101604418,
+      "learning_rate": 3.955715778335984e-06,
+      "loss": 0.0528,
+      "step": 230
+    },
+    {
+      "epoch": 0.7463651050080775,
+      "grad_norm": 0.8045709480783542,
+      "learning_rate": 3.944208501255796e-06,
+      "loss": 0.0821,
+      "step": 231
+    },
+    {
+      "epoch": 0.7495961227786753,
+      "grad_norm": 0.752743452245411,
+      "learning_rate": 3.932655115897513e-06,
+      "loss": 0.0916,
+      "step": 232
+    },
+    {
+      "epoch": 0.752827140549273,
+      "grad_norm": 0.9726656057832658,
+      "learning_rate": 3.9210559911183345e-06,
+      "loss": 0.0826,
+      "step": 233
+    },
+    {
+      "epoch": 0.7560581583198708,
+      "grad_norm": 0.8247240601878854,
+      "learning_rate": 3.909411497235752e-06,
+      "loss": 0.0712,
+      "step": 234
+    },
+    {
+      "epoch": 0.7592891760904685,
+      "grad_norm": 0.5846761529324903,
+      "learning_rate": 3.89772200601573e-06,
+      "loss": 0.0497,
+      "step": 235
+    },
+    {
+      "epoch": 0.7625201938610663,
+      "grad_norm": 1.215906602352585,
+      "learning_rate": 3.885987890660828e-06,
+      "loss": 0.096,
+      "step": 236
+    },
+    {
+      "epoch": 0.7657512116316639,
+      "grad_norm": 0.7228133512999668,
+      "learning_rate": 3.874209525798293e-06,
+      "loss": 0.0676,
+      "step": 237
+    },
+    {
+      "epoch": 0.7689822294022617,
+      "grad_norm": 0.628874981721426,
+      "learning_rate": 3.862387287468095e-06,
+      "loss": 0.0638,
+      "step": 238
+    },
+    {
+      "epoch": 0.7722132471728594,
+      "grad_norm": 0.8665871369668172,
+      "learning_rate": 3.850521553110924e-06,
+      "loss": 0.0676,
+      "step": 239
+    },
+    {
+      "epoch": 0.7754442649434572,
+      "grad_norm": 0.7726368504889838,
+      "learning_rate": 3.838612701556138e-06,
+      "loss": 0.0617,
+      "step": 240
+    },
+    {
+      "epoch": 0.778675282714055,
+      "grad_norm": 0.8259876486738208,
+      "learning_rate": 3.826661113009671e-06,
+      "loss": 0.0827,
+      "step": 241
+    },
+    {
+      "epoch": 0.7819063004846527,
+      "grad_norm": 0.5825223238179703,
+      "learning_rate": 3.814667169041887e-06,
+      "loss": 0.0536,
+      "step": 242
+    },
+    {
+      "epoch": 0.7851373182552503,
+      "grad_norm": 0.8337660440150237,
+      "learning_rate": 3.8026312525754095e-06,
+      "loss": 0.0805,
+      "step": 243
+    },
+    {
+      "epoch": 0.7883683360258481,
+      "grad_norm": 1.2051840693103852,
+      "learning_rate": 3.790553747872885e-06,
+      "loss": 0.0876,
+      "step": 244
+    },
+    {
+      "epoch": 0.7915993537964459,
+      "grad_norm": 1.0177651978897424,
+      "learning_rate": 3.778435040524722e-06,
+      "loss": 0.088,
+      "step": 245
+    },
+    {
+      "epoch": 0.7948303715670436,
+      "grad_norm": 0.5088711298667009,
+      "learning_rate": 3.766275517436779e-06,
+      "loss": 0.0436,
+      "step": 246
+    },
+    {
+      "epoch": 0.7980613893376414,
+      "grad_norm": 0.7256417368432645,
+      "learning_rate": 3.75407556681801e-06,
+      "loss": 0.0567,
+      "step": 247
+    },
+    {
+      "epoch": 0.8012924071082391,
+      "grad_norm": 0.6647991839006098,
+      "learning_rate": 3.741835578168071e-06,
+      "loss": 0.0682,
+      "step": 248
+    },
+    {
+      "epoch": 0.8045234248788369,
+      "grad_norm": 0.6160939622053359,
+      "learning_rate": 3.7295559422648874e-06,
+      "loss": 0.0471,
+      "step": 249
+    },
+    {
+      "epoch": 0.8077544426494345,
+      "grad_norm": 0.9179464461903228,
+      "learning_rate": 3.717237051152175e-06,
+      "loss": 0.0774,
+      "step": 250
+    },
+    {
+      "epoch": 0.8109854604200323,
+      "grad_norm": 1.0639899619677535,
+      "learning_rate": 3.7048792981269245e-06,
+      "loss": 0.076,
+      "step": 251
+    },
+    {
+      "epoch": 0.81421647819063,
+      "grad_norm": 0.6225213021803089,
+      "learning_rate": 3.692483077726843e-06,
+      "loss": 0.0499,
+      "step": 252
+    },
+    {
+      "epoch": 0.8174474959612278,
+      "grad_norm": 0.6507281120193775,
+      "learning_rate": 3.6800487857177636e-06,
+      "loss": 0.0542,
+      "step": 253
+    },
+    {
+      "epoch": 0.8206785137318255,
+      "grad_norm": 0.8342956252285098,
+      "learning_rate": 3.6675768190810023e-06,
+      "loss": 0.0563,
+      "step": 254
+    },
+    {
+      "epoch": 0.8239095315024233,
+      "grad_norm": 0.8643206085017306,
+      "learning_rate": 3.6550675760006904e-06,
+      "loss": 0.0739,
+      "step": 255
+    },
+    {
+      "epoch": 0.827140549273021,
+      "grad_norm": 0.7946001867850625,
+      "learning_rate": 3.642521455851058e-06,
+      "loss": 0.0714,
+      "step": 256
+    },
+    {
+      "epoch": 0.8303715670436187,
+      "grad_norm": 0.8926582386596207,
+      "learning_rate": 3.629938859183686e-06,
+      "loss": 0.0713,
+      "step": 257
+    },
+    {
+      "epoch": 0.8336025848142165,
+      "grad_norm": 0.7491907200888189,
+      "learning_rate": 3.6173201877147134e-06,
+      "loss": 0.068,
+      "step": 258
+    },
+    {
+      "epoch": 0.8368336025848142,
+      "grad_norm": 1.1183078136862983,
+      "learning_rate": 3.6046658443120196e-06,
+      "loss": 0.1055,
+      "step": 259
+    },
+    {
+      "epoch": 0.840064620355412,
+      "grad_norm": 0.7909225982481742,
+      "learning_rate": 3.5919762329823556e-06,
+      "loss": 0.0695,
+      "step": 260
+    },
+    {
+      "epoch": 0.8432956381260097,
+      "grad_norm": 0.8739075343916631,
+      "learning_rate": 3.579251758858447e-06,
+      "loss": 0.069,
+      "step": 261
+    },
+    {
+      "epoch": 0.8465266558966075,
+      "grad_norm": 0.7730576890810616,
+      "learning_rate": 3.566492828186063e-06,
+      "loss": 0.0666,
+      "step": 262
+    },
+    {
+      "epoch": 0.8497576736672051,
+      "grad_norm": 0.9066328657599784,
+      "learning_rate": 3.5536998483110418e-06,
+      "loss": 0.0707,
+      "step": 263
+    },
+    {
+      "epoch": 0.8529886914378029,
+      "grad_norm": 0.713312105314591,
+      "learning_rate": 3.5408732276662882e-06,
+      "loss": 0.0913,
+      "step": 264
+    },
+    {
+      "epoch": 0.8562197092084006,
+      "grad_norm": 0.7057871040611122,
+      "learning_rate": 3.5280133757587343e-06,
+      "loss": 0.0591,
+      "step": 265
+    },
+    {
+      "epoch": 0.8594507269789984,
+      "grad_norm": 1.2427283512085103,
+      "learning_rate": 3.515120703156264e-06,
+      "loss": 0.0839,
+      "step": 266
+    },
+    {
+      "epoch": 0.8626817447495961,
+      "grad_norm": 0.7653071583973694,
+      "learning_rate": 3.5021956214746046e-06,
+      "loss": 0.0645,
+      "step": 267
+    },
+    {
+      "epoch": 0.8659127625201939,
+      "grad_norm": 0.6634481800817704,
+      "learning_rate": 3.4892385433641875e-06,
+      "loss": 0.0459,
+      "step": 268
+    },
+    {
+      "epoch": 0.8691437802907916,
+      "grad_norm": 0.6953364982312946,
+      "learning_rate": 3.4762498824969726e-06,
+      "loss": 0.0632,
+      "step": 269
+    },
+    {
+      "epoch": 0.8723747980613893,
+      "grad_norm": 0.6532166155768556,
+      "learning_rate": 3.4632300535532415e-06,
+      "loss": 0.053,
+      "step": 270
+    },
+    {
+      "epoch": 0.875605815831987,
+      "grad_norm": 0.685089930364244,
+      "learning_rate": 3.450179472208356e-06,
+      "loss": 0.0586,
+      "step": 271
+    },
+    {
+      "epoch": 0.8788368336025848,
+      "grad_norm": 0.79809625541097,
+      "learning_rate": 3.437098555119493e-06,
+      "loss": 0.053,
+      "step": 272
+    },
+    {
+      "epoch": 0.8820678513731826,
+      "grad_norm": 1.001360047718671,
+      "learning_rate": 3.4239877199123343e-06,
+      "loss": 0.0856,
+      "step": 273
+    },
+    {
+      "epoch": 0.8852988691437803,
+      "grad_norm": 0.9793506266219179,
+      "learning_rate": 3.4108473851677408e-06,
+      "loss": 0.0603,
+      "step": 274
+    },
+    {
+      "epoch": 0.8885298869143781,
+      "grad_norm": 1.1216710753897985,
+      "learning_rate": 3.397677970408384e-06,
+      "loss": 0.091,
+      "step": 275
+    },
+    {
+      "epoch": 0.8917609046849758,
+      "grad_norm": 1.0349146688296054,
+      "learning_rate": 3.3844798960853533e-06,
+      "loss": 0.0712,
+      "step": 276
+    },
+    {
+      "epoch": 0.8949919224555735,
+      "grad_norm": 1.3307977354919367,
+      "learning_rate": 3.3712535835647326e-06,
+      "loss": 0.0617,
+      "step": 277
+    },
+    {
+      "epoch": 0.8982229402261712,
+      "grad_norm": 0.7132616053518929,
+      "learning_rate": 3.357999455114148e-06,
+      "loss": 0.0659,
+      "step": 278
+    },
+    {
+      "epoch": 0.901453957996769,
