diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6517 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.993252361673415,
+  "eval_steps": 500,
+  "global_step": 925,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.005398110661268556,
+      "grad_norm": 6.807432191601978,
+      "learning_rate": 4.301075268817205e-07,
+      "loss": 0.9574,
+      "step": 1
+    },
+    {
+      "epoch": 0.010796221322537112,
+      "grad_norm": 7.154196232139333,
+      "learning_rate": 8.60215053763441e-07,
+      "loss": 1.0004,
+      "step": 2
+    },
+    {
+      "epoch": 0.016194331983805668,
+      "grad_norm": 6.8534903536674365,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 1.3255,
+      "step": 3
+    },
+    {
+      "epoch": 0.021592442645074223,
+      "grad_norm": 8.487770394838087,
+      "learning_rate": 1.720430107526882e-06,
+      "loss": 1.0104,
+      "step": 4
+    },
+    {
+      "epoch": 0.02699055330634278,
+      "grad_norm": 6.270336418965172,
+      "learning_rate": 2.1505376344086023e-06,
+      "loss": 0.9156,
+      "step": 5
+    },
+    {
+      "epoch": 0.032388663967611336,
+      "grad_norm": 4.950501420555945,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 0.9515,
+      "step": 6
+    },
+    {
+      "epoch": 0.037786774628879895,
+      "grad_norm": 4.854264899680687,
+      "learning_rate": 3.0107526881720433e-06,
+      "loss": 1.1282,
+      "step": 7
+    },
+    {
+      "epoch": 0.043184885290148446,
+      "grad_norm": 3.939770315104577,
+      "learning_rate": 3.440860215053764e-06,
+      "loss": 0.8315,
+      "step": 8
+    },
+    {
+      "epoch": 0.048582995951417005,
+      "grad_norm": 4.010235086280238,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 0.8276,
+      "step": 9
+    },
+    {
+      "epoch": 0.05398110661268556,
+      "grad_norm": 2.6452813194614913,
+      "learning_rate": 4.3010752688172045e-06,
+      "loss": 0.8049,
+      "step": 10
+    },
+    {
+      "epoch": 0.059379217273954114,
+      "grad_norm": 2.1256473577976744,
+      "learning_rate": 4.731182795698925e-06,
+      "loss": 0.7715,
+      "step": 11
+    },
+    {
+      "epoch": 0.06477732793522267,
+      "grad_norm": 1.9385445088532463,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 0.7011,
+      "step": 12
+    },
+    {
+      "epoch": 0.07017543859649122,
+      "grad_norm": 1.840269705646139,
+      "learning_rate": 5.591397849462365e-06,
+      "loss": 0.6846,
+      "step": 13
+    },
+    {
+      "epoch": 0.07557354925775979,
+      "grad_norm": 2.197108380669797,
+      "learning_rate": 6.021505376344087e-06,
+      "loss": 0.8333,
+      "step": 14
+    },
+    {
+      "epoch": 0.08097165991902834,
+      "grad_norm": 2.7001092221365814,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.739,
+      "step": 15
+    },
+    {
+      "epoch": 0.08636977058029689,
+      "grad_norm": 2.1323750470470237,
+      "learning_rate": 6.881720430107528e-06,
+      "loss": 0.7759,
+      "step": 16
+    },
+    {
+      "epoch": 0.09176788124156546,
+      "grad_norm": 1.7491411378335646,
+      "learning_rate": 7.311827956989248e-06,
+      "loss": 0.6942,
+      "step": 17
+    },
+    {
+      "epoch": 0.09716599190283401,
+      "grad_norm": 1.3975365903682948,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 0.7317,
+      "step": 18
+    },
+    {
+      "epoch": 0.10256410256410256,
+      "grad_norm": 1.6319090485240784,
+      "learning_rate": 8.172043010752689e-06,
+      "loss": 0.7492,
+      "step": 19
+    },
+    {
+      "epoch": 0.10796221322537113,
+      "grad_norm": 1.3853317061054715,
+      "learning_rate": 8.602150537634409e-06,
+      "loss": 0.6592,
+      "step": 20
+    },
+    {
+      "epoch": 0.11336032388663968,
+      "grad_norm": 1.4855270503881244,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 0.6964,
+      "step": 21
+    },
+    {
+      "epoch": 0.11875843454790823,
+      "grad_norm": 1.5172910137865936,
+      "learning_rate": 9.46236559139785e-06,
+      "loss": 0.6902,
+      "step": 22
+    },
+    {
+      "epoch": 0.1241565452091768,
+      "grad_norm": 1.5930353043485512,
+      "learning_rate": 9.89247311827957e-06,
+      "loss": 0.7194,
+      "step": 23
+    },
+    {
+      "epoch": 0.12955465587044535,
+      "grad_norm": 1.5067818069552623,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 0.7384,
+      "step": 24
+    },
+    {
+      "epoch": 0.1349527665317139,
+      "grad_norm": 1.2650800025908264,
+      "learning_rate": 1.0752688172043012e-05,
+      "loss": 0.7906,
+      "step": 25
+    },
+    {
+      "epoch": 0.14035087719298245,
+      "grad_norm": 1.359410295321697,
+      "learning_rate": 1.118279569892473e-05,
+      "loss": 0.7055,
+      "step": 26
+    },
+    {
+      "epoch": 0.145748987854251,
+      "grad_norm": 1.4212635924731631,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 0.6539,
+      "step": 27
+    },
+    {
+      "epoch": 0.15114709851551958,
+      "grad_norm": 1.3794060569987638,
+      "learning_rate": 1.2043010752688173e-05,
+      "loss": 0.6785,
+      "step": 28
+    },
+    {
+      "epoch": 0.15654520917678813,
+      "grad_norm": 1.3043065755711913,
+      "learning_rate": 1.2473118279569894e-05,
+      "loss": 0.673,
+      "step": 29
+    },
+    {
+      "epoch": 0.16194331983805668,
+      "grad_norm": 1.1956862581839012,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.6981,
+      "step": 30
+    },
+    {
+      "epoch": 0.16734143049932523,
+      "grad_norm": 1.156945332555027,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.653,
+      "step": 31
+    },
+    {
+      "epoch": 0.17273954116059378,
+      "grad_norm": 1.2677289672684375,
+      "learning_rate": 1.3763440860215056e-05,
+      "loss": 0.9307,
+      "step": 32
+    },
+    {
+      "epoch": 0.17813765182186234,
+      "grad_norm": 1.2958447679390157,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 0.6477,
+      "step": 33
+    },
+    {
+      "epoch": 0.18353576248313092,
+      "grad_norm": 1.4198092756978424,
+      "learning_rate": 1.4623655913978497e-05,
+      "loss": 0.7098,
+      "step": 34
+    },
+    {
+      "epoch": 0.18893387314439947,
+      "grad_norm": 1.1771747136833552,
+      "learning_rate": 1.5053763440860215e-05,
+      "loss": 0.64,
+      "step": 35
+    },
+    {
+      "epoch": 0.19433198380566802,
+      "grad_norm": 1.1219114629409634,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 0.6617,
+      "step": 36
+    },
+    {
+      "epoch": 0.19973009446693657,
+      "grad_norm": 1.1631216376404652,
+      "learning_rate": 1.5913978494623657e-05,
+      "loss": 0.7625,
+      "step": 37
+    },
+    {
+      "epoch": 0.20512820512820512,
+      "grad_norm": 1.1711859490984557,
+      "learning_rate": 1.6344086021505377e-05,
+      "loss": 0.7097,
+      "step": 38
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 1.0115891973598403,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 0.5863,
+      "step": 39
+    },
+    {
+      "epoch": 0.21592442645074225,
+      "grad_norm": 1.1861663211200362,
+      "learning_rate": 1.7204301075268818e-05,
+      "loss": 0.6906,
+      "step": 40
+    },
+    {
+      "epoch": 0.2213225371120108,
+      "grad_norm": 1.1564670498049654,
+      "learning_rate": 1.763440860215054e-05,
+      "loss": 0.6751,
+      "step": 41
+    },
+    {
+      "epoch": 0.22672064777327935,
+      "grad_norm": 1.1584918051767952,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 0.6512,
+      "step": 42
+    },
+    {
+      "epoch": 0.2321187584345479,
+      "grad_norm": 1.1679139274141883,
+      "learning_rate": 1.849462365591398e-05,
+      "loss": 0.67,
+      "step": 43
+    },
+    {
+      "epoch": 0.23751686909581646,
+      "grad_norm": 1.1276690612438414,
+      "learning_rate": 1.89247311827957e-05,
+      "loss": 0.6484,
+      "step": 44
+    },
+    {
+      "epoch": 0.242914979757085,
+      "grad_norm": 1.1173417597702537,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.6416,
+      "step": 45
+    },
+    {
+      "epoch": 0.2483130904183536,
+      "grad_norm": 1.1571204225768013,
+      "learning_rate": 1.978494623655914e-05,
+      "loss": 0.6386,
+      "step": 46
+    },
+    {
+      "epoch": 0.25371120107962214,
+      "grad_norm": 1.2572680699570142,
+      "learning_rate": 2.0215053763440862e-05,
+      "loss": 0.7161,
+      "step": 47
+    },
+    {
+      "epoch": 0.2591093117408907,
+      "grad_norm": 1.0515938088008163,
+      "learning_rate": 2.0645161290322582e-05,
+      "loss": 0.6631,
+      "step": 48
+    },
+    {
+      "epoch": 0.26450742240215924,
+      "grad_norm": 1.0935333846184119,
+      "learning_rate": 2.1075268817204303e-05,
+      "loss": 0.6592,
+      "step": 49
+    },
+    {
+      "epoch": 0.2699055330634278,
+      "grad_norm": 1.151245960586739,
+      "learning_rate": 2.1505376344086024e-05,
+      "loss": 0.6274,
+      "step": 50
+    },
+    {
+      "epoch": 0.27530364372469635,
+      "grad_norm": 1.0058062006554418,
+      "learning_rate": 2.193548387096774e-05,
+      "loss": 0.6653,
+      "step": 51
+    },
+    {
+      "epoch": 0.2807017543859649,
+      "grad_norm": 1.0467393720802025,
+      "learning_rate": 2.236559139784946e-05,
+      "loss": 0.6407,
+      "step": 52
+    },
+    {
+      "epoch": 0.28609986504723345,
+      "grad_norm": 1.087001687450635,
+      "learning_rate": 2.2795698924731185e-05,
+      "loss": 0.6526,
+      "step": 53
+    },
+    {
+      "epoch": 0.291497975708502,
+      "grad_norm": 1.0579130252667992,
+      "learning_rate": 2.3225806451612906e-05,
+      "loss": 0.6579,
+      "step": 54
+    },
+    {
+      "epoch": 0.2968960863697706,
+      "grad_norm": 1.036531765831115,
+      "learning_rate": 2.3655913978494626e-05,
+      "loss": 0.6274,
+      "step": 55
+    },
+    {
+      "epoch": 0.30229419703103916,
+      "grad_norm": 1.0632744213126575,
+      "learning_rate": 2.4086021505376347e-05,
+      "loss": 0.8142,
+      "step": 56
+    },
+    {
+      "epoch": 0.3076923076923077,
