diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4535 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9992229992229992,
+  "eval_steps": 500,
+  "global_step": 643,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001554001554001554,
+      "grad_norm": 1.0753202631674548,
+      "learning_rate": 1e-05,
+      "loss": 0.5508,
+      "step": 1
+    },
+    {
+      "epoch": 0.003108003108003108,
+      "grad_norm": 1.394325863785031,
+      "learning_rate": 9.999940321631158e-06,
+      "loss": 0.6286,
+      "step": 2
+    },
+    {
+      "epoch": 0.004662004662004662,
+      "grad_norm": 0.8138734995659078,
+      "learning_rate": 9.999761287949237e-06,
+      "loss": 0.4343,
+      "step": 3
+    },
+    {
+      "epoch": 0.006216006216006216,
+      "grad_norm": 1.4834629168746152,
+      "learning_rate": 9.99946290322801e-06,
+      "loss": 0.5391,
+      "step": 4
+    },
+    {
+      "epoch": 0.00777000777000777,
+      "grad_norm": 0.6636994391139434,
+      "learning_rate": 9.999045174590324e-06,
+      "loss": 0.5696,
+      "step": 5
+    },
+    {
+      "epoch": 0.009324009324009324,
+      "grad_norm": 0.8144572025546154,
+      "learning_rate": 9.998508112007925e-06,
+      "loss": 0.8279,
+      "step": 6
+    },
+    {
+      "epoch": 0.010878010878010878,
+      "grad_norm": 0.6969071538976613,
+      "learning_rate": 9.997851728301219e-06,
+      "loss": 0.537,
+      "step": 7
+    },
+    {
+      "epoch": 0.012432012432012432,
+      "grad_norm": 0.8735249365273472,
+      "learning_rate": 9.99707603913897e-06,
+      "loss": 0.6761,
+      "step": 8
+    },
+    {
+      "epoch": 0.013986013986013986,
+      "grad_norm": 0.7027334655844945,
+      "learning_rate": 9.996181063037924e-06,
+      "loss": 0.5301,
+      "step": 9
+    },
+    {
+      "epoch": 0.01554001554001554,
+      "grad_norm": 0.8373746380293126,
+      "learning_rate": 9.995166821362368e-06,
+      "loss": 0.6674,
+      "step": 10
+    },
+    {
+      "epoch": 0.017094017094017096,
+      "grad_norm": 0.984420779214608,
+      "learning_rate": 9.994033338323612e-06,
+      "loss": 0.5113,
+      "step": 11
+    },
+    {
+      "epoch": 0.018648018648018648,
+      "grad_norm": 0.7622743125468773,
+      "learning_rate": 9.99278064097943e-06,
+      "loss": 0.6431,
+      "step": 12
+    },
+    {
+      "epoch": 0.020202020202020204,
+      "grad_norm": 0.7377189725798596,
+      "learning_rate": 9.991408759233394e-06,
+      "loss": 0.6131,
+      "step": 13
+    },
+    {
+      "epoch": 0.021756021756021756,
+      "grad_norm": 0.9016901262329209,
+      "learning_rate": 9.989917725834166e-06,
+      "loss": 0.5208,
+      "step": 14
+    },
+    {
+      "epoch": 0.023310023310023312,
+      "grad_norm": 0.7060988613807518,
+      "learning_rate": 9.988307576374727e-06,
+      "loss": 0.5591,
+      "step": 15
+    },
+    {
+      "epoch": 0.024864024864024864,
+      "grad_norm": 1.965431280490754,
+      "learning_rate": 9.986578349291514e-06,
+      "loss": 0.6333,
+      "step": 16
+    },
+    {
+      "epoch": 0.02641802641802642,
+      "grad_norm": 0.6204824016696763,
+      "learning_rate": 9.984730085863504e-06,
+      "loss": 0.4729,
+      "step": 17
+    },
+    {
+      "epoch": 0.027972027972027972,
+      "grad_norm": 0.6629661100589302,
+      "learning_rate": 9.982762830211239e-06,
+      "loss": 0.41,
+      "step": 18
+    },
+    {
+      "epoch": 0.029526029526029528,
+      "grad_norm": 0.679676793160211,
+      "learning_rate": 9.980676629295763e-06,
+      "loss": 0.5619,
+      "step": 19
+    },
+    {
+      "epoch": 0.03108003108003108,
+      "grad_norm": 0.7228425066613519,
+      "learning_rate": 9.9784715329175e-06,
+      "loss": 0.5637,
+      "step": 20
+    },
+    {
+      "epoch": 0.03263403263403263,
+      "grad_norm": 1.4494500979247202,
+      "learning_rate": 9.976147593715074e-06,
+      "loss": 0.4784,
+      "step": 21
+    },
+    {
+      "epoch": 0.03418803418803419,
+      "grad_norm": 0.9490720634807803,
+      "learning_rate": 9.973704867164044e-06,
+      "loss": 0.775,
+      "step": 22
+    },
+    {
+      "epoch": 0.035742035742035744,
+      "grad_norm": 0.7173074724185691,
+      "learning_rate": 9.971143411575585e-06,
+      "loss": 0.5456,
+      "step": 23
+    },
+    {
+      "epoch": 0.037296037296037296,
+      "grad_norm": 1.813950013627374,
+      "learning_rate": 9.968463288095096e-06,
+      "loss": 0.5494,
+      "step": 24
+    },
+    {
+      "epoch": 0.03885003885003885,
+      "grad_norm": 0.764359484896435,
+      "learning_rate": 9.965664560700734e-06,
+      "loss": 0.6135,
+      "step": 25
+    },
+    {
+      "epoch": 0.04040404040404041,
+      "grad_norm": 0.8961015996035883,
+      "learning_rate": 9.962747296201891e-06,
+      "loss": 0.4535,
+      "step": 26
+    },
+    {
+      "epoch": 0.04195804195804196,
+      "grad_norm": 0.7061951973387899,
+      "learning_rate": 9.959711564237603e-06,
+      "loss": 0.5766,
+      "step": 27
+    },
+    {
+      "epoch": 0.04351204351204351,
+      "grad_norm": 0.9186274326988478,
+      "learning_rate": 9.956557437274887e-06,
+      "loss": 0.4126,
+      "step": 28
+    },
+    {
+      "epoch": 0.045066045066045064,
+      "grad_norm": 1.0005571619677003,
+      "learning_rate": 9.953284990607e-06,
+      "loss": 0.5344,
+      "step": 29
+    },
+    {
+      "epoch": 0.046620046620046623,
+      "grad_norm": 1.0384488624813175,
+      "learning_rate": 9.949894302351653e-06,
+      "loss": 0.686,
+      "step": 30
+    },
+    {
+      "epoch": 0.048174048174048176,
+      "grad_norm": 1.1091587053056344,
+      "learning_rate": 9.946385453449145e-06,
+      "loss": 0.7023,
+      "step": 31
+    },
+    {
+      "epoch": 0.04972804972804973,
+      "grad_norm": 0.8221842041481426,
+      "learning_rate": 9.942758527660429e-06,
+      "loss": 0.4769,
+      "step": 32
+    },
+    {
+      "epoch": 0.05128205128205128,
+      "grad_norm": 0.8497884172706046,
+      "learning_rate": 9.93901361156511e-06,
+      "loss": 0.5504,
+      "step": 33
+    },
+    {
+      "epoch": 0.05283605283605284,
+      "grad_norm": 0.6872331365190385,
+      "learning_rate": 9.935150794559379e-06,
+      "loss": 0.5369,
+      "step": 34
+    },
+    {
+      "epoch": 0.05439005439005439,
+      "grad_norm": 0.8777872780147628,
+      "learning_rate": 9.931170168853886e-06,
+      "loss": 0.657,
+      "step": 35
+    },
+    {
+      "epoch": 0.055944055944055944,
+      "grad_norm": 0.6487410029703777,
+      "learning_rate": 9.927071829471531e-06,
+      "loss": 0.4006,
+      "step": 36
+    },
+    {
+      "epoch": 0.057498057498057496,
+      "grad_norm": 0.6468395282548087,
+      "learning_rate": 9.922855874245197e-06,
+      "loss": 0.4699,
+      "step": 37
+    },
+    {
+      "epoch": 0.059052059052059055,
+      "grad_norm": 0.821140351961526,
+      "learning_rate": 9.918522403815414e-06,
+      "loss": 0.5927,
+      "step": 38
+    },
+    {
+      "epoch": 0.06060606060606061,
+      "grad_norm": 0.7910492113722175,
+      "learning_rate": 9.914071521627964e-06,
+      "loss": 0.4792,
+      "step": 39
+    },
+    {
+      "epoch": 0.06216006216006216,
+      "grad_norm": 1.036910491968146,
+      "learning_rate": 9.909503333931402e-06,
+      "loss": 0.699,
+      "step": 40
+    },
+    {
+      "epoch": 0.06371406371406371,
+      "grad_norm": 0.8196334764356895,
+      "learning_rate": 9.904817949774524e-06,
+      "loss": 0.7333,
+      "step": 41
+    },
+    {
+      "epoch": 0.06526806526806526,
+      "grad_norm": 0.6798779713989535,
+      "learning_rate": 9.900015481003762e-06,
+      "loss": 0.6433,
+      "step": 42
+    },
+    {
+      "epoch": 0.06682206682206682,
+      "grad_norm": 0.9642768659293228,
+      "learning_rate": 9.895096042260517e-06,
+      "loss": 0.5257,
+      "step": 43
+    },
+    {
+      "epoch": 0.06837606837606838,
+      "grad_norm": 1.7689631037060651,
+      "learning_rate": 9.890059750978425e-06,
+      "loss": 0.6034,
+      "step": 44
+    },
+    {
+      "epoch": 0.06993006993006994,
+      "grad_norm": 0.7443934495191462,
+      "learning_rate": 9.88490672738054e-06,
+      "loss": 0.6089,
+      "step": 45
+    },
+    {
+      "epoch": 0.07148407148407149,
+      "grad_norm": 1.3800385551010166,
+      "learning_rate": 9.879637094476482e-06,
+      "loss": 0.555,
+      "step": 46
+    },
+    {
+      "epoch": 0.07303807303807304,
+      "grad_norm": 0.6934112455074183,
+      "learning_rate": 9.874250978059489e-06,
+      "loss": 0.5253,
+      "step": 47
+    },
+    {
+      "epoch": 0.07459207459207459,
+      "grad_norm": 0.728895851371575,
+      "learning_rate": 9.86874850670342e-06,
+      "loss": 0.5111,
+      "step": 48
+    },
+    {
+      "epoch": 0.07614607614607614,
+      "grad_norm": 0.705671764219133,
+      "learning_rate": 9.863129811759678e-06,
+      "loss": 0.5166,
+      "step": 49
+    },
+    {
+      "epoch": 0.0777000777000777,
+      "grad_norm": 0.9004429060865304,
+      "learning_rate": 9.857395027354085e-06,
+      "loss": 0.4981,
+      "step": 50
+    },
+    {
+      "epoch": 0.07925407925407925,
+      "grad_norm": 0.6186440018603123,
+      "learning_rate": 9.85154429038367e-06,
+      "loss": 0.3806,
+      "step": 51
+    },
+    {
+      "epoch": 0.08080808080808081,
+      "grad_norm": 0.7725834636904632,
+      "learning_rate": 9.845577740513409e-06,
+      "loss": 0.4948,
+      "step": 52
+    },
+    {
+      "epoch": 0.08236208236208237,
+      "grad_norm": 0.9253638464116425,
+      "learning_rate": 9.83949552017289e-06,
+      "loss": 0.5285,
+      "step": 53
+    },
+    {
+      "epoch": 0.08391608391608392,
+      "grad_norm": 0.7656584472395311,
+      "learning_rate": 9.833297774552905e-06,
+      "loss": 0.6342,
+      "step": 54
+    },
+    {
+      "epoch": 0.08547008547008547,
+      "grad_norm": 0.8998211569830732,
+      "learning_rate": 9.826984651601998e-06,
+      "loss": 0.6178,
+      "step": 55
+    },
+    {
+      "epoch": 0.08702408702408702,
+      "grad_norm": 0.8685832029444266,
+      "learning_rate": 9.820556302022916e-06,
+      "loss": 0.6903,
+      "step": 56
+    },
+    {
+      "epoch": 0.08857808857808858,
+      "grad_norm": 0.7304781373527222,
+      "learning_rate": 9.814012879269031e-06,
+      "loss": 0.5054,
+      "step": 57
+    },
+    {
+      "epoch": 0.09013209013209013,
+      "grad_norm": 0.9941249676710022,
+      "learning_rate": 9.80735453954066e-06,
+      "loss": 0.5752,
+      "step": 58
+    },
+    {
+      "epoch": 0.09168609168609168,
+      "grad_norm": 0.8636774782845266,
+      "learning_rate": 9.800581441781342e-06,
+      "loss": 0.472,
+      "step": 59
+    },
+    {
+      "epoch": 0.09324009324009325,
+      "grad_norm": 0.8037019830533219,
+      "learning_rate": 9.79369374767405e-06,
+      "loss": 0.451,
+      "step": 60
+    },
+    {
+      "epoch": 0.0947940947940948,
+      "grad_norm": 0.7569996171253018,
+      "learning_rate": 9.786691621637322e-06,
+      "loss": 0.5862,
+      "step": 61
+    },
+    {
+      "epoch": 0.09634809634809635,
+      "grad_norm": 0.718132158990271,
+      "learning_rate": 9.779575230821344e-06,
+      "loss": 0.57,
+      "step": 62
+    },
+    {
+      "epoch": 0.0979020979020979,
+      "grad_norm": 0.8861885775634778,
+      "learning_rate": 9.772344745103955e-06,
+      "loss": 0.661,
+      "step": 63
+    },
+    {
+      "epoch": 0.09945609945609946,
+      "grad_norm": 0.7647017941329705,
+      "learning_rate": 9.76500033708659e-06,
+      "loss": 0.5032,
+      "step": 64
+    },
+    {
+      "epoch": 0.10101010101010101,
+      "grad_norm": 0.6503958573653673,
+      "learning_rate": 9.757542182090165e-06,
+      "loss": 0.5113,
+      "step": 65
+    },
+    {
+      "epoch": 0.10256410256410256,
+      "grad_norm": 0.8514535986961426,
+      "learning_rate": 9.749970458150893e-06,
+      "loss": 0.4582,
+      "step": 66
+    },
+    {
+      "epoch": 0.10411810411810411,
+      "grad_norm": 0.8115540624516915,
+      "learning_rate": 9.742285346016024e-06,
+      "loss": 0.5324,
+      "step": 67
+    },
+    {
+      "epoch": 0.10567210567210568,
+      "grad_norm": 0.8716261250005453,
+      "learning_rate": 9.734487029139544e-06,
+      "loss": 0.5622,
+      "step": 68
+    },
+    {
+      "epoch": 0.10722610722610723,
+      "grad_norm": 0.6674304497161411,
+      "learning_rate": 9.726575693677782e-06,
+      "loss": 0.4194,
+      "step": 69
+    },
+    {
+      "epoch": 0.10878010878010878,
+      "grad_norm": 1.0158040848497383,
+      "learning_rate": 9.718551528484979e-06,
+      "loss": 0.4802,
+      "step": 70
+    },
+    {
+      "epoch": 0.11033411033411034,
+      "grad_norm": 0.595576459482197,
+      "learning_rate": 9.710414725108771e-06,
+      "loss": 0.4594,
+      "step": 71
+    },
+    {
+      "epoch": 0.11188811188811189,
+      "grad_norm": 0.9680714108769358,
+      "learning_rate": 9.702165477785618e-06,
+      "loss": 0.5109,
+      "step": 72
+    },
+    {
+      "epoch": 0.11344211344211344,
+      "grad_norm": 0.7314195273632106,
+      "learning_rate": 9.69380398343617e-06,
+      "loss": 0.4759,
+      "step": 73
+    },
+    {
+      "epoch": 0.11499611499611499,
+      "grad_norm": 0.8578361036597735,
+      "learning_rate": 9.685330441660564e-06,
+      "loss": 0.5533,
+      "step": 74
+    },
+    {
+      "epoch": 0.11655011655011654,
+      "grad_norm": 1.890522114467283,
+      "learning_rate": 9.676745054733661e-06,
+      "loss": 0.6364,
+      "step": 75
+    },
+    {
+      "epoch": 0.11810411810411811,
+      "grad_norm": 0.8238570929416833,
+      "learning_rate": 9.668048027600217e-06,
+      "loss": 0.5262,
+      "step": 76
+    },
+    {
+      "epoch": 0.11965811965811966,
+      "grad_norm": 0.7000587405463369,
+      "learning_rate": 9.659239567869989e-06,
+      "loss": 0.5541,
+      "step": 77
+    },
+    {
+      "epoch": 0.12121212121212122,
+      "grad_norm": 1.0275499560212522,
+      "learning_rate": 9.650319885812777e-06,
+      "loss": 0.6356,
+      "step": 78
+    },
+    {
+      "epoch": 0.12276612276612277,
+      "grad_norm": 0.9584707284083808,
+      "learning_rate": 9.641289194353418e-06,
+      "loss": 0.5393,
+      "step": 79
+    },
+    {
+      "epoch": 0.12432012432012432,
+      "grad_norm": 0.7661012737973905,
+      "learning_rate": 9.632147709066682e-06,