+      "grad_norm": 0.9273582042526977,
+      "learning_rate": 3.344717933889289e-06,
+      "loss": 0.0641,
+      "step": 279
+    },
+    {
+      "epoch": 0.9046849757673667,
+      "grad_norm": 0.9302335920003337,
+      "learning_rate": 3.3314094439203903e-06,
+      "loss": 0.0626,
+      "step": 280
+    },
+    {
+      "epoch": 0.9079159935379645,
+      "grad_norm": 0.45357719437194755,
+      "learning_rate": 3.3180744100987045e-06,
+      "loss": 0.0431,
+      "step": 281
+    },
+    {
+      "epoch": 0.9111470113085622,
+      "grad_norm": 0.8331062557625248,
+      "learning_rate": 3.3047132581629297e-06,
+      "loss": 0.0755,
+      "step": 282
+    },
+    {
+      "epoch": 0.9143780290791599,
+      "grad_norm": 0.7394449396136501,
+      "learning_rate": 3.29132641468562e-06,
+      "loss": 0.0566,
+      "step": 283
+    },
+    {
+      "epoch": 0.9176090468497576,
+      "grad_norm": 0.9737240770489647,
+      "learning_rate": 3.277914307059566e-06,
+      "loss": 0.0801,
+      "step": 284
+    },
+    {
+      "epoch": 0.9208400646203554,
+      "grad_norm": 0.7702804759383343,
+      "learning_rate": 3.264477363484151e-06,
+      "loss": 0.0573,
+      "step": 285
+    },
+    {
+      "epoch": 0.9240710823909531,
+      "grad_norm": 0.5835013641966551,
+      "learning_rate": 3.251016012951678e-06,
+      "loss": 0.0562,
+      "step": 286
+    },
+    {
+      "epoch": 0.9273021001615509,
+      "grad_norm": 0.5371255642799613,
+      "learning_rate": 3.237530685233673e-06,
+      "loss": 0.0413,
+      "step": 287
+    },
+    {
+      "epoch": 0.9305331179321487,
+      "grad_norm": 0.816748025986031,
+      "learning_rate": 3.2240218108671683e-06,
+      "loss": 0.0534,
+      "step": 288
+    },
+    {
+      "epoch": 0.9337641357027464,
+      "grad_norm": 0.712574290987934,
+      "learning_rate": 3.2104898211409546e-06,
+      "loss": 0.0635,
+      "step": 289
+    },
+    {
+      "epoch": 0.9369951534733441,
+      "grad_norm": 0.7596552800433577,
+      "learning_rate": 3.196935148081808e-06,
+      "loss": 0.0562,
+      "step": 290
+    },
+    {
+      "epoch": 0.9402261712439418,
+      "grad_norm": 0.6394269281528583,
+      "learning_rate": 3.1833582244407036e-06,
+      "loss": 0.0569,
+      "step": 291
+    },
+    {
+      "epoch": 0.9434571890145396,
+      "grad_norm": 0.8847550945975203,
+      "learning_rate": 3.1697594836789924e-06,
+      "loss": 0.0854,
+      "step": 292
+    },
+    {
+      "epoch": 0.9466882067851373,
+      "grad_norm": 0.6380616258246674,
+      "learning_rate": 3.156139359954569e-06,
+      "loss": 0.0603,
+      "step": 293
+    },
+    {
+      "epoch": 0.9499192245557351,
+      "grad_norm": 0.8054064646081707,
+      "learning_rate": 3.142498288108007e-06,
+      "loss": 0.0811,
+      "step": 294
+    },
+    {
+      "epoch": 0.9531502423263328,
+      "grad_norm": 0.6206438760553665,
+      "learning_rate": 3.128836703648676e-06,
+      "loss": 0.0702,
+      "step": 295
+    },
+    {
+      "epoch": 0.9563812600969306,
+      "grad_norm": 0.925371456511262,
+      "learning_rate": 3.1151550427408383e-06,
+      "loss": 0.0663,
+      "step": 296
+    },
+    {
+      "epoch": 0.9596122778675282,
+      "grad_norm": 0.8737367079669791,
+      "learning_rate": 3.1014537421897222e-06,
+      "loss": 0.078,
+      "step": 297
+    },
+    {
+      "epoch": 0.962843295638126,
+      "grad_norm": 0.8790452308237618,
+      "learning_rate": 3.0877332394275806e-06,
+      "loss": 0.0796,
+      "step": 298
+    },
+    {
+      "epoch": 0.9660743134087237,
+      "grad_norm": 0.8200476689531796,
+      "learning_rate": 3.0739939724997205e-06,
+      "loss": 0.0777,
+      "step": 299
+    },
+    {
+      "epoch": 0.9693053311793215,
+      "grad_norm": 0.91054817725634,
+      "learning_rate": 3.0602363800505198e-06,
+      "loss": 0.0823,
+      "step": 300
+    },
+    {
+      "epoch": 0.9725363489499192,
+      "grad_norm": 1.0055625552384646,
+      "learning_rate": 3.0464609013094244e-06,
+      "loss": 0.0723,
+      "step": 301
+    },
+    {
+      "epoch": 0.975767366720517,
+      "grad_norm": 0.8948435648737145,
+      "learning_rate": 3.032667976076923e-06,
+      "loss": 0.0776,
+      "step": 302
+    },
+    {
+      "epoch": 0.9789983844911146,
+      "grad_norm": 0.6947875633713175,
+      "learning_rate": 3.0188580447105055e-06,
+      "loss": 0.0576,
+      "step": 303
+    },
+    {
+      "epoch": 0.9822294022617124,
+      "grad_norm": 0.710748833707824,
+      "learning_rate": 3.0050315481106074e-06,
+      "loss": 0.0606,
+      "step": 304
+    },
+    {
+      "epoch": 0.9854604200323102,
+      "grad_norm": 0.4721520789494047,
+      "learning_rate": 2.9911889277065314e-06,
+      "loss": 0.043,
+      "step": 305
+    },
+    {
+      "epoch": 0.9886914378029079,
+      "grad_norm": 0.7878237728564794,
+      "learning_rate": 2.977330625442352e-06,
+      "loss": 0.0832,
+      "step": 306
+    },
+    {
+      "epoch": 0.9919224555735057,
+      "grad_norm": 0.818350538492891,
+      "learning_rate": 2.963457083762809e-06,
+      "loss": 0.0727,
+      "step": 307
+    },
+    {
+      "epoch": 0.9951534733441034,
+      "grad_norm": 0.8240104583346066,
+      "learning_rate": 2.949568745599182e-06,
+      "loss": 0.0698,
+      "step": 308
+    },
+    {
+      "epoch": 0.9983844911147012,
+      "grad_norm": 0.6820819240767714,
+      "learning_rate": 2.935666054355146e-06,
+      "loss": 0.0604,
+      "step": 309
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6820819240767714,
+      "learning_rate": 2.921749453892618e-06,
+      "loss": 0.0453,
+      "step": 310
+    },
+    {
+      "epoch": 1.0032310177705976,
+      "grad_norm": 0.8574995005049099,
+      "learning_rate": 2.9078193885175875e-06,
+      "loss": 0.0401,
+      "step": 311
+    },
+    {
+      "epoch": 1.0064620355411955,
+      "grad_norm": 0.7517375531460555,
+      "learning_rate": 2.893876302965925e-06,
+      "loss": 0.0646,
+      "step": 312
+    },
+    {
+      "epoch": 1.0096930533117932,
+      "grad_norm": 0.7030688450315203,
+      "learning_rate": 2.8799206423891895e-06,
+      "loss": 0.0428,
+      "step": 313
+    },
+    {
+      "epoch": 1.012924071082391,
+      "grad_norm": 0.6810197305641746,
+      "learning_rate": 2.865952852340417e-06,
+      "loss": 0.0491,
+      "step": 314
+    },
+    {
+      "epoch": 1.0161550888529887,
+      "grad_norm": 0.6457251139138377,
+      "learning_rate": 2.8519733787598887e-06,
+      "loss": 0.0527,
+      "step": 315
+    },
+    {
+      "epoch": 1.0193861066235865,
+      "grad_norm": 0.6750520782194651,
+      "learning_rate": 2.8379826679609e-06,
+      "loss": 0.0456,
+      "step": 316
+    },
+    {
+      "epoch": 1.0226171243941842,
+      "grad_norm": 0.7287551036819017,
+      "learning_rate": 2.8239811666155105e-06,
+      "loss": 0.0703,
+      "step": 317
+    },
+    {
+      "epoch": 1.0258481421647818,
+      "grad_norm": 0.5506534914240341,
+      "learning_rate": 2.8099693217402807e-06,
+      "loss": 0.0461,
+      "step": 318
+    },
+    {
+      "epoch": 1.0290791599353797,
+      "grad_norm": 0.6716945269140621,
+      "learning_rate": 2.795947580682003e-06,
+      "loss": 0.0424,
+      "step": 319
+    },
+    {
+      "epoch": 1.0323101777059773,
+      "grad_norm": 0.6029067986975312,
+      "learning_rate": 2.7819163911034175e-06,
+      "loss": 0.032,
+      "step": 320
+    },
+    {
+      "epoch": 1.0355411954765752,
+      "grad_norm": 0.5754461947410837,
+      "learning_rate": 2.767876200968923e-06,
+      "loss": 0.0417,
+      "step": 321
+    },
+    {
+      "epoch": 1.0387722132471728,
+      "grad_norm": 0.5670203427921292,
+      "learning_rate": 2.7538274585302707e-06,
+      "loss": 0.0676,
+      "step": 322
+    },
+    {
+      "epoch": 1.0420032310177707,
+      "grad_norm": 0.6371474674240052,
+      "learning_rate": 2.7397706123122563e-06,
+      "loss": 0.0401,
+      "step": 323
+    },
+    {
+      "epoch": 1.0452342487883683,
+      "grad_norm": 0.633563817351995,
+      "learning_rate": 2.7257061110984005e-06,
+      "loss": 0.0507,
+      "step": 324
+    },
+    {
+      "epoch": 1.048465266558966,
+      "grad_norm": 0.455254402627076,
+      "learning_rate": 2.7116344039166192e-06,
+      "loss": 0.0346,
+      "step": 325
+    },
+    {
+      "epoch": 1.0516962843295639,
+      "grad_norm": 0.46808216871215624,
+      "learning_rate": 2.6975559400248876e-06,
+      "loss": 0.0447,
+      "step": 326
+    },
+    {
+      "epoch": 1.0549273021001615,
+      "grad_norm": 0.5421960059682478,
+      "learning_rate": 2.683471168896899e-06,
+      "loss": 0.0412,
+      "step": 327
+    },
+    {
+      "epoch": 1.0581583198707594,