+      "grad_norm": 1.0540844469974375,
+      "learning_rate": 2.4516129032258067e-05,
+      "loss": 0.6416,
+      "step": 57
+    },
+    {
+      "epoch": 0.31309041835357626,
+      "grad_norm": 1.0955841227061578,
+      "learning_rate": 2.4946236559139788e-05,
+      "loss": 0.6253,
+      "step": 58
+    },
+    {
+      "epoch": 0.3184885290148448,
+      "grad_norm": 1.1064443188106945,
+      "learning_rate": 2.537634408602151e-05,
+      "loss": 0.6696,
+      "step": 59
+    },
+    {
+      "epoch": 0.32388663967611336,
+      "grad_norm": 1.1588742913133638,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.6937,
+      "step": 60
+    },
+    {
+      "epoch": 0.3292847503373819,
+      "grad_norm": 1.0990955649880314,
+      "learning_rate": 2.6236559139784946e-05,
+      "loss": 0.7085,
+      "step": 61
+    },
+    {
+      "epoch": 0.33468286099865047,
+      "grad_norm": 1.084316800394684,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.6714,
+      "step": 62
+    },
+    {
+      "epoch": 0.340080971659919,
+      "grad_norm": 0.9906103041168199,
+      "learning_rate": 2.7096774193548387e-05,
+      "loss": 0.6972,
+      "step": 63
+    },
+    {
+      "epoch": 0.34547908232118757,
+      "grad_norm": 1.054655977322489,
+      "learning_rate": 2.752688172043011e-05,
+      "loss": 0.6927,
+      "step": 64
+    },
+    {
+      "epoch": 0.3508771929824561,
+      "grad_norm": 1.0749901143869052,
+      "learning_rate": 2.795698924731183e-05,
+      "loss": 0.683,
+      "step": 65
+    },
+    {
+      "epoch": 0.3562753036437247,
+      "grad_norm": 0.9933924944456113,
+      "learning_rate": 2.8387096774193552e-05,
+      "loss": 0.6669,
+      "step": 66
+    },
+    {
+      "epoch": 0.3616734143049933,
+      "grad_norm": 1.1374688961251551,
+      "learning_rate": 2.8817204301075273e-05,
+      "loss": 0.6998,
+      "step": 67
+    },
+    {
+      "epoch": 0.36707152496626183,
+      "grad_norm": 1.0407303668897125,
+      "learning_rate": 2.9247311827956993e-05,
+      "loss": 0.6497,
+      "step": 68
+    },
+    {
+      "epoch": 0.3724696356275304,
+      "grad_norm": 1.103124863702835,
+      "learning_rate": 2.9677419354838714e-05,
+      "loss": 0.8667,
+      "step": 69
+    },
+    {
+      "epoch": 0.37786774628879893,
+      "grad_norm": 1.0102782599067142,
+      "learning_rate": 3.010752688172043e-05,
+      "loss": 0.6354,
+      "step": 70
+    },
+    {
+      "epoch": 0.3832658569500675,
+      "grad_norm": 1.0913980163625263,
+      "learning_rate": 3.053763440860215e-05,
+      "loss": 0.6616,
+      "step": 71
+    },
+    {
+      "epoch": 0.38866396761133604,
+      "grad_norm": 0.9583647530026241,
+      "learning_rate": 3.096774193548387e-05,
+      "loss": 0.6186,
+      "step": 72
+    },
+    {
+      "epoch": 0.3940620782726046,
+      "grad_norm": 1.052465884877506,
+      "learning_rate": 3.139784946236559e-05,
+      "loss": 0.6599,
+      "step": 73
+    },
+    {
+      "epoch": 0.39946018893387314,
+      "grad_norm": 1.0515134291579666,
+      "learning_rate": 3.182795698924731e-05,
+      "loss": 0.7066,
+      "step": 74
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 1.0419495962684775,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.6239,
+      "step": 75
+    },
+    {
+      "epoch": 0.41025641025641024,
+      "grad_norm": 1.0890766504906937,
+      "learning_rate": 3.2688172043010754e-05,
+      "loss": 0.7229,
+      "step": 76
+    },
+    {
+      "epoch": 0.4156545209176788,
+      "grad_norm": 1.0490537282957513,
+      "learning_rate": 3.3118279569892475e-05,
+      "loss": 0.7329,
+      "step": 77
+    },
+    {
+      "epoch": 0.42105263157894735,
+      "grad_norm": 1.1084001513261095,
+      "learning_rate": 3.3548387096774195e-05,
+      "loss": 0.7218,
+      "step": 78
+    },
+    {
+      "epoch": 0.42645074224021595,
+      "grad_norm": 0.9654691704116004,
+      "learning_rate": 3.3978494623655916e-05,
+      "loss": 0.6217,
+      "step": 79
+    },
+    {
+      "epoch": 0.4318488529014845,
+      "grad_norm": 1.0309859110617534,
+      "learning_rate": 3.4408602150537636e-05,
+      "loss": 0.7064,
+      "step": 80
+    },
+    {
+      "epoch": 0.43724696356275305,
+      "grad_norm": 0.9536382238095081,
+      "learning_rate": 3.483870967741936e-05,
+      "loss": 0.661,
+      "step": 81
+    },
+    {
+      "epoch": 0.4426450742240216,
+      "grad_norm": 1.0523169841990292,
+      "learning_rate": 3.526881720430108e-05,
+      "loss": 0.6921,
+      "step": 82
+    },
+    {
+      "epoch": 0.44804318488529016,
+      "grad_norm": 1.0338096927516947,
+      "learning_rate": 3.56989247311828e-05,
+      "loss": 0.6932,
+      "step": 83
+    },
+    {
+      "epoch": 0.4534412955465587,
+      "grad_norm": 0.9595198324524808,
+      "learning_rate": 3.612903225806452e-05,
+      "loss": 0.6532,
+      "step": 84
+    },
+    {
+      "epoch": 0.45883940620782726,
+      "grad_norm": 0.9947987677712621,
+      "learning_rate": 3.655913978494624e-05,
+      "loss": 0.8222,
+      "step": 85
+    },
+    {
+      "epoch": 0.4642375168690958,
+      "grad_norm": 1.0250201420497596,
+      "learning_rate": 3.698924731182796e-05,
+      "loss": 0.9273,
+      "step": 86
+    },
+    {
+      "epoch": 0.46963562753036436,
+      "grad_norm": 1.0947389325525194,
+      "learning_rate": 3.741935483870968e-05,
+      "loss": 0.7369,
+      "step": 87
+    },
+    {
+      "epoch": 0.4750337381916329,
+      "grad_norm": 1.0165976773778116,
+      "learning_rate": 3.78494623655914e-05,
+      "loss": 0.7342,
+      "step": 88
+    },
+    {
+      "epoch": 0.48043184885290147,
+      "grad_norm": 0.9116781997576249,
+      "learning_rate": 3.827956989247312e-05,
+      "loss": 0.6436,
+      "step": 89
+    },
+    {
+      "epoch": 0.48582995951417,
+      "grad_norm": 0.96184054847529,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.66,
+      "step": 90
+    },
+    {
+      "epoch": 0.49122807017543857,
+      "grad_norm": 1.0035315302654166,
+      "learning_rate": 3.913978494623656e-05,
+      "loss": 0.7024,
+      "step": 91
+    },
+    {
+      "epoch": 0.4966261808367072,
+      "grad_norm": 0.9751688070116965,
+      "learning_rate": 3.956989247311828e-05,
+      "loss": 0.6338,
+      "step": 92
+    },
+    {
+      "epoch": 0.5020242914979757,
+      "grad_norm": 0.9697341224230857,
+      "learning_rate": 4e-05,
+      "loss": 0.7448,
+      "step": 93
+    },
+    {
+      "epoch": 0.5074224021592443,
+      "grad_norm": 0.9194411086999839,
+      "learning_rate": 3.999985742198083e-05,
+      "loss": 0.7091,
+      "step": 94
+    },
+    {
+      "epoch": 0.5128205128205128,
+      "grad_norm": 0.9694916079484717,
+      "learning_rate": 3.999942968995616e-05,
+      "loss": 0.6797,
+      "step": 95
+    },
+    {
+      "epoch": 0.5182186234817814,
+      "grad_norm": 1.0181481075774792,
+      "learning_rate": 3.99987168100245e-05,
+      "loss": 0.7829,
+      "step": 96
+    },
+    {
+      "epoch": 0.5236167341430499,
+      "grad_norm": 1.0785524566284714,
+      "learning_rate": 3.9997718792349965e-05,
+      "loss": 0.8447,
+      "step": 97
+    },
+    {
+      "epoch": 0.5290148448043185,
+      "grad_norm": 1.0979759880180078,
+      "learning_rate": 3.9996435651162085e-05,
+      "loss": 0.7234,
+      "step": 98
+    },
+    {
+      "epoch": 0.5344129554655871,
+      "grad_norm": 1.0328240510235125,
+      "learning_rate": 3.999486740475564e-05,
+      "loss": 0.7069,
+      "step": 99
+    },
+    {
+      "epoch": 0.5398110661268556,
+      "grad_norm": 1.07482749731842,
+      "learning_rate": 3.999301407549037e-05,
+      "loss": 0.723,
+      "step": 100
+    },
+    {
+      "epoch": 0.5452091767881242,
+      "grad_norm": 1.1252207718206446,
+      "learning_rate": 3.9990875689790674e-05,
+      "loss": 0.7369,
+      "step": 101
+    },
+    {
+      "epoch": 0.5506072874493927,
+      "grad_norm": 1.0500151754278362,
+      "learning_rate": 3.998845227814524e-05,
+      "loss": 0.6824,
+      "step": 102
+    },
+    {
+      "epoch": 0.5560053981106613,
+      "grad_norm": 1.0080759941105553,
+      "learning_rate": 3.9985743875106584e-05,
+      "loss": 0.6872,
+      "step": 103
+    },
+    {
+      "epoch": 0.5614035087719298,
+      "grad_norm": 0.9141499316846765,
+      "learning_rate": 3.9982750519290587e-05,
+      "loss": 0.615,
+      "step": 104
+    },
+    {
+      "epoch": 0.5668016194331984,
+      "grad_norm": 1.0153954421898292,
+      "learning_rate": 3.997947225337592e-05,
+      "loss": 0.8052,
+      "step": 105
+    },
+    {
+      "epoch": 0.5721997300944669,
+      "grad_norm": 1.0767523652137478,
+      "learning_rate": 3.997590912410345e-05,
+      "loss": 0.7559,
+      "step": 106
+    },
+    {
+      "epoch": 0.5775978407557355,
+      "grad_norm": 0.8733116794243336,
+      "learning_rate": 3.997206118227557e-05,
+      "loss": 0.7033,
+      "step": 107
+    },
+    {
+      "epoch": 0.582995951417004,
+      "grad_norm": 0.9284538711864947,
+      "learning_rate": 3.996792848275546e-05,
+      "loss": 0.7147,
+      "step": 108
+    },
+    {
+      "epoch": 0.5883940620782726,
+      "grad_norm": 1.034139453978068,
+      "learning_rate": 3.996351108446635e-05,
+      "loss": 0.8897,
+      "step": 109
+    },
+    {
+      "epoch": 0.5937921727395412,
+      "grad_norm": 0.953999434143935,
+      "learning_rate": 3.9958809050390626e-05,
+      "loss": 0.6739,
+      "step": 110
+    },
+    {
+      "epoch": 0.5991902834008097,
+      "grad_norm": 1.0225124910703658,
+      "learning_rate": 3.995382244756895e-05,
+      "loss": 0.6902,
+      "step": 111
+    },
+    {
+      "epoch": 0.6045883940620783,
+      "grad_norm": 1.0580009090642508,
+      "learning_rate": 3.994855134709931e-05,
+      "loss": 0.7694,
+      "step": 112
+    },
+    {
+      "epoch": 0.6099865047233468,
+      "grad_norm": 1.0267380802655353,
+      "learning_rate": 3.994299582413603e-05,
+      "loss": 0.7656,
+      "step": 113
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 0.973579702494221,