+      "loss": 0.5864,
+      "step": 80
+    },
+    {
+      "epoch": 0.1258741258741259,
+      "grad_norm": 0.8981339358976846,
+      "learning_rate": 9.622895648172141e-06,
+      "loss": 0.5001,
+      "step": 81
+    },
+    {
+      "epoch": 0.12742812742812742,
+      "grad_norm": 0.9136632663354124,
+      "learning_rate": 9.613533232528956e-06,
+      "loss": 0.5989,
+      "step": 82
+    },
+    {
+      "epoch": 0.128982128982129,
+      "grad_norm": 1.0752006190192398,
+      "learning_rate": 9.604060685630608e-06,
+      "loss": 0.6551,
+      "step": 83
+    },
+    {
+      "epoch": 0.13053613053613053,
+      "grad_norm": 0.6746196580403023,
+      "learning_rate": 9.594478233599551e-06,
+      "loss": 0.5763,
+      "step": 84
+    },
+    {
+      "epoch": 0.1320901320901321,
+      "grad_norm": 1.0560596133360578,
+      "learning_rate": 9.584786105181831e-06,
+      "loss": 0.626,
+      "step": 85
+    },
+    {
+      "epoch": 0.13364413364413363,
+      "grad_norm": 1.9322491414214575,
+      "learning_rate": 9.574984531741613e-06,
+      "loss": 0.6028,
+      "step": 86
+    },
+    {
+      "epoch": 0.1351981351981352,
+      "grad_norm": 1.1199274751748922,
+      "learning_rate": 9.565073747255665e-06,
+      "loss": 0.5536,
+      "step": 87
+    },
+    {
+      "epoch": 0.13675213675213677,
+      "grad_norm": 1.2142928266989612,
+      "learning_rate": 9.555053988307764e-06,
+      "loss": 0.5881,
+      "step": 88
+    },
+    {
+      "epoch": 0.1383061383061383,
+      "grad_norm": 0.886265217836807,
+      "learning_rate": 9.544925494083062e-06,
+      "loss": 0.4521,
+      "step": 89
+    },
+    {
+      "epoch": 0.13986013986013987,
+      "grad_norm": 0.7758293999026016,
+      "learning_rate": 9.53468850636236e-06,
+      "loss": 0.6255,
+      "step": 90
+    },
+    {
+      "epoch": 0.1414141414141414,
+      "grad_norm": 0.8222144113614714,
+      "learning_rate": 9.524343269516354e-06,
+      "loss": 0.5643,
+      "step": 91
+    },
+    {
+      "epoch": 0.14296814296814297,
+      "grad_norm": 0.7971504019254859,
+      "learning_rate": 9.513890030499786e-06,
+      "loss": 0.5469,
+      "step": 92
+    },
+    {
+      "epoch": 0.1445221445221445,
+      "grad_norm": 1.0988108202778002,
+      "learning_rate": 9.503329038845556e-06,
+      "loss": 0.6065,
+      "step": 93
+    },
+    {
+      "epoch": 0.14607614607614608,
+      "grad_norm": 0.9770953296063096,
+      "learning_rate": 9.492660546658771e-06,
+      "loss": 0.4678,
+      "step": 94
+    },
+    {
+      "epoch": 0.14763014763014762,
+      "grad_norm": 0.7001359835404304,
+      "learning_rate": 9.481884808610712e-06,
+      "loss": 0.5186,
+      "step": 95
+    },
+    {
+      "epoch": 0.14918414918414918,
+      "grad_norm": 1.028392902002761,
+      "learning_rate": 9.471002081932767e-06,
+      "loss": 0.5347,
+      "step": 96
+    },
+    {
+      "epoch": 0.15073815073815075,
+      "grad_norm": 0.6722329878713879,
+      "learning_rate": 9.460012626410286e-06,
+      "loss": 0.5138,
+      "step": 97
+    },
+    {
+      "epoch": 0.1522921522921523,
+      "grad_norm": 1.0396069509533319,
+      "learning_rate": 9.448916704376384e-06,
+      "loss": 0.5388,
+      "step": 98
+    },
+    {
+      "epoch": 0.15384615384615385,
+      "grad_norm": 0.788486690631997,
+      "learning_rate": 9.437714580705671e-06,
+      "loss": 0.4044,
+      "step": 99
+    },
+    {
+      "epoch": 0.1554001554001554,
+      "grad_norm": 0.7359447373312309,
+      "learning_rate": 9.426406522807932e-06,
+      "loss": 0.6637,
+      "step": 100
+    },
+    {
+      "epoch": 0.15695415695415696,
+      "grad_norm": 1.0558555591940144,
+      "learning_rate": 9.414992800621749e-06,
+      "loss": 0.4842,
+      "step": 101
+    },
+    {
+      "epoch": 0.1585081585081585,
+      "grad_norm": 0.8538402867666735,
+      "learning_rate": 9.40347368660805e-06,
+      "loss": 0.51,
+      "step": 102
+    },
+    {
+      "epoch": 0.16006216006216006,
+      "grad_norm": 0.5016420650271773,
+      "learning_rate": 9.39184945574361e-06,
+      "loss": 0.3733,
+      "step": 103
+    },
+    {
+      "epoch": 0.16161616161616163,
+      "grad_norm": 0.8043995407764333,
+      "learning_rate": 9.380120385514484e-06,
+      "loss": 0.6424,
+      "step": 104
+    },
+    {
+      "epoch": 0.16317016317016317,
+      "grad_norm": 0.7432969785115889,
+      "learning_rate": 9.368286755909383e-06,
+      "loss": 0.5918,
+      "step": 105
+    },
+    {
+      "epoch": 0.16472416472416473,
+      "grad_norm": 0.7800858860573905,
+      "learning_rate": 9.356348849412991e-06,
+      "loss": 0.5557,
+      "step": 106
+    },
+    {
+      "epoch": 0.16627816627816627,
+      "grad_norm": 0.8792710300993529,
+      "learning_rate": 9.344306950999226e-06,
+      "loss": 0.5359,
+      "step": 107
+    },
+    {
+      "epoch": 0.16783216783216784,
+      "grad_norm": 1.0983876112460325,
+      "learning_rate": 9.332161348124426e-06,
+      "loss": 0.6833,
+      "step": 108
+    },
+    {
+      "epoch": 0.16938616938616938,
+      "grad_norm": 0.8026939716617563,
+      "learning_rate": 9.319912330720502e-06,
+      "loss": 0.5503,
+      "step": 109
+    },
+    {
+      "epoch": 0.17094017094017094,
+      "grad_norm": 0.7137107427499185,
+      "learning_rate": 9.307560191188e-06,
+      "loss": 0.4826,
+      "step": 110
+    },
+    {
+      "epoch": 0.17249417249417248,
+      "grad_norm": 0.8145843383585528,
+      "learning_rate": 9.295105224389144e-06,
+      "loss": 0.5508,
+      "step": 111
+    },
+    {
+      "epoch": 0.17404817404817405,
+      "grad_norm": 0.6415967276646262,
+      "learning_rate": 9.282547727640767e-06,
+      "loss": 0.6075,
+      "step": 112
+    },
+    {
+      "epoch": 0.17560217560217561,
+      "grad_norm": 0.6695998673267839,
+      "learning_rate": 9.269888000707243e-06,
+      "loss": 0.4768,
+      "step": 113
+    },
+    {
+      "epoch": 0.17715617715617715,
+      "grad_norm": 0.692586482200001,
+      "learning_rate": 9.25712634579331e-06,
+      "loss": 0.5651,
+      "step": 114
+    },
+    {
+      "epoch": 0.17871017871017872,
+      "grad_norm": 0.9490969710584504,
+      "learning_rate": 9.244263067536872e-06,
+      "loss": 0.4607,
+      "step": 115
+    },
+    {
+      "epoch": 0.18026418026418026,
+      "grad_norm": 0.7448331565841488,
+      "learning_rate": 9.23129847300171e-06,
+      "loss": 0.5287,
+      "step": 116
+    },
+    {
+      "epoch": 0.18181818181818182,
+      "grad_norm": 0.7568020783154017,
+      "learning_rate": 9.218232871670168e-06,
+      "loss": 0.5876,
+      "step": 117
+    },
+    {
+      "epoch": 0.18337218337218336,
+      "grad_norm": 0.703790461691835,
+      "learning_rate": 9.205066575435754e-06,
+      "loss": 0.5477,
+      "step": 118
+    },
+    {
+      "epoch": 0.18492618492618493,
+      "grad_norm": 0.7355757023568825,
+      "learning_rate": 9.191799898595706e-06,
+      "loss": 0.4819,
+      "step": 119
+    },
+    {
+      "epoch": 0.1864801864801865,
+      "grad_norm": 0.7718259998249563,
+      "learning_rate": 9.178433157843474e-06,
+      "loss": 0.5529,
+      "step": 120
+    },
+    {
+      "epoch": 0.18803418803418803,
+      "grad_norm": 0.7325746328877619,
+      "learning_rate": 9.164966672261171e-06,
+      "loss": 0.471,
+      "step": 121
+    },
+    {
+      "epoch": 0.1895881895881896,
+      "grad_norm": 0.7807440267754963,
+      "learning_rate": 9.151400763311958e-06,
+      "loss": 0.5616,
+      "step": 122
+    },
+    {
+      "epoch": 0.19114219114219114,
+      "grad_norm": 0.6694238548360774,
+      "learning_rate": 9.13773575483236e-06,
+      "loss": 0.5637,
+      "step": 123
+    },
+    {
+      "epoch": 0.1926961926961927,
+      "grad_norm": 0.7619298527686099,
+      "learning_rate": 9.123971973024543e-06,
+      "loss": 0.5222,
+      "step": 124
+    },
+    {
+      "epoch": 0.19425019425019424,
+      "grad_norm": 0.8077572349157733,
+      "learning_rate": 9.110109746448527e-06,
+      "loss": 0.4792,
+      "step": 125
+    },
+    {
+      "epoch": 0.1958041958041958,
+      "grad_norm": 0.7156882422608634,
+      "learning_rate": 9.09614940601434e-06,
+      "loss": 0.5079,
+      "step": 126
+    },
+    {
+      "epoch": 0.19735819735819735,
+      "grad_norm": 0.8839790656748958,
+      "learning_rate": 9.08209128497412e-06,
+      "loss": 0.6167,
+      "step": 127
+    },
+    {
+      "epoch": 0.1989121989121989,
+      "grad_norm": 0.7361442782777586,
+      "learning_rate": 9.06793571891416e-06,
+      "loss": 0.573,
+      "step": 128
+    },
+    {
+      "epoch": 0.20046620046620048,
+      "grad_norm": 0.6874615113965296,
+      "learning_rate": 9.053683045746897e-06,
+      "loss": 0.5625,
+      "step": 129
+    },
+    {
+      "epoch": 0.20202020202020202,
+      "grad_norm": 1.1187739918068127,
+      "learning_rate": 9.039333605702844e-06,
+      "loss": 0.5386,
+      "step": 130
+    },
+    {
+      "epoch": 0.20357420357420358,
+      "grad_norm": 0.7935185219891059,
+      "learning_rate": 9.024887741322475e-06,
+      "loss": 0.4354,
+      "step": 131
+    },
+    {
+      "epoch": 0.20512820512820512,
+      "grad_norm": 0.6755878221667591,
+      "learning_rate": 9.010345797448037e-06,
+      "loss": 0.4871,
+      "step": 132
+    },
+    {
+      "epoch": 0.2066822066822067,
+      "grad_norm": 0.8821400973796808,
+      "learning_rate": 8.995708121215325e-06,
+      "loss": 0.5988,
+      "step": 133
+    },
+    {
+      "epoch": 0.20823620823620823,
+      "grad_norm": 0.8963468623381899,
+      "learning_rate": 8.980975062045398e-06,
+      "loss": 0.4808,
+      "step": 134
+    },
+    {
+      "epoch": 0.2097902097902098,
+      "grad_norm": 0.7605330355025913,
+      "learning_rate": 8.96614697163623e-06,
+      "loss": 0.5307,
+      "step": 135
+    },
+    {
+      "epoch": 0.21134421134421136,
+      "grad_norm": 0.8377449496608493,
+      "learning_rate": 8.95122420395432e-06,
+      "loss": 0.5855,
+      "step": 136
+    },
+    {
+      "epoch": 0.2128982128982129,
+      "grad_norm": 0.6874533009837056,
+      "learning_rate": 8.936207115226242e-06,
+      "loss": 0.5025,
+      "step": 137
+    },
+    {
+      "epoch": 0.21445221445221446,
+      "grad_norm": 1.0615886301023565,
+      "learning_rate": 8.921096063930141e-06,
+      "loss": 0.5858,
+      "step": 138
+    },
+    {
+      "epoch": 0.216006216006216,
+      "grad_norm": 0.7440356443459726,
+      "learning_rate": 8.905891410787174e-06,
+      "loss": 0.4721,
+      "step": 139
+    },
+    {
+      "epoch": 0.21756021756021757,
+      "grad_norm": 0.9780044181786849,
+      "learning_rate": 8.8905935187529e-06,
+      "loss": 0.5506,
+      "step": 140
+    },
+    {
+      "epoch": 0.2191142191142191,
+      "grad_norm": 0.7269242314361574,
+      "learning_rate": 8.875202753008614e-06,
+      "loss": 0.5308,
+      "step": 141
+    },
+    {
+      "epoch": 0.22066822066822067,
+      "grad_norm": 0.8667053692132669,
+      "learning_rate": 8.859719480952637e-06,
+      "loss": 0.5679,
+      "step": 142
+    },
+    {
+      "epoch": 0.2222222222222222,
+      "grad_norm": 0.7257081688438616,
+      "learning_rate": 8.844144072191537e-06,
+      "loss": 0.4858,
+      "step": 143
+    },
+    {
+      "epoch": 0.22377622377622378,
+      "grad_norm": 0.6542430837557276,
+      "learning_rate": 8.828476898531308e-06,
+      "loss": 0.5328,
+      "step": 144
+    },
+    {
+      "epoch": 0.22533022533022534,
+      "grad_norm": 0.7971170503328316,
+      "learning_rate": 8.812718333968498e-06,
+      "loss": 0.5695,
+      "step": 145
+    },
+    {
+      "epoch": 0.22688422688422688,
+      "grad_norm": 0.8486913376544397,
+      "learning_rate": 8.79686875468128e-06,
+      "loss": 0.5336,
+      "step": 146
+    },
+    {
+      "epoch": 0.22843822843822845,
+      "grad_norm": 1.0063509221057145,
+      "learning_rate": 8.780928539020467e-06,
+      "loss": 0.4731,
+      "step": 147
+    },
+    {
+      "epoch": 0.22999222999222999,
+      "grad_norm": 1.084598842228243,
+      "learning_rate": 8.764898067500488e-06,
+      "loss": 0.5336,
+      "step": 148
+    },
+    {
+      "epoch": 0.23154623154623155,
+      "grad_norm": 0.8477855123394982,
+      "learning_rate": 8.7487777227903e-06,
+      "loss": 0.7133,
+      "step": 149
+    },
+    {
+      "epoch": 0.2331002331002331,
+      "grad_norm": 0.7433328633942683,
+      "learning_rate": 8.732567889704253e-06,
+      "loss": 0.4906,
+      "step": 150
+    },
+    {
+      "epoch": 0.23465423465423466,
+      "grad_norm": 1.3157111994706745,
+      "learning_rate": 8.716268955192908e-06,
+      "loss": 0.4115,
+      "step": 151
+    },
+    {
+      "epoch": 0.23620823620823622,
+      "grad_norm": 0.7915085431469301,
+      "learning_rate": 8.699881308333794e-06,
+      "loss": 0.4529,
+      "step": 152
+    },
+    {
+      "epoch": 0.23776223776223776,
+      "grad_norm": 0.7093953259845076,
+      "learning_rate": 8.683405340322123e-06,
+      "loss": 0.5795,
+      "step": 153
+    },
+    {
+      "epoch": 0.23931623931623933,
+      "grad_norm": 0.9148926328978699,
+      "learning_rate": 8.666841444461456e-06,
+      "loss": 0.5371,
+      "step": 154
+    },
+    {
+      "epoch": 0.24087024087024086,
+      "grad_norm": 0.8451708726467831,
+      "learning_rate": 8.650190016154307e-06,
+      "loss": 0.6872,
+      "step": 155
+    },
+    {
+      "epoch": 0.24242424242424243,
+      "grad_norm": 0.7575098139209872,
+      "learning_rate": 8.633451452892707e-06,
+      "loss": 0.5262,
+      "step": 156
+    },
+    {
+      "epoch": 0.24397824397824397,
+      "grad_norm": 0.8039436989228282,
+      "learning_rate": 8.616626154248717e-06,
+      "loss": 0.5936,
+      "step": 157
+    },
+    {
+      "epoch": 0.24553224553224554,
+      "grad_norm": 0.9111018315972883,
+      "learning_rate": 8.59971452186489e-06,
+      "loss": 0.433,
+      "step": 158
+    },
+    {
+      "epoch": 0.24708624708624707,
+      "grad_norm": 0.7730492291768862,
+      "learning_rate": 8.582716959444679e-06,
+      "loss": 0.4906,
+      "step": 159
+    },
+    {
+      "epoch": 0.24864024864024864,
+      "grad_norm": 0.7508532423751979,
+      "learning_rate": 8.565633872742803e-06,
+      "loss": 0.5951,
+      "step": 160
+    },
+    {
+      "epoch": 0.2501942501942502,