+      "grad_norm": 0.6365474014428788,
+      "learning_rate": 2.6693805402077123e-06,
+      "loss": 0.0482,
+      "step": 328
+    },
+    {
+      "epoch": 1.061389337641357,
+      "grad_norm": 0.5544336402631056,
+      "learning_rate": 2.6552845038193977e-06,
+      "loss": 0.0392,
+      "step": 329
+    },
+    {
+      "epoch": 1.0646203554119547,
+      "grad_norm": 0.49713090653008446,
+      "learning_rate": 2.641183509766675e-06,
+      "loss": 0.028,
+      "step": 330
+    },
+    {
+      "epoch": 1.0678513731825525,
+      "grad_norm": 0.42542651226656086,
+      "learning_rate": 2.627078008242541e-06,
+      "loss": 0.0361,
+      "step": 331
+    },
+    {
+      "epoch": 1.0710823909531502,
+      "grad_norm": 0.5295471778809274,
+      "learning_rate": 2.6129684495839013e-06,
+      "loss": 0.0657,
+      "step": 332
+    },
+    {
+      "epoch": 1.074313408723748,
+      "grad_norm": 0.5910271655601097,
+      "learning_rate": 2.5988552842571903e-06,
+      "loss": 0.0447,
+      "step": 333
+    },
+    {
+      "epoch": 1.0775444264943457,
+      "grad_norm": 0.5391865465723893,
+      "learning_rate": 2.5847389628439905e-06,
+      "loss": 0.0468,
+      "step": 334
+    },
+    {
+      "epoch": 1.0807754442649435,
+      "grad_norm": 0.7945092160241262,
+      "learning_rate": 2.570619936026647e-06,
+      "loss": 0.0526,
+      "step": 335
+    },
+    {
+      "epoch": 1.0840064620355412,
+      "grad_norm": 0.6384237762747144,
+      "learning_rate": 2.5564986545738767e-06,
+      "loss": 0.0435,
+      "step": 336
+    },
+    {
+      "epoch": 1.0872374798061388,
+      "grad_norm": 0.6166409813836169,
+      "learning_rate": 2.542375569326382e-06,
+      "loss": 0.0405,
+      "step": 337
+    },
+    {
+      "epoch": 1.0904684975767367,
+      "grad_norm": 0.5773480639364901,
+      "learning_rate": 2.52825113118245e-06,
+      "loss": 0.0534,
+      "step": 338
+    },
+    {
+      "epoch": 1.0936995153473343,
+      "grad_norm": 0.5202558074438228,
+      "learning_rate": 2.514125791083563e-06,
+      "loss": 0.0507,
+      "step": 339
+    },
+    {
+      "epoch": 1.0969305331179322,
+      "grad_norm": 0.6263121460011797,
+      "learning_rate": 2.5e-06,
+      "loss": 0.0512,
+      "step": 340
+    },
+    {
+      "epoch": 1.1001615508885298,
+      "grad_norm": 0.7123374131643287,
+      "learning_rate": 2.485874208916438e-06,
+      "loss": 0.0332,
+      "step": 341
+    },
+    {
+      "epoch": 1.1033925686591277,
+      "grad_norm": 0.4202479639460712,
+      "learning_rate": 2.4717488688175513e-06,
+      "loss": 0.0407,
+      "step": 342
+    },
+    {
+      "epoch": 1.1066235864297254,
+      "grad_norm": 0.642912112013506,
+      "learning_rate": 2.4576244306736193e-06,
+      "loss": 0.0454,
+      "step": 343
+    },
+    {
+      "epoch": 1.109854604200323,
+      "grad_norm": 0.5278433369392868,
+      "learning_rate": 2.4435013454261246e-06,
+      "loss": 0.0454,
+      "step": 344
+    },
+    {
+      "epoch": 1.1130856219709209,
+      "grad_norm": 0.49367590472244555,
+      "learning_rate": 2.4293800639733537e-06,
+      "loss": 0.0425,
+      "step": 345
+    },
+    {
+      "epoch": 1.1163166397415185,
+      "grad_norm": 0.663577683371772,
+      "learning_rate": 2.4152610371560095e-06,
+      "loss": 0.0502,
+      "step": 346
+    },
+    {
+      "epoch": 1.1195476575121164,
+      "grad_norm": 0.6258554932784628,
+      "learning_rate": 2.40114471574281e-06,
+      "loss": 0.0503,
+      "step": 347
+    },
+    {
+      "epoch": 1.122778675282714,
+      "grad_norm": 0.6117692229579965,
+      "learning_rate": 2.3870315504160995e-06,
+      "loss": 0.0357,
+      "step": 348
+    },
+    {
+      "epoch": 1.1260096930533119,
+      "grad_norm": 0.698709357123754,
+      "learning_rate": 2.3729219917574597e-06,
+      "loss": 0.0398,
+      "step": 349
+    },
+    {
+      "epoch": 1.1292407108239095,
+      "grad_norm": 0.5509108322599596,
+      "learning_rate": 2.358816490233326e-06,
+      "loss": 0.0504,
+      "step": 350
+    },
+    {
+      "epoch": 1.1324717285945072,
+      "grad_norm": 0.5944772383426618,
+      "learning_rate": 2.3447154961806027e-06,
+      "loss": 0.0417,
+      "step": 351
+    },
+    {
+      "epoch": 1.135702746365105,
+      "grad_norm": 0.7597386477794142,
+      "learning_rate": 2.330619459792289e-06,
+      "loss": 0.0469,
+      "step": 352
+    },
+    {
+      "epoch": 1.1389337641357027,
+      "grad_norm": 0.5719437487716487,
+      "learning_rate": 2.3165288311031024e-06,
+      "loss": 0.0294,
+      "step": 353
+    },
+    {
+      "epoch": 1.1421647819063006,
+      "grad_norm": 0.6213273588952939,
+      "learning_rate": 2.3024440599751132e-06,
+      "loss": 0.0384,
+      "step": 354
+    },
+    {
+      "epoch": 1.1453957996768982,
+      "grad_norm": 0.6354291435615703,
+      "learning_rate": 2.288365596083381e-06,
+      "loss": 0.0349,
+      "step": 355
+    },
+    {
+      "epoch": 1.148626817447496,
+      "grad_norm": 0.7486710035630336,
+      "learning_rate": 2.274293888901599e-06,
+      "loss": 0.0406,
+      "step": 356
+    },
+    {
+      "epoch": 1.1518578352180937,
+      "grad_norm": 0.46539243165120686,
+      "learning_rate": 2.260229387687744e-06,
+      "loss": 0.0406,
+      "step": 357
+    },
+    {
+      "epoch": 1.1550888529886914,
+      "grad_norm": 0.8582980219163947,
+      "learning_rate": 2.24617254146973e-06,
+      "loss": 0.0751,
+      "step": 358
+    },
+    {
+      "epoch": 1.1583198707592892,
+      "grad_norm": 0.5134947351957787,
+      "learning_rate": 2.232123799031078e-06,
+      "loss": 0.0257,
+      "step": 359
+    },
+    {
+      "epoch": 1.1615508885298869,
+      "grad_norm": 0.6414286514892072,
+      "learning_rate": 2.2180836088965833e-06,
+      "loss": 0.053,
+      "step": 360
+    },
+    {
+      "epoch": 1.1647819063004847,
+      "grad_norm": 0.6893327762289948,
+      "learning_rate": 2.204052419317998e-06,
+      "loss": 0.0518,
+      "step": 361
+    },
+    {
+      "epoch": 1.1680129240710824,
+      "grad_norm": 0.6315851441021518,
+      "learning_rate": 2.19003067825972e-06,
+      "loss": 0.0311,
+      "step": 362
+    },
+    {
+      "epoch": 1.1712439418416802,
+      "grad_norm": 0.506389589217495,
+      "learning_rate": 2.1760188333844907e-06,
+      "loss": 0.03,
+      "step": 363
+    },
+    {
+      "epoch": 1.1744749596122779,
+      "grad_norm": 0.48074703533770824,
+      "learning_rate": 2.1620173320391007e-06,
+      "loss": 0.0575,
+      "step": 364
+    },
+    {
+      "epoch": 1.1777059773828755,
+      "grad_norm": 0.7338858093688776,
+      "learning_rate": 2.1480266212401117e-06,
+      "loss": 0.0652,
+      "step": 365
+    },
+    {
+      "epoch": 1.1809369951534734,
+      "grad_norm": 0.5215250199335827,
+      "learning_rate": 2.1340471476595836e-06,
+      "loss": 0.0404,
+      "step": 366
+    },
+    {
+      "epoch": 1.184168012924071,
+      "grad_norm": 0.5101940305735346,
+      "learning_rate": 2.1200793576108105e-06,
+      "loss": 0.04,
+      "step": 367
+    },
+    {
+      "epoch": 1.187399030694669,
+      "grad_norm": 0.6072945357387388,
+      "learning_rate": 2.1061236970340756e-06,
+      "loss": 0.0471,
+      "step": 368
+    },
+    {
+      "epoch": 1.1906300484652665,
+      "grad_norm": 0.6485948720338923,
+      "learning_rate": 2.0921806114824134e-06,
+      "loss": 0.0402,
+      "step": 369
+    },
+    {
+      "epoch": 1.1938610662358644,
+      "grad_norm": 0.5618119097496929,
+      "learning_rate": 2.0782505461073822e-06,
+      "loss": 0.0309,
+      "step": 370
+    },
+    {
+      "epoch": 1.197092084006462,
+      "grad_norm": 0.5710364970799129,
+      "learning_rate": 2.0643339456448547e-06,
+      "loss": 0.0534,
+      "step": 371
+    },
+    {
+      "epoch": 1.2003231017770597,
+      "grad_norm": 0.5166825620149433,
+      "learning_rate": 2.0504312544008193e-06,
+      "loss": 0.0443,
+      "step": 372
+    },
+    {
+      "epoch": 1.2035541195476576,
+      "grad_norm": 0.6582279195349342,
+      "learning_rate": 2.0365429162371922e-06,
+      "loss": 0.042,
+      "step": 373
+    },
+    {
+      "epoch": 1.2067851373182552,
+      "grad_norm": 0.5628923740662667,
+      "learning_rate": 2.0226693745576494e-06,
+      "loss": 0.0338,
+      "step": 374
+    },
+    {
+      "epoch": 1.210016155088853,
+      "grad_norm": 0.5269299240421488,
+      "learning_rate": 2.008811072293469e-06,
+      "loss": 0.0333,
+      "step": 375
+    },
+    {
+      "epoch": 1.2132471728594507,
+      "grad_norm": 0.37904232204217875,
+      "learning_rate": 1.9949684518893926e-06,
+      "loss": 0.0321,
+      "step": 376
+    },
+    {
+      "epoch": 1.2164781906300486,