+      "learning_rate": 3.9937155957888646e-05,
+      "loss": 0.7677,
+      "step": 114
+    },
+    {
+      "epoch": 0.6207827260458839,
+      "grad_norm": 0.9578211850462446,
+      "learning_rate": 3.9931031831620816e-05,
+      "loss": 0.6577,
+      "step": 115
+    },
+    {
+      "epoch": 0.6261808367071525,
+      "grad_norm": 0.8906311067096349,
+      "learning_rate": 3.9924623532649124e-05,
+      "loss": 0.6594,
+      "step": 116
+    },
+    {
+      "epoch": 0.631578947368421,
+      "grad_norm": 1.0623119165643196,
+      "learning_rate": 3.991793115234182e-05,
+      "loss": 0.7637,
+      "step": 117
+    },
+    {
+      "epoch": 0.6369770580296896,
+      "grad_norm": 0.9060230192223652,
+      "learning_rate": 3.991095478611755e-05,
+      "loss": 0.6749,
+      "step": 118
+    },
+    {
+      "epoch": 0.6423751686909581,
+      "grad_norm": 1.0676225134345259,
+      "learning_rate": 3.990369453344394e-05,
+      "loss": 0.7092,
+      "step": 119
+    },
+    {
+      "epoch": 0.6477732793522267,
+      "grad_norm": 0.8987761262752798,
+      "learning_rate": 3.989615049783625e-05,
+      "loss": 0.7309,
+      "step": 120
+    },
+    {
+      "epoch": 0.6531713900134952,
+      "grad_norm": 0.8594763504734059,
+      "learning_rate": 3.9888322786855846e-05,
+      "loss": 0.6883,
+      "step": 121
+    },
+    {
+      "epoch": 0.6585695006747638,
+      "grad_norm": 0.9830202476329205,
+      "learning_rate": 3.9880211512108674e-05,
+      "loss": 0.7604,
+      "step": 122
+    },
+    {
+      "epoch": 0.6639676113360324,
+      "grad_norm": 0.9053121994382571,
+      "learning_rate": 3.987181678924369e-05,
+      "loss": 0.8748,
+      "step": 123
+    },
+    {
+      "epoch": 0.6693657219973009,
+      "grad_norm": 0.8898560961592561,
+      "learning_rate": 3.9863138737951175e-05,
+      "loss": 0.6792,
+      "step": 124
+    },
+    {
+      "epoch": 0.6747638326585695,
+      "grad_norm": 0.9147601883304166,
+      "learning_rate": 3.985417748196108e-05,
+      "loss": 0.7001,
+      "step": 125
+    },
+    {
+      "epoch": 0.680161943319838,
+      "grad_norm": 0.9009775248244547,
+      "learning_rate": 3.984493314904122e-05,
+      "loss": 0.6655,
+      "step": 126
+    },
+    {
+      "epoch": 0.6855600539811066,
+      "grad_norm": 0.9286279151444712,
+      "learning_rate": 3.983540587099545e-05,
+      "loss": 0.7159,
+      "step": 127
+    },
+    {
+      "epoch": 0.6909581646423751,
+      "grad_norm": 0.8938878575332024,
+      "learning_rate": 3.982559578366182e-05,
+      "loss": 0.8353,
+      "step": 128
+    },
+    {
+      "epoch": 0.6963562753036437,
+      "grad_norm": 0.8927016057836781,
+      "learning_rate": 3.98155030269106e-05,
+      "loss": 0.6327,
+      "step": 129
+    },
+    {
+      "epoch": 0.7017543859649122,
+      "grad_norm": 0.9351871429989481,
+      "learning_rate": 3.980512774464235e-05,
+      "loss": 0.7138,
+      "step": 130
+    },
+    {
+      "epoch": 0.7071524966261808,
+      "grad_norm": 0.8483519367428406,
+      "learning_rate": 3.979447008478575e-05,
+      "loss": 0.6275,
+      "step": 131
+    },
+    {
+      "epoch": 0.7125506072874493,
+      "grad_norm": 0.9706267575303317,
+      "learning_rate": 3.978353019929562e-05,
+      "loss": 0.7038,
+      "step": 132
+    },
+    {
+      "epoch": 0.717948717948718,
+      "grad_norm": 1.0375565226008272,
+      "learning_rate": 3.977230824415069e-05,
+      "loss": 0.7927,
+      "step": 133
+    },
+    {
+      "epoch": 0.7233468286099866,
+      "grad_norm": 0.953128885896293,
+      "learning_rate": 3.9760804379351354e-05,
+      "loss": 0.7174,
+      "step": 134
+    },
+    {
+      "epoch": 0.728744939271255,
+      "grad_norm": 1.0191595715624093,
+      "learning_rate": 3.974901876891745e-05,
+      "loss": 0.7215,
+      "step": 135
+    },
+    {
+      "epoch": 0.7341430499325237,
+      "grad_norm": 1.0166497719104675,
+      "learning_rate": 3.973695158088588e-05,
+      "loss": 0.7036,
+      "step": 136
+    },
+    {
+      "epoch": 0.7395411605937922,
+      "grad_norm": 0.8999962758004968,
+      "learning_rate": 3.972460298730822e-05,
+      "loss": 0.6583,
+      "step": 137
+    },
+    {
+      "epoch": 0.7449392712550608,
+      "grad_norm": 1.068304194336641,
+      "learning_rate": 3.9711973164248255e-05,
+      "loss": 0.776,
+      "step": 138
+    },
+    {
+      "epoch": 0.7503373819163293,
+      "grad_norm": 0.9469850180645717,
+      "learning_rate": 3.9699062291779516e-05,
+      "loss": 0.6935,
+      "step": 139
+    },
+    {
+      "epoch": 0.7557354925775979,
+      "grad_norm": 0.9373619020903246,
+      "learning_rate": 3.9685870553982665e-05,
+      "loss": 0.7552,
+      "step": 140
+    },
+    {
+      "epoch": 0.7611336032388664,
+      "grad_norm": 0.9597314115710233,
+      "learning_rate": 3.967239813894288e-05,
+      "loss": 0.6639,
+      "step": 141
+    },
+    {
+      "epoch": 0.766531713900135,
+      "grad_norm": 0.9917963708096413,
+      "learning_rate": 3.965864523874718e-05,
+      "loss": 0.7001,
+      "step": 142
+    },
+    {
+      "epoch": 0.7719298245614035,
+      "grad_norm": 0.9933806782956979,
+      "learning_rate": 3.964461204948171e-05,
+      "loss": 0.6524,
+      "step": 143
+    },
+    {
+      "epoch": 0.7773279352226721,
+      "grad_norm": 1.0289492267546245,
+      "learning_rate": 3.963029877122889e-05,
+      "loss": 0.6852,
+      "step": 144
+    },
+    {
+      "epoch": 0.7827260458839406,
+      "grad_norm": 1.0807691676224107,
+      "learning_rate": 3.961570560806461e-05,
+      "loss": 0.7367,
+      "step": 145
+    },
+    {
+      "epoch": 0.7881241565452092,
+      "grad_norm": 0.9698768218538012,
+      "learning_rate": 3.96008327680553e-05,
+      "loss": 0.7537,
+      "step": 146
+    },
+    {
+      "epoch": 0.7935222672064778,
+      "grad_norm": 0.8573500783646442,
+      "learning_rate": 3.9585680463254965e-05,
+      "loss": 0.6721,
+      "step": 147
+    },
+    {
+      "epoch": 0.7989203778677463,
+      "grad_norm": 0.9207124724229049,
+      "learning_rate": 3.9570248909702165e-05,
+      "loss": 0.6855,
+      "step": 148
+    },
+    {
+      "epoch": 0.8043184885290149,
+      "grad_norm": 0.9237323443960497,
+      "learning_rate": 3.955453832741694e-05,
+      "loss": 0.6344,
+      "step": 149
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 0.8827377832405502,
+      "learning_rate": 3.9538548940397654e-05,
+      "loss": 0.9605,
+      "step": 150
+    },
+    {
+      "epoch": 0.815114709851552,
+      "grad_norm": 1.001158615881627,
+      "learning_rate": 3.952228097661782e-05,
+      "loss": 0.7557,
+      "step": 151
+    },
+    {
+      "epoch": 0.8205128205128205,
+      "grad_norm": 0.9114023428447467,
+      "learning_rate": 3.9505734668022844e-05,
+      "loss": 0.7095,
+      "step": 152
+    },
+    {
+      "epoch": 0.8259109311740891,
+      "grad_norm": 1.0334572501302381,
+      "learning_rate": 3.948891025052672e-05,
+      "loss": 0.6735,
+      "step": 153
+    },
+    {
+      "epoch": 0.8313090418353576,
+      "grad_norm": 0.8884012995992501,
+      "learning_rate": 3.947180796400865e-05,
+      "loss": 0.6968,
+      "step": 154
+    },
+    {
+      "epoch": 0.8367071524966262,
+      "grad_norm": 0.9669368902851598,
+      "learning_rate": 3.9454428052309654e-05,
+      "loss": 0.6716,
+      "step": 155
+    },
+    {
+      "epoch": 0.8421052631578947,
+      "grad_norm": 0.9239616303764089,
+      "learning_rate": 3.943677076322908e-05,
+      "loss": 0.7839,
+      "step": 156
+    },
+    {
+      "epoch": 0.8475033738191633,
+      "grad_norm": 0.9451380007035719,
+      "learning_rate": 3.9418836348521045e-05,
+      "loss": 0.7045,
+      "step": 157
+    },
+    {
+      "epoch": 0.8529014844804319,
+      "grad_norm": 1.063003592125855,
+      "learning_rate": 3.940062506389089e-05,
+      "loss": 0.6854,
+      "step": 158
+    },
+    {
+      "epoch": 0.8582995951417004,
+      "grad_norm": 1.8021078869971958,
+      "learning_rate": 3.938213716899149e-05,
+      "loss": 0.7488,
+      "step": 159
+    },
+    {
+      "epoch": 0.863697705802969,
+      "grad_norm": 0.9871503858629542,
+      "learning_rate": 3.93633729274196e-05,
+      "loss": 0.735,
+      "step": 160
+    },
+    {
+      "epoch": 0.8690958164642375,
+      "grad_norm": 0.9932972593804917,
+      "learning_rate": 3.934433260671206e-05,
+      "loss": 0.6599,
+      "step": 161
+    },
+    {
+      "epoch": 0.8744939271255061,
+      "grad_norm": 0.9961454664914111,
+      "learning_rate": 3.9325016478341986e-05,
+      "loss": 0.7225,
+      "step": 162
+    },
+    {
+      "epoch": 0.8798920377867746,
+      "grad_norm": 0.9673270740440703,
+      "learning_rate": 3.9305424817714905e-05,
+      "loss": 0.7151,
+      "step": 163
+    },
+    {
+      "epoch": 0.8852901484480432,
+      "grad_norm": 5.451095634597877,
+      "learning_rate": 3.928555790416485e-05,
+      "loss": 0.8751,
+      "step": 164
+    },
+    {
+      "epoch": 0.8906882591093117,
+      "grad_norm": 0.9428314240010024,
+      "learning_rate": 3.926541602095033e-05,
+      "loss": 0.7981,
+      "step": 165
+    },
+    {
+      "epoch": 0.8960863697705803,
+      "grad_norm": 0.9096664280699019,
+      "learning_rate": 3.9244999455250324e-05,
+      "loss": 0.6851,
+      "step": 166
+    },
+    {
+      "epoch": 0.9014844804318488,
+      "grad_norm": 0.931366867737836,
+      "learning_rate": 3.922430849816018e-05,
+      "loss": 0.7066,
+      "step": 167
+    },
+    {
+      "epoch": 0.9068825910931174,
+      "grad_norm": 0.9121538158817324,
+      "learning_rate": 3.9203343444687475e-05,
+      "loss": 0.6972,
+      "step": 168
+    },
+    {
+      "epoch": 0.9122807017543859,
+      "grad_norm": 0.8824698034055396,
+      "learning_rate": 3.918210459374778e-05,
+      "loss": 0.6528,
+      "step": 169
+    },
+    {
+      "epoch": 0.9176788124156545,
+      "grad_norm": 0.9125560943728268,
+      "learning_rate": 3.916059224816043e-05,