+      "grad_norm": 0.7979075113160532,
+      "learning_rate": 8.548465669555564e-06,
+      "loss": 0.5487,
+      "step": 161
+    },
+    {
+      "epoch": 0.2517482517482518,
+      "grad_norm": 2.822907214340435,
+      "learning_rate": 8.531212759711103e-06,
+      "loss": 0.5006,
+      "step": 162
+    },
+    {
+      "epoch": 0.2533022533022533,
+      "grad_norm": 0.7951356785770967,
+      "learning_rate": 8.51387555505963e-06,
+      "loss": 0.4786,
+      "step": 163
+    },
+    {
+      "epoch": 0.25485625485625485,
+      "grad_norm": 0.7944890316200554,
+      "learning_rate": 8.496454469463583e-06,
+      "loss": 0.5428,
+      "step": 164
+    },
+    {
+      "epoch": 0.2564102564102564,
+      "grad_norm": 0.885097829322448,
+      "learning_rate": 8.478949918787746e-06,
+      "loss": 0.5559,
+      "step": 165
+    },
+    {
+      "epoch": 0.257964257964258,
+      "grad_norm": 0.8765537416075171,
+      "learning_rate": 8.461362320889338e-06,
+      "loss": 0.6328,
+      "step": 166
+    },
+    {
+      "epoch": 0.2595182595182595,
+      "grad_norm": 0.7600338630468001,
+      "learning_rate": 8.443692095608019e-06,
+      "loss": 0.4644,
+      "step": 167
+    },
+    {
+      "epoch": 0.26107226107226106,
+      "grad_norm": 0.7771144701749865,
+      "learning_rate": 8.425939664755874e-06,
+      "loss": 0.4637,
+      "step": 168
+    },
+    {
+      "epoch": 0.26262626262626265,
+      "grad_norm": 1.066507788727023,
+      "learning_rate": 8.408105452107353e-06,
+      "loss": 0.4977,
+      "step": 169
+    },
+    {
+      "epoch": 0.2641802641802642,
+      "grad_norm": 1.0331827740490165,
+      "learning_rate": 8.390189883389143e-06,
+      "loss": 0.4936,
+      "step": 170
+    },
+    {
+      "epoch": 0.26573426573426573,
+      "grad_norm": 0.6598501232172324,
+      "learning_rate": 8.37219338627001e-06,
+      "loss": 0.4534,
+      "step": 171
+    },
+    {
+      "epoch": 0.26728826728826727,
+      "grad_norm": 0.7813638132356793,
+      "learning_rate": 8.354116390350594e-06,
+      "loss": 0.5658,
+      "step": 172
+    },
+    {
+      "epoch": 0.26884226884226886,
+      "grad_norm": 0.9169947439806584,
+      "learning_rate": 8.335959327153148e-06,
+      "loss": 0.6649,
+      "step": 173
+    },
+    {
+      "epoch": 0.2703962703962704,
+      "grad_norm": 0.9173506379895737,
+      "learning_rate": 8.317722630111233e-06,
+      "loss": 0.6389,
+      "step": 174
+    },
+    {
+      "epoch": 0.27195027195027194,
+      "grad_norm": 0.9071927842627939,
+      "learning_rate": 8.299406734559385e-06,
+      "loss": 0.428,
+      "step": 175
+    },
+    {
+      "epoch": 0.27350427350427353,
+      "grad_norm": 0.7688021245301622,
+      "learning_rate": 8.281012077722712e-06,
+      "loss": 0.5133,
+      "step": 176
+    },
+    {
+      "epoch": 0.27505827505827507,
+      "grad_norm": 0.6639447866723064,
+      "learning_rate": 8.26253909870646e-06,
+      "loss": 0.5077,
+      "step": 177
+    },
+    {
+      "epoch": 0.2766122766122766,
+      "grad_norm": 0.6947258135733688,
+      "learning_rate": 8.24398823848553e-06,
+      "loss": 0.5527,
+      "step": 178
+    },
+    {
+      "epoch": 0.27816627816627815,
+      "grad_norm": 0.8705601245175114,
+      "learning_rate": 8.225359939893954e-06,
+      "loss": 0.4838,
+      "step": 179
+    },
+    {
+      "epoch": 0.27972027972027974,
+      "grad_norm": 0.6618526787787719,
+      "learning_rate": 8.206654647614323e-06,
+      "loss": 0.3804,
+      "step": 180
+    },
+    {
+      "epoch": 0.2812742812742813,
+      "grad_norm": 1.0338275501153558,
+      "learning_rate": 8.18787280816717e-06,
+      "loss": 0.5362,
+      "step": 181
+    },
+    {
+      "epoch": 0.2828282828282828,
+      "grad_norm": 0.7923177578469405,
+      "learning_rate": 8.169014869900308e-06,
+      "loss": 0.5842,
+      "step": 182
+    },
+    {
+      "epoch": 0.28438228438228436,
+      "grad_norm": 0.6396399402988676,
+      "learning_rate": 8.150081282978139e-06,
+      "loss": 0.4271,
+      "step": 183
+    },
+    {
+      "epoch": 0.28593628593628595,
+      "grad_norm": 0.8865011660122173,
+      "learning_rate": 8.131072499370897e-06,
+      "loss": 0.5489,
+      "step": 184
+    },
+    {
+      "epoch": 0.2874902874902875,
+      "grad_norm": 0.9138845420481327,
+      "learning_rate": 8.111988972843859e-06,
+      "loss": 0.5943,
+      "step": 185
+    },
+    {
+      "epoch": 0.289044289044289,
+      "grad_norm": 1.1268837902712043,
+      "learning_rate": 8.09283115894652e-06,
+      "loss": 0.5491,
+      "step": 186
+    },
+    {
+      "epoch": 0.2905982905982906,
+      "grad_norm": 0.7444814008859322,
+      "learning_rate": 8.073599515001713e-06,
+      "loss": 0.4997,
+      "step": 187
+    },
+    {
+      "epoch": 0.29215229215229216,
+      "grad_norm": 0.8044881882512651,
+      "learning_rate": 8.054294500094697e-06,
+      "loss": 0.5325,
+      "step": 188
+    },
+    {
+      "epoch": 0.2937062937062937,
+      "grad_norm": 0.8146942609420201,
+      "learning_rate": 8.034916575062188e-06,
+      "loss": 0.5437,
+      "step": 189
+    },
+    {
+      "epoch": 0.29526029526029524,
+      "grad_norm": 0.7872597021358523,
+      "learning_rate": 8.015466202481371e-06,
+      "loss": 0.4712,
+      "step": 190
+    },
+    {
+      "epoch": 0.29681429681429683,
+      "grad_norm": 0.7193071970265177,
+      "learning_rate": 7.995943846658852e-06,
+      "loss": 0.5397,
+      "step": 191
+    },
+    {
+      "epoch": 0.29836829836829837,
+      "grad_norm": 0.8391823684267,
+      "learning_rate": 7.976349973619567e-06,
+      "loss": 0.5765,
+      "step": 192
+    },
+    {
+      "epoch": 0.2999222999222999,
+      "grad_norm": 0.6845170782635249,
+      "learning_rate": 7.956685051095672e-06,
+      "loss": 0.56,
+      "step": 193
+    },
+    {
+      "epoch": 0.3014763014763015,
+      "grad_norm": 0.7252420419163962,
+      "learning_rate": 7.936949548515364e-06,
+      "loss": 0.605,
+      "step": 194
+    },
+    {
+      "epoch": 0.30303030303030304,
+      "grad_norm": 0.7875283859706272,
+      "learning_rate": 7.917143936991688e-06,
+      "loss": 0.5004,
+      "step": 195
+    },
+    {
+      "epoch": 0.3045843045843046,
+      "grad_norm": 0.6047235742618552,
+      "learning_rate": 7.897268689311278e-06,
+      "loss": 0.4562,
+      "step": 196
+    },
+    {
+      "epoch": 0.3061383061383061,
+      "grad_norm": 0.5825984478127255,
+      "learning_rate": 7.877324279923078e-06,
+      "loss": 0.451,
+      "step": 197
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 0.7369125343894832,
+      "learning_rate": 7.857311184927015e-06,
+      "loss": 0.5195,
+      "step": 198
+    },
+    {
+      "epoch": 0.30924630924630925,
+      "grad_norm": 0.6104305536066545,
+      "learning_rate": 7.837229882062638e-06,
+      "loss": 0.531,
+      "step": 199
+    },
+    {
+      "epoch": 0.3108003108003108,
+      "grad_norm": 0.9139389379071885,
+      "learning_rate": 7.817080850697705e-06,
+      "loss": 0.4996,
+      "step": 200
+    },
+    {
+      "epoch": 0.3123543123543124,
+      "grad_norm": 0.8513425536888274,
+      "learning_rate": 7.796864571816745e-06,
+      "loss": 0.3977,
+      "step": 201
+    },
+    {
+      "epoch": 0.3139083139083139,
+      "grad_norm": 0.7739897197051825,
+      "learning_rate": 7.77658152800958e-06,
+      "loss": 0.5569,
+      "step": 202
+    },
+    {
+      "epoch": 0.31546231546231546,
+      "grad_norm": 0.7242668166031772,
+      "learning_rate": 7.756232203459794e-06,
+      "loss": 0.4652,
+      "step": 203
+    },
+    {
+      "epoch": 0.317016317016317,
+      "grad_norm": 0.7649888193815342,
+      "learning_rate": 7.735817083933189e-06,
+      "loss": 0.4699,
+      "step": 204
+    },
+    {
+      "epoch": 0.3185703185703186,
+      "grad_norm": 1.0274699034423136,
+      "learning_rate": 7.715336656766176e-06,
+      "loss": 0.5284,
+      "step": 205
+    },
+    {
+      "epoch": 0.3201243201243201,
+      "grad_norm": 0.6976554827980787,
+      "learning_rate": 7.69479141085415e-06,
+      "loss": 0.4875,
+      "step": 206
+    },
+    {
+      "epoch": 0.32167832167832167,
+      "grad_norm": 1.130786196712028,
+      "learning_rate": 7.674181836639819e-06,
+      "loss": 0.4214,
+      "step": 207
+    },
+    {
+      "epoch": 0.32323232323232326,
+      "grad_norm": 0.7607350344632271,
+      "learning_rate": 7.653508426101488e-06,
+      "loss": 0.4803,
+      "step": 208
+    },
+    {
+      "epoch": 0.3247863247863248,
+      "grad_norm": 0.7396245146929729,
+      "learning_rate": 7.632771672741326e-06,
+      "loss": 0.5967,
+      "step": 209
+    },
+    {
+      "epoch": 0.32634032634032634,
+      "grad_norm": 0.9687489401453343,
+      "learning_rate": 7.611972071573579e-06,
+      "loss": 0.5753,
+      "step": 210
+    },
+    {
+      "epoch": 0.3278943278943279,
+      "grad_norm": 0.7739827653387099,
+      "learning_rate": 7.591110119112757e-06,
+      "loss": 0.495,
+      "step": 211
+    },
+    {
+      "epoch": 0.32944832944832947,
+      "grad_norm": 0.8201270448540147,
+      "learning_rate": 7.5701863133617735e-06,
+      "loss": 0.562,
+      "step": 212
+    },
+    {
+      "epoch": 0.331002331002331,
+      "grad_norm": 0.8576029389701612,
+      "learning_rate": 7.549201153800073e-06,
+      "loss": 0.5381,
+      "step": 213
+    },
+    {
+      "epoch": 0.33255633255633255,
+      "grad_norm": 0.6841266898837828,
+      "learning_rate": 7.528155141371688e-06,
+      "loss": 0.467,
+      "step": 214
+    },
+    {
+      "epoch": 0.3341103341103341,
+      "grad_norm": 0.7802991357810948,
+      "learning_rate": 7.507048778473296e-06,
+      "loss": 0.4966,
+      "step": 215
+    },
+    {
+      "epoch": 0.3356643356643357,
+      "grad_norm": 0.7218564126997288,
+      "learning_rate": 7.485882568942222e-06,
+      "loss": 0.5112,
+      "step": 216
+    },
+    {
+      "epoch": 0.3372183372183372,
+      "grad_norm": 0.7400107616507341,
+      "learning_rate": 7.464657018044411e-06,
+      "loss": 0.6446,
+      "step": 217
+    },
+    {
+      "epoch": 0.33877233877233875,
+      "grad_norm": 0.9014902134428892,
+      "learning_rate": 7.443372632462363e-06,
+      "loss": 0.5053,
+      "step": 218
+    },
+    {
+      "epoch": 0.34032634032634035,
+      "grad_norm": 0.7329909522650422,
+      "learning_rate": 7.422029920283044e-06,
+      "loss": 0.5443,
+      "step": 219
+    },
+    {
+      "epoch": 0.3418803418803419,
+      "grad_norm": 0.6892541253744358,
+      "learning_rate": 7.400629390985753e-06,
+      "loss": 0.4837,
+      "step": 220
+    },
+    {
+      "epoch": 0.3434343434343434,
+      "grad_norm": 0.7723348206252342,
+      "learning_rate": 7.379171555429965e-06,
+      "loss": 0.5755,
+      "step": 221
+    },
+    {
+      "epoch": 0.34498834498834496,
+      "grad_norm": 7.664546431285865,
+      "learning_rate": 7.357656925843125e-06,
+      "loss": 0.4786,
+      "step": 222
+    },
+    {
+      "epoch": 0.34654234654234656,
+      "grad_norm": 0.7853221971786,
+      "learning_rate": 7.336086015808439e-06,
+      "loss": 0.4729,
+      "step": 223
+    },
+    {
+      "epoch": 0.3480963480963481,
+      "grad_norm": 0.7864246657192466,
+      "learning_rate": 7.314459340252593e-06,
+      "loss": 0.5848,
+      "step": 224
+    },
+    {
+      "epoch": 0.34965034965034963,
+      "grad_norm": 0.7588019542197073,
+      "learning_rate": 7.2927774154334765e-06,
+      "loss": 0.4884,
+      "step": 225
+    },
+    {
+      "epoch": 0.35120435120435123,
+      "grad_norm": 0.6496638579237419,
+      "learning_rate": 7.271040758927852e-06,
+      "loss": 0.4323,
+      "step": 226
+    },
+    {
+      "epoch": 0.35275835275835277,
+      "grad_norm": 0.6618044442655401,
+      "learning_rate": 7.2492498896190015e-06,
+      "loss": 0.5797,
+      "step": 227
+    },
+    {
+      "epoch": 0.3543123543123543,
+      "grad_norm": 0.8288426526834968,
+      "learning_rate": 7.227405327684339e-06,
+      "loss": 0.5113,
+      "step": 228
+    },
+    {
+      "epoch": 0.35586635586635584,
+      "grad_norm": 0.6693277294293136,
+      "learning_rate": 7.205507594582994e-06,
+      "loss": 0.4158,
+      "step": 229
+    },
+    {
+      "epoch": 0.35742035742035744,
+      "grad_norm": 0.834098167457632,
+      "learning_rate": 7.183557213043365e-06,
+      "loss": 0.4826,
+      "step": 230
+    },
+    {
+      "epoch": 0.358974358974359,
+      "grad_norm": 0.7563577310691308,
+      "learning_rate": 7.161554707050637e-06,
+      "loss": 0.4767,
+      "step": 231
+    },
+    {
+      "epoch": 0.3605283605283605,
+      "grad_norm": 0.718266653620817,
+      "learning_rate": 7.1395006018342774e-06,
+      "loss": 0.5423,
+      "step": 232
+    },
+    {
+      "epoch": 0.3620823620823621,
+      "grad_norm": 0.7794408706255476,
+      "learning_rate": 7.117395423855496e-06,
+      "loss": 0.5394,
+      "step": 233
+    },
+    {
+      "epoch": 0.36363636363636365,
+      "grad_norm": 0.7269215366898196,
+      "learning_rate": 7.09523970079468e-06,
+      "loss": 0.5705,
+      "step": 234
+    },
+    {
+      "epoch": 0.3651903651903652,
+      "grad_norm": 0.9408390684558995,
+      "learning_rate": 7.073033961538793e-06,
+      "loss": 0.6389,
+      "step": 235
+    },
+    {
+      "epoch": 0.3667443667443667,
+      "grad_norm": 0.6638828664952492,
+      "learning_rate": 7.050778736168757e-06,
+      "loss": 0.5542,
+      "step": 236
+    },
+    {
+      "epoch": 0.3682983682983683,
+      "grad_norm": 0.8460797947106127,
+      "learning_rate": 7.028474555946787e-06,
+      "loss": 0.5444,
+      "step": 237
+    },
+    {
+      "epoch": 0.36985236985236986,
+      "grad_norm": 0.7665734815326051,
+      "learning_rate": 7.006121953303724e-06,
+      "loss": 0.5229,
+      "step": 238
+    },
+    {
+      "epoch": 0.3714063714063714,
+      "grad_norm": 0.8309718789678989,
+      "learning_rate": 6.983721461826312e-06,
+      "loss": 0.5784,
+      "step": 239
+    },
+    {
+      "epoch": 0.372960372960373,
+      "grad_norm": 0.8222921080795965,
+      "learning_rate": 6.9612736162444695e-06,
+      "loss": 0.4292,
+      "step": 240
+    },
+    {
+      "epoch": 0.3745143745143745,
+      "grad_norm": 0.7527448252012365,
+      "learning_rate": 6.938778952418519e-06,
+      "loss": 0.4729,
+      "step": 241