+      "grad_norm": 0.5084850707267878,
+      "learning_rate": 1.9811419552894953e-06,
+      "loss": 0.0408,
+      "step": 377
+    },
+    {
+      "epoch": 1.2197092084006462,
+      "grad_norm": 0.4746627443509293,
+      "learning_rate": 1.9673320239230783e-06,
+      "loss": 0.0359,
+      "step": 378
+    },
+    {
+      "epoch": 1.2229402261712439,
+      "grad_norm": 0.6346489110095358,
+      "learning_rate": 1.9535390986905764e-06,
+      "loss": 0.0505,
+      "step": 379
+    },
+    {
+      "epoch": 1.2261712439418417,
+      "grad_norm": 0.6162762793044799,
+      "learning_rate": 1.939763619949481e-06,
+      "loss": 0.0327,
+      "step": 380
+    },
+    {
+      "epoch": 1.2294022617124394,
+      "grad_norm": 0.7282286592219928,
+      "learning_rate": 1.92600602750028e-06,
+      "loss": 0.0664,
+      "step": 381
+    },
+    {
+      "epoch": 1.2326332794830372,
+      "grad_norm": 0.687246463772339,
+      "learning_rate": 1.9122667605724202e-06,
+      "loss": 0.0485,
+      "step": 382
+    },
+    {
+      "epoch": 1.235864297253635,
+      "grad_norm": 0.6284714240874483,
+      "learning_rate": 1.8985462578102786e-06,
+      "loss": 0.0505,
+      "step": 383
+    },
+    {
+      "epoch": 1.2390953150242328,
+      "grad_norm": 0.6820536891979064,
+      "learning_rate": 1.884844957259163e-06,
+      "loss": 0.0311,
+      "step": 384
+    },
+    {
+      "epoch": 1.2423263327948304,
+      "grad_norm": 0.549881437893592,
+      "learning_rate": 1.8711632963513237e-06,
+      "loss": 0.0375,
+      "step": 385
+    },
+    {
+      "epoch": 1.245557350565428,
+      "grad_norm": 0.6650268250589539,
+      "learning_rate": 1.857501711891993e-06,
+      "loss": 0.054,
+      "step": 386
+    },
+    {
+      "epoch": 1.248788368336026,
+      "grad_norm": 0.5113954409802595,
+      "learning_rate": 1.8438606400454312e-06,
+      "loss": 0.031,
+      "step": 387
+    },
+    {
+      "epoch": 1.2520193861066236,
+      "grad_norm": 0.5332778481364839,
+      "learning_rate": 1.830240516321008e-06,
+      "loss": 0.0463,
+      "step": 388
+    },
+    {
+      "epoch": 1.2552504038772212,
+      "grad_norm": 0.5928101900055703,
+      "learning_rate": 1.8166417755592975e-06,
+      "loss": 0.0353,
+      "step": 389
+    },
+    {
+      "epoch": 1.258481421647819,
+      "grad_norm": 0.4719442237818212,
+      "learning_rate": 1.8030648519181926e-06,
+      "loss": 0.037,
+      "step": 390
+    },
+    {
+      "epoch": 1.261712439418417,
+      "grad_norm": 0.6879135991203472,
+      "learning_rate": 1.789510178859046e-06,
+      "loss": 0.0523,
+      "step": 391
+    },
+    {
+      "epoch": 1.2649434571890146,
+      "grad_norm": 0.6862575200798275,
+      "learning_rate": 1.7759781891328321e-06,
+      "loss": 0.0344,
+      "step": 392
+    },
+    {
+      "epoch": 1.2681744749596122,
+      "grad_norm": 0.4157263515958759,
+      "learning_rate": 1.762469314766328e-06,
+      "loss": 0.0373,
+      "step": 393
+    },
+    {
+      "epoch": 1.27140549273021,
+      "grad_norm": 0.6888877331536797,
+      "learning_rate": 1.7489839870483236e-06,
+      "loss": 0.0493,
+      "step": 394
+    },
+    {
+      "epoch": 1.2746365105008077,
+      "grad_norm": 0.7225121433676083,
+      "learning_rate": 1.7355226365158489e-06,
+      "loss": 0.0685,
+      "step": 395
+    },
+    {
+      "epoch": 1.2778675282714054,
+      "grad_norm": 0.7380932165691942,
+      "learning_rate": 1.7220856929404342e-06,
+      "loss": 0.068,
+      "step": 396
+    },
+    {
+      "epoch": 1.2810985460420032,
+      "grad_norm": 0.8373452926500001,
+      "learning_rate": 1.7086735853143803e-06,
+      "loss": 0.0263,
+      "step": 397
+    },
+    {
+      "epoch": 1.284329563812601,
+      "grad_norm": 0.4057578810529795,
+      "learning_rate": 1.6952867418370707e-06,
+      "loss": 0.0435,
+      "step": 398
+    },
+    {
+      "epoch": 1.2875605815831987,
+      "grad_norm": 0.8172156898788309,
+      "learning_rate": 1.6819255899012963e-06,
+      "loss": 0.0521,
+      "step": 399
+    },
+    {
+      "epoch": 1.2907915993537964,
+      "grad_norm": 0.5540166018870007,
+      "learning_rate": 1.6685905560796101e-06,
+      "loss": 0.0301,
+      "step": 400
+    },
+    {
+      "epoch": 1.2940226171243943,
+      "grad_norm": 0.5152650366438487,
+      "learning_rate": 1.6552820661107119e-06,
+      "loss": 0.0357,
+      "step": 401
+    },
+    {
+      "epoch": 1.297253634894992,
+      "grad_norm": 0.5108482274841849,
+      "learning_rate": 1.6420005448858522e-06,
+      "loss": 0.0431,
+      "step": 402
+    },
+    {
+      "epoch": 1.3004846526655895,
+      "grad_norm": 0.4240316233127095,
+      "learning_rate": 1.6287464164352684e-06,
+      "loss": 0.0342,
+      "step": 403
+    },
+    {
+      "epoch": 1.3037156704361874,
+      "grad_norm": 0.5910340203210401,
+      "learning_rate": 1.6155201039146478e-06,
+      "loss": 0.0466,
+      "step": 404
+    },
+    {
+      "epoch": 1.306946688206785,
+      "grad_norm": 0.4716663558161469,
+      "learning_rate": 1.6023220295916162e-06,
+      "loss": 0.0247,
+      "step": 405
+    },
+    {
+      "epoch": 1.310177705977383,
+      "grad_norm": 0.5348908739290964,
+      "learning_rate": 1.5891526148322594e-06,
+      "loss": 0.0448,
+      "step": 406
+    },
+    {
+      "epoch": 1.3134087237479806,
+      "grad_norm": 0.7405657033049722,
+      "learning_rate": 1.576012280087666e-06,
+      "loss": 0.0425,
+      "step": 407
+    },
+    {
+      "epoch": 1.3166397415185784,
+      "grad_norm": 0.5749817118431164,
+      "learning_rate": 1.562901444880508e-06,
+      "loss": 0.0415,
+      "step": 408
+    },
+    {
+      "epoch": 1.319870759289176,
+      "grad_norm": 0.659084728824521,
+      "learning_rate": 1.5498205277916444e-06,
+      "loss": 0.0525,
+      "step": 409
+    },
+    {
+      "epoch": 1.3231017770597737,
+      "grad_norm": 0.4660035971953124,
+      "learning_rate": 1.5367699464467596e-06,
+      "loss": 0.0352,
+      "step": 410
+    },
+    {
+      "epoch": 1.3263327948303716,
+      "grad_norm": 0.6397348475129173,
+      "learning_rate": 1.523750117503028e-06,
+      "loss": 0.0456,
+      "step": 411
+    },
+    {
+      "epoch": 1.3295638126009692,
+      "grad_norm": 0.6385407976586193,
+      "learning_rate": 1.5107614566358136e-06,
+      "loss": 0.0562,
+      "step": 412
+    },
+    {
+      "epoch": 1.332794830371567,
+      "grad_norm": 0.5922977287717205,
+      "learning_rate": 1.4978043785253964e-06,
+      "loss": 0.0333,
+      "step": 413
+    },
+    {
+      "epoch": 1.3360258481421647,
+      "grad_norm": 0.47241930141595956,
+      "learning_rate": 1.4848792968437376e-06,
+      "loss": 0.0311,
+      "step": 414
+    },
+    {
+      "epoch": 1.3392568659127626,
+      "grad_norm": 0.43287945232089914,
+      "learning_rate": 1.4719866242412661e-06,
+      "loss": 0.0384,
+      "step": 415
+    },
+    {
+      "epoch": 1.3424878836833603,
+      "grad_norm": 0.5476421812276494,
+      "learning_rate": 1.4591267723337122e-06,
+      "loss": 0.0285,
+      "step": 416
+    },
+    {
+      "epoch": 1.345718901453958,
+      "grad_norm": 0.6654675392000585,
+      "learning_rate": 1.4463001516889597e-06,
+      "loss": 0.0546,
+      "step": 417
+    },
+    {
+      "epoch": 1.3489499192245558,
+      "grad_norm": 0.5265404309231011,
+      "learning_rate": 1.4335071718139379e-06,
+      "loss": 0.0563,
+      "step": 418
+    },
+    {
+      "epoch": 1.3521809369951534,
+      "grad_norm": 0.7273168490119778,
+      "learning_rate": 1.4207482411415532e-06,
+      "loss": 0.0319,
+      "step": 419
+    },
+    {
+      "epoch": 1.3554119547657513,
+      "grad_norm": 0.5515284785247014,
+      "learning_rate": 1.4080237670176456e-06,
+      "loss": 0.0465,
+      "step": 420
+    },
+    {
+      "epoch": 1.358642972536349,
+      "grad_norm": 0.6815587580127385,
+      "learning_rate": 1.395334155687981e-06,
+      "loss": 0.0396,
+      "step": 421
+    },
+    {
+      "epoch": 1.3618739903069468,
+      "grad_norm": 0.5046530996438253,
+      "learning_rate": 1.382679812285287e-06,
+      "loss": 0.0359,
+      "step": 422
+    },
+    {
+      "epoch": 1.3651050080775444,
+      "grad_norm": 0.636616018716514,
+      "learning_rate": 1.3700611408163158e-06,
+      "loss": 0.0364,
+      "step": 423
+    },
+    {
+      "epoch": 1.368336025848142,
+      "grad_norm": 0.8016482324734947,
+      "learning_rate": 1.357478544148943e-06,
+      "loss": 0.0548,
+      "step": 424
+    },
+    {
+      "epoch": 1.37156704361874,
+      "grad_norm": 0.7335385001009505,
+      "learning_rate": 1.3449324239993094e-06,
+      "loss": 0.0307,
+      "step": 425
+    },