+      "loss": 0.8869,
+      "step": 170
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 0.920393509862747,
+      "learning_rate": 3.913880671464418e-05,
+      "loss": 0.6674,
+      "step": 171
+    },
+    {
+      "epoch": 0.9284750337381916,
+      "grad_norm": 1.015400533440785,
+      "learning_rate": 3.911674830381286e-05,
+      "loss": 0.7454,
+      "step": 172
+    },
+    {
+      "epoch": 0.9338731443994602,
+      "grad_norm": 0.8834748729526964,
+      "learning_rate": 3.909441733017092e-05,
+      "loss": 0.6617,
+      "step": 173
+    },
+    {
+      "epoch": 0.9392712550607287,
+      "grad_norm": 1.0375353699953684,
+      "learning_rate": 3.907181411210895e-05,
+      "loss": 0.8789,
+      "step": 174
+    },
+    {
+      "epoch": 0.9446693657219973,
+      "grad_norm": 0.8880343085641559,
+      "learning_rate": 3.904893897189917e-05,
+      "loss": 0.7352,
+      "step": 175
+    },
+    {
+      "epoch": 0.9500674763832658,
+      "grad_norm": 0.8671333074476734,
+      "learning_rate": 3.9025792235690787e-05,
+      "loss": 0.6831,
+      "step": 176
+    },
+    {
+      "epoch": 0.9554655870445344,
+      "grad_norm": 0.8868491545985568,
+      "learning_rate": 3.900237423350539e-05,
+      "loss": 0.6898,
+      "step": 177
+    },
+    {
+      "epoch": 0.9608636977058029,
+      "grad_norm": 0.8479323521306729,
+      "learning_rate": 3.89786852992322e-05,
+      "loss": 0.6624,
+      "step": 178
+    },
+    {
+      "epoch": 0.9662618083670715,
+      "grad_norm": 0.9483917825593386,
+      "learning_rate": 3.895472577062337e-05,
+      "loss": 0.693,
+      "step": 179
+    },
+    {
+      "epoch": 0.97165991902834,
+      "grad_norm": 0.9612236822036284,
+      "learning_rate": 3.89304959892891e-05,
+      "loss": 0.6777,
+      "step": 180
+    },
+    {
+      "epoch": 0.9770580296896086,
+      "grad_norm": 0.9286076341189001,
+      "learning_rate": 3.8905996300692806e-05,
+      "loss": 0.797,
+      "step": 181
+    },
+    {
+      "epoch": 0.9824561403508771,
+      "grad_norm": 0.9127213840777594,
+      "learning_rate": 3.888122705414621e-05,
+      "loss": 0.6602,
+      "step": 182
+    },
+    {
+      "epoch": 0.9878542510121457,
+      "grad_norm": 0.9493385128697333,
+      "learning_rate": 3.885618860280433e-05,
+      "loss": 0.7122,
+      "step": 183
+    },
+    {
+      "epoch": 0.9932523616734144,
+      "grad_norm": 0.9006976633339124,
+      "learning_rate": 3.883088130366042e-05,
+      "loss": 0.6757,
+      "step": 184
+    },
+    {
+      "epoch": 0.9986504723346828,
+      "grad_norm": 0.888522082847371,
+      "learning_rate": 3.880530551754095e-05,
+      "loss": 0.6921,
+      "step": 185
+    },
+    {
+      "epoch": 1.0040485829959513,
+      "grad_norm": 2.107623724071206,
+      "learning_rate": 3.877946160910042e-05,
+      "loss": 0.8971,
+      "step": 186
+    },
+    {
+      "epoch": 1.00944669365722,
+      "grad_norm": 0.9189393752172309,
+      "learning_rate": 3.8753349946816154e-05,
+      "loss": 0.5301,
+      "step": 187
+    },
+    {
+      "epoch": 1.0148448043184886,
+      "grad_norm": 0.8747156107255568,
+      "learning_rate": 3.8726970902983046e-05,
+      "loss": 0.3892,
+      "step": 188
+    },
+    {
+      "epoch": 1.0202429149797572,
+      "grad_norm": 0.9099854731930261,
+      "learning_rate": 3.8700324853708304e-05,
+      "loss": 0.3632,
+      "step": 189
+    },
+    {
+      "epoch": 1.0256410256410255,
+      "grad_norm": 0.9647540571546368,
+      "learning_rate": 3.867341217890599e-05,
+      "loss": 0.3434,
+      "step": 190
+    },
+    {
+      "epoch": 1.0310391363022942,
+      "grad_norm": 1.0825287977741263,
+      "learning_rate": 3.864623326229172e-05,
+      "loss": 0.33,
+      "step": 191
+    },
+    {
+      "epoch": 1.0364372469635628,
+      "grad_norm": 1.04692971678771,
+      "learning_rate": 3.861878849137708e-05,
+      "loss": 0.3468,
+      "step": 192
+    },
+    {
+      "epoch": 1.0418353576248314,
+      "grad_norm": 0.9621311057987273,
+      "learning_rate": 3.859107825746419e-05,
+      "loss": 0.3685,
+      "step": 193
+    },
+    {
+      "epoch": 1.0472334682861,
+      "grad_norm": 0.8951929289556881,
+      "learning_rate": 3.8563102955640076e-05,
+      "loss": 0.5484,
+      "step": 194
+    },
+    {
+      "epoch": 1.0526315789473684,
+      "grad_norm": 0.8449031648070203,
+      "learning_rate": 3.853486298477105e-05,
+      "loss": 0.434,
+      "step": 195
+    },
+    {
+      "epoch": 1.058029689608637,
+      "grad_norm": 0.7908350281563535,
+      "learning_rate": 3.850635874749701e-05,
+      "loss": 0.3216,
+      "step": 196
+    },
+    {
+      "epoch": 1.0634278002699056,
+      "grad_norm": 0.7658178085112315,
+      "learning_rate": 3.8477590650225735e-05,
+      "loss": 0.3463,
+      "step": 197
+    },
+    {
+      "epoch": 1.0688259109311742,
+      "grad_norm": 0.8293601353750603,
+      "learning_rate": 3.8448559103127065e-05,
+      "loss": 0.5209,
+      "step": 198
+    },
+    {
+      "epoch": 1.0742240215924426,
+      "grad_norm": 0.9324084250572037,
+      "learning_rate": 3.841926452012704e-05,
+      "loss": 0.3883,
+      "step": 199
+    },
+    {
+      "epoch": 1.0796221322537112,
+      "grad_norm": 0.9751349766349348,
+      "learning_rate": 3.838970731890202e-05,
+      "loss": 0.3814,
+      "step": 200
+    },
+    {
+      "epoch": 1.0850202429149798,
+      "grad_norm": 0.9158339914827958,
+      "learning_rate": 3.835988792087272e-05,
+      "loss": 0.5085,
+      "step": 201
+    },
+    {
+      "epoch": 1.0904183535762484,
+      "grad_norm": 0.969846087023057,
+      "learning_rate": 3.832980675119823e-05,
+      "loss": 0.4278,
+      "step": 202
+    },
+    {
+      "epoch": 1.0958164642375168,
+      "grad_norm": 0.9652923092534054,
+      "learning_rate": 3.8299464238769883e-05,
+      "loss": 0.3848,
+      "step": 203
+    },
+    {
+      "epoch": 1.1012145748987854,
+      "grad_norm": 0.7607007987891418,
+      "learning_rate": 3.826886081620523e-05,
+      "loss": 0.3147,
+      "step": 204
+    },
+    {
+      "epoch": 1.106612685560054,
+      "grad_norm": 0.8495962762794291,
+      "learning_rate": 3.82379969198418e-05,
+      "loss": 0.5328,
+      "step": 205
+    },
+    {
+      "epoch": 1.1120107962213226,
+      "grad_norm": 0.8289863223276595,
+      "learning_rate": 3.8206872989730926e-05,
+      "loss": 0.3223,
+      "step": 206
+    },
+    {
+      "epoch": 1.117408906882591,
+      "grad_norm": 1.222040602702584,
+      "learning_rate": 3.817548946963142e-05,
+      "loss": 0.3392,
+      "step": 207
+    },
+    {
+      "epoch": 1.1228070175438596,
+      "grad_norm": 0.8657089069286453,
+      "learning_rate": 3.81438468070033e-05,
+      "loss": 0.3371,
+      "step": 208
+    },
+    {
+      "epoch": 1.1282051282051282,
+      "grad_norm": 0.8119296657125807,
+      "learning_rate": 3.811194545300139e-05,
+      "loss": 0.3571,
+      "step": 209
+    },
+    {
+      "epoch": 1.1336032388663968,
+      "grad_norm": 0.8233951136336931,
+      "learning_rate": 3.807978586246887e-05,
+      "loss": 0.3206,
+      "step": 210
+    },
+    {
+      "epoch": 1.1390013495276654,
+      "grad_norm": 0.8401268303490904,
+      "learning_rate": 3.8047368493930814e-05,
+      "loss": 0.3526,
+      "step": 211
+    },
+    {
+      "epoch": 1.1443994601889338,
+      "grad_norm": 0.8710349516226565,
+      "learning_rate": 3.8014693809587635e-05,
+      "loss": 0.4329,
+      "step": 212
+    },
+    {
+      "epoch": 1.1497975708502024,
+      "grad_norm": 0.753648162212496,
+      "learning_rate": 3.798176227530852e-05,
+      "loss": 0.2964,
+      "step": 213
+    },
+    {
+      "epoch": 1.155195681511471,
+      "grad_norm": 0.7555733624437558,
+      "learning_rate": 3.794857436062476e-05,
+      "loss": 0.3081,
+      "step": 214
+    },
+    {
+      "epoch": 1.1605937921727396,
+      "grad_norm": 0.8388358754392063,
+      "learning_rate": 3.791513053872306e-05,
+      "loss": 0.3723,
+      "step": 215
+    },
+    {
+      "epoch": 1.165991902834008,
+      "grad_norm": 0.8639315626899287,
+      "learning_rate": 3.788143128643881e-05,
+      "loss": 0.3351,
+      "step": 216
+    },
+    {
+      "epoch": 1.1713900134952766,
+      "grad_norm": 0.8636653829702634,
+      "learning_rate": 3.784747708424929e-05,
+      "loss": 0.3533,
+      "step": 217
+    },
+    {
+      "epoch": 1.1767881241565452,
+      "grad_norm": 0.925338274642717,
+      "learning_rate": 3.781326841626677e-05,
+      "loss": 0.359,
+      "step": 218
+    },
+    {
+      "epoch": 1.1821862348178138,
+      "grad_norm": 0.837482775212241,
+      "learning_rate": 3.777880577023167e-05,
+      "loss": 0.3305,
+      "step": 219
+    },
+    {
+      "epoch": 1.1875843454790824,
+      "grad_norm": 2.1021757730771076,
+      "learning_rate": 3.7744089637505565e-05,
+      "loss": 0.5936,
+      "step": 220
+    },
+    {
+      "epoch": 1.1929824561403508,
+      "grad_norm": 0.9296562522439475,
+      "learning_rate": 3.7709120513064196e-05,
+      "loss": 0.3762,
+      "step": 221
+    },
+    {
+      "epoch": 1.1983805668016194,
+      "grad_norm": 0.9309903763944034,
+      "learning_rate": 3.7673898895490435e-05,
+      "loss": 0.3366,
+      "step": 222
+    },
+    {
+      "epoch": 1.203778677462888,
+      "grad_norm": 0.9325456251013327,
+      "learning_rate": 3.76384252869671e-05,
+      "loss": 0.4013,
+      "step": 223
+    },
+    {
+      "epoch": 1.2091767881241566,
+      "grad_norm": 0.831945837860431,
+      "learning_rate": 3.760270019326989e-05,
+      "loss": 0.3515,
+      "step": 224
+    },
+    {
+      "epoch": 1.214574898785425,
+      "grad_norm": 0.8691957746034139,
+      "learning_rate": 3.7566724123760126e-05,
+      "loss": 0.3192,
+      "step": 225
+    },
+    {
+      "epoch": 1.2199730094466936,
+      "grad_norm": 0.9139101759145002,
+      "learning_rate": 3.753049759137745e-05,
+      "loss": 0.3682,
+      "step": 226
+    },
+    {
+      "epoch": 1.2253711201079622,