+    },
+    {
+      "epoch": 0.37606837606837606,
+      "grad_norm": 0.8985906116004181,
+      "learning_rate": 6.916238007326399e-06,
+      "loss": 0.5655,
+      "step": 242
+    },
+    {
+      "epoch": 0.3776223776223776,
+      "grad_norm": 0.6113924311221383,
+      "learning_rate": 6.893651319050842e-06,
+      "loss": 0.5263,
+      "step": 243
+    },
+    {
+      "epoch": 0.3791763791763792,
+      "grad_norm": 0.7817006095897586,
+      "learning_rate": 6.871019426766537e-06,
+      "loss": 0.4632,
+      "step": 244
+    },
+    {
+      "epoch": 0.38073038073038074,
+      "grad_norm": 1.7226586227932226,
+      "learning_rate": 6.8483428707272456e-06,
+      "loss": 0.5609,
+      "step": 245
+    },
+    {
+      "epoch": 0.3822843822843823,
+      "grad_norm": 0.8080544757443882,
+      "learning_rate": 6.825622192252922e-06,
+      "loss": 0.5424,
+      "step": 246
+    },
+    {
+      "epoch": 0.3838383838383838,
+      "grad_norm": 0.8602129359285826,
+      "learning_rate": 6.802857933716774e-06,
+      "loss": 0.4986,
+      "step": 247
+    },
+    {
+      "epoch": 0.3853923853923854,
+      "grad_norm": 0.7014559071633292,
+      "learning_rate": 6.7800506385323335e-06,
+      "loss": 0.5079,
+      "step": 248
+    },
+    {
+      "epoch": 0.38694638694638694,
+      "grad_norm": 0.8917239030684251,
+      "learning_rate": 6.757200851140468e-06,
+      "loss": 0.5281,
+      "step": 249
+    },
+    {
+      "epoch": 0.3885003885003885,
+      "grad_norm": 0.7447990452878435,
+      "learning_rate": 6.734309116996392e-06,
+      "loss": 0.6071,
+      "step": 250
+    },
+    {
+      "epoch": 0.3900543900543901,
+      "grad_norm": 0.7472396252981742,
+      "learning_rate": 6.711375982556648e-06,
+      "loss": 0.5418,
+      "step": 251
+    },
+    {
+      "epoch": 0.3916083916083916,
+      "grad_norm": 0.6987079676338328,
+      "learning_rate": 6.688401995266061e-06,
+      "loss": 0.4444,
+      "step": 252
+    },
+    {
+      "epoch": 0.39316239316239315,
+      "grad_norm": 0.8025151202499369,
+      "learning_rate": 6.665387703544661e-06,
+      "loss": 0.5082,
+      "step": 253
+    },
+    {
+      "epoch": 0.3947163947163947,
+      "grad_norm": 0.8425567278002095,
+      "learning_rate": 6.642333656774607e-06,
+      "loss": 0.4212,
+      "step": 254
+    },
+    {
+      "epoch": 0.3962703962703963,
+      "grad_norm": 0.7875874739440923,
+      "learning_rate": 6.61924040528706e-06,
+      "loss": 0.4969,
+      "step": 255
+    },
+    {
+      "epoch": 0.3978243978243978,
+      "grad_norm": 0.645057270258725,
+      "learning_rate": 6.596108500349054e-06,
+      "loss": 0.4857,
+      "step": 256
+    },
+    {
+      "epoch": 0.39937839937839936,
+      "grad_norm": 0.670409599043594,
+      "learning_rate": 6.572938494150332e-06,
+      "loss": 0.5544,
+      "step": 257
+    },
+    {
+      "epoch": 0.40093240093240096,
+      "grad_norm": 0.6198181436713772,
+      "learning_rate": 6.54973093979016e-06,
+      "loss": 0.4725,
+      "step": 258
+    },
+    {
+      "epoch": 0.4024864024864025,
+      "grad_norm": 0.7662890791531904,
+      "learning_rate": 6.526486391264137e-06,
+      "loss": 0.5041,
+      "step": 259
+    },
+    {
+      "epoch": 0.40404040404040403,
+      "grad_norm": 0.71665149094791,
+      "learning_rate": 6.503205403450957e-06,
+      "loss": 0.5193,
+      "step": 260
+    },
+    {
+      "epoch": 0.40559440559440557,
+      "grad_norm": 0.8140997620947699,
+      "learning_rate": 6.479888532099175e-06,
+      "loss": 0.5243,
+      "step": 261
+    },
+    {
+      "epoch": 0.40714840714840717,
+      "grad_norm": 0.624858848374034,
+      "learning_rate": 6.4565363338139245e-06,
+      "loss": 0.3996,
+      "step": 262
+    },
+    {
+      "epoch": 0.4087024087024087,
+      "grad_norm": 0.7647591367675671,
+      "learning_rate": 6.433149366043652e-06,
+      "loss": 0.4658,
+      "step": 263
+    },
+    {
+      "epoch": 0.41025641025641024,
+      "grad_norm": 0.6633892543296002,
+      "learning_rate": 6.409728187066789e-06,
+      "loss": 0.4452,
+      "step": 264
+    },
+    {
+      "epoch": 0.41181041181041184,
+      "grad_norm": 0.9241797213745755,
+      "learning_rate": 6.386273355978442e-06,
+      "loss": 0.5173,
+      "step": 265
+    },
+    {
+      "epoch": 0.4133644133644134,
+      "grad_norm": 0.9790400835539028,
+      "learning_rate": 6.3627854326770326e-06,
+      "loss": 0.5611,
+      "step": 266
+    },
+    {
+      "epoch": 0.4149184149184149,
+      "grad_norm": 0.7010295847973484,
+      "learning_rate": 6.339264977850943e-06,
+      "loss": 0.6185,
+      "step": 267
+    },
+    {
+      "epoch": 0.41647241647241645,
+      "grad_norm": 0.9024237646983181,
+      "learning_rate": 6.3157125529651205e-06,
+      "loss": 0.4375,
+      "step": 268
+    },
+    {
+      "epoch": 0.41802641802641805,
+      "grad_norm": 0.728576137938516,
+      "learning_rate": 6.292128720247692e-06,
+      "loss": 0.416,
+      "step": 269
+    },
+    {
+      "epoch": 0.4195804195804196,
+      "grad_norm": 0.7209803232506808,
+      "learning_rate": 6.268514042676519e-06,
+      "loss": 0.4931,
+      "step": 270
+    },
+    {
+      "epoch": 0.4211344211344211,
+      "grad_norm": 0.9852718978800553,
+      "learning_rate": 6.244869083965777e-06,
+      "loss": 0.6138,
+      "step": 271
+    },
+    {
+      "epoch": 0.4226884226884227,
+      "grad_norm": 0.8544813586101946,
+      "learning_rate": 6.221194408552494e-06,
+      "loss": 0.4828,
+      "step": 272
+    },
+    {
+      "epoch": 0.42424242424242425,
+      "grad_norm": 0.5339139693029649,
+      "learning_rate": 6.197490581583078e-06,
+      "loss": 0.3763,
+      "step": 273
+    },
+    {
+      "epoch": 0.4257964257964258,
+      "grad_norm": 0.7029889613516929,
+      "learning_rate": 6.173758168899814e-06,
+      "loss": 0.5003,
+      "step": 274
+    },
+    {
+      "epoch": 0.42735042735042733,
+      "grad_norm": 0.6697709635144842,
+      "learning_rate": 6.149997737027377e-06,
+      "loss": 0.5023,
+      "step": 275
+    },
+    {
+      "epoch": 0.4289044289044289,
+      "grad_norm": 0.7711140122998004,
+      "learning_rate": 6.126209853159293e-06,
+      "loss": 0.5458,
+      "step": 276
+    },
+    {
+      "epoch": 0.43045843045843046,
+      "grad_norm": 0.7959237764492787,
+      "learning_rate": 6.102395085144406e-06,
+      "loss": 0.5436,
+      "step": 277
+    },
+    {
+      "epoch": 0.432012432012432,
+      "grad_norm": 0.7725608290998155,
+      "learning_rate": 6.078554001473317e-06,
+      "loss": 0.5174,
+      "step": 278
+    },
+    {
+      "epoch": 0.43356643356643354,
+      "grad_norm": 1.0803231102952509,
+      "learning_rate": 6.054687171264822e-06,
+      "loss": 0.6228,
+      "step": 279
+    },
+    {
+      "epoch": 0.43512043512043513,
+      "grad_norm": 0.8198164427592709,
+      "learning_rate": 6.030795164252321e-06,
+      "loss": 0.4489,
+      "step": 280
+    },
+    {
+      "epoch": 0.4366744366744367,
+      "grad_norm": 0.9826854880546106,
+      "learning_rate": 6.006878550770213e-06,
+      "loss": 0.4776,
+      "step": 281
+    },
+    {
+      "epoch": 0.4382284382284382,
+      "grad_norm": 0.6853341082656256,
+      "learning_rate": 5.982937901740296e-06,
+      "loss": 0.5168,
+      "step": 282
+    },
+    {
+      "epoch": 0.4397824397824398,
+      "grad_norm": 0.8527642374724453,
+      "learning_rate": 5.958973788658115e-06,
+      "loss": 0.4427,
+      "step": 283
+    },
+    {
+      "epoch": 0.44133644133644134,
+      "grad_norm": 0.820049645998347,
+      "learning_rate": 5.934986783579349e-06,
+      "loss": 0.5325,
+      "step": 284
+    },
+    {
+      "epoch": 0.4428904428904429,
+      "grad_norm": 0.8635505053006802,
+      "learning_rate": 5.91097745910613e-06,
+      "loss": 0.5158,
+      "step": 285
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 0.6163515215798748,
+      "learning_rate": 5.886946388373387e-06,
+      "loss": 0.5613,
+      "step": 286
+    },
+    {
+      "epoch": 0.445998445998446,
+      "grad_norm": 0.7114308768259194,
+      "learning_rate": 5.862894145035158e-06,
+      "loss": 0.5812,
+      "step": 287
+    },
+    {
+      "epoch": 0.44755244755244755,
+      "grad_norm": 0.7695278464891274,
+      "learning_rate": 5.83882130325091e-06,
+      "loss": 0.5216,
+      "step": 288
+    },
+    {
+      "epoch": 0.4491064491064491,
+      "grad_norm": 0.7953550029299292,
+      "learning_rate": 5.814728437671808e-06,
+      "loss": 0.3821,
+      "step": 289
+    },
+    {
+      "epoch": 0.4506604506604507,
+      "grad_norm": 0.8080489538709861,
+      "learning_rate": 5.7906161234270234e-06,
+      "loss": 0.4982,
+      "step": 290
+    },
+    {
+      "epoch": 0.4522144522144522,
+      "grad_norm": 0.7349261540661857,
+      "learning_rate": 5.766484936109987e-06,
+      "loss": 0.4715,
+      "step": 291
+    },
+    {
+      "epoch": 0.45376845376845376,
+      "grad_norm": 1.0100168672849925,
+      "learning_rate": 5.7423354517646616e-06,
+      "loss": 0.4585,
+      "step": 292
+    },
+    {
+      "epoch": 0.4553224553224553,
+      "grad_norm": 0.7711000819084326,
+      "learning_rate": 5.718168246871775e-06,
+      "loss": 0.5841,
+      "step": 293
+    },
+    {
+      "epoch": 0.4568764568764569,
+      "grad_norm": 0.9388643582133742,
+      "learning_rate": 5.69398389833508e-06,
+      "loss": 0.5857,
+      "step": 294
+    },
+    {
+      "epoch": 0.45843045843045843,
+      "grad_norm": 0.6391229423057062,
+      "learning_rate": 5.669782983467562e-06,
+      "loss": 0.5786,
+      "step": 295
+    },
+    {
+      "epoch": 0.45998445998445997,
+      "grad_norm": 0.7381704493689959,
+      "learning_rate": 5.645566079977673e-06,
+      "loss": 0.4531,
+      "step": 296
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 0.6427062310992607,
+      "learning_rate": 5.621333765955529e-06,
+      "loss": 0.4516,
+      "step": 297
+    },
+    {
+      "epoch": 0.4630924630924631,
+      "grad_norm": 0.6045207793918917,
+      "learning_rate": 5.5970866198591235e-06,
+      "loss": 0.5904,
+      "step": 298
+    },
+    {
+      "epoch": 0.46464646464646464,
+      "grad_norm": 0.7486268299657493,
+      "learning_rate": 5.572825220500505e-06,
+      "loss": 0.5559,
+      "step": 299
+    },
+    {
+      "epoch": 0.4662004662004662,
+      "grad_norm": 0.8393377063443249,
+      "learning_rate": 5.548550147031971e-06,
+      "loss": 0.2951,
+      "step": 300
+    },
+    {
+      "epoch": 0.4677544677544678,
+      "grad_norm": 0.5563021839864102,
+      "learning_rate": 5.524261978932234e-06,
+      "loss": 0.4705,
+      "step": 301
+    },
+    {
+      "epoch": 0.4693084693084693,
+      "grad_norm": 0.5244277581252728,
+      "learning_rate": 5.4999612959925995e-06,
+      "loss": 0.5826,
+      "step": 302
+    },
+    {
+      "epoch": 0.47086247086247085,
+      "grad_norm": 0.8967222238475693,
+      "learning_rate": 5.475648678303112e-06,
+      "loss": 0.4326,
+      "step": 303
+    },
+    {
+      "epoch": 0.47241647241647244,
+      "grad_norm": 0.8597788121495409,
+      "learning_rate": 5.451324706238721e-06,
+      "loss": 0.5024,
+      "step": 304
+    },
+    {
+      "epoch": 0.473970473970474,
+      "grad_norm": 0.6219585515878757,
+      "learning_rate": 5.426989960445415e-06,
+      "loss": 0.4228,
+      "step": 305
+    },
+    {
+      "epoch": 0.4755244755244755,
+      "grad_norm": 0.87442477445858,
+      "learning_rate": 5.402645021826367e-06,
+      "loss": 0.5451,
+      "step": 306
+    },
+    {
+      "epoch": 0.47707847707847706,
+      "grad_norm": 0.7018801509337828,
+      "learning_rate": 5.3782904715280705e-06,
+      "loss": 0.454,
+      "step": 307
+    },
+    {
+      "epoch": 0.47863247863247865,
+      "grad_norm": 0.7870901274907686,
+      "learning_rate": 5.35392689092646e-06,
+      "loss": 0.6209,
+      "step": 308
+    },
+    {
+      "epoch": 0.4801864801864802,
+      "grad_norm": 0.8404067379482166,
+      "learning_rate": 5.329554861613031e-06,
+      "loss": 0.4577,
+      "step": 309
+    },
+    {
+      "epoch": 0.48174048174048173,
+      "grad_norm": 0.8609659868189161,
+      "learning_rate": 5.3051749653809685e-06,
+      "loss": 0.516,
+      "step": 310
+    },
+    {
+      "epoch": 0.48329448329448327,
+      "grad_norm": 0.6451369228669567,
+      "learning_rate": 5.2807877842112475e-06,
+      "loss": 0.529,
+      "step": 311
+    },
+    {
+      "epoch": 0.48484848484848486,
+      "grad_norm": 0.7089279571799519,
+      "learning_rate": 5.256393900258747e-06,
+      "loss": 0.55,
+      "step": 312
+    },
+    {
+      "epoch": 0.4864024864024864,
+      "grad_norm": 0.6009435199914196,
+      "learning_rate": 5.231993895838348e-06,
+      "loss": 0.4524,
+      "step": 313
+    },
+    {
+      "epoch": 0.48795648795648794,
+      "grad_norm": 0.7399390807603545,
+      "learning_rate": 5.207588353411032e-06,
+      "loss": 0.5826,
+      "step": 314
+    },
+    {
+      "epoch": 0.48951048951048953,
+      "grad_norm": 0.6358169721821743,
+      "learning_rate": 5.183177855569989e-06,
+      "loss": 0.4686,
+      "step": 315
+    },
+    {
+      "epoch": 0.49106449106449107,
+      "grad_norm": 0.834159373189627,
+      "learning_rate": 5.158762985026694e-06,
+      "loss": 0.6194,
+      "step": 316
+    },
+    {
+      "epoch": 0.4926184926184926,
+      "grad_norm": 0.672932649828031,
+      "learning_rate": 5.1343443245970095e-06,
+      "loss": 0.5593,
+      "step": 317
+    },
+    {
+      "epoch": 0.49417249417249415,
+      "grad_norm": 0.7305995766455743,
+      "learning_rate": 5.10992245718726e-06,
+      "loss": 0.4943,
+      "step": 318
+    },
+    {
+      "epoch": 0.49572649572649574,
+      "grad_norm": 1.1658079579901208,
+      "learning_rate": 5.085497965780335e-06,
+      "loss": 0.5031,
+      "step": 319
+    },
+    {
+      "epoch": 0.4972804972804973,
+      "grad_norm": 0.9841093713953577,
+      "learning_rate": 5.061071433421754e-06,
+      "loss": 0.5508,
+      "step": 320
+    },
+    {
+      "epoch": 0.4988344988344988,
+      "grad_norm": 0.7841115815090715,
+      "learning_rate": 5.0366434432057624e-06,
+      "loss": 0.563,
+      "step": 321
+    },
+    {
+      "epoch": 0.5003885003885004,
+      "grad_norm": 0.5645683948415432,