+    {
+      "epoch": 1.3747980613893376,
+      "grad_norm": 0.5405071350061603,
+      "learning_rate": 1.3324231809189985e-06,
+      "loss": 0.0435,
+      "step": 426
+    },
+    {
+      "epoch": 1.3780290791599354,
+      "grad_norm": 0.5968950584538801,
+      "learning_rate": 1.3199512142822374e-06,
+      "loss": 0.0442,
+      "step": 427
+    },
+    {
+      "epoch": 1.381260096930533,
+      "grad_norm": 0.5586370191823509,
+      "learning_rate": 1.3075169222731573e-06,
+      "loss": 0.034,
+      "step": 428
+    },
+    {
+      "epoch": 1.384491114701131,
+      "grad_norm": 0.6462203526051947,
+      "learning_rate": 1.2951207018730772e-06,
+      "loss": 0.0591,
+      "step": 429
+    },
+    {
+      "epoch": 1.3877221324717286,
+      "grad_norm": 0.4892957539044203,
+      "learning_rate": 1.2827629488478254e-06,
+      "loss": 0.0345,
+      "step": 430
+    },
+    {
+      "epoch": 1.3909531502423262,
+      "grad_norm": 0.543453607383533,
+      "learning_rate": 1.270444057735113e-06,
+      "loss": 0.0275,
+      "step": 431
+    },
+    {
+      "epoch": 1.394184168012924,
+      "grad_norm": 0.6522770107038732,
+      "learning_rate": 1.25816442183193e-06,
+      "loss": 0.0446,
+      "step": 432
+    },
+    {
+      "epoch": 1.3974151857835218,
+      "grad_norm": 0.54367348046309,
+      "learning_rate": 1.2459244331819912e-06,
+      "loss": 0.0381,
+      "step": 433
+    },
+    {
+      "epoch": 1.4006462035541196,
+      "grad_norm": 0.6775444066593574,
+      "learning_rate": 1.2337244825632217e-06,
+      "loss": 0.0507,
+      "step": 434
+    },
+    {
+      "epoch": 1.4038772213247173,
+      "grad_norm": 0.7125046807656238,
+      "learning_rate": 1.2215649594752782e-06,
+      "loss": 0.0564,
+      "step": 435
+    },
+    {
+      "epoch": 1.4071082390953151,
+      "grad_norm": 0.6675281925628074,
+      "learning_rate": 1.2094462521271156e-06,
+      "loss": 0.0337,
+      "step": 436
+    },
+    {
+      "epoch": 1.4103392568659128,
+      "grad_norm": 0.5401273336581622,
+      "learning_rate": 1.197368747424592e-06,
+      "loss": 0.0398,
+      "step": 437
+    },
+    {
+      "epoch": 1.4135702746365104,
+      "grad_norm": 0.7065350638753161,
+      "learning_rate": 1.1853328309581139e-06,
+      "loss": 0.0435,
+      "step": 438
+    },
+    {
+      "epoch": 1.4168012924071083,
+      "grad_norm": 0.7095730881777876,
+      "learning_rate": 1.17333888699033e-06,
+      "loss": 0.0297,
+      "step": 439
+    },
+    {
+      "epoch": 1.420032310177706,
+      "grad_norm": 0.8419216282460489,
+      "learning_rate": 1.161387298443863e-06,
+      "loss": 0.0516,
+      "step": 440
+    },
+    {
+      "epoch": 1.4232633279483038,
+      "grad_norm": 0.6570671193798376,
+      "learning_rate": 1.149478446889077e-06,
+      "loss": 0.0499,
+      "step": 441
+    },
+    {
+      "epoch": 1.4264943457189014,
+      "grad_norm": 0.4516949531869673,
+      "learning_rate": 1.1376127125319065e-06,
+      "loss": 0.0332,
+      "step": 442
+    },
+    {
+      "epoch": 1.4297253634894993,
+      "grad_norm": 0.6804270748390296,
+      "learning_rate": 1.125790474201708e-06,
+      "loss": 0.0554,
+      "step": 443
+    },
+    {
+      "epoch": 1.432956381260097,
+      "grad_norm": 0.6237302316612817,
+      "learning_rate": 1.1140121093391736e-06,
+      "loss": 0.0406,
+      "step": 444
+    },
+    {
+      "epoch": 1.4361873990306946,
+      "grad_norm": 0.5492979077070937,
+      "learning_rate": 1.1022779939842704e-06,
+      "loss": 0.0359,
+      "step": 445
+    },
+    {
+      "epoch": 1.4394184168012925,
+      "grad_norm": 0.5435741772086434,
+      "learning_rate": 1.0905885027642484e-06,
+      "loss": 0.0524,
+      "step": 446
+    },
+    {
+      "epoch": 1.44264943457189,
+      "grad_norm": 0.6801306673227177,
+      "learning_rate": 1.0789440088816666e-06,
+      "loss": 0.0385,
+      "step": 447
+    },
+    {
+      "epoch": 1.445880452342488,
+      "grad_norm": 0.4434075942240935,
+      "learning_rate": 1.0673448841024875e-06,
+      "loss": 0.048,
+      "step": 448
+    },
+    {
+      "epoch": 1.4491114701130856,
+      "grad_norm": 0.6540515881701509,
+      "learning_rate": 1.0557914987442048e-06,
+      "loss": 0.0471,
+      "step": 449
+    },
+    {
+      "epoch": 1.4523424878836835,
+      "grad_norm": 0.6361305209643866,
+      "learning_rate": 1.0442842216640168e-06,
+      "loss": 0.0388,
+      "step": 450
+    },
+    {
+      "epoch": 1.4555735056542811,
+      "grad_norm": 0.5896519670655204,
+      "learning_rate": 1.0328234202470574e-06,
+      "loss": 0.0466,
+      "step": 451
+    },
+    {
+      "epoch": 1.4588045234248788,
+      "grad_norm": 0.49886003535682916,
+      "learning_rate": 1.021409460394663e-06,
+      "loss": 0.024,
+      "step": 452
+    },
+    {
+      "epoch": 1.4620355411954766,
+      "grad_norm": 0.4865979088073283,
+      "learning_rate": 1.0100427065126874e-06,
+      "loss": 0.0461,
+      "step": 453
+    },
+    {
+      "epoch": 1.4652665589660743,
+      "grad_norm": 0.6304447558055624,
+      "learning_rate": 9.987235214998741e-07,
+      "loss": 0.0303,
+      "step": 454
+    },
+    {
+      "epoch": 1.468497576736672,
+      "grad_norm": 0.5998572755132743,
+      "learning_rate": 9.87452266736266e-07,
+      "loss": 0.0413,
+      "step": 455
+    },
+    {
+      "epoch": 1.4717285945072698,
+      "grad_norm": 0.475261837629553,
+      "learning_rate": 9.762293020716696e-07,
+      "loss": 0.051,
+      "step": 456
+    },
+    {
+      "epoch": 1.4749596122778676,
+      "grad_norm": 0.5717008410694144,
+      "learning_rate": 9.650549858141646e-07,
+      "loss": 0.0325,
+      "step": 457
+    },
+    {
+      "epoch": 1.4781906300484653,
+      "grad_norm": 0.5180031308985593,
+      "learning_rate": 9.53929674718668e-07,
+      "loss": 0.0296,
+      "step": 458
+    },
+    {
+      "epoch": 1.481421647819063,
+      "grad_norm": 0.5815736580452745,
+      "learning_rate": 9.428537239755381e-07,
+      "loss": 0.0462,
+      "step": 459
+    },
+    {
+      "epoch": 1.4846526655896608,
+      "grad_norm": 0.6313068151131468,
+      "learning_rate": 9.318274871992408e-07,
+      "loss": 0.0726,
+      "step": 460
+    },
+    {
+      "epoch": 1.4878836833602584,
+      "grad_norm": 1.221847458822884,
+      "learning_rate": 9.208513164170579e-07,
+      "loss": 0.0405,
+      "step": 461
+    },
+    {
+      "epoch": 1.491114701130856,
+      "grad_norm": 0.5310642493149078,
+      "learning_rate": 9.099255620578451e-07,
+      "loss": 0.0296,
+      "step": 462
+    },
+    {
+      "epoch": 1.494345718901454,
+      "grad_norm": 0.5177627615454572,
+      "learning_rate": 8.990505729408494e-07,
+      "loss": 0.0651,
+      "step": 463
+    },
+    {
+      "epoch": 1.4975767366720518,
+      "grad_norm": 0.6991644226087359,
+      "learning_rate": 8.882266962645695e-07,
+      "loss": 0.0403,
+      "step": 464
+    },
+    {
+      "epoch": 1.5008077544426495,
+      "grad_norm": 0.5744181895807364,
+      "learning_rate": 8.774542775956679e-07,
+      "loss": 0.0512,
+      "step": 465
+    },
+    {
+      "epoch": 1.504038772213247,
+      "grad_norm": 0.5233586593423929,
+      "learning_rate": 8.667336608579488e-07,
+      "loss": 0.0408,
+      "step": 466
+    },
+    {
+      "epoch": 1.507269789983845,
+      "grad_norm": 0.569083608134158,
+      "learning_rate": 8.560651883213633e-07,
+      "loss": 0.0325,
+      "step": 467
+    },
+    {
+      "epoch": 1.5105008077544426,
+      "grad_norm": 0.5065403358211885,
+      "learning_rate": 8.454492005910942e-07,
+      "loss": 0.039,
+      "step": 468
+    },
+    {
+      "epoch": 1.5137318255250403,
+      "grad_norm": 0.5538367181328561,
+      "learning_rate": 8.34886036596676e-07,
+      "loss": 0.0461,
+      "step": 469
+    },
+    {
+      "epoch": 1.5169628432956381,
+      "grad_norm": 0.5214647871849324,
+      "learning_rate": 8.243760335811734e-07,
+      "loss": 0.0284,
+      "step": 470
+    },
+    {
+      "epoch": 1.520193861066236,
+      "grad_norm": 0.6202345630142632,
+      "learning_rate": 8.139195270904182e-07,
+      "loss": 0.0448,
+      "step": 471
+    },
+    {
+      "epoch": 1.5234248788368336,
+      "grad_norm": 0.596159859282425,
+      "learning_rate": 8.035168509622948e-07,
+      "loss": 0.0743,
+      "step": 472
+    },
+    {
+      "epoch": 1.5266558966074313,
+      "grad_norm": 0.5888141535757658,
+      "learning_rate": 7.931683373160789e-07,
+      "loss": 0.0379,
+      "step": 473
+    },
+    {
+      "epoch": 1.5298869143780292,
+      "grad_norm": 0.7088883925452755,
+      "learning_rate": 7.828743165418393e-07,
+      "loss": 0.0676,
+      "step": 474