+      "grad_norm": 0.9052414221450297,
+      "learning_rate": 3.749402111263261e-05,
+      "loss": 0.4985,
+      "step": 227
+    },
+    {
+      "epoch": 1.2307692307692308,
+      "grad_norm": 0.8601788432649373,
+      "learning_rate": 3.745729520760001e-05,
+      "loss": 0.5243,
+      "step": 228
+    },
+    {
+      "epoch": 1.2361673414304994,
+      "grad_norm": 0.9082317660460365,
+      "learning_rate": 3.7420320399910315e-05,
+      "loss": 0.487,
+      "step": 229
+    },
+    {
+      "epoch": 1.2415654520917678,
+      "grad_norm": 0.8370456976476759,
+      "learning_rate": 3.738309721674302e-05,
+      "loss": 0.3438,
+      "step": 230
+    },
+    {
+      "epoch": 1.2469635627530364,
+      "grad_norm": 0.8831885906992842,
+      "learning_rate": 3.7345626188818906e-05,
+      "loss": 0.4309,
+      "step": 231
+    },
+    {
+      "epoch": 1.252361673414305,
+      "grad_norm": 0.8451679762528965,
+      "learning_rate": 3.730790785039245e-05,
+      "loss": 0.3743,
+      "step": 232
+    },
+    {
+      "epoch": 1.2577597840755734,
+      "grad_norm": 0.8332770375826859,
+      "learning_rate": 3.726994273924426e-05,
+      "loss": 0.34,
+      "step": 233
+    },
+    {
+      "epoch": 1.263157894736842,
+      "grad_norm": 0.7907219046982563,
+      "learning_rate": 3.7231731396673365e-05,
+      "loss": 0.3425,
+      "step": 234
+    },
+    {
+      "epoch": 1.2685560053981106,
+      "grad_norm": 0.8212370028839864,
+      "learning_rate": 3.7193274367489524e-05,
+      "loss": 0.3323,
+      "step": 235
+    },
+    {
+      "epoch": 1.2739541160593792,
+      "grad_norm": 0.9273758975891271,
+      "learning_rate": 3.7154572200005446e-05,
+      "loss": 0.3162,
+      "step": 236
+    },
+    {
+      "epoch": 1.2793522267206479,
+      "grad_norm": 0.8394862351712985,
+      "learning_rate": 3.711562544602895e-05,
+      "loss": 0.3536,
+      "step": 237
+    },
+    {
+      "epoch": 1.2847503373819165,
+      "grad_norm": 0.8760920284227975,
+      "learning_rate": 3.707643466085516e-05,
+      "loss": 0.3506,
+      "step": 238
+    },
+    {
+      "epoch": 1.2901484480431848,
+      "grad_norm": 1.2688148905957715,
+      "learning_rate": 3.703700040325852e-05,
+      "loss": 0.3955,
+      "step": 239
+    },
+    {
+      "epoch": 1.2955465587044535,
+      "grad_norm": 0.804670809424361,
+      "learning_rate": 3.699732323548485e-05,
+      "loss": 0.3319,
+      "step": 240
+    },
+    {
+      "epoch": 1.300944669365722,
+      "grad_norm": 0.8412796947360813,
+      "learning_rate": 3.695740372324337e-05,
+      "loss": 0.3594,
+      "step": 241
+    },
+    {
+      "epoch": 1.3063427800269904,
+      "grad_norm": 0.9589921139509276,
+      "learning_rate": 3.691724243569857e-05,
+      "loss": 0.3702,
+      "step": 242
+    },
+    {
+      "epoch": 1.311740890688259,
+      "grad_norm": 0.8714916983803426,
+      "learning_rate": 3.687683994546213e-05,
+      "loss": 0.3308,
+      "step": 243
+    },
+    {
+      "epoch": 1.3171390013495277,
+      "grad_norm": 0.8285733278761981,
+      "learning_rate": 3.683619682858476e-05,
+      "loss": 0.347,
+      "step": 244
+    },
+    {
+      "epoch": 1.3225371120107963,
+      "grad_norm": 0.8429506895539456,
+      "learning_rate": 3.6795313664547965e-05,
+      "loss": 0.3526,
+      "step": 245
+    },
+    {
+      "epoch": 1.3279352226720649,
+      "grad_norm": 0.7519445012666636,
+      "learning_rate": 3.675419103625579e-05,
+      "loss": 0.3162,
+      "step": 246
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.8723567333640136,
+      "learning_rate": 3.6712829530026535e-05,
+      "loss": 0.3946,
+      "step": 247
+    },
+    {
+      "epoch": 1.3387314439946019,
+      "grad_norm": 0.7465599134338717,
+      "learning_rate": 3.6671229735584365e-05,
+      "loss": 0.292,
+      "step": 248
+    },
+    {
+      "epoch": 1.3441295546558705,
+      "grad_norm": 0.9610089965759107,
+      "learning_rate": 3.662939224605091e-05,
+      "loss": 0.5651,
+      "step": 249
+    },
+    {
+      "epoch": 1.349527665317139,
+      "grad_norm": 0.7958636434458861,
+      "learning_rate": 3.658731765793679e-05,
+      "loss": 0.3456,
+      "step": 250
+    },
+    {
+      "epoch": 1.3549257759784075,
+      "grad_norm": 0.9025425406607611,
+      "learning_rate": 3.6545006571133174e-05,
+      "loss": 0.3962,
+      "step": 251
+    },
+    {
+      "epoch": 1.360323886639676,
+      "grad_norm": 1.1188923175275363,
+      "learning_rate": 3.650245958890314e-05,
+      "loss": 0.4168,
+      "step": 252
+    },
+    {
+      "epoch": 1.3657219973009447,
+      "grad_norm": 0.9085928678920568,
+      "learning_rate": 3.645967731787313e-05,
+      "loss": 0.4299,
+      "step": 253
+    },
+    {
+      "epoch": 1.3711201079622133,
+      "grad_norm": 0.8811462614926182,
+      "learning_rate": 3.64166603680243e-05,
+      "loss": 0.3348,
+      "step": 254
+    },
+    {
+      "epoch": 1.376518218623482,
+      "grad_norm": 0.841371629214431,
+      "learning_rate": 3.63734093526838e-05,
+      "loss": 0.3329,
+      "step": 255
+    },
+    {
+      "epoch": 1.3819163292847503,
+      "grad_norm": 0.8034694226086229,
+      "learning_rate": 3.632992488851603e-05,
+      "loss": 0.3251,
+      "step": 256
+    },
+    {
+      "epoch": 1.3873144399460189,
+      "grad_norm": 0.9123778149424828,
+      "learning_rate": 3.6286207595513884e-05,
+      "loss": 0.4415,
+      "step": 257
+    },
+    {
+      "epoch": 1.3927125506072875,
+      "grad_norm": 0.8477543880054812,
+      "learning_rate": 3.624225809698984e-05,
+      "loss": 0.356,
+      "step": 258
+    },
+    {
+      "epoch": 1.3981106612685559,
+      "grad_norm": 0.8724561309775043,
+      "learning_rate": 3.619807701956717e-05,
+      "loss": 0.6079,
+      "step": 259
+    },
+    {
+      "epoch": 1.4035087719298245,
+      "grad_norm": 0.8788948245658205,
+      "learning_rate": 3.61536649931709e-05,
+      "loss": 0.379,
+      "step": 260
+    },
+    {
+      "epoch": 1.408906882591093,
+      "grad_norm": 0.8084132959877686,
+      "learning_rate": 3.610902265101892e-05,
+      "loss": 0.3206,
+      "step": 261
+    },
+    {
+      "epoch": 1.4143049932523617,
+      "grad_norm": 1.128225187623626,
+      "learning_rate": 3.60641506296129e-05,
+      "loss": 0.3713,
+      "step": 262
+    },
+    {
+      "epoch": 1.4197031039136303,
+      "grad_norm": 0.8487023363953061,
+      "learning_rate": 3.601904956872923e-05,
+      "loss": 0.4068,
+      "step": 263
+    },
+    {
+      "epoch": 1.425101214574899,
+      "grad_norm": 0.7735840238208479,
+      "learning_rate": 3.5973720111409904e-05,
+      "loss": 0.3027,
+      "step": 264
+    },
+    {
+      "epoch": 1.4304993252361673,
+      "grad_norm": 0.8387835714414273,
+      "learning_rate": 3.5928162903953347e-05,
+      "loss": 0.3316,
+      "step": 265
+    },
+    {
+      "epoch": 1.435897435897436,
+      "grad_norm": 0.8931911914493708,
+      "learning_rate": 3.5882378595905195e-05,
+      "loss": 0.4559,
+      "step": 266
+    },
+    {
+      "epoch": 1.4412955465587045,
+      "grad_norm": 0.7862491291050567,
+      "learning_rate": 3.583636784004904e-05,
+      "loss": 0.51,
+      "step": 267
+    },
+    {
+      "epoch": 1.446693657219973,
+      "grad_norm": 0.8277281328416171,
+      "learning_rate": 3.5790131292397135e-05,
+      "loss": 0.3729,
+      "step": 268
+    },
+    {
+      "epoch": 1.4520917678812415,
+      "grad_norm": 0.9650920391532519,
+      "learning_rate": 3.5743669612181004e-05,
+      "loss": 0.367,
+      "step": 269
+    },
+    {
+      "epoch": 1.45748987854251,
+      "grad_norm": 0.8553135521291629,
+      "learning_rate": 3.569698346184209e-05,
+      "loss": 0.5576,
+      "step": 270
+    },
+    {
+      "epoch": 1.4628879892037787,
+      "grad_norm": 0.9299003613730609,
+      "learning_rate": 3.5650073507022286e-05,
+      "loss": 0.3706,
+      "step": 271
+    },
+    {
+      "epoch": 1.4682860998650473,
+      "grad_norm": 0.8793888895679159,
+      "learning_rate": 3.560294041655442e-05,
+      "loss": 0.3762,
+      "step": 272
+    },
+    {
+      "epoch": 1.4736842105263157,
+      "grad_norm": 0.8467989291039462,
+      "learning_rate": 3.555558486245277e-05,
+      "loss": 0.3465,
+      "step": 273
+    },
+    {
+      "epoch": 1.4790823211875843,
+      "grad_norm": 0.8637631419963459,
+      "learning_rate": 3.550800751990343e-05,
+      "loss": 0.3475,
+      "step": 274
+    },
+    {
+      "epoch": 1.484480431848853,
+      "grad_norm": 0.7364752903466767,
+      "learning_rate": 3.546020906725474e-05,
+      "loss": 0.3164,
+      "step": 275
+    },
+    {
+      "epoch": 1.4898785425101215,
+      "grad_norm": 0.7765670613963818,
+      "learning_rate": 3.5412190186007573e-05,
+      "loss": 0.3317,
+      "step": 276
+    },
+    {
+      "epoch": 1.49527665317139,
+      "grad_norm": 0.8621566950599185,
+      "learning_rate": 3.5363951560805615e-05,
+      "loss": 0.3959,
+      "step": 277
+    },
+    {
+      "epoch": 1.5006747638326585,
+      "grad_norm": 0.8422602307178598,
+      "learning_rate": 3.5315493879425634e-05,
+      "loss": 0.4035,
+      "step": 278
+    },
+    {
+      "epoch": 1.5060728744939271,
+      "grad_norm": 0.7596899423538056,
+      "learning_rate": 3.526681783276765e-05,
+      "loss": 0.4266,
+      "step": 279
+    },
+    {
+      "epoch": 1.5114709851551957,
+      "grad_norm": 0.8514772881962402,
+      "learning_rate": 3.52179241148451e-05,
+      "loss": 0.3525,
+      "step": 280
+    },
+    {
+      "epoch": 1.5168690958164643,
+      "grad_norm": 0.931617785975809,
+      "learning_rate": 3.5168813422774923e-05,
+      "loss": 0.3883,
+      "step": 281
+    },
+    {
+      "epoch": 1.522267206477733,
+      "grad_norm": 0.8127891358761098,
+      "learning_rate": 3.511948645676764e-05,
+      "loss": 0.3389,
+      "step": 282
+    },
+    {
+      "epoch": 1.5276653171390013,
+      "grad_norm": 0.7851309128836562,
+      "learning_rate": 3.506994392011737e-05,