+      "learning_rate": 5.012214578261402e-06,
+      "loss": 0.4011,
+      "step": 322
+    },
+    {
+      "epoch": 0.5019425019425019,
+      "grad_norm": 0.8438933215498013,
+      "learning_rate": 4.987785421738599e-06,
+      "loss": 0.5832,
+      "step": 323
+    },
+    {
+      "epoch": 0.5034965034965035,
+      "grad_norm": 0.5640735635986946,
+      "learning_rate": 4.963356556794238e-06,
+      "loss": 0.401,
+      "step": 324
+    },
+    {
+      "epoch": 0.5050505050505051,
+      "grad_norm": 1.2747685452336512,
+      "learning_rate": 4.938928566578247e-06,
+      "loss": 0.5392,
+      "step": 325
+    },
+    {
+      "epoch": 0.5066045066045066,
+      "grad_norm": 0.6877059270681696,
+      "learning_rate": 4.914502034219667e-06,
+      "loss": 0.439,
+      "step": 326
+    },
+    {
+      "epoch": 0.5081585081585082,
+      "grad_norm": 0.705842825427767,
+      "learning_rate": 4.890077542812742e-06,
+      "loss": 0.5187,
+      "step": 327
+    },
+    {
+      "epoch": 0.5097125097125097,
+      "grad_norm": 0.5947821340173784,
+      "learning_rate": 4.865655675402993e-06,
+      "loss": 0.4178,
+      "step": 328
+    },
+    {
+      "epoch": 0.5112665112665112,
+      "grad_norm": 0.6757902851130476,
+      "learning_rate": 4.841237014973305e-06,
+      "loss": 0.493,
+      "step": 329
+    },
+    {
+      "epoch": 0.5128205128205128,
+      "grad_norm": 0.6256933230220865,
+      "learning_rate": 4.8168221444300124e-06,
+      "loss": 0.4499,
+      "step": 330
+    },
+    {
+      "epoch": 0.5143745143745144,
+      "grad_norm": 0.6390396367290697,
+      "learning_rate": 4.7924116465889684e-06,
+      "loss": 0.4783,
+      "step": 331
+    },
+    {
+      "epoch": 0.515928515928516,
+      "grad_norm": 0.7039920639754556,
+      "learning_rate": 4.768006104161655e-06,
+      "loss": 0.4862,
+      "step": 332
+    },
+    {
+      "epoch": 0.5174825174825175,
+      "grad_norm": 0.6996136377759004,
+      "learning_rate": 4.743606099741255e-06,
+      "loss": 0.43,
+      "step": 333
+    },
+    {
+      "epoch": 0.519036519036519,
+      "grad_norm": 0.5635901754632139,
+      "learning_rate": 4.719212215788753e-06,
+      "loss": 0.3489,
+      "step": 334
+    },
+    {
+      "epoch": 0.5205905205905206,
+      "grad_norm": 1.0174320744839043,
+      "learning_rate": 4.6948250346190315e-06,
+      "loss": 0.4664,
+      "step": 335
+    },
+    {
+      "epoch": 0.5221445221445221,
+      "grad_norm": 0.7281418344534993,
+      "learning_rate": 4.670445138386971e-06,
+      "loss": 0.4357,
+      "step": 336
+    },
+    {
+      "epoch": 0.5236985236985237,
+      "grad_norm": 0.7444913580894218,
+      "learning_rate": 4.646073109073542e-06,
+      "loss": 0.5164,
+      "step": 337
+    },
+    {
+      "epoch": 0.5252525252525253,
+      "grad_norm": 0.6453883418849363,
+      "learning_rate": 4.621709528471931e-06,
+      "loss": 0.4876,
+      "step": 338
+    },
+    {
+      "epoch": 0.5268065268065268,
+      "grad_norm": 0.6332116311781928,
+      "learning_rate": 4.5973549781736335e-06,
+      "loss": 0.4512,
+      "step": 339
+    },
+    {
+      "epoch": 0.5283605283605284,
+      "grad_norm": 0.7839883177892899,
+      "learning_rate": 4.573010039554587e-06,
+      "loss": 0.4864,
+      "step": 340
+    },
+    {
+      "epoch": 0.5299145299145299,
+      "grad_norm": 0.574123724971613,
+      "learning_rate": 4.548675293761281e-06,
+      "loss": 0.5942,
+      "step": 341
+    },
+    {
+      "epoch": 0.5314685314685315,
+      "grad_norm": 0.7341359415325991,
+      "learning_rate": 4.524351321696889e-06,
+      "loss": 0.6479,
+      "step": 342
+    },
+    {
+      "epoch": 0.533022533022533,
+      "grad_norm": 0.7042514483644169,
+      "learning_rate": 4.500038704007402e-06,
+      "loss": 0.5163,
+      "step": 343
+    },
+    {
+      "epoch": 0.5345765345765345,
+      "grad_norm": 0.7905795068611852,
+      "learning_rate": 4.475738021067768e-06,
+      "loss": 0.4238,
+      "step": 344
+    },
+    {
+      "epoch": 0.5361305361305362,
+      "grad_norm": 0.6659802197095699,
+      "learning_rate": 4.451449852968031e-06,
+      "loss": 0.5456,
+      "step": 345
+    },
+    {
+      "epoch": 0.5376845376845377,
+      "grad_norm": 0.790921006825035,
+      "learning_rate": 4.427174779499498e-06,
+      "loss": 0.3841,
+      "step": 346
+    },
+    {
+      "epoch": 0.5392385392385393,
+      "grad_norm": 0.7752025731889072,
+      "learning_rate": 4.402913380140878e-06,
+      "loss": 0.5737,
+      "step": 347
+    },
+    {
+      "epoch": 0.5407925407925408,
+      "grad_norm": 0.69589495472164,
+      "learning_rate": 4.378666234044471e-06,
+      "loss": 0.5731,
+      "step": 348
+    },
+    {
+      "epoch": 0.5423465423465423,
+      "grad_norm": 0.6620238929419696,
+      "learning_rate": 4.354433920022328e-06,
+      "loss": 0.6002,
+      "step": 349
+    },
+    {
+      "epoch": 0.5439005439005439,
+      "grad_norm": 0.7640658489542932,
+      "learning_rate": 4.3302170165324385e-06,
+      "loss": 0.5681,
+      "step": 350
+    },
+    {
+      "epoch": 0.5454545454545454,
+      "grad_norm": 0.8733971833765328,
+      "learning_rate": 4.306016101664921e-06,
+      "loss": 0.5128,
+      "step": 351
+    },
+    {
+      "epoch": 0.5470085470085471,
+      "grad_norm": 1.0078351927972078,
+      "learning_rate": 4.281831753128226e-06,
+      "loss": 0.5179,
+      "step": 352
+    },
+    {
+      "epoch": 0.5485625485625486,
+      "grad_norm": 0.5776801607508973,
+      "learning_rate": 4.257664548235341e-06,
+      "loss": 0.5191,
+      "step": 353
+    },
+    {
+      "epoch": 0.5501165501165501,
+      "grad_norm": 0.7595818677545988,
+      "learning_rate": 4.233515063890013e-06,
+      "loss": 0.4912,
+      "step": 354
+    },
+    {
+      "epoch": 0.5516705516705517,
+      "grad_norm": 0.6606204189468279,
+      "learning_rate": 4.209383876572977e-06,
+      "loss": 0.5694,
+      "step": 355
+    },
+    {
+      "epoch": 0.5532245532245532,
+      "grad_norm": 0.7228853812417981,
+      "learning_rate": 4.1852715623281934e-06,
+      "loss": 0.5009,
+      "step": 356
+    },
+    {
+      "epoch": 0.5547785547785548,
+      "grad_norm": 0.679061167051694,
+      "learning_rate": 4.161178696749092e-06,
+      "loss": 0.4821,
+      "step": 357
+    },
+    {
+      "epoch": 0.5563325563325563,
+      "grad_norm": 1.034495766490782,
+      "learning_rate": 4.1371058549648425e-06,
+      "loss": 0.5308,
+      "step": 358
+    },
+    {
+      "epoch": 0.5578865578865578,
+      "grad_norm": 0.7686980644117477,
+      "learning_rate": 4.1130536116266155e-06,
+      "loss": 0.489,
+      "step": 359
+    },
+    {
+      "epoch": 0.5594405594405595,
+      "grad_norm": 0.9681286219055543,
+      "learning_rate": 4.089022540893871e-06,
+      "loss": 0.5282,
+      "step": 360
+    },
+    {
+      "epoch": 0.560994560994561,
+      "grad_norm": 0.8743676508733846,
+      "learning_rate": 4.0650132164206515e-06,
+      "loss": 0.5241,
+      "step": 361
+    },
+    {
+      "epoch": 0.5625485625485626,
+      "grad_norm": 0.6062796295682528,
+      "learning_rate": 4.041026211341886e-06,
+      "loss": 0.5007,
+      "step": 362
+    },
+    {
+      "epoch": 0.5641025641025641,
+      "grad_norm": 0.8866048922703949,
+      "learning_rate": 4.017062098259707e-06,
+      "loss": 0.5478,
+      "step": 363
+    },
+    {
+      "epoch": 0.5656565656565656,
+      "grad_norm": 0.6320850311198043,
+      "learning_rate": 3.9931214492297875e-06,
+      "loss": 0.5351,
+      "step": 364
+    },
+    {
+      "epoch": 0.5672105672105672,
+      "grad_norm": 0.6435052023704503,
+      "learning_rate": 3.969204835747681e-06,
+      "loss": 0.4535,
+      "step": 365
+    },
+    {
+      "epoch": 0.5687645687645687,
+      "grad_norm": 0.6824957419803489,
+      "learning_rate": 3.945312828735179e-06,
+      "loss": 0.4962,
+      "step": 366
+    },
+    {
+      "epoch": 0.5703185703185704,
+      "grad_norm": 0.6630522920673861,
+      "learning_rate": 3.921445998526684e-06,
+      "loss": 0.5484,
+      "step": 367
+    },
+    {
+      "epoch": 0.5718725718725719,
+      "grad_norm": 0.6203973778780169,
+      "learning_rate": 3.897604914855596e-06,
+      "loss": 0.4307,
+      "step": 368
+    },
+    {
+      "epoch": 0.5734265734265734,
+      "grad_norm": 0.8691644246672308,
+      "learning_rate": 3.873790146840709e-06,
+      "loss": 0.4494,
+      "step": 369
+    },
+    {
+      "epoch": 0.574980574980575,
+      "grad_norm": 0.6583038284757291,
+      "learning_rate": 3.8500022629726246e-06,
+      "loss": 0.5461,
+      "step": 370
+    },
+    {
+      "epoch": 0.5765345765345765,
+      "grad_norm": 0.6205452786275579,
+      "learning_rate": 3.8262418311001884e-06,
+      "loss": 0.441,
+      "step": 371
+    },
+    {
+      "epoch": 0.578088578088578,
+      "grad_norm": 0.8090078519631057,
+      "learning_rate": 3.8025094184169254e-06,
+      "loss": 0.4396,
+      "step": 372
+    },
+    {
+      "epoch": 0.5796425796425796,
+      "grad_norm": 3.7824694886306442,
+      "learning_rate": 3.778805591447505e-06,
+      "loss": 0.5243,
+      "step": 373
+    },
+    {
+      "epoch": 0.5811965811965812,
+      "grad_norm": 0.7728659065744149,
+      "learning_rate": 3.7551309160342233e-06,
+      "loss": 0.492,
+      "step": 374
+    },
+    {
+      "epoch": 0.5827505827505828,
+      "grad_norm": 0.6144384200403961,
+      "learning_rate": 3.731485957323483e-06,
+      "loss": 0.5093,
+      "step": 375
+    },
+    {
+      "epoch": 0.5843045843045843,
+      "grad_norm": 0.7022871161510597,
+      "learning_rate": 3.707871279752309e-06,
+      "loss": 0.5471,
+      "step": 376
+    },
+    {
+      "epoch": 0.5858585858585859,
+      "grad_norm": 0.7563905294320689,
+      "learning_rate": 3.68428744703488e-06,
+      "loss": 0.5169,
+      "step": 377
+    },
+    {
+      "epoch": 0.5874125874125874,
+      "grad_norm": 0.7546808388382039,
+      "learning_rate": 3.6607350221490593e-06,
+      "loss": 0.5465,
+      "step": 378
+    },
+    {
+      "epoch": 0.5889665889665889,
+      "grad_norm": 2.017091506116899,
+      "learning_rate": 3.6372145673229683e-06,
+      "loss": 0.5379,
+      "step": 379
+    },
+    {
+      "epoch": 0.5905205905205905,
+      "grad_norm": 1.0384918566794876,
+      "learning_rate": 3.613726644021559e-06,
+      "loss": 0.5533,
+      "step": 380
+    },
+    {
+      "epoch": 0.5920745920745921,
+      "grad_norm": 0.7843829856867381,
+      "learning_rate": 3.590271812933212e-06,
+      "loss": 0.5192,
+      "step": 381
+    },
+    {
+      "epoch": 0.5936285936285937,
+      "grad_norm": 0.8917403925552857,
+      "learning_rate": 3.5668506339563502e-06,
+      "loss": 0.5273,
+      "step": 382
+    },
+    {
+      "epoch": 0.5951825951825952,
+      "grad_norm": 0.8080693974342459,
+      "learning_rate": 3.5434636661860776e-06,
+      "loss": 0.4502,
+      "step": 383
+    },
+    {
+      "epoch": 0.5967365967365967,
+      "grad_norm": 0.8589025653140446,
+      "learning_rate": 3.5201114679008286e-06,
+      "loss": 0.4597,
+      "step": 384
+    },
+    {
+      "epoch": 0.5982905982905983,
+      "grad_norm": 0.8451337336955351,
+      "learning_rate": 3.4967945965490434e-06,
+      "loss": 0.5479,
+      "step": 385
+    },
+    {
+      "epoch": 0.5998445998445998,
+      "grad_norm": 0.6022399299860023,
+      "learning_rate": 3.4735136087358646e-06,
+      "loss": 0.3617,
+      "step": 386
+    },
+    {
+      "epoch": 0.6013986013986014,
+      "grad_norm": 1.091419134601723,
+      "learning_rate": 3.450269060209841e-06,
+      "loss": 0.4422,
+      "step": 387
+    },
+    {
+      "epoch": 0.602952602952603,
+      "grad_norm": 0.7663320014407212,
+      "learning_rate": 3.42706150584967e-06,
+      "loss": 0.5793,
+      "step": 388
+    },
+    {
+      "epoch": 0.6045066045066045,
+      "grad_norm": 0.8338372079331721,
+      "learning_rate": 3.4038914996509464e-06,
+      "loss": 0.6015,
+      "step": 389
+    },
+    {
+      "epoch": 0.6060606060606061,
+      "grad_norm": 0.8601272441851022,
+      "learning_rate": 3.3807595947129405e-06,
+      "loss": 0.6391,
+      "step": 390
+    },
+    {
+      "epoch": 0.6076146076146076,
+      "grad_norm": 0.8548829786482054,
+      "learning_rate": 3.357666343225396e-06,
+      "loss": 0.5055,
+      "step": 391
+    },
+    {
+      "epoch": 0.6091686091686092,
+      "grad_norm": 0.7149586625595873,
+      "learning_rate": 3.3346122964553407e-06,
+      "loss": 0.4305,
+      "step": 392
+    },
+    {
+      "epoch": 0.6107226107226107,
+      "grad_norm": 0.7734311233562257,
+      "learning_rate": 3.3115980047339415e-06,
+      "loss": 0.4676,
+      "step": 393
+    },
+    {
+      "epoch": 0.6122766122766122,
+      "grad_norm": 0.7524053415700651,
+      "learning_rate": 3.288624017443353e-06,
+      "loss": 0.6177,
+      "step": 394
+    },
+    {
+      "epoch": 0.6138306138306139,
+      "grad_norm": 0.6577165811691974,
+      "learning_rate": 3.265690883003609e-06,
+      "loss": 0.4062,
+      "step": 395
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.7818702560735775,
+      "learning_rate": 3.2427991488595334e-06,
+      "loss": 0.7139,
+      "step": 396
+    },
+    {
+      "epoch": 0.616938616938617,
+      "grad_norm": 0.8337417209233989,
+      "learning_rate": 3.219949361467668e-06,
+      "loss": 0.3658,
+      "step": 397
+    },
+    {
+      "epoch": 0.6184926184926185,
+      "grad_norm": 0.6408051420484582,
+      "learning_rate": 3.197142066283225e-06,
+      "loss": 0.4037,
+      "step": 398
+    },
+    {
+      "epoch": 0.62004662004662,
+      "grad_norm": 0.6924590018612481,
+      "learning_rate": 3.174377807747079e-06,
+      "loss": 0.3915,
+      "step": 399
+    },
+    {
+      "epoch": 0.6216006216006216,
+      "grad_norm": 0.7521284979964744,
+      "learning_rate": 3.1516571292727553e-06,
+      "loss": 0.5316,
+      "step": 400
+    },
+    {
+      "epoch": 0.6231546231546231,
+      "grad_norm": 0.6902359570029046,
+      "learning_rate": 3.128980573233465e-06,
+      "loss": 0.3485,
+      "step": 401
+    },
+    {
+      "epoch": 0.6247086247086248,
+      "grad_norm": 0.8616508180715082,
+      "learning_rate": 3.1063486809491595e-06,
+      "loss": 0.5077,
+      "step": 402
+    },
+    {
+      "epoch": 0.6262626262626263,