+    },
+    {
+      "epoch": 1.5331179321486268,
+      "grad_norm": 0.7275077844858919,
+      "learning_rate": 7.726351172898869e-07,
+      "loss": 0.0372,
+      "step": 475
+    },
+    {
+      "epoch": 1.5363489499192244,
+      "grad_norm": 0.4568256055898612,
+      "learning_rate": 7.624510664602819e-07,
+      "loss": 0.037,
+      "step": 476
+    },
+    {
+      "epoch": 1.5395799676898223,
+      "grad_norm": 0.5900479097481395,
+      "learning_rate": 7.523224891923983e-07,
+      "loss": 0.0335,
+      "step": 477
+    },
+    {
+      "epoch": 1.5428109854604202,
+      "grad_norm": 0.5945624497711766,
+      "learning_rate": 7.422497088545436e-07,
+      "loss": 0.0473,
+      "step": 478
+    },
+    {
+      "epoch": 1.5460420032310178,
+      "grad_norm": 0.6053475731541442,
+      "learning_rate": 7.322330470336314e-07,
+      "loss": 0.0355,
+      "step": 479
+    },
+    {
+      "epoch": 1.5492730210016155,
+      "grad_norm": 0.5924482508513196,
+      "learning_rate": 7.222728235249196e-07,
+      "loss": 0.0395,
+      "step": 480
+    },
+    {
+      "epoch": 1.5525040387722133,
+      "grad_norm": 0.6450291567303781,
+      "learning_rate": 7.123693563217978e-07,
+      "loss": 0.0597,
+      "step": 481
+    },
+    {
+      "epoch": 1.555735056542811,
+      "grad_norm": 0.8328909187805514,
+      "learning_rate": 7.025229616056326e-07,
+      "loss": 0.0472,
+      "step": 482
+    },
+    {
+      "epoch": 1.5589660743134086,
+      "grad_norm": 0.6806153349540757,
+      "learning_rate": 6.927339537356778e-07,
+      "loss": 0.0484,
+      "step": 483
+    },
+    {
+      "epoch": 1.5621970920840065,
+      "grad_norm": 0.6602101596852485,
+      "learning_rate": 6.830026452390354e-07,
+      "loss": 0.0477,
+      "step": 484
+    },
+    {
+      "epoch": 1.5654281098546043,
+      "grad_norm": 0.5937543973198223,
+      "learning_rate": 6.733293468006774e-07,
+      "loss": 0.042,
+      "step": 485
+    },
+    {
+      "epoch": 1.568659127625202,
+      "grad_norm": 0.6210912711298314,
+      "learning_rate": 6.637143672535282e-07,
+      "loss": 0.0411,
+      "step": 486
+    },
+    {
+      "epoch": 1.5718901453957996,
+      "grad_norm": 0.563220202424066,
+      "learning_rate": 6.541580135686046e-07,
+      "loss": 0.0287,
+      "step": 487
+    },
+    {
+      "epoch": 1.5751211631663975,
+      "grad_norm": 0.4909721841694075,
+      "learning_rate": 6.446605908452122e-07,
+      "loss": 0.0341,
+      "step": 488
+    },
+    {
+      "epoch": 1.5783521809369951,
+      "grad_norm": 0.6948201563174986,
+      "learning_rate": 6.352224023012096e-07,
+      "loss": 0.046,
+      "step": 489
+    },
+    {
+      "epoch": 1.5815831987075928,
+      "grad_norm": 0.46668729207255394,
+      "learning_rate": 6.258437492633254e-07,
+      "loss": 0.0597,
+      "step": 490
+    },
+    {
+      "epoch": 1.5848142164781907,
+      "grad_norm": 0.7349838782964371,
+      "learning_rate": 6.165249311575361e-07,
+      "loss": 0.047,
+      "step": 491
+    },
+    {
+      "epoch": 1.5880452342487885,
+      "grad_norm": 0.6329932642072914,
+      "learning_rate": 6.072662454995101e-07,
+      "loss": 0.0402,
+      "step": 492
+    },
+    {
+      "epoch": 1.5912762520193862,
+      "grad_norm": 1.1075017060647618,
+      "learning_rate": 5.980679878851076e-07,
+      "loss": 0.0644,
+      "step": 493
+    },
+    {
+      "epoch": 1.5945072697899838,
+      "grad_norm": 0.6709210038908787,
+      "learning_rate": 5.889304519809402e-07,
+      "loss": 0.0405,
+      "step": 494
+    },
+    {
+      "epoch": 1.5977382875605817,
+      "grad_norm": 0.6039830090199084,
+      "learning_rate": 5.798539295150027e-07,
+      "loss": 0.0515,
+      "step": 495
+    },
+    {
+      "epoch": 1.6009693053311793,
+      "grad_norm": 0.47306097411821496,
+      "learning_rate": 5.708387102673507e-07,
+      "loss": 0.0452,
+      "step": 496
+    },
+    {
+      "epoch": 1.604200323101777,
+      "grad_norm": 0.7299041964334542,
+      "learning_rate": 5.618850820608548e-07,
+      "loss": 0.0386,
+      "step": 497
+    },
+    {
+      "epoch": 1.6074313408723748,
+      "grad_norm": 0.527708763111458,
+      "learning_rate": 5.529933307520102e-07,
+      "loss": 0.037,
+      "step": 498
+    },
+    {
+      "epoch": 1.6106623586429727,
+      "grad_norm": 0.6276271955216034,
+      "learning_rate": 5.441637402218077e-07,
+      "loss": 0.0566,
+      "step": 499
+    },
+    {
+      "epoch": 1.6138933764135701,
+      "grad_norm": 0.5498853546865076,
+      "learning_rate": 5.353965923666743e-07,
+      "loss": 0.0373,
+      "step": 500
+    },
+    {
+      "epoch": 1.617124394184168,
+      "grad_norm": 0.6058328263346706,
+      "learning_rate": 5.26692167089472e-07,
+      "loss": 0.0269,
+      "step": 501
+    },
+    {
+      "epoch": 1.6203554119547658,
+      "grad_norm": 0.4158541379346223,
+      "learning_rate": 5.180507422905585e-07,
+      "loss": 0.0393,
+      "step": 502
+    },
+    {
+      "epoch": 1.6235864297253635,
+      "grad_norm": 0.5063280186973653,
+      "learning_rate": 5.094725938589193e-07,
+      "loss": 0.0299,
+      "step": 503
+    },
+    {
+      "epoch": 1.6268174474959611,
+      "grad_norm": 0.42841992008265245,
+      "learning_rate": 5.009579956633578e-07,
+      "loss": 0.0301,
+      "step": 504
+    },
+    {
+      "epoch": 1.630048465266559,
+      "grad_norm": 0.7945718206145054,
+      "learning_rate": 4.925072195437511e-07,
+      "loss": 0.0488,
+      "step": 505
+    },
+    {
+      "epoch": 1.6332794830371569,
+      "grad_norm": 0.507794598401741,
+      "learning_rate": 4.841205353023715e-07,
+      "loss": 0.0617,
+      "step": 506
+    },
+    {
+      "epoch": 1.6365105008077543,
+      "grad_norm": 0.5915868711150505,
+      "learning_rate": 4.757982106952735e-07,
+      "loss": 0.0357,
+      "step": 507
+    },
+    {
+      "epoch": 1.6397415185783522,
+      "grad_norm": 0.5886571451007635,
+      "learning_rate": 4.6754051142374275e-07,
+      "loss": 0.0383,
+      "step": 508
+    },
+    {
+      "epoch": 1.64297253634895,
+      "grad_norm": 0.7114424474339083,
+      "learning_rate": 4.5934770112581713e-07,
+      "loss": 0.0299,
+      "step": 509
+    },
+    {
+      "epoch": 1.6462035541195477,
+      "grad_norm": 0.4631064089704488,
+      "learning_rate": 4.512200413678672e-07,
+      "loss": 0.04,
+      "step": 510
+    },
+    {
+      "epoch": 1.6494345718901453,
+      "grad_norm": 0.4403449444039548,
+      "learning_rate": 4.4315779163624476e-07,
+      "loss": 0.0329,
+      "step": 511
+    },
+    {
+      "epoch": 1.6526655896607432,
+      "grad_norm": 0.5420749756119222,
+      "learning_rate": 4.351612093290006e-07,
+      "loss": 0.0423,
+      "step": 512
+    },
+    {
+      "epoch": 1.655896607431341,
+      "grad_norm": 0.6010914237823689,
+      "learning_rate": 4.2723054974766585e-07,
+      "loss": 0.0381,
+      "step": 513
+    },
+    {
+      "epoch": 1.6591276252019385,
+      "grad_norm": 0.7377468114121554,
+      "learning_rate": 4.1936606608909887e-07,
+      "loss": 0.0515,
+      "step": 514
+    },
+    {
+      "epoch": 1.6623586429725363,
+      "grad_norm": 0.3875573660249705,
+      "learning_rate": 4.115680094374075e-07,
+      "loss": 0.0375,
+      "step": 515
+    },
+    {
+      "epoch": 1.6655896607431342,
+      "grad_norm": 0.605903732271668,
+      "learning_rate": 4.038366287559245e-07,
+      "loss": 0.0359,
+      "step": 516
+    },
+    {
+      "epoch": 1.6688206785137318,
+      "grad_norm": 0.5676552999273701,
+      "learning_rate": 3.961721708792662e-07,
+      "loss": 0.0352,
+      "step": 517
+    },
+    {
+      "epoch": 1.6720516962843295,
+      "grad_norm": 0.49258629677672283,
+      "learning_rate": 3.8857488050544903e-07,
+      "loss": 0.0427,
+      "step": 518
+    },
+    {
+      "epoch": 1.6752827140549273,
+      "grad_norm": 0.6179633227247718,
+      "learning_rate": 3.8104500018807806e-07,
+      "loss": 0.0436,
+      "step": 519
+    },
+    {
+      "epoch": 1.678513731825525,
+      "grad_norm": 0.6093007583811607,
+      "learning_rate": 3.7358277032860016e-07,
+      "loss": 0.0285,
+      "step": 520
+    },
+    {
+      "epoch": 1.6817447495961226,
+      "grad_norm": 0.5650933346174037,
+      "learning_rate": 3.6618842916863377e-07,
+      "loss": 0.062,
+      "step": 521
+    },
+    {
+      "epoch": 1.6849757673667205,
+      "grad_norm": 0.938334017983649,
+      "learning_rate": 3.5886221278236045e-07,
+      "loss": 0.0533,
+      "step": 522
+    },
+    {
+      "epoch": 1.6882067851373184,
+      "grad_norm": 0.5004878719352018,
+      "learning_rate": 3.5160435506898514e-07,
+      "loss": 0.0357,
+      "step": 523