+      "loss": 0.5101,
+      "step": 283
+    },
+    {
+      "epoch": 1.53306342780027,
+      "grad_norm": 0.8808007068286364,
+      "learning_rate": 3.5020186519191775e-05,
+      "loss": 0.3914,
+      "step": 284
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 0.8733322559194485,
+      "learning_rate": 3.497021496342203e-05,
+      "loss": 0.4108,
+      "step": 285
+    },
+    {
+      "epoch": 1.543859649122807,
+      "grad_norm": 0.7737338680770315,
+      "learning_rate": 3.492002996529267e-05,
+      "loss": 0.305,
+      "step": 286
+    },
+    {
+      "epoch": 1.5492577597840755,
+      "grad_norm": 0.8418178027753623,
+      "learning_rate": 3.486963224033146e-05,
+      "loss": 0.5515,
+      "step": 287
+    },
+    {
+      "epoch": 1.5546558704453441,
+      "grad_norm": 0.8420303033592552,
+      "learning_rate": 3.4819022507099184e-05,
+      "loss": 0.3534,
+      "step": 288
+    },
+    {
+      "epoch": 1.5600539811066128,
+      "grad_norm": 1.0620805768424724,
+      "learning_rate": 3.4768201487179395e-05,
+      "loss": 0.3353,
+      "step": 289
+    },
+    {
+      "epoch": 1.5654520917678814,
+      "grad_norm": 1.024773159315396,
+      "learning_rate": 3.471716990516812e-05,
+      "loss": 0.4392,
+      "step": 290
+    },
+    {
+      "epoch": 1.5708502024291497,
+      "grad_norm": 0.8521039631280201,
+      "learning_rate": 3.466592848866356e-05,
+      "loss": 0.3592,
+      "step": 291
+    },
+    {
+      "epoch": 1.5762483130904184,
+      "grad_norm": 0.929491372833879,
+      "learning_rate": 3.4614477968255663e-05,
+      "loss": 0.4302,
+      "step": 292
+    },
+    {
+      "epoch": 1.581646423751687,
+      "grad_norm": 0.848771342694289,
+      "learning_rate": 3.456281907751577e-05,
+      "loss": 0.3386,
+      "step": 293
+    },
+    {
+      "epoch": 1.5870445344129553,
+      "grad_norm": 0.8184995822192409,
+      "learning_rate": 3.4510952552986114e-05,
+      "loss": 0.3173,
+      "step": 294
+    },
+    {
+      "epoch": 1.592442645074224,
+      "grad_norm": 0.8927131276185085,
+      "learning_rate": 3.445887913416932e-05,
+      "loss": 0.3615,
+      "step": 295
+    },
+    {
+      "epoch": 1.5978407557354926,
+      "grad_norm": 0.8767018575427654,
+      "learning_rate": 3.440659956351788e-05,
+      "loss": 0.3297,
+      "step": 296
+    },
+    {
+      "epoch": 1.6032388663967612,
+      "grad_norm": 1.1261411903507468,
+      "learning_rate": 3.435411458642357e-05,
+      "loss": 0.3861,
+      "step": 297
+    },
+    {
+      "epoch": 1.6086369770580298,
+      "grad_norm": 1.178065026924561,
+      "learning_rate": 3.430142495120678e-05,
+      "loss": 0.3637,
+      "step": 298
+    },
+    {
+      "epoch": 1.6140350877192984,
+      "grad_norm": 0.925225267197288,
+      "learning_rate": 3.4248531409105896e-05,
+      "loss": 0.3435,
+      "step": 299
+    },
+    {
+      "epoch": 1.6194331983805668,
+      "grad_norm": 1.123072007122771,
+      "learning_rate": 3.419543471426657e-05,
+      "loss": 0.3607,
+      "step": 300
+    },
+    {
+      "epoch": 1.6248313090418354,
+      "grad_norm": 0.8402779580473496,
+      "learning_rate": 3.4142135623730954e-05,
+      "loss": 0.372,
+      "step": 301
+    },
+    {
+      "epoch": 1.6302294197031038,
+      "grad_norm": 0.8432024190788434,
+      "learning_rate": 3.408863489742692e-05,
+      "loss": 0.3797,
+      "step": 302
+    },
+    {
+      "epoch": 1.6356275303643724,
+      "grad_norm": 0.762768499335429,
+      "learning_rate": 3.403493329815724e-05,
+      "loss": 0.3165,
+      "step": 303
+    },
+    {
+      "epoch": 1.641025641025641,
+      "grad_norm": 0.896166920092516,
+      "learning_rate": 3.3981031591588665e-05,
+      "loss": 0.3498,
+      "step": 304
+    },
+    {
+      "epoch": 1.6464237516869096,
+      "grad_norm": 0.8344661963797614,
+      "learning_rate": 3.392693054624106e-05,
+      "loss": 0.3377,
+      "step": 305
+    },
+    {
+      "epoch": 1.6518218623481782,
+      "grad_norm": 0.8250102849016384,
+      "learning_rate": 3.387263093347641e-05,
+      "loss": 0.3569,
+      "step": 306
+    },
+    {
+      "epoch": 1.6572199730094468,
+      "grad_norm": 0.8506372433524317,
+      "learning_rate": 3.381813352748783e-05,
+      "loss": 0.3639,
+      "step": 307
+    },
+    {
+      "epoch": 1.6626180836707154,
+      "grad_norm": 0.8257724438772815,
+      "learning_rate": 3.376343910528855e-05,
+      "loss": 0.3874,
+      "step": 308
+    },
+    {
+      "epoch": 1.6680161943319838,
+      "grad_norm": 0.8022963652236939,
+      "learning_rate": 3.37085484467008e-05,
+      "loss": 0.3492,
+      "step": 309
+    },
+    {
+      "epoch": 1.6734143049932524,
+      "grad_norm": 0.905234518186738,
+      "learning_rate": 3.365346233434472e-05,
+      "loss": 0.384,
+      "step": 310
+    },
+    {
+      "epoch": 1.6788124156545208,
+      "grad_norm": 0.8591813855171073,
+      "learning_rate": 3.359818155362719e-05,
+      "loss": 0.4369,
+      "step": 311
+    },
+    {
+      "epoch": 1.6842105263157894,
+      "grad_norm": 0.9410857979791486,
+      "learning_rate": 3.354270689273064e-05,
+      "loss": 0.3781,
+      "step": 312
+    },
+    {
+      "epoch": 1.689608636977058,
+      "grad_norm": 0.8405404929752105,
+      "learning_rate": 3.348703914260178e-05,
+      "loss": 0.3411,
+      "step": 313
+    },
+    {
+      "epoch": 1.6950067476383266,
+      "grad_norm": 0.8905814682326707,
+      "learning_rate": 3.3431179096940375e-05,
+      "loss": 0.3693,
+      "step": 314
+    },
+    {
+      "epoch": 1.7004048582995952,
+      "grad_norm": 0.8941324324160825,
+      "learning_rate": 3.3375127552187885e-05,
+      "loss": 0.3785,
+      "step": 315
+    },
+    {
+      "epoch": 1.7058029689608638,
+      "grad_norm": 0.7616216142868617,
+      "learning_rate": 3.331888530751613e-05,
+      "loss": 0.3039,
+      "step": 316
+    },
+    {
+      "epoch": 1.7112010796221324,
+      "grad_norm": 0.9084079682135392,
+      "learning_rate": 3.326245316481591e-05,
+      "loss": 0.4713,
+      "step": 317
+    },
+    {
+      "epoch": 1.7165991902834008,
+      "grad_norm": 0.7825014568707412,
+      "learning_rate": 3.320583192868552e-05,
+      "loss": 0.3262,
+      "step": 318
+    },
+    {
+      "epoch": 1.7219973009446694,
+      "grad_norm": 0.861815212658462,
+      "learning_rate": 3.3149022406419335e-05,
+      "loss": 0.3782,
+      "step": 319
+    },
+    {
+      "epoch": 1.7273954116059378,
+      "grad_norm": 0.8862028154187571,
+      "learning_rate": 3.309202540799628e-05,
+      "loss": 0.4097,
+      "step": 320
+    },
+    {
+      "epoch": 1.7327935222672064,
+      "grad_norm": 0.8186946479204251,
+      "learning_rate": 3.303484174606824e-05,
+      "loss": 0.3588,
+      "step": 321
+    },
+    {
+      "epoch": 1.738191632928475,
+      "grad_norm": 0.776241019569162,
+      "learning_rate": 3.297747223594858e-05,
+      "loss": 0.3342,
+      "step": 322
+    },
+    {
+      "epoch": 1.7435897435897436,
+      "grad_norm": 0.8452001526797029,
+      "learning_rate": 3.291991769560038e-05,
+      "loss": 0.3545,
+      "step": 323
+    },
+    {
+      "epoch": 1.7489878542510122,
+      "grad_norm": 0.8961982534144177,
+      "learning_rate": 3.286217894562489e-05,
+      "loss": 0.414,
+      "step": 324
+    },
+    {
+      "epoch": 1.7543859649122808,
+      "grad_norm": 0.8118847490614682,
+      "learning_rate": 3.280425680924976e-05,
+      "loss": 0.3305,
+      "step": 325
+    },
+    {
+      "epoch": 1.7597840755735492,
+      "grad_norm": 0.8073762456721245,
+      "learning_rate": 3.274615211231735e-05,
+      "loss": 0.3282,
+      "step": 326
+    },
+    {
+      "epoch": 1.7651821862348178,
+      "grad_norm": 0.8314384392452843,
+      "learning_rate": 3.268786568327291e-05,
+      "loss": 0.3698,
+      "step": 327
+    },
+    {
+      "epoch": 1.7705802968960864,
+      "grad_norm": 0.925358774753271,
+      "learning_rate": 3.262939835315281e-05,
+      "loss": 0.3987,
+      "step": 328
+    },
+    {
+      "epoch": 1.7759784075573548,
+      "grad_norm": 0.8013523350987026,
+      "learning_rate": 3.2570750955572643e-05,
+      "loss": 0.3757,
+      "step": 329
+    },
+    {
+      "epoch": 1.7813765182186234,
+      "grad_norm": 0.7324707115191926,
+      "learning_rate": 3.25119243267154e-05,
+      "loss": 0.2965,
+      "step": 330
+    },
+    {
+      "epoch": 1.786774628879892,
+      "grad_norm": 0.8262424130157302,
+      "learning_rate": 3.2452919305319505e-05,
+      "loss": 0.3442,
+      "step": 331
+    },
+    {
+      "epoch": 1.7921727395411606,
+      "grad_norm": 0.8246556345266295,
+      "learning_rate": 3.239373673266686e-05,
+      "loss": 0.3468,
+      "step": 332
+    },
+    {
+      "epoch": 1.7975708502024292,
+      "grad_norm": 0.8777196939219886,
+      "learning_rate": 3.2334377452570866e-05,
+      "loss": 0.3921,
+      "step": 333
+    },
+    {
+      "epoch": 1.8029689608636978,
+      "grad_norm": 0.7459489151831635,
+      "learning_rate": 3.227484231136437e-05,
+      "loss": 0.2964,
+      "step": 334
+    },
+    {
+      "epoch": 1.8083670715249662,
+      "grad_norm": 0.828491144813016,
+      "learning_rate": 3.2215132157887636e-05,
+      "loss": 0.3421,
+      "step": 335
+    },
+    {
+      "epoch": 1.8137651821862348,
+      "grad_norm": 0.8526117164203312,
+      "learning_rate": 3.2155247843476204e-05,
+      "loss": 0.3378,
+      "step": 336
+    },
+    {
+      "epoch": 1.8191632928475032,
+      "grad_norm": 0.7963280690247899,
+      "learning_rate": 3.209519022194875e-05,
+      "loss": 0.3651,
+      "step": 337
+    },
+    {
+      "epoch": 1.8245614035087718,
+      "grad_norm": 0.8934492004331263,
+      "learning_rate": 3.203496014959497e-05,
+      "loss": 0.4337,
+      "step": 338
+    },
+    {
+      "epoch": 1.8299595141700404,
+      "grad_norm": 0.8985086086718561,
+      "learning_rate": 3.197455848516328e-05,
+      "loss": 0.388,
+      "step": 339
+    },
+    {
+      "epoch": 1.835357624831309,