+      "grad_norm": 0.7958497698850551,
+      "learning_rate": 3.0837619926736027e-06,
+      "loss": 0.4225,
+      "step": 403
+    },
+    {
+      "epoch": 0.6278166278166278,
+      "grad_norm": 0.7109428935226295,
+      "learning_rate": 3.061221047581482e-06,
+      "loss": 0.6677,
+      "step": 404
+    },
+    {
+      "epoch": 0.6293706293706294,
+      "grad_norm": 0.7470001438392854,
+      "learning_rate": 3.038726383755531e-06,
+      "loss": 0.4242,
+      "step": 405
+    },
+    {
+      "epoch": 0.6309246309246309,
+      "grad_norm": 0.6742001176378809,
+      "learning_rate": 3.0162785381736893e-06,
+      "loss": 0.4731,
+      "step": 406
+    },
+    {
+      "epoch": 0.6324786324786325,
+      "grad_norm": 0.8148615875421817,
+      "learning_rate": 2.9938780466962768e-06,
+      "loss": 0.5124,
+      "step": 407
+    },
+    {
+      "epoch": 0.634032634032634,
+      "grad_norm": 0.629279814363528,
+      "learning_rate": 2.9715254440532147e-06,
+      "loss": 0.5542,
+      "step": 408
+    },
+    {
+      "epoch": 0.6355866355866356,
+      "grad_norm": 0.7893368478824807,
+      "learning_rate": 2.9492212638312458e-06,
+      "loss": 0.4712,
+      "step": 409
+    },
+    {
+      "epoch": 0.6371406371406372,
+      "grad_norm": 0.700072426705422,
+      "learning_rate": 2.9269660384612064e-06,
+      "loss": 0.5445,
+      "step": 410
+    },
+    {
+      "epoch": 0.6386946386946387,
+      "grad_norm": 0.6730347447205193,
+      "learning_rate": 2.90476029920532e-06,
+      "loss": 0.569,
+      "step": 411
+    },
+    {
+      "epoch": 0.6402486402486403,
+      "grad_norm": 1.8040910895752915,
+      "learning_rate": 2.882604576144505e-06,
+      "loss": 0.5713,
+      "step": 412
+    },
+    {
+      "epoch": 0.6418026418026418,
+      "grad_norm": 0.6061113607030476,
+      "learning_rate": 2.8604993981657247e-06,
+      "loss": 0.4493,
+      "step": 413
+    },
+    {
+      "epoch": 0.6433566433566433,
+      "grad_norm": 1.350624074574079,
+      "learning_rate": 2.8384452929493645e-06,
+      "loss": 0.4243,
+      "step": 414
+    },
+    {
+      "epoch": 0.6449106449106449,
+      "grad_norm": 1.020418478286547,
+      "learning_rate": 2.8164427869566367e-06,
+      "loss": 0.4695,
+      "step": 415
+    },
+    {
+      "epoch": 0.6464646464646465,
+      "grad_norm": 1.073163390573267,
+      "learning_rate": 2.7944924054170087e-06,
+      "loss": 0.4878,
+      "step": 416
+    },
+    {
+      "epoch": 0.6480186480186481,
+      "grad_norm": 0.8939568549949852,
+      "learning_rate": 2.7725946723156626e-06,
+      "loss": 0.564,
+      "step": 417
+    },
+    {
+      "epoch": 0.6495726495726496,
+      "grad_norm": 0.7664017459877794,
+      "learning_rate": 2.750750110381001e-06,
+      "loss": 0.499,
+      "step": 418
+    },
+    {
+      "epoch": 0.6511266511266511,
+      "grad_norm": 0.8801185511706636,
+      "learning_rate": 2.728959241072149e-06,
+      "loss": 0.4769,
+      "step": 419
+    },
+    {
+      "epoch": 0.6526806526806527,
+      "grad_norm": 0.645638892422625,
+      "learning_rate": 2.7072225845665256e-06,
+      "loss": 0.5266,
+      "step": 420
+    },
+    {
+      "epoch": 0.6542346542346542,
+      "grad_norm": 0.8936830405965801,
+      "learning_rate": 2.6855406597474098e-06,
+      "loss": 0.5236,
+      "step": 421
+    },
+    {
+      "epoch": 0.6557886557886557,
+      "grad_norm": 0.8974442709328094,
+      "learning_rate": 2.6639139841915628e-06,
+      "loss": 0.5072,
+      "step": 422
+    },
+    {
+      "epoch": 0.6573426573426573,
+      "grad_norm": 0.8176617831786887,
+      "learning_rate": 2.6423430741568746e-06,
+      "loss": 0.6122,
+      "step": 423
+    },
+    {
+      "epoch": 0.6588966588966589,
+      "grad_norm": 1.027696072983406,
+      "learning_rate": 2.6208284445700373e-06,
+      "loss": 0.5373,
+      "step": 424
+    },
+    {
+      "epoch": 0.6604506604506605,
+      "grad_norm": 1.254318562097477,
+      "learning_rate": 2.5993706090142484e-06,
+      "loss": 0.5659,
+      "step": 425
+    },
+    {
+      "epoch": 0.662004662004662,
+      "grad_norm": 0.9412050785188167,
+      "learning_rate": 2.577970079716959e-06,
+      "loss": 0.6532,
+      "step": 426
+    },
+    {
+      "epoch": 0.6635586635586636,
+      "grad_norm": 0.669172885578478,
+      "learning_rate": 2.5566273675376386e-06,
+      "loss": 0.4321,
+      "step": 427
+    },
+    {
+      "epoch": 0.6651126651126651,
+      "grad_norm": 0.7839761385934717,
+      "learning_rate": 2.535342981955591e-06,
+      "loss": 0.5029,
+      "step": 428
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.6291916677121909,
+      "learning_rate": 2.5141174310577774e-06,
+      "loss": 0.5619,
+      "step": 429
+    },
+    {
+      "epoch": 0.6682206682206682,
+      "grad_norm": 0.6009527872848712,
+      "learning_rate": 2.492951221526705e-06,
+      "loss": 0.4838,
+      "step": 430
+    },
+    {
+      "epoch": 0.6697746697746698,
+      "grad_norm": 0.7557317812271531,
+      "learning_rate": 2.4718448586283126e-06,
+      "loss": 0.5686,
+      "step": 431
+    },
+    {
+      "epoch": 0.6713286713286714,
+      "grad_norm": 0.6594827809287767,
+      "learning_rate": 2.4507988461999283e-06,
+      "loss": 0.4434,
+      "step": 432
+    },
+    {
+      "epoch": 0.6728826728826729,
+      "grad_norm": 0.6721258490052082,
+      "learning_rate": 2.429813686638227e-06,
+      "loss": 0.5448,
+      "step": 433
+    },
+    {
+      "epoch": 0.6744366744366744,
+      "grad_norm": 0.6253425210970757,
+      "learning_rate": 2.408889880887246e-06,
+      "loss": 0.5399,
+      "step": 434
+    },
+    {
+      "epoch": 0.675990675990676,
+      "grad_norm": 0.7132068704320651,
+      "learning_rate": 2.38802792842642e-06,
+      "loss": 0.4131,
+      "step": 435
+    },
+    {
+      "epoch": 0.6775446775446775,
+      "grad_norm": 0.9992884448946338,
+      "learning_rate": 2.3672283272586745e-06,
+      "loss": 0.6681,
+      "step": 436
+    },
+    {
+      "epoch": 0.679098679098679,
+      "grad_norm": 0.6736697588900878,
+      "learning_rate": 2.346491573898513e-06,
+      "loss": 0.5672,
+      "step": 437
+    },
+    {
+      "epoch": 0.6806526806526807,
+      "grad_norm": 0.8547599476784933,
+      "learning_rate": 2.3258181633601836e-06,
+      "loss": 0.5186,
+      "step": 438
+    },
+    {
+      "epoch": 0.6822066822066822,
+      "grad_norm": 0.7608589363389933,
+      "learning_rate": 2.30520858914585e-06,
+      "loss": 0.4367,
+      "step": 439
+    },
+    {
+      "epoch": 0.6837606837606838,
+      "grad_norm": 0.6900628130745625,
+      "learning_rate": 2.2846633432338256e-06,
+      "loss": 0.5083,
+      "step": 440
+    },
+    {
+      "epoch": 0.6853146853146853,
+      "grad_norm": 0.7581972429614234,
+      "learning_rate": 2.2641829160668137e-06,
+      "loss": 0.4485,
+      "step": 441
+    },
+    {
+      "epoch": 0.6868686868686869,
+      "grad_norm": 0.7973753083321776,
+      "learning_rate": 2.243767796540207e-06,
+      "loss": 0.4911,
+      "step": 442
+    },
+    {
+      "epoch": 0.6884226884226884,
+      "grad_norm": 0.8430463031780523,
+      "learning_rate": 2.223418471990421e-06,
+      "loss": 0.6334,
+      "step": 443
+    },
+    {
+      "epoch": 0.6899766899766899,
+      "grad_norm": 0.6780160144046566,
+      "learning_rate": 2.2031354281832555e-06,
+      "loss": 0.4788,
+      "step": 444
+    },
+    {
+      "epoch": 0.6915306915306916,
+      "grad_norm": 0.8261099182920753,
+      "learning_rate": 2.1829191493022974e-06,
+      "loss": 0.5132,
+      "step": 445
+    },
+    {
+      "epoch": 0.6930846930846931,
+      "grad_norm": 0.9687007131393034,
+      "learning_rate": 2.1627701179373645e-06,
+      "loss": 0.5751,
+      "step": 446
+    },
+    {
+      "epoch": 0.6946386946386947,
+      "grad_norm": 0.6461088689056241,
+      "learning_rate": 2.142688815072986e-06,
+      "loss": 0.5866,
+      "step": 447
+    },
+    {
+      "epoch": 0.6961926961926962,
+      "grad_norm": 0.8088646308620774,
+      "learning_rate": 2.1226757200769225e-06,
+      "loss": 0.6883,
+      "step": 448
+    },
+    {
+      "epoch": 0.6977466977466977,
+      "grad_norm": 1.6821696576244451,
+      "learning_rate": 2.102731310688723e-06,
+      "loss": 0.5542,
+      "step": 449
+    },
+    {
+      "epoch": 0.6993006993006993,
+      "grad_norm": 0.6856586399011556,
+      "learning_rate": 2.0828560630083127e-06,
+      "loss": 0.3613,
+      "step": 450
+    },
+    {
+      "epoch": 0.7008547008547008,
+      "grad_norm": 0.8799640778401495,
+      "learning_rate": 2.0630504514846372e-06,
+      "loss": 0.6021,
+      "step": 451
+    },
+    {
+      "epoch": 0.7024087024087025,
+      "grad_norm": 1.0346792136769154,
+      "learning_rate": 2.0433149489043296e-06,
+      "loss": 0.6368,
+      "step": 452
+    },
+    {
+      "epoch": 0.703962703962704,
+      "grad_norm": 0.7233064199559888,
+      "learning_rate": 2.0236500263804355e-06,
+      "loss": 0.5101,
+      "step": 453
+    },
+    {
+      "epoch": 0.7055167055167055,
+      "grad_norm": 0.6101595125764867,
+      "learning_rate": 2.0040561533411494e-06,
+      "loss": 0.4283,
+      "step": 454
+    },
+    {
+      "epoch": 0.7070707070707071,
+      "grad_norm": 0.6838739790158097,
+      "learning_rate": 1.9845337975186297e-06,
+      "loss": 0.4044,
+      "step": 455
+    },
+    {
+      "epoch": 0.7086247086247086,
+      "grad_norm": 0.6512670498735073,
+      "learning_rate": 1.9650834249378125e-06,
+      "loss": 0.5122,
+      "step": 456
+    },
+    {
+      "epoch": 0.7101787101787101,
+      "grad_norm": 0.7448486863111607,
+      "learning_rate": 1.945705499905305e-06,
+      "loss": 0.4664,
+      "step": 457
+    },
+    {
+      "epoch": 0.7117327117327117,
+      "grad_norm": 0.6714430637930826,
+      "learning_rate": 1.926400484998289e-06,
+      "loss": 0.4605,
+      "step": 458
+    },
+    {
+      "epoch": 0.7132867132867133,
+      "grad_norm": 0.7034702671220221,
+      "learning_rate": 1.9071688410534828e-06,
+      "loss": 0.4332,
+      "step": 459
+    },
+    {
+      "epoch": 0.7148407148407149,
+      "grad_norm": 0.920215915504864,
+      "learning_rate": 1.8880110271561415e-06,
+      "loss": 0.582,
+      "step": 460
+    },
+    {
+      "epoch": 0.7163947163947164,
+      "grad_norm": 0.7704916182387226,
+      "learning_rate": 1.8689275006291035e-06,
+      "loss": 0.3934,
+      "step": 461
+    },
+    {
+      "epoch": 0.717948717948718,
+      "grad_norm": 0.7275954100685235,
+      "learning_rate": 1.8499187170218614e-06,
+      "loss": 0.6786,
+      "step": 462
+    },
+    {
+      "epoch": 0.7195027195027195,
+      "grad_norm": 0.9130677264258454,
+      "learning_rate": 1.8309851300996934e-06,
+      "loss": 0.4669,
+      "step": 463
+    },
+    {
+      "epoch": 0.721056721056721,
+      "grad_norm": 0.627613808696502,
+      "learning_rate": 1.8121271918328314e-06,
+      "loss": 0.3791,
+      "step": 464
+    },
+    {
+      "epoch": 0.7226107226107226,
+      "grad_norm": 0.7966353111195642,
+      "learning_rate": 1.793345352385678e-06,
+      "loss": 0.5131,
+      "step": 465
+    },
+    {
+      "epoch": 0.7241647241647242,
+      "grad_norm": 0.7257869097486319,
+      "learning_rate": 1.7746400601060476e-06,
+      "loss": 0.4698,
+      "step": 466
+    },
+    {
+      "epoch": 0.7257187257187258,
+      "grad_norm": 0.8274666640783022,
+      "learning_rate": 1.7560117615144717e-06,
+      "loss": 0.5127,
+      "step": 467
+    },
+    {
+      "epoch": 0.7272727272727273,
+      "grad_norm": 0.919410912868769,
+      "learning_rate": 1.7374609012935412e-06,
+      "loss": 0.5077,
+      "step": 468
+    },
+    {
+      "epoch": 0.7288267288267288,
+      "grad_norm": 0.8104780902360221,
+      "learning_rate": 1.7189879222772894e-06,
+      "loss": 0.4689,
+      "step": 469
+    },
+    {
+      "epoch": 0.7303807303807304,
+      "grad_norm": 0.7461709338199918,
+      "learning_rate": 1.7005932654406165e-06,
+      "loss": 0.3507,
+      "step": 470
+    },
+    {
+      "epoch": 0.7319347319347319,
+      "grad_norm": 0.6643952482627672,
+      "learning_rate": 1.682277369888769e-06,
+      "loss": 0.4538,
+      "step": 471
+    },
+    {
+      "epoch": 0.7334887334887334,
+      "grad_norm": 0.8395310070807439,
+      "learning_rate": 1.6640406728468534e-06,
+      "loss": 0.4401,
+      "step": 472
+    },
+    {
+      "epoch": 0.7350427350427351,
+      "grad_norm": 0.9088380178834886,
+      "learning_rate": 1.6458836096494046e-06,
+      "loss": 0.6829,
+      "step": 473
+    },
+    {
+      "epoch": 0.7365967365967366,
+      "grad_norm": 0.8158423552015206,
+      "learning_rate": 1.6278066137299898e-06,
+      "loss": 0.5124,
+      "step": 474
+    },
+    {
+      "epoch": 0.7381507381507382,
+      "grad_norm": 0.6628148397288011,
+      "learning_rate": 1.6098101166108593e-06,
+      "loss": 0.4556,
+      "step": 475
+    },
+    {
+      "epoch": 0.7397047397047397,
+      "grad_norm": 0.7157818223192728,
+      "learning_rate": 1.5918945478926484e-06,
+      "loss": 0.5349,
+      "step": 476
+    },
+    {
+      "epoch": 0.7412587412587412,
+      "grad_norm": 0.9140000377609874,
+      "learning_rate": 1.5740603352441281e-06,
+      "loss": 0.6558,
+      "step": 477
+    },
+    {
+      "epoch": 0.7428127428127428,
+      "grad_norm": 1.8219839216709126,
+      "learning_rate": 1.5563079043919843e-06,
+      "loss": 0.6314,
+      "step": 478
+    },
+    {
+      "epoch": 0.7443667443667443,
+      "grad_norm": 0.6702786236421273,
+      "learning_rate": 1.5386376791106627e-06,
+      "loss": 0.4835,
+      "step": 479
+    },
+    {
+      "epoch": 0.745920745920746,
+      "grad_norm": 0.8589486289983319,
+      "learning_rate": 1.5210500812122548e-06,
+      "loss": 0.28,
+      "step": 480
+    },
+    {
+      "epoch": 0.7474747474747475,
+      "grad_norm": 0.6696692189402571,
+      "learning_rate": 1.5035455305364188e-06,
+      "loss": 0.4483,
+      "step": 481
+    },
+    {
+      "epoch": 0.749028749028749,
+      "grad_norm": 0.832181293716284,
+      "learning_rate": 1.4861244449403717e-06,
+      "loss": 0.5884,
+      "step": 482
+    },
+    {
+      "epoch": 0.7505827505827506,
+      "grad_norm": 0.9644123682005706,
+      "learning_rate": 1.4687872402888991e-06,