+    },
+    {
+      "epoch": 1.691437802907916,
+      "grad_norm": 0.5331645289198377,
+      "learning_rate": 3.4441508774527345e-07,
+      "loss": 0.049,
+      "step": 524
+    },
+    {
+      "epoch": 1.6946688206785137,
+      "grad_norm": 0.6457764842086132,
+      "learning_rate": 3.3729464033815077e-07,
+      "loss": 0.0428,
+      "step": 525
+    },
+    {
+      "epoch": 1.6978998384491115,
+      "grad_norm": 0.5850353162840484,
+      "learning_rate": 3.3024324017737555e-07,
+      "loss": 0.0353,
+      "step": 526
+    },
+    {
+      "epoch": 1.7011308562197092,
+      "grad_norm": 0.5309296730801445,
+      "learning_rate": 3.232611123882809e-07,
+      "loss": 0.0423,
+      "step": 527
+    },
+    {
+      "epoch": 1.7043618739903068,
+      "grad_norm": 0.5673210523979683,
+      "learning_rate": 3.163484798845862e-07,
+      "loss": 0.0426,
+      "step": 528
+    },
+    {
+      "epoch": 1.7075928917609047,
+      "grad_norm": 0.6324074485656012,
+      "learning_rate": 3.0950556336128255e-07,
+      "loss": 0.0306,
+      "step": 529
+    },
+    {
+      "epoch": 1.7108239095315025,
+      "grad_norm": 0.44362940233898746,
+      "learning_rate": 3.0273258128758585e-07,
+      "loss": 0.0356,
+      "step": 530
+    },
+    {
+      "epoch": 1.7140549273021002,
+      "grad_norm": 0.6204310781031833,
+      "learning_rate": 2.960297498999601e-07,
+      "loss": 0.0375,
+      "step": 531
+    },
+    {
+      "epoch": 1.7172859450726978,
+      "grad_norm": 0.5062348215221578,
+      "learning_rate": 2.893972831952166e-07,
+      "loss": 0.0345,
+      "step": 532
+    },
+    {
+      "epoch": 1.7205169628432957,
+      "grad_norm": 0.5125908206118261,
+      "learning_rate": 2.82835392923681e-07,
+      "loss": 0.0284,
+      "step": 533
+    },
+    {
+      "epoch": 1.7237479806138933,
+      "grad_norm": 0.4768669694675723,
+      "learning_rate": 2.7634428858242995e-07,
+      "loss": 0.0466,
+      "step": 534
+    },
+    {
+      "epoch": 1.726978998384491,
+      "grad_norm": 0.6399021669002876,
+      "learning_rate": 2.699241774086081e-07,
+      "loss": 0.0561,
+      "step": 535
+    },
+    {
+      "epoch": 1.7302100161550888,
+      "grad_norm": 0.6404329572646896,
+      "learning_rate": 2.6357526437280764e-07,
+      "loss": 0.0322,
+      "step": 536
+    },
+    {
+      "epoch": 1.7334410339256867,
+      "grad_norm": 0.594233151602921,
+      "learning_rate": 2.572977521725242e-07,
+      "loss": 0.0343,
+      "step": 537
+    },
+    {
+      "epoch": 1.7366720516962844,
+      "grad_norm": 0.6970207221495152,
+      "learning_rate": 2.5109184122568797e-07,
+      "loss": 0.0676,
+      "step": 538
+    },
+    {
+      "epoch": 1.739903069466882,
+      "grad_norm": 0.5196706248162594,
+      "learning_rate": 2.449577296642647e-07,
+      "loss": 0.0306,
+      "step": 539
+    },
+    {
+      "epoch": 1.7431340872374799,
+      "grad_norm": 0.4727031774493694,
+      "learning_rate": 2.388956133279266e-07,
+      "loss": 0.0318,
+      "step": 540
+    },
+    {
+      "epoch": 1.7463651050080775,
+      "grad_norm": 0.5101584744111383,
+      "learning_rate": 2.329056857578049e-07,
+      "loss": 0.0475,
+      "step": 541
+    },
+    {
+      "epoch": 1.7495961227786752,
+      "grad_norm": 0.47288331779975473,
+      "learning_rate": 2.2698813819030802e-07,
+      "loss": 0.0349,
+      "step": 542
+    },
+    {
+      "epoch": 1.752827140549273,
+      "grad_norm": 0.5306522527872187,
+      "learning_rate": 2.2114315955101495e-07,
+      "loss": 0.0404,
+      "step": 543
+    },
+    {
+      "epoch": 1.7560581583198709,
+      "grad_norm": 0.6700784665030379,
+      "learning_rate": 2.153709364486467e-07,
+      "loss": 0.0535,
+      "step": 544
+    },
+    {
+      "epoch": 1.7592891760904685,
+      "grad_norm": 0.5532660459489857,
+      "learning_rate": 2.0967165316910675e-07,
+      "loss": 0.0362,
+      "step": 545
+    },
+    {
+      "epoch": 1.7625201938610662,
+      "grad_norm": 0.5690415151498087,
+      "learning_rate": 2.040454916695972e-07,
+      "loss": 0.0416,
+      "step": 546
+    },
+    {
+      "epoch": 1.765751211631664,
+      "grad_norm": 0.581032791692991,
+      "learning_rate": 1.9849263157281057e-07,
+      "loss": 0.0489,
+      "step": 547
+    },
+    {
+      "epoch": 1.7689822294022617,
+      "grad_norm": 0.5121050632247067,
+      "learning_rate": 1.9301325016119338e-07,
+      "loss": 0.0326,
+      "step": 548
+    },
+    {
+      "epoch": 1.7722132471728593,
+      "grad_norm": 0.4844742553742546,
+      "learning_rate": 1.8760752237128864e-07,
+      "loss": 0.0495,
+      "step": 549
+    },
+    {
+      "epoch": 1.7754442649434572,
+      "grad_norm": 0.6114242878876927,
+      "learning_rate": 1.8227562078814903e-07,
+      "loss": 0.0332,
+      "step": 550
+    },
+    {
+      "epoch": 1.778675282714055,
+      "grad_norm": 0.5118426883532984,
+      "learning_rate": 1.7701771563982757e-07,
+      "loss": 0.0369,
+      "step": 551
+    },
+    {
+      "epoch": 1.7819063004846527,
+      "grad_norm": 0.4094343968094117,
+      "learning_rate": 1.7183397479194175e-07,
+      "loss": 0.0449,
+      "step": 552
+    },
+    {
+      "epoch": 1.7851373182552503,
+      "grad_norm": 1.1161654981261413,
+      "learning_rate": 1.667245637423162e-07,
+      "loss": 0.0461,
+      "step": 553
+    },
+    {
+      "epoch": 1.7883683360258482,
+      "grad_norm": 0.5663414514839669,
+      "learning_rate": 1.6168964561569716e-07,
+      "loss": 0.0341,
+      "step": 554
+    },
+    {
+      "epoch": 1.7915993537964459,
+      "grad_norm": 0.5593424866536396,
+      "learning_rate": 1.5672938115854546e-07,
+      "loss": 0.0415,
+      "step": 555
+    },
+    {
+      "epoch": 1.7948303715670435,
+      "grad_norm": 0.6317941412989588,
+      "learning_rate": 1.5184392873390463e-07,
+      "loss": 0.0393,
+      "step": 556
+    },
+    {
+      "epoch": 1.7980613893376414,
+      "grad_norm": 0.4921503369201142,
+      "learning_rate": 1.470334443163432e-07,
+      "loss": 0.0338,
+      "step": 557
+    },
+    {
+      "epoch": 1.8012924071082392,
+      "grad_norm": 0.45350637218576434,
+      "learning_rate": 1.4229808148697732e-07,
+      "loss": 0.0307,
+      "step": 558
+    },
+    {
+      "epoch": 1.8045234248788369,
+      "grad_norm": 0.5558616504683307,
+      "learning_rate": 1.3763799142856693e-07,
+      "loss": 0.0394,
+      "step": 559
+    },
+    {
+      "epoch": 1.8077544426494345,
+      "grad_norm": 0.5602513622205471,
+      "learning_rate": 1.3305332292068706e-07,
+      "loss": 0.0461,
+      "step": 560
+    },
+    {
+      "epoch": 1.8109854604200324,
+      "grad_norm": 0.4909817316983247,
+      "learning_rate": 1.285442223349806e-07,
+      "loss": 0.0391,
+      "step": 561
+    },
+    {
+      "epoch": 1.81421647819063,
+      "grad_norm": 0.6251726656596086,
+      "learning_rate": 1.2411083363048386e-07,
+      "loss": 0.0384,
+      "step": 562
+    },
+    {
+      "epoch": 1.8174474959612277,
+      "grad_norm": 0.4604846376779599,
+      "learning_rate": 1.1975329834903017e-07,
+      "loss": 0.0271,
+      "step": 563
+    },
+    {
+      "epoch": 1.8206785137318255,
+      "grad_norm": 0.5538018269789088,
+      "learning_rate": 1.1547175561073154e-07,
+      "loss": 0.0606,
+      "step": 564
+    },
+    {
+      "epoch": 1.8239095315024234,
+      "grad_norm": 0.716319643356168,
+      "learning_rate": 1.1126634210953751e-07,
+      "loss": 0.0455,
+      "step": 565
+    },
+    {
+      "epoch": 1.827140549273021,
+      "grad_norm": 0.5640740061628162,
+      "learning_rate": 1.071371921088693e-07,
+      "loss": 0.0243,
+      "step": 566
+    },
+    {
+      "epoch": 1.8303715670436187,
+      "grad_norm": 0.31893744573237565,
+      "learning_rate": 1.0308443743733548e-07,
+      "loss": 0.0239,
+      "step": 567
+    },
+    {
+      "epoch": 1.8336025848142166,
+      "grad_norm": 0.5743187176077259,
+      "learning_rate": 9.91082074845215e-08,
+      "loss": 0.0399,
+      "step": 568
+    },
+    {
+      "epoch": 1.8368336025848142,
+      "grad_norm": 0.555173849824351,
+      "learning_rate": 9.520862919685903e-08,
+      "loss": 0.0565,
+      "step": 569
+    },
+    {
+      "epoch": 1.8400646203554119,
+      "grad_norm": 0.5247191749483705,
+      "learning_rate": 9.138582707357429e-08,
+      "loss": 0.0317,
+      "step": 570
+    },
+    {
+      "epoch": 1.8432956381260097,
+      "grad_norm": 0.6462019405718445,
+      "learning_rate": 8.763992316271175e-08,
+      "loss": 0.0443,
+      "step": 571
+    },
+    {
+      "epoch": 1.8465266558966076,
+      "grad_norm": 0.5199310882556472,
+      "learning_rate": 8.397103705723774e-08,