+      "grad_norm": 0.8915332910088548,
+      "learning_rate": 3.191398608984867e-05,
+      "loss": 0.3913,
+      "step": 340
+    },
+    {
+      "epoch": 1.8407557354925776,
+      "grad_norm": 0.7721439545372889,
+      "learning_rate": 3.185324382728034e-05,
+      "loss": 0.3086,
+      "step": 341
+    },
+    {
+      "epoch": 1.8461538461538463,
+      "grad_norm": 0.8184460892756168,
+      "learning_rate": 3.179233256350944e-05,
+      "loss": 0.3676,
+      "step": 342
+    },
+    {
+      "epoch": 1.8515519568151149,
+      "grad_norm": 0.9226807093965883,
+      "learning_rate": 3.173125316699671e-05,
+      "loss": 0.5641,
+      "step": 343
+    },
+    {
+      "epoch": 1.8569500674763832,
+      "grad_norm": 0.8525374078537165,
+      "learning_rate": 3.1670006508600076e-05,
+      "loss": 0.3127,
+      "step": 344
+    },
+    {
+      "epoch": 1.8623481781376519,
+      "grad_norm": 0.9153664740398068,
+      "learning_rate": 3.160859346156227e-05,
+      "loss": 0.3624,
+      "step": 345
+    },
+    {
+      "epoch": 1.8677462887989202,
+      "grad_norm": 0.8653270177477282,
+      "learning_rate": 3.1547014901498344e-05,
+      "loss": 0.3762,
+      "step": 346
+    },
+    {
+      "epoch": 1.8731443994601888,
+      "grad_norm": 0.8652049537921908,
+      "learning_rate": 3.148527170638322e-05,
+      "loss": 0.4097,
+      "step": 347
+    },
+    {
+      "epoch": 1.8785425101214575,
+      "grad_norm": 0.8195695433857648,
+      "learning_rate": 3.1423364756539135e-05,
+      "loss": 0.3777,
+      "step": 348
+    },
+    {
+      "epoch": 1.883940620782726,
+      "grad_norm": 0.8298301686755263,
+      "learning_rate": 3.136129493462312e-05,
+      "loss": 0.3709,
+      "step": 349
+    },
+    {
+      "epoch": 1.8893387314439947,
+      "grad_norm": 0.8730439467118726,
+      "learning_rate": 3.12990631256144e-05,
+      "loss": 0.369,
+      "step": 350
+    },
+    {
+      "epoch": 1.8947368421052633,
+      "grad_norm": 0.8811562346740207,
+      "learning_rate": 3.1236670216801786e-05,
+      "loss": 0.378,
+      "step": 351
+    },
+    {
+      "epoch": 1.9001349527665317,
+      "grad_norm": 0.9029600963348642,
+      "learning_rate": 3.117411709777101e-05,
+      "loss": 0.4677,
+      "step": 352
+    },
+    {
+      "epoch": 1.9055330634278003,
+      "grad_norm": 0.8241134002145012,
+      "learning_rate": 3.111140466039205e-05,
+      "loss": 0.3297,
+      "step": 353
+    },
+    {
+      "epoch": 1.9109311740890689,
+      "grad_norm": 0.807276924572608,
+      "learning_rate": 3.104853379880641e-05,
+      "loss": 0.5004,
+      "step": 354
+    },
+    {
+      "epoch": 1.9163292847503373,
+      "grad_norm": 0.8204486996495141,
+      "learning_rate": 3.0985505409414395e-05,
+      "loss": 0.3344,
+      "step": 355
+    },
+    {
+      "epoch": 1.9217273954116059,
+      "grad_norm": 0.7969004086168736,
+      "learning_rate": 3.09223203908623e-05,
+      "loss": 0.5046,
+      "step": 356
+    },
+    {
+      "epoch": 1.9271255060728745,
+      "grad_norm": 0.8150046291130755,
+      "learning_rate": 3.085897964402958e-05,
+      "loss": 0.3913,
+      "step": 357
+    },
+    {
+      "epoch": 1.932523616734143,
+      "grad_norm": 0.7986421126905342,
+      "learning_rate": 3.0795484072016066e-05,
+      "loss": 0.3366,
+      "step": 358
+    },
+    {
+      "epoch": 1.9379217273954117,
+      "grad_norm": 0.8792903034368934,
+      "learning_rate": 3.073183458012906e-05,
+      "loss": 0.4179,
+      "step": 359
+    },
+    {
+      "epoch": 1.9433198380566803,
+      "grad_norm": 0.779085071848146,
+      "learning_rate": 3.0668032075870394e-05,
+      "loss": 0.3046,
+      "step": 360
+    },
+    {
+      "epoch": 1.9487179487179487,
+      "grad_norm": 0.7875096169267145,
+      "learning_rate": 3.060407746892354e-05,
+      "loss": 0.3429,
+      "step": 361
+    },
+    {
+      "epoch": 1.9541160593792173,
+      "grad_norm": 2.54950881345751,
+      "learning_rate": 3.053997167114062e-05,
+      "loss": 0.3572,
+      "step": 362
+    },
+    {
+      "epoch": 1.9595141700404857,
+      "grad_norm": 0.8007107726963351,
+      "learning_rate": 3.0475715596529402e-05,
+      "loss": 0.3873,
+      "step": 363
+    },
+    {
+      "epoch": 1.9649122807017543,
+      "grad_norm": 0.7836754358043907,
+      "learning_rate": 3.041131016124026e-05,
+      "loss": 0.3389,
+      "step": 364
+    },
+    {
+      "epoch": 1.9703103913630229,
+      "grad_norm": 0.7678617425606538,
+      "learning_rate": 3.0346756283553138e-05,
+      "loss": 0.3084,
+      "step": 365
+    },
+    {
+      "epoch": 1.9757085020242915,
+      "grad_norm": 0.8721820631412847,
+      "learning_rate": 3.0282054883864434e-05,
+      "loss": 0.4878,
+      "step": 366
+    },
+    {
+      "epoch": 1.98110661268556,
+      "grad_norm": 0.7880172313158853,
+      "learning_rate": 3.0217206884673898e-05,
+      "loss": 0.4642,
+      "step": 367
+    },
+    {
+      "epoch": 1.9865047233468287,
+      "grad_norm": 0.9355221602685624,
+      "learning_rate": 3.015221321057145e-05,
+      "loss": 0.4395,
+      "step": 368
+    },
+    {
+      "epoch": 1.9919028340080973,
+      "grad_norm": 0.8976894020686896,
+      "learning_rate": 3.0087074788224016e-05,
+      "loss": 0.5688,
+      "step": 369
+    },
+    {
+      "epoch": 1.9973009446693657,
+      "grad_norm": 0.8067307264795585,
+      "learning_rate": 3.0021792546362332e-05,
+      "loss": 0.3345,
+      "step": 370
+    },
+    {
+      "epoch": 2.002699055330634,
+      "grad_norm": 1.879871601659991,
+      "learning_rate": 2.9956367415767657e-05,
+      "loss": 0.5265,
+      "step": 371
+    },
+    {
+      "epoch": 2.0080971659919027,
+      "grad_norm": 0.9492101152184929,
+      "learning_rate": 2.9890800329258554e-05,
+      "loss": 0.1471,
+      "step": 372
+    },
+    {
+      "epoch": 2.0134952766531713,
+      "grad_norm": 0.6102897197309013,
+      "learning_rate": 2.982509222167755e-05,
+      "loss": 0.1186,
+      "step": 373
+    },
+    {
+      "epoch": 2.01889338731444,
+      "grad_norm": 0.708621483290416,
+      "learning_rate": 2.975924402987783e-05,
+      "loss": 0.1269,
+      "step": 374
+    },
+    {
+      "epoch": 2.0242914979757085,
+      "grad_norm": 0.7564499045617382,
+      "learning_rate": 2.969325669270987e-05,
+      "loss": 0.1258,
+      "step": 375
+    },
+    {
+      "epoch": 2.029689608636977,
+      "grad_norm": 0.7375719154959449,
+      "learning_rate": 2.9627131151008046e-05,
+      "loss": 0.1205,
+      "step": 376
+    },
+    {
+      "epoch": 2.0350877192982457,
+      "grad_norm": 0.8730624102839428,
+      "learning_rate": 2.9560868347577235e-05,
+      "loss": 0.1419,
+      "step": 377
+    },
+    {
+      "epoch": 2.0404858299595143,
+      "grad_norm": 1.0659419319620667,
+      "learning_rate": 2.9494469227179375e-05,
+      "loss": 0.2593,
+      "step": 378
+    },
+    {
+      "epoch": 2.045883940620783,
+      "grad_norm": 0.6610071937298646,
+      "learning_rate": 2.9427934736519962e-05,
+      "loss": 0.1263,
+      "step": 379
+    },
+    {
+      "epoch": 2.051282051282051,
+      "grad_norm": 0.6137997704208841,
+      "learning_rate": 2.9361265824234575e-05,
+      "loss": 0.1198,
+      "step": 380
+    },
+    {
+      "epoch": 2.0566801619433197,
+      "grad_norm": 0.7021778207773588,
+      "learning_rate": 2.9294463440875375e-05,
+      "loss": 0.1469,
+      "step": 381
+    },
+    {
+      "epoch": 2.0620782726045883,
+      "grad_norm": 0.6404458024178316,
+      "learning_rate": 2.9227528538897503e-05,
+      "loss": 0.1742,
+      "step": 382
+    },
+    {
+      "epoch": 2.067476383265857,
+      "grad_norm": 0.6191203308201093,
+      "learning_rate": 2.916046207264554e-05,
+      "loss": 0.1248,
+      "step": 383
+    },
+    {
+      "epoch": 2.0728744939271255,
+      "grad_norm": 0.6083615788044857,
+      "learning_rate": 2.9093264998339875e-05,
+      "loss": 0.1252,
+      "step": 384
+    },
+    {
+      "epoch": 2.078272604588394,
+      "grad_norm": 0.6908299717391188,
+      "learning_rate": 2.9025938274063077e-05,
+      "loss": 0.153,
+      "step": 385
+    },
+    {
+      "epoch": 2.0836707152496627,
+      "grad_norm": 0.7567808418973763,
+      "learning_rate": 2.8958482859746242e-05,
+      "loss": 0.1863,
+      "step": 386
+    },
+    {
+      "epoch": 2.0890688259109313,
+      "grad_norm": 0.6523366424540843,
+      "learning_rate": 2.889089971715532e-05,
+      "loss": 0.1354,
+      "step": 387
+    },
+    {
+      "epoch": 2.0944669365722,
+      "grad_norm": 0.6756790959278772,
+      "learning_rate": 2.8823189809877365e-05,
+      "loss": 0.1313,
+      "step": 388
+    },
+    {
+      "epoch": 2.099865047233468,
+      "grad_norm": 0.6078081219040387,
+      "learning_rate": 2.8755354103306808e-05,
+      "loss": 0.2153,
+      "step": 389
+    },
+    {
+      "epoch": 2.1052631578947367,
+      "grad_norm": 1.0167045690969643,
+      "learning_rate": 2.868739356463174e-05,
+      "loss": 0.1251,
+      "step": 390
+    },
+    {
+      "epoch": 2.1106612685560053,
+      "grad_norm": 0.6967631702757965,
+      "learning_rate": 2.8619309162820048e-05,
+      "loss": 0.1293,
+      "step": 391
+    },
+    {
+      "epoch": 2.116059379217274,
+      "grad_norm": 0.6759765116278269,
+      "learning_rate": 2.8551101868605644e-05,
+      "loss": 0.2055,
+      "step": 392
+    },
+    {
+      "epoch": 2.1214574898785425,
+      "grad_norm": 0.6811842481145785,
+      "learning_rate": 2.8482772654474624e-05,
+      "loss": 0.2664,
+      "step": 393
+    },
+    {
+      "epoch": 2.126855600539811,
+      "grad_norm": 0.6345308055561072,
+      "learning_rate": 2.841432249465138e-05,
+      "loss": 0.1222,
+      "step": 394
+    },
+    {
+      "epoch": 2.1322537112010798,
+      "grad_norm": 0.619234065227691,
+      "learning_rate": 2.8345752365084744e-05,
+      "loss": 0.1256,
+      "step": 395
+    },
+    {
+      "epoch": 2.1376518218623484,
+      "grad_norm": 0.6325904065714351,