+      "loss": 0.4757,
+      "step": 483
+    },
+    {
+      "epoch": 0.7521367521367521,
+      "grad_norm": 0.8646333036074157,
+      "learning_rate": 1.451534330444438e-06,
+      "loss": 0.4525,
+      "step": 484
+    },
+    {
+      "epoch": 0.7536907536907537,
+      "grad_norm": 0.7543556415051755,
+      "learning_rate": 1.4343661272571967e-06,
+      "loss": 0.453,
+      "step": 485
+    },
+    {
+      "epoch": 0.7552447552447552,
+      "grad_norm": 0.7262516356092494,
+      "learning_rate": 1.4172830405553216e-06,
+      "loss": 0.3922,
+      "step": 486
+    },
+    {
+      "epoch": 0.7567987567987567,
+      "grad_norm": 0.7925066399718954,
+      "learning_rate": 1.4002854781351104e-06,
+      "loss": 0.517,
+      "step": 487
+    },
+    {
+      "epoch": 0.7583527583527584,
+      "grad_norm": 0.7148060275837002,
+      "learning_rate": 1.3833738457512842e-06,
+      "loss": 0.3847,
+      "step": 488
+    },
+    {
+      "epoch": 0.7599067599067599,
+      "grad_norm": 0.6309982281142955,
+      "learning_rate": 1.3665485471072937e-06,
+      "loss": 0.4778,
+      "step": 489
+    },
+    {
+      "epoch": 0.7614607614607615,
+      "grad_norm": 0.7642090785940528,
+      "learning_rate": 1.3498099838456947e-06,
+      "loss": 0.4309,
+      "step": 490
+    },
+    {
+      "epoch": 0.763014763014763,
+      "grad_norm": 0.7640039314162583,
+      "learning_rate": 1.3331585555385458e-06,
+      "loss": 0.4696,
+      "step": 491
+    },
+    {
+      "epoch": 0.7645687645687645,
+      "grad_norm": 0.8643751125446875,
+      "learning_rate": 1.3165946596778773e-06,
+      "loss": 0.4918,
+      "step": 492
+    },
+    {
+      "epoch": 0.7661227661227661,
+      "grad_norm": 0.6804715835256383,
+      "learning_rate": 1.3001186916662066e-06,
+      "loss": 0.4724,
+      "step": 493
+    },
+    {
+      "epoch": 0.7676767676767676,
+      "grad_norm": 1.0793870282462463,
+      "learning_rate": 1.2837310448070929e-06,
+      "loss": 0.5909,
+      "step": 494
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 0.7827368042523198,
+      "learning_rate": 1.2674321102957476e-06,
+      "loss": 0.5138,
+      "step": 495
+    },
+    {
+      "epoch": 0.7707847707847708,
+      "grad_norm": 0.847544394369445,
+      "learning_rate": 1.251222277209702e-06,
+      "loss": 0.603,
+      "step": 496
+    },
+    {
+      "epoch": 0.7723387723387724,
+      "grad_norm": 0.9723527581718034,
+      "learning_rate": 1.2351019324995128e-06,
+      "loss": 0.6304,
+      "step": 497
+    },
+    {
+      "epoch": 0.7738927738927739,
+      "grad_norm": 0.6468908302032275,
+      "learning_rate": 1.2190714609795334e-06,
+      "loss": 0.521,
+      "step": 498
+    },
+    {
+      "epoch": 0.7754467754467754,
+      "grad_norm": 0.7114004263355455,
+      "learning_rate": 1.203131245318721e-06,
+      "loss": 0.4607,
+      "step": 499
+    },
+    {
+      "epoch": 0.777000777000777,
+      "grad_norm": 0.6224249787588289,
+      "learning_rate": 1.1872816660315029e-06,
+      "loss": 0.3998,
+      "step": 500
+    },
+    {
+      "epoch": 0.7785547785547785,
+      "grad_norm": 0.8157807813857092,
+      "learning_rate": 1.171523101468693e-06,
+      "loss": 0.5267,
+      "step": 501
+    },
+    {
+      "epoch": 0.7801087801087802,
+      "grad_norm": 0.799774654796459,
+      "learning_rate": 1.1558559278084647e-06,
+      "loss": 0.4909,
+      "step": 502
+    },
+    {
+      "epoch": 0.7816627816627817,
+      "grad_norm": 0.712975958103021,
+      "learning_rate": 1.1402805190473649e-06,
+      "loss": 0.4833,
+      "step": 503
+    },
+    {
+      "epoch": 0.7832167832167832,
+      "grad_norm": 0.7414921854529274,
+      "learning_rate": 1.124797246991387e-06,
+      "loss": 0.4083,
+      "step": 504
+    },
+    {
+      "epoch": 0.7847707847707848,
+      "grad_norm": 1.0834283945062337,
+      "learning_rate": 1.1094064812471028e-06,
+      "loss": 0.5327,
+      "step": 505
+    },
+    {
+      "epoch": 0.7863247863247863,
+      "grad_norm": 0.7386660114893862,
+      "learning_rate": 1.0941085892128272e-06,
+      "loss": 0.4922,
+      "step": 506
+    },
+    {
+      "epoch": 0.7878787878787878,
+      "grad_norm": 0.5879874025861683,
+      "learning_rate": 1.07890393606986e-06,
+      "loss": 0.3765,
+      "step": 507
+    },
+    {
+      "epoch": 0.7894327894327894,
+      "grad_norm": 0.8290075934295431,
+      "learning_rate": 1.0637928847737594e-06,
+      "loss": 0.5165,
+      "step": 508
+    },
+    {
+      "epoch": 0.790986790986791,
+      "grad_norm": 0.9222992948806642,
+      "learning_rate": 1.0487757960456812e-06,
+      "loss": 0.5239,
+      "step": 509
+    },
+    {
+      "epoch": 0.7925407925407926,
+      "grad_norm": 0.894510374178912,
+      "learning_rate": 1.0338530283637704e-06,
+      "loss": 0.5294,
+      "step": 510
+    },
+    {
+      "epoch": 0.7940947940947941,
+      "grad_norm": 0.9067683678931795,
+      "learning_rate": 1.0190249379546024e-06,
+      "loss": 0.6271,
+      "step": 511
+    },
+    {
+      "epoch": 0.7956487956487956,
+      "grad_norm": 0.8420348253099458,
+      "learning_rate": 1.0042918787846757e-06,
+      "loss": 0.4476,
+      "step": 512
+    },
+    {
+      "epoch": 0.7972027972027972,
+      "grad_norm": 2.2314012317895555,
+      "learning_rate": 9.896542025519645e-07,
+      "loss": 0.4875,
+      "step": 513
+    },
+    {
+      "epoch": 0.7987567987567987,
+      "grad_norm": 1.112553561220701,
+      "learning_rate": 9.751122586775253e-07,
+      "loss": 0.4668,
+      "step": 514
+    },
+    {
+      "epoch": 0.8003108003108003,
+      "grad_norm": 0.8631165830420672,
+      "learning_rate": 9.606663942971568e-07,
+      "loss": 0.4823,
+      "step": 515
+    },
+    {
+      "epoch": 0.8018648018648019,
+      "grad_norm": 0.8259570757220709,
+      "learning_rate": 9.463169542531059e-07,
+      "loss": 0.5339,
+      "step": 516
+    },
+    {
+      "epoch": 0.8034188034188035,
+      "grad_norm": 1.166843823225754,
+      "learning_rate": 9.320642810858421e-07,
+      "loss": 0.5133,
+      "step": 517
+    },
+    {
+      "epoch": 0.804972804972805,
+      "grad_norm": 0.9069458506573143,
+      "learning_rate": 9.179087150258814e-07,
+      "loss": 0.4574,
+      "step": 518
+    },
+    {
+      "epoch": 0.8065268065268065,
+      "grad_norm": 0.6083374762695839,
+      "learning_rate": 9.038505939856612e-07,
+      "loss": 0.5352,
+      "step": 519
+    },
+    {
+      "epoch": 0.8080808080808081,
+      "grad_norm": 0.7051533869734491,
+      "learning_rate": 8.898902535514747e-07,
+      "loss": 0.486,
+      "step": 520
+    },
+    {
+      "epoch": 0.8096348096348096,
+      "grad_norm": 1.1361715723258188,
+      "learning_rate": 8.76028026975459e-07,
+      "loss": 0.5395,
+      "step": 521
+    },
+    {
+      "epoch": 0.8111888111888111,
+      "grad_norm": 0.7251363387868152,
+      "learning_rate": 8.62264245167641e-07,
+      "loss": 0.4734,
+      "step": 522
+    },
+    {
+      "epoch": 0.8127428127428128,
+      "grad_norm": 0.7015732791695378,
+      "learning_rate": 8.485992366880419e-07,
+      "loss": 0.5281,
+      "step": 523
+    },
+    {
+      "epoch": 0.8142968142968143,
+      "grad_norm": 1.5424433933643296,
+      "learning_rate": 8.35033327738829e-07,
+      "loss": 0.4043,
+      "step": 524
+    },
+    {
+      "epoch": 0.8158508158508159,
+      "grad_norm": 0.9218463524060198,
+      "learning_rate": 8.215668421565276e-07,
+      "loss": 0.5614,
+      "step": 525
+    },
+    {
+      "epoch": 0.8174048174048174,
+      "grad_norm": 0.764325971788183,
+      "learning_rate": 8.082001014042945e-07,
+      "loss": 0.5032,
+      "step": 526
+    },
+    {
+      "epoch": 0.818958818958819,
+      "grad_norm": 0.652486590264396,
+      "learning_rate": 7.949334245642459e-07,
+      "loss": 0.4544,
+      "step": 527
+    },
+    {
+      "epoch": 0.8205128205128205,
+      "grad_norm": 0.7176940824971066,
+      "learning_rate": 7.817671283298345e-07,
+      "loss": 0.5098,
+      "step": 528
+    },
+    {
+      "epoch": 0.822066822066822,
+      "grad_norm": 1.823594644970129,
+      "learning_rate": 7.687015269982917e-07,
+      "loss": 0.4841,
+      "step": 529
+    },
+    {
+      "epoch": 0.8236208236208237,
+      "grad_norm": 0.7669592956201216,
+      "learning_rate": 7.557369324631303e-07,
+      "loss": 0.5751,
+      "step": 530
+    },
+    {
+      "epoch": 0.8251748251748252,
+      "grad_norm": 0.7791697796343787,
+      "learning_rate": 7.4287365420669e-07,
+      "loss": 0.5334,
+      "step": 531
+    },
+    {
+      "epoch": 0.8267288267288267,
+      "grad_norm": 0.7675193579164078,
+      "learning_rate": 7.301119992927585e-07,
+      "loss": 0.4749,
+      "step": 532
+    },
+    {
+      "epoch": 0.8282828282828283,
+      "grad_norm": 0.8274131519826066,
+      "learning_rate": 7.174522723592342e-07,
+      "loss": 0.5556,
+      "step": 533
+    },
+    {
+      "epoch": 0.8298368298368298,
+      "grad_norm": 0.6248155156332447,
+      "learning_rate": 7.048947756108576e-07,
+      "loss": 0.4449,
+      "step": 534
+    },
+    {
+      "epoch": 0.8313908313908314,
+      "grad_norm": 0.764818832164719,
+      "learning_rate": 6.924398088119988e-07,
+      "loss": 0.5908,
+      "step": 535
+    },
+    {
+      "epoch": 0.8329448329448329,
+      "grad_norm": 0.7925584720042309,
+      "learning_rate": 6.800876692794994e-07,
+      "loss": 0.5557,
+      "step": 536
+    },
+    {
+      "epoch": 0.8344988344988346,
+      "grad_norm": 0.8639156535546165,
+      "learning_rate": 6.678386518755747e-07,
+      "loss": 0.5577,
+      "step": 537
+    },
+    {
+      "epoch": 0.8360528360528361,
+      "grad_norm": 0.8390195478931005,
+      "learning_rate": 6.556930490007762e-07,
+      "loss": 0.5484,
+      "step": 538
+    },
+    {
+      "epoch": 0.8376068376068376,
+      "grad_norm": 0.63339201522811,
+      "learning_rate": 6.436511505870091e-07,
+      "loss": 0.5014,
+      "step": 539
+    },
+    {
+      "epoch": 0.8391608391608392,
+      "grad_norm": 0.889532359805588,
+      "learning_rate": 6.317132440906188e-07,
+      "loss": 0.5266,
+      "step": 540
+    },
+    {
+      "epoch": 0.8407148407148407,
+      "grad_norm": 0.5624920627332638,
+      "learning_rate": 6.198796144855168e-07,
+      "loss": 0.4393,
+      "step": 541
+    },
+    {
+      "epoch": 0.8422688422688422,
+      "grad_norm": 0.6994504769690357,
+      "learning_rate": 6.081505442563912e-07,
+      "loss": 0.4758,
+      "step": 542
+    },
+    {
+      "epoch": 0.8438228438228438,
+      "grad_norm": 0.677345489523822,
+      "learning_rate": 5.965263133919508e-07,
+      "loss": 0.4376,
+      "step": 543
+    },
+    {
+      "epoch": 0.8453768453768454,
+      "grad_norm": 1.1792412849625011,
+      "learning_rate": 5.850071993782525e-07,
+      "loss": 0.557,
+      "step": 544
+    },
+    {
+      "epoch": 0.846930846930847,
+      "grad_norm": 0.8383140877354333,
+      "learning_rate": 5.735934771920704e-07,
+      "loss": 0.4129,
+      "step": 545
+    },
+    {
+      "epoch": 0.8484848484848485,
+      "grad_norm": 0.7877126864302514,
+      "learning_rate": 5.622854192943317e-07,
+      "loss": 0.5008,
+      "step": 546
+    },
+    {
+      "epoch": 0.85003885003885,
+      "grad_norm": 0.9108037190809487,
+      "learning_rate": 5.510832956236173e-07,
+      "loss": 0.4783,
+      "step": 547
+    },
+    {
+      "epoch": 0.8515928515928516,
+      "grad_norm": 1.0094511865353648,
+      "learning_rate": 5.399873735897137e-07,
+      "loss": 0.4469,
+      "step": 548
+    },
+    {
+      "epoch": 0.8531468531468531,
+      "grad_norm": 0.663812075111233,
+      "learning_rate": 5.289979180672344e-07,
+      "loss": 0.6368,
+      "step": 549
+    },
+    {
+      "epoch": 0.8547008547008547,
+      "grad_norm": 0.7540552593487783,
+      "learning_rate": 5.181151913892896e-07,
+      "loss": 0.4343,
+      "step": 550
+    },
+    {
+      "epoch": 0.8562548562548562,
+      "grad_norm": 1.0037008402911451,
+      "learning_rate": 5.073394533412296e-07,
+      "loss": 0.5469,
+      "step": 551
+    },
+    {
+      "epoch": 0.8578088578088578,
+      "grad_norm": 0.5605077409329356,
+      "learning_rate": 4.966709611544435e-07,
+      "loss": 0.4462,
+      "step": 552
+    },
+    {
+      "epoch": 0.8593628593628594,
+      "grad_norm": 0.8732352825291733,
+      "learning_rate": 4.861099695002158e-07,
+      "loss": 0.5251,
+      "step": 553
+    },
+    {
+      "epoch": 0.8609168609168609,
+      "grad_norm": 0.7053119794249835,
+      "learning_rate": 4.7565673048364735e-07,
+      "loss": 0.4842,
+      "step": 554
+    },
+    {
+      "epoch": 0.8624708624708625,
+      "grad_norm": 0.821971296932637,
+      "learning_rate": 4.6531149363764126e-07,
+      "loss": 0.4487,
+      "step": 555
+    },
+    {
+      "epoch": 0.864024864024864,
+      "grad_norm": 1.2794248867833895,
+      "learning_rate": 4.550745059169398e-07,
+      "loss": 0.5328,
+      "step": 556
+    },
+    {
+      "epoch": 0.8655788655788655,
+      "grad_norm": 0.7489672108536733,
+      "learning_rate": 4.4494601169223715e-07,
+      "loss": 0.5069,
+      "step": 557
+    },
+    {
+      "epoch": 0.8671328671328671,
+      "grad_norm": 0.6659188193571587,
+      "learning_rate": 4.349262527443371e-07,
+      "loss": 0.5734,
+      "step": 558
+    },
+    {
+      "epoch": 0.8686868686868687,
+      "grad_norm": 0.7454236276103242,
+      "learning_rate": 4.2501546825838735e-07,
+      "loss": 0.358,
+      "step": 559
+    },
+    {
+      "epoch": 0.8702408702408703,
+      "grad_norm": 0.9570678482900348,
+      "learning_rate": 4.152138948181689e-07,
+      "loss": 0.538,
+      "step": 560
+    },
+    {
+      "epoch": 0.8717948717948718,
+      "grad_norm": 0.9593262810731289,
+      "learning_rate": 4.0552176640045017e-07,
+      "loss": 0.6003,
+      "step": 561
+    },
+    {
+      "epoch": 0.8733488733488733,
+      "grad_norm": 0.6814712625667729,
+      "learning_rate": 3.959393143693946e-07,
+      "loss": 0.4296,
+      "step": 562
+    },
+    {
+      "epoch": 0.8749028749028749,
+      "grad_norm": 0.8468228674134194,
+      "learning_rate": 3.864667674710454e-07,
+      "loss": 0.478,
+      "step": 563
+    },
+    {
+      "epoch": 0.8764568764568764,
+      "grad_norm": 0.8214210844483506,