+      "loss": 0.0496,
+      "step": 572
+    },
+    {
+      "epoch": 1.849757673667205,
+      "grad_norm": 0.5797413283647508,
+      "learning_rate": 8.037928589122306e-08,
+      "loss": 0.0362,
+      "step": 573
+    },
+    {
+      "epoch": 1.8529886914378029,
+      "grad_norm": 0.5916717165607657,
+      "learning_rate": 7.686478433610339e-08,
+      "loss": 0.036,
+      "step": 574
+    },
+    {
+      "epoch": 1.8562197092084007,
+      "grad_norm": 0.6872719525671662,
+      "learning_rate": 7.342764459701723e-08,
+      "loss": 0.0433,
+      "step": 575
+    },
+    {
+      "epoch": 1.8594507269789984,
+      "grad_norm": 0.5719774164295217,
+      "learning_rate": 7.006797640922436e-08,
+      "loss": 0.045,
+      "step": 576
+    },
+    {
+      "epoch": 1.862681744749596,
+      "grad_norm": 0.4607000516397503,
+      "learning_rate": 6.678588703460165e-08,
+      "loss": 0.0308,
+      "step": 577
+    },
+    {
+      "epoch": 1.865912762520194,
+      "grad_norm": 0.47224203890648525,
+      "learning_rate": 6.358148125822e-08,
+      "loss": 0.0263,
+      "step": 578
+    },
+    {
+      "epoch": 1.8691437802907918,
+      "grad_norm": 0.4659556647523011,
+      "learning_rate": 6.045486138499756e-08,
+      "loss": 0.0331,
+      "step": 579
+    },
+    {
+      "epoch": 1.8723747980613892,
+      "grad_norm": 0.48850435913378176,
+      "learning_rate": 5.7406127236434016e-08,
+      "loss": 0.0385,
+      "step": 580
+    },
+    {
+      "epoch": 1.875605815831987,
+      "grad_norm": 0.527755833512547,
+      "learning_rate": 5.4435376147423945e-08,
+      "loss": 0.0361,
+      "step": 581
+    },
+    {
+      "epoch": 1.878836833602585,
+      "grad_norm": 0.5906290141554674,
+      "learning_rate": 5.154270296314878e-08,
+      "loss": 0.0341,
+      "step": 582
+    },
+    {
+      "epoch": 1.8820678513731826,
+      "grad_norm": 0.5773632179611871,
+      "learning_rate": 4.872820003604922e-08,
+      "loss": 0.0436,
+      "step": 583
+    },
+    {
+      "epoch": 1.8852988691437802,
+      "grad_norm": 0.5457918200937637,
+      "learning_rate": 4.599195722287536e-08,
+      "loss": 0.0316,
+      "step": 584
+    },
+    {
+      "epoch": 1.888529886914378,
+      "grad_norm": 0.48490285444923714,
+      "learning_rate": 4.3334061881820934e-08,
+      "loss": 0.0367,
+      "step": 585
+    },
+    {
+      "epoch": 1.891760904684976,
+      "grad_norm": 0.6250593429748459,
+      "learning_rate": 4.0754598869730824e-08,
+      "loss": 0.048,
+      "step": 586
+    },
+    {
+      "epoch": 1.8949919224555734,
+      "grad_norm": 0.7932835837622026,
+      "learning_rate": 3.825365053939406e-08,
+      "loss": 0.0444,
+      "step": 587
+    },
+    {
+      "epoch": 1.8982229402261712,
+      "grad_norm": 0.694824220848745,
+      "learning_rate": 3.583129673691427e-08,
+      "loss": 0.046,
+      "step": 588
+    },
+    {
+      "epoch": 1.901453957996769,
+      "grad_norm": 0.6400347943270668,
+      "learning_rate": 3.3487614799159186e-08,
+      "loss": 0.0447,
+      "step": 589
+    },
+    {
+      "epoch": 1.9046849757673667,
+      "grad_norm": 0.6428387216930922,
+      "learning_rate": 3.1222679551293486e-08,
+      "loss": 0.0378,
+      "step": 590
+    },
+    {
+      "epoch": 1.9079159935379644,
+      "grad_norm": 0.5181562999964454,
+      "learning_rate": 2.9036563304389032e-08,
+      "loss": 0.0447,
+      "step": 591
+    },
+    {
+      "epoch": 1.9111470113085622,
+      "grad_norm": 0.626996440091958,
+      "learning_rate": 2.6929335853115302e-08,
+      "loss": 0.0549,
+      "step": 592
+    },
+    {
+      "epoch": 1.9143780290791599,
+      "grad_norm": 0.46884001067775155,
+      "learning_rate": 2.490106447351315e-08,
+      "loss": 0.0277,
+      "step": 593
+    },
+    {
+      "epoch": 1.9176090468497575,
+      "grad_norm": 0.6158232854353491,
+      "learning_rate": 2.295181392084511e-08,
+      "loss": 0.0601,
+      "step": 594
+    },
+    {
+      "epoch": 1.9208400646203554,
+      "grad_norm": 0.9359381094535966,
+      "learning_rate": 2.1081646427528468e-08,
+      "loss": 0.0318,
+      "step": 595
+    },
+    {
+      "epoch": 1.9240710823909533,
+      "grad_norm": 0.49020001937141494,
+      "learning_rate": 1.9290621701149315e-08,
+      "loss": 0.0426,
+      "step": 596
+    },
+    {
+      "epoch": 1.927302100161551,
+      "grad_norm": 0.6612576122540997,
+      "learning_rate": 1.757879692255493e-08,
+      "loss": 0.0425,
+      "step": 597
+    },
+    {
+      "epoch": 1.9305331179321485,
+      "grad_norm": 0.6686158894140516,
+      "learning_rate": 1.5946226744029402e-08,
+      "loss": 0.0513,
+      "step": 598
+    },
+    {
+      "epoch": 1.9337641357027464,
+      "grad_norm": 0.570200319629059,
+      "learning_rate": 1.43929632875478e-08,
+      "loss": 0.0433,
+      "step": 599
+    },
+    {
+      "epoch": 1.936995153473344,
+      "grad_norm": 0.7735949555352855,
+      "learning_rate": 1.2919056143113062e-08,
+      "loss": 0.0626,
+      "step": 600
+    },
+    {
+      "epoch": 1.9402261712439417,
+      "grad_norm": 0.870645903135215,
+      "learning_rate": 1.1524552367171705e-08,
+      "loss": 0.0325,
+      "step": 601
+    },
+    {
+      "epoch": 1.9434571890145396,
+      "grad_norm": 0.7457181182787564,
+      "learning_rate": 1.0209496481112247e-08,
+      "loss": 0.0429,
+      "step": 602
+    },
+    {
+      "epoch": 1.9466882067851374,
+      "grad_norm": 0.4589602142690698,
+      "learning_rate": 8.973930469844127e-09,
+      "loss": 0.0456,
+      "step": 603
+    },
+    {
+      "epoch": 1.949919224555735,
+      "grad_norm": 0.6365759321501641,
+      "learning_rate": 7.81789378045572e-09,
+      "loss": 0.0444,
+      "step": 604
+    },
+    {
+      "epoch": 1.9531502423263327,
+      "grad_norm": 0.5685501801373803,
+      "learning_rate": 6.741423320957286e-09,
+      "loss": 0.0389,
+      "step": 605
+    },
+    {
+      "epoch": 1.9563812600969306,
+      "grad_norm": 0.6248328079624079,
+      "learning_rate": 5.7445534591002435e-09,
+      "loss": 0.0359,
+      "step": 606
+    },
+    {
+      "epoch": 1.9596122778675282,
+      "grad_norm": 0.5081816835827963,
+      "learning_rate": 4.8273160212811145e-09,
+      "loss": 0.0372,
+      "step": 607
+    },
+    {
+      "epoch": 1.9628432956381259,
+      "grad_norm": 0.6052821334268623,
+      "learning_rate": 3.989740291526212e-09,
+      "loss": 0.0342,
+      "step": 608
+    },
+    {
+      "epoch": 1.9660743134087237,
+      "grad_norm": 0.6069914651621039,
+      "learning_rate": 3.2318530105546198e-09,
+      "loss": 0.0562,
+      "step": 609
+    },
+    {
+      "epoch": 1.9693053311793216,
+      "grad_norm": 0.6247816362826526,
+      "learning_rate": 2.553678374926649e-09,
+      "loss": 0.0345,
+      "step": 610
+    },
+    {
+      "epoch": 1.9725363489499192,
+      "grad_norm": 0.4803333487137274,
+      "learning_rate": 1.9552380362697355e-09,
+      "loss": 0.0389,
+      "step": 611
+    },
+    {
+      "epoch": 1.975767366720517,
+      "grad_norm": 0.4064456733315355,
+      "learning_rate": 1.4365511005878796e-09,
+      "loss": 0.038,
+      "step": 612
+    },
+    {
+      "epoch": 1.9789983844911148,
+      "grad_norm": 0.6627884418012588,
+      "learning_rate": 9.976341276521361e-10,
+      "loss": 0.0499,
+      "step": 613
+    },
+    {
+      "epoch": 1.9822294022617124,
+      "grad_norm": 0.5188386174906152,
+      "learning_rate": 6.385011304704814e-10,
+      "loss": 0.0504,
+      "step": 614
+    },
+    {
+      "epoch": 1.98546042003231,
+      "grad_norm": 0.4923517024917629,
+      "learning_rate": 3.59163574841781e-10,
+      "loss": 0.0313,
+      "step": 615
+    },
+    {
+      "epoch": 1.988691437802908,
+      "grad_norm": 1.0015315278245285,
+      "learning_rate": 1.5963037898913957e-10,
+      "loss": 0.0512,
+      "step": 616
+    },
+    {
+      "epoch": 1.9919224555735058,
+      "grad_norm": 0.6617805237667193,
+      "learning_rate": 3.9907913275683e-11,
+      "loss": 0.0487,
+      "step": 617
+    },
+    {
+      "epoch": 1.9951534733441034,
+      "grad_norm": 0.5338181066973476,
+      "learning_rate": 0.0,
+      "loss": 0.0393,
+      "step": 618
+    },
+    {
+      "epoch": 1.9951534733441034,
+      "step": 618,
+      "total_flos": 52320870432768.0,
+      "train_loss": 0.06418457114014135,
+      "train_runtime": 2400.3631,
+      "train_samples_per_second": 4.126,
+      "train_steps_per_second": 0.257
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 618,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "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": 52320870432768.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}