+      "learning_rate": 3.7710435182786053e-07,
+      "loss": 0.4921,
+      "step": 564
+    },
+    {
+      "epoch": 0.878010878010878,
+      "grad_norm": 0.7775961282959882,
+      "learning_rate": 3.6785229093331987e-07,
+      "loss": 0.5158,
+      "step": 565
+    },
+    {
+      "epoch": 0.8795648795648796,
+      "grad_norm": 0.7706689173433338,
+      "learning_rate": 3.587108056465827e-07,
+      "loss": 0.4987,
+      "step": 566
+    },
+    {
+      "epoch": 0.8811188811188811,
+      "grad_norm": 0.7115920841531194,
+      "learning_rate": 3.496801141872225e-07,
+      "loss": 0.5508,
+      "step": 567
+    },
+    {
+      "epoch": 0.8826728826728827,
+      "grad_norm": 0.7360248498356703,
+      "learning_rate": 3.407604321300123e-07,
+      "loss": 0.4622,
+      "step": 568
+    },
+    {
+      "epoch": 0.8842268842268842,
+      "grad_norm": 0.6656952275189396,
+      "learning_rate": 3.3195197239978384e-07,
+      "loss": 0.4616,
+      "step": 569
+    },
+    {
+      "epoch": 0.8857808857808858,
+      "grad_norm": 1.0438892609670738,
+      "learning_rate": 3.232549452663403e-07,
+      "loss": 0.6725,
+      "step": 570
+    },
+    {
+      "epoch": 0.8873348873348873,
+      "grad_norm": 0.9087915959905298,
+      "learning_rate": 3.146695583394377e-07,
+      "loss": 0.5397,
+      "step": 571
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 0.7459812770941501,
+      "learning_rate": 3.061960165638317e-07,
+      "loss": 0.4763,
+      "step": 572
+    },
+    {
+      "epoch": 0.8904428904428905,
+      "grad_norm": 0.9368206700984494,
+      "learning_rate": 2.9783452221438304e-07,
+      "loss": 0.4894,
+      "step": 573
+    },
+    {
+      "epoch": 0.891996891996892,
+      "grad_norm": 0.7581701388162561,
+      "learning_rate": 2.895852748912298e-07,
+      "loss": 0.4285,
+      "step": 574
+    },
+    {
+      "epoch": 0.8935508935508936,
+      "grad_norm": 0.7295548456707804,
+      "learning_rate": 2.814484715150212e-07,
+      "loss": 0.4201,
+      "step": 575
+    },
+    {
+      "epoch": 0.8951048951048951,
+      "grad_norm": 0.6762252893935657,
+      "learning_rate": 2.734243063222181e-07,
+      "loss": 0.4842,
+      "step": 576
+    },
+    {
+      "epoch": 0.8966588966588966,
+      "grad_norm": 0.7273733934803849,
+      "learning_rate": 2.655129708604576e-07,
+      "loss": 0.3651,
+      "step": 577
+    },
+    {
+      "epoch": 0.8982128982128982,
+      "grad_norm": 0.9666761581157529,
+      "learning_rate": 2.5771465398397757e-07,
+      "loss": 0.7778,
+      "step": 578
+    },
+    {
+      "epoch": 0.8997668997668997,
+      "grad_norm": 0.6776269970027221,
+      "learning_rate": 2.5002954184910887e-07,
+      "loss": 0.5259,
+      "step": 579
+    },
+    {
+      "epoch": 0.9013209013209014,
+      "grad_norm": 0.5934838544460401,
+      "learning_rate": 2.424578179098358e-07,
+      "loss": 0.5417,
+      "step": 580
+    },
+    {
+      "epoch": 0.9028749028749029,
+      "grad_norm": 0.8173118127489211,
+      "learning_rate": 2.3499966291341213e-07,
+      "loss": 0.4431,
+      "step": 581
+    },
+    {
+      "epoch": 0.9044289044289044,
+      "grad_norm": 0.6926431286616653,
+      "learning_rate": 2.2765525489604702e-07,
+      "loss": 0.3946,
+      "step": 582
+    },
+    {
+      "epoch": 0.905982905982906,
+      "grad_norm": 0.8058462854601025,
+      "learning_rate": 2.2042476917865706e-07,
+      "loss": 0.4716,
+      "step": 583
+    },
+    {
+      "epoch": 0.9075369075369075,
+      "grad_norm": 0.735345958366895,
+      "learning_rate": 2.1330837836267882e-07,
+      "loss": 0.4664,
+      "step": 584
+    },
+    {
+      "epoch": 0.9090909090909091,
+      "grad_norm": 0.6326682057182371,
+      "learning_rate": 2.0630625232595126e-07,
+      "loss": 0.395,
+      "step": 585
+    },
+    {
+      "epoch": 0.9106449106449106,
+      "grad_norm": 0.7934491290140667,
+      "learning_rate": 1.9941855821865918e-07,
+      "loss": 0.4923,
+      "step": 586
+    },
+    {
+      "epoch": 0.9121989121989122,
+      "grad_norm": 0.7620750821145263,
+      "learning_rate": 1.9264546045934196e-07,
+      "loss": 0.4928,
+      "step": 587
+    },
+    {
+      "epoch": 0.9137529137529138,
+      "grad_norm": 0.6744132607863633,
+      "learning_rate": 1.859871207309688e-07,
+      "loss": 0.544,
+      "step": 588
+    },
+    {
+      "epoch": 0.9153069153069153,
+      "grad_norm": 0.850939097017279,
+      "learning_rate": 1.7944369797708362e-07,
+      "loss": 0.4553,
+      "step": 589
+    },
+    {
+      "epoch": 0.9168609168609169,
+      "grad_norm": 0.6424420190959574,
+      "learning_rate": 1.7301534839800348e-07,
+      "loss": 0.3915,
+      "step": 590
+    },
+    {
+      "epoch": 0.9184149184149184,
+      "grad_norm": 0.7046012463166428,
+      "learning_rate": 1.6670222544709515e-07,
+      "loss": 0.4708,
+      "step": 591
+    },
+    {
+      "epoch": 0.9199689199689199,
+      "grad_norm": 1.0364283887763577,
+      "learning_rate": 1.6050447982711214e-07,
+      "loss": 0.4111,
+      "step": 592
+    },
+    {
+      "epoch": 0.9215229215229215,
+      "grad_norm": 0.7049190344022591,
+      "learning_rate": 1.5442225948659183e-07,
+      "loss": 0.5696,
+      "step": 593
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.8155103152236225,
+      "learning_rate": 1.4845570961633192e-07,
+      "loss": 0.5059,
+      "step": 594
+    },
+    {
+      "epoch": 0.9246309246309247,
+      "grad_norm": 0.5949094353172898,
+      "learning_rate": 1.426049726459172e-07,
+      "loss": 0.4638,
+      "step": 595
+    },
+    {
+      "epoch": 0.9261849261849262,
+      "grad_norm": 0.6815208521528887,
+      "learning_rate": 1.368701882403234e-07,
+      "loss": 0.4551,
+      "step": 596
+    },
+    {
+      "epoch": 0.9277389277389277,
+      "grad_norm": 0.9296974247938309,
+      "learning_rate": 1.3125149329658083e-07,
+      "loss": 0.4026,
+      "step": 597
+    },
+    {
+      "epoch": 0.9292929292929293,
+      "grad_norm": 1.2038986087224606,
+      "learning_rate": 1.2574902194050996e-07,
+      "loss": 0.4656,
+      "step": 598
+    },
+    {
+      "epoch": 0.9308469308469308,
+      "grad_norm": 1.0581872023301784,
+      "learning_rate": 1.2036290552351838e-07,
+      "loss": 0.6128,
+      "step": 599
+    },
+    {
+      "epoch": 0.9324009324009324,
+      "grad_norm": 0.9158733284924282,
+      "learning_rate": 1.150932726194609e-07,
+      "loss": 0.4848,
+      "step": 600
+    },
+    {
+      "epoch": 0.933954933954934,
+      "grad_norm": 0.7543888048930436,
+      "learning_rate": 1.0994024902157674e-07,
+      "loss": 0.4701,
+      "step": 601
+    },
+    {
+      "epoch": 0.9355089355089355,
+      "grad_norm": 0.7081866182048879,
+      "learning_rate": 1.0490395773948336e-07,
+      "loss": 0.5253,
+      "step": 602
+    },
+    {
+      "epoch": 0.9370629370629371,
+      "grad_norm": 0.6949901814621636,
+      "learning_rate": 9.998451899624007e-08,
+      "loss": 0.4287,
+      "step": 603
+    },
+    {
+      "epoch": 0.9386169386169386,
+      "grad_norm": 0.8441421847918561,
+      "learning_rate": 9.5182050225478e-08,
+      "loss": 0.5441,
+      "step": 604
+    },
+    {
+      "epoch": 0.9401709401709402,
+      "grad_norm": 0.7034318164083617,
+      "learning_rate": 9.049666606859852e-08,
+      "loss": 0.4834,
+      "step": 605
+    },
+    {
+      "epoch": 0.9417249417249417,
+      "grad_norm": 0.8895517754609739,
+      "learning_rate": 8.592847837203655e-08,
+      "loss": 0.4564,
+      "step": 606
+    },
+    {
+      "epoch": 0.9432789432789432,
+      "grad_norm": 0.7017208181125264,
+      "learning_rate": 8.147759618458706e-08,
+      "loss": 0.4238,
+      "step": 607
+    },
+    {
+      "epoch": 0.9448329448329449,
+      "grad_norm": 0.7595723199577458,
+      "learning_rate": 7.714412575480556e-08,
+      "loss": 0.541,
+      "step": 608
+    },
+    {
+      "epoch": 0.9463869463869464,
+      "grad_norm": 0.701349856315337,
+      "learning_rate": 7.292817052847068e-08,
+      "loss": 0.5784,
+      "step": 609
+    },
+    {
+      "epoch": 0.947940947940948,
+      "grad_norm": 0.6902057779636126,
+      "learning_rate": 6.882983114611497e-08,
+      "loss": 0.5877,
+      "step": 610
+    },
+    {
+      "epoch": 0.9494949494949495,
+      "grad_norm": 0.7053000221431716,
+      "learning_rate": 6.484920544062245e-08,
+      "loss": 0.3299,
+      "step": 611
+    },
+    {
+      "epoch": 0.951048951048951,
+      "grad_norm": 0.7668065733213358,
+      "learning_rate": 6.098638843489213e-08,
+      "loss": 0.6096,
+      "step": 612
+    },
+    {
+      "epoch": 0.9526029526029526,
+      "grad_norm": 0.9203339601219532,
+      "learning_rate": 5.7241472339572e-08,
+      "loss": 0.6287,
+      "step": 613
+    },
+    {
+      "epoch": 0.9541569541569541,
+      "grad_norm": 0.7745409282534764,
+      "learning_rate": 5.361454655085529e-08,
+      "loss": 0.4677,
+      "step": 614
+    },
+    {
+      "epoch": 0.9557109557109557,
+      "grad_norm": 0.7385280228665362,
+      "learning_rate": 5.0105697648347716e-08,
+      "loss": 0.5869,
+      "step": 615
+    },
+    {
+      "epoch": 0.9572649572649573,
+      "grad_norm": 0.7825960999337372,
+      "learning_rate": 4.671500939300133e-08,
+      "loss": 0.498,
+      "step": 616
+    },
+    {
+      "epoch": 0.9588189588189588,
+      "grad_norm": 0.753511461886924,
+      "learning_rate": 4.344256272511338e-08,
+      "loss": 0.4964,
+      "step": 617
+    },
+    {
+      "epoch": 0.9603729603729604,
+      "grad_norm": 0.6840524141170308,
+      "learning_rate": 4.0288435762396164e-08,
+      "loss": 0.4294,
+      "step": 618
+    },
+    {
+      "epoch": 0.9619269619269619,
+      "grad_norm": 1.035162046106296,
+      "learning_rate": 3.725270379811019e-08,
+      "loss": 0.5584,
+      "step": 619
+    },
+    {
+      "epoch": 0.9634809634809635,
+      "grad_norm": 0.9156660131303417,
+      "learning_rate": 3.4335439299268414e-08,
+      "loss": 0.4166,
+      "step": 620
+    },
+    {
+      "epoch": 0.965034965034965,
+      "grad_norm": 0.7719462038166263,
+      "learning_rate": 3.1536711904904816e-08,
+      "loss": 0.4776,
+      "step": 621
+    },
+    {
+      "epoch": 0.9665889665889665,
+      "grad_norm": 0.6723067728580475,
+      "learning_rate": 2.8856588424414632e-08,
+      "loss": 0.4692,
+      "step": 622
+    },
+    {
+      "epoch": 0.9681429681429682,
+      "grad_norm": 0.8273069565821325,
+      "learning_rate": 2.6295132835956748e-08,
+      "loss": 0.4465,
+      "step": 623
+    },
+    {
+      "epoch": 0.9696969696969697,
+      "grad_norm": 0.7689357247441089,
+      "learning_rate": 2.3852406284927687e-08,
+      "loss": 0.4544,
+      "step": 624
+    },
+    {
+      "epoch": 0.9712509712509713,
+      "grad_norm": 0.7619653211518902,
+      "learning_rate": 2.152846708250167e-08,
+      "loss": 0.5586,
+      "step": 625
+    },
+    {
+      "epoch": 0.9728049728049728,
+      "grad_norm": 1.5148882415320697,
+      "learning_rate": 1.93233707042384e-08,
+      "loss": 0.4416,
+      "step": 626
+    },
+    {
+      "epoch": 0.9743589743589743,
+      "grad_norm": 0.723885981000404,
+      "learning_rate": 1.723716978876133e-08,
+      "loss": 0.366,
+      "step": 627
+    },
+    {
+      "epoch": 0.9759129759129759,
+      "grad_norm": 0.9896178053638355,
+      "learning_rate": 1.5269914136497033e-08,
+      "loss": 0.6383,
+      "step": 628
+    },
+    {
+      "epoch": 0.9774669774669774,
+      "grad_norm": 0.9450685315761042,
+      "learning_rate": 1.3421650708487777e-08,
+      "loss": 0.4261,
+      "step": 629
+    },
+    {
+      "epoch": 0.9790209790209791,
+      "grad_norm": 0.7776928320411156,
+      "learning_rate": 1.1692423625273563e-08,
+      "loss": 0.3718,
+      "step": 630
+    },
+    {
+      "epoch": 0.9805749805749806,
+      "grad_norm": 2.709950931518374,
+      "learning_rate": 1.0082274165834627e-08,
+      "loss": 0.4794,
+      "step": 631
+    },
+    {
+      "epoch": 0.9821289821289821,
+      "grad_norm": 0.812744833469983,
+      "learning_rate": 8.591240766607779e-09,
+      "loss": 0.5243,
+      "step": 632
+    },
+    {
+      "epoch": 0.9836829836829837,
+      "grad_norm": 0.7580063402130408,
+      "learning_rate": 7.219359020570471e-09,
+      "loss": 0.4821,
+      "step": 633
+    },
+    {
+      "epoch": 0.9852369852369852,
+      "grad_norm": 1.012231114852923,
+      "learning_rate": 5.966661676388152e-09,
+      "loss": 0.4589,
+      "step": 634
+    },
+    {
+      "epoch": 0.9867909867909868,
+      "grad_norm": 0.7714041439030666,
+      "learning_rate": 4.833178637633773e-09,
+      "loss": 0.5265,
+      "step": 635
+    },
+    {
+      "epoch": 0.9883449883449883,
+      "grad_norm": 0.7785451239864247,
+      "learning_rate": 3.818936962076136e-09,
+      "loss": 0.5888,
+      "step": 636
+    },
+    {
+      "epoch": 0.98989898989899,
+      "grad_norm": 0.8911993119168611,
+      "learning_rate": 2.9239608610298618e-09,
+      "loss": 0.415,
+      "step": 637
+    },
+    {
+      "epoch": 0.9914529914529915,
+      "grad_norm": 0.9270156128048929,
+      "learning_rate": 2.148271698781401e-09,
+      "loss": 0.4057,
+      "step": 638
+    },
+    {
+      "epoch": 0.993006993006993,
+      "grad_norm": 0.7334067869030746,
+      "learning_rate": 1.4918879920750029e-09,
+      "loss": 0.4826,
+      "step": 639
+    },
+    {
+      "epoch": 0.9945609945609946,
+      "grad_norm": 0.7633076997899867,
+      "learning_rate": 9.548254096752862e-10,
+      "loss": 0.6209,
+      "step": 640
+    },
+    {
+      "epoch": 0.9961149961149961,
+      "grad_norm": 1.0315005661760475,
+      "learning_rate": 5.370967719897646e-10,
+      "loss": 0.4908,
+      "step": 641
+    },
+    {
+      "epoch": 0.9976689976689976,
+      "grad_norm": 0.8907517067968038,
+      "learning_rate": 2.387120507629792e-10,
+      "loss": 0.467,
+      "step": 642
+    },
+    {
+      "epoch": 0.9992229992229992,
+      "grad_norm": 0.9387605136360067,
+      "learning_rate": 5.967836884168687e-11,
+      "loss": 0.5826,
+      "step": 643
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 643,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "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": 70581768388608.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}