+      "learning_rate": 2.8277063243434036e-05,
+      "loss": 0.1274,
+      "step": 396
+    },
+    {
+      "epoch": 2.1430499325236165,
+      "grad_norm": 1.3888097472274337,
+      "learning_rate": 2.820825610905514e-05,
+      "loss": 0.1428,
+      "step": 397
+    },
+    {
+      "epoch": 2.148448043184885,
+      "grad_norm": 0.6574983703756084,
+      "learning_rate": 2.8139331942986552e-05,
+      "loss": 0.1187,
+      "step": 398
+    },
+    {
+      "epoch": 2.1538461538461537,
+      "grad_norm": 0.6676000331741341,
+      "learning_rate": 2.8070291727935385e-05,
+      "loss": 0.1221,
+      "step": 399
+    },
+    {
+      "epoch": 2.1592442645074224,
+      "grad_norm": 0.5845820509379751,
+      "learning_rate": 2.8001136448263345e-05,
+      "loss": 0.1104,
+      "step": 400
+    },
+    {
+      "epoch": 2.164642375168691,
+      "grad_norm": 0.6285772845826897,
+      "learning_rate": 2.7931867089972703e-05,
+      "loss": 0.2991,
+      "step": 401
+    },
+    {
+      "epoch": 2.1700404858299596,
+      "grad_norm": 0.622942079890072,
+      "learning_rate": 2.7862484640692265e-05,
+      "loss": 0.1268,
+      "step": 402
+    },
+    {
+      "epoch": 2.175438596491228,
+      "grad_norm": 0.7258077146788737,
+      "learning_rate": 2.7792990089663244e-05,
+      "loss": 0.2378,
+      "step": 403
+    },
+    {
+      "epoch": 2.180836707152497,
+      "grad_norm": 0.6729130160854355,
+      "learning_rate": 2.772338442772517e-05,
+      "loss": 0.1285,
+      "step": 404
+    },
+    {
+      "epoch": 2.1862348178137654,
+      "grad_norm": 0.6993929265761061,
+      "learning_rate": 2.7653668647301797e-05,
+      "loss": 0.1418,
+      "step": 405
+    },
+    {
+      "epoch": 2.1916329284750335,
+      "grad_norm": 0.6367301270364276,
+      "learning_rate": 2.758384374238691e-05,
+      "loss": 0.1185,
+      "step": 406
+    },
+    {
+      "epoch": 2.197031039136302,
+      "grad_norm": 0.6650512164895968,
+      "learning_rate": 2.751391070853017e-05,
+      "loss": 0.1255,
+      "step": 407
+    },
+    {
+      "epoch": 2.2024291497975708,
+      "grad_norm": 0.6726822863599137,
+      "learning_rate": 2.744387054282293e-05,
+      "loss": 0.1401,
+      "step": 408
+    },
+    {
+      "epoch": 2.2078272604588394,
+      "grad_norm": 0.6322927923442171,
+      "learning_rate": 2.737372424388398e-05,
+      "loss": 0.2242,
+      "step": 409
+    },
+    {
+      "epoch": 2.213225371120108,
+      "grad_norm": 0.7139601265211752,
+      "learning_rate": 2.7303472811845373e-05,
+      "loss": 0.1276,
+      "step": 410
+    },
+    {
+      "epoch": 2.2186234817813766,
+      "grad_norm": 0.6840865587063936,
+      "learning_rate": 2.72331172483381e-05,
+      "loss": 0.1677,
+      "step": 411
+    },
+    {
+      "epoch": 2.224021592442645,
+      "grad_norm": 0.586515521368017,
+      "learning_rate": 2.7162658556477856e-05,
+      "loss": 0.1309,
+      "step": 412
+    },
+    {
+      "epoch": 2.229419703103914,
+      "grad_norm": 0.6652777071072453,
+      "learning_rate": 2.7092097740850712e-05,
+      "loss": 0.1278,
+      "step": 413
+    },
+    {
+      "epoch": 2.234817813765182,
+      "grad_norm": 0.5965513260042465,
+      "learning_rate": 2.7021435807498803e-05,
+      "loss": 0.1184,
+      "step": 414
+    },
+    {
+      "epoch": 2.2402159244264506,
+      "grad_norm": 0.6168490779643224,
+      "learning_rate": 2.6950673763905976e-05,
+      "loss": 0.1297,
+      "step": 415
+    },
+    {
+      "epoch": 2.245614035087719,
+      "grad_norm": 0.8196897290258232,
+      "learning_rate": 2.6879812618983427e-05,
+      "loss": 0.156,
+      "step": 416
+    },
+    {
+      "epoch": 2.251012145748988,
+      "grad_norm": 0.6574593517521166,
+      "learning_rate": 2.680885338305532e-05,
+      "loss": 0.1243,
+      "step": 417
+    },
+    {
+      "epoch": 2.2564102564102564,
+      "grad_norm": 0.5734169523498336,
+      "learning_rate": 2.6737797067844403e-05,
+      "loss": 0.1085,
+      "step": 418
+    },
+    {
+      "epoch": 2.261808367071525,
+      "grad_norm": 0.6486922562667219,
+      "learning_rate": 2.6666644686457527e-05,
+      "loss": 0.2638,
+      "step": 419
+    },
+    {
+      "epoch": 2.2672064777327936,
+      "grad_norm": 0.6085204978403612,
+      "learning_rate": 2.6595397253371263e-05,
+      "loss": 0.1167,
+      "step": 420
+    },
+    {
+      "epoch": 2.272604588394062,
+      "grad_norm": 0.6506248265768584,
+      "learning_rate": 2.652405578441739e-05,
+      "loss": 0.1222,
+      "step": 421
+    },
+    {
+      "epoch": 2.278002699055331,
+      "grad_norm": 0.7121470667213897,
+      "learning_rate": 2.6452621296768444e-05,
+      "loss": 0.1528,
+      "step": 422
+    },
+    {
+      "epoch": 2.283400809716599,
+      "grad_norm": 0.6497799274343271,
+      "learning_rate": 2.63810948089232e-05,
+      "loss": 0.1334,
+      "step": 423
+    },
+    {
+      "epoch": 2.2887989203778676,
+      "grad_norm": 0.6396475010884002,
+      "learning_rate": 2.630947734069216e-05,
+      "loss": 0.1243,
+      "step": 424
+    },
+    {
+      "epoch": 2.294197031039136,
+      "grad_norm": 0.5821855633698776,
+      "learning_rate": 2.623776991318298e-05,
+      "loss": 0.1072,
+      "step": 425
+    },
+    {
+      "epoch": 2.299595141700405,
+      "grad_norm": 0.5914718298060722,
+      "learning_rate": 2.6165973548785985e-05,
+      "loss": 0.1353,
+      "step": 426
+    },
+    {
+      "epoch": 2.3049932523616734,
+      "grad_norm": 0.5706413579516998,
+      "learning_rate": 2.6094089271159497e-05,
+      "loss": 0.107,
+      "step": 427
+    },
+    {
+      "epoch": 2.310391363022942,
+      "grad_norm": 0.6462522629941394,
+      "learning_rate": 2.6022118105215314e-05,
+      "loss": 0.1224,
+      "step": 428
+    },
+    {
+      "epoch": 2.3157894736842106,
+      "grad_norm": 0.6643268530758245,
+      "learning_rate": 2.595006107710406e-05,
+      "loss": 0.1315,
+      "step": 429
+    },
+    {
+      "epoch": 2.3211875843454792,
+      "grad_norm": 0.5483753514158015,
+      "learning_rate": 2.5877919214200575e-05,
+      "loss": 0.1082,
+      "step": 430
+    },
+    {
+      "epoch": 2.326585695006748,
+      "grad_norm": 0.6224759778845899,
+      "learning_rate": 2.580569354508925e-05,
+      "loss": 0.1221,
+      "step": 431
+    },
+    {
+      "epoch": 2.331983805668016,
+      "grad_norm": 0.7339251866382231,
+      "learning_rate": 2.5733385099549365e-05,
+      "loss": 0.1485,
+      "step": 432
+    },
+    {
+      "epoch": 2.3373819163292846,
+      "grad_norm": 0.6162937360533036,
+      "learning_rate": 2.566099490854041e-05,
+      "loss": 0.1177,
+      "step": 433
+    },
+    {
+      "epoch": 2.342780026990553,
+      "grad_norm": 0.8019402799911678,
+      "learning_rate": 2.5588524004187395e-05,
+      "loss": 0.1408,
+      "step": 434
+    },
+    {
+      "epoch": 2.348178137651822,
+      "grad_norm": 0.6500650250390327,
+      "learning_rate": 2.5515973419766117e-05,
+      "loss": 0.3431,
+      "step": 435
+    },
+    {
+      "epoch": 2.3535762483130904,
+      "grad_norm": 0.7639880727710019,
+      "learning_rate": 2.5443344189688442e-05,
+      "loss": 0.2162,
+      "step": 436
+    },
+    {
+      "epoch": 2.358974358974359,
+      "grad_norm": 0.634855618923187,
+      "learning_rate": 2.5370637349487537e-05,
+      "loss": 0.1327,
+      "step": 437
+    },
+    {
+      "epoch": 2.3643724696356276,
+      "grad_norm": 0.6549886936910365,
+      "learning_rate": 2.5297853935803134e-05,
+      "loss": 0.1399,
+      "step": 438
+    },
+    {
+      "epoch": 2.3697705802968962,
+      "grad_norm": 0.6151850437919503,
+      "learning_rate": 2.522499498636673e-05,
+      "loss": 0.1231,
+      "step": 439
+    },
+    {
+      "epoch": 2.375168690958165,
+      "grad_norm": 0.7032621813927611,
+      "learning_rate": 2.5152061539986786e-05,
+      "loss": 0.1245,
+      "step": 440
+    },
+    {
+      "epoch": 2.380566801619433,
+      "grad_norm": 0.6457747219764639,
+      "learning_rate": 2.5079054636533943e-05,
+      "loss": 0.1388,
+      "step": 441
+    },
+    {
+      "epoch": 2.3859649122807016,
+      "grad_norm": 0.6062768562560231,
+      "learning_rate": 2.5005975316926155e-05,
+      "loss": 0.1073,
+      "step": 442
+    },
+    {
+      "epoch": 2.3913630229419702,
+      "grad_norm": 0.6415372015530586,
+      "learning_rate": 2.4932824623113904e-05,
+      "loss": 0.1631,
+      "step": 443
+    },
+    {
+      "epoch": 2.396761133603239,
+      "grad_norm": 0.573880476798231,
+      "learning_rate": 2.485960359806528e-05,
+      "loss": 0.1998,
+      "step": 444
+    },
+    {
+      "epoch": 2.4021592442645074,
+      "grad_norm": 0.6781677388644582,
+      "learning_rate": 2.4786313285751158e-05,
+      "loss": 0.1504,
+      "step": 445
+    },
+    {
+      "epoch": 2.407557354925776,
+      "grad_norm": 0.6437900937327667,
+      "learning_rate": 2.4712954731130295e-05,
+      "loss": 0.163,
+      "step": 446
+    },
+    {
+      "epoch": 2.4129554655870447,
+      "grad_norm": 0.6588269585671177,
+      "learning_rate": 2.4639528980134424e-05,
+      "loss": 0.1274,
+      "step": 447
+    },
+    {
+      "epoch": 2.4183535762483133,
+      "grad_norm": 0.6007239148865836,
+      "learning_rate": 2.4566037079653372e-05,
+      "loss": 0.1083,
+      "step": 448
+    },
+    {
+      "epoch": 2.423751686909582,
+      "grad_norm": 0.5952155204313362,
+      "learning_rate": 2.4492480077520084e-05,
+      "loss": 0.1192,
+      "step": 449
+    },
+    {
+      "epoch": 2.42914979757085,
+      "grad_norm": 0.6626867675379507,
+      "learning_rate": 2.441885902249573e-05,
+      "loss": 0.121,
+      "step": 450
+    },
+    {
+      "epoch": 2.4345479082321186,
+      "grad_norm": 0.6080214134317363,
+      "learning_rate": 2.434517496425474e-05,
+      "loss": 0.1126,
+      "step": 451
+    },
+    {
+      "epoch": 2.4399460188933872,
+      "grad_norm": 0.6599813084853793,
+      "learning_rate": 2.4271428953369805e-05,
+      "loss": 0.3167,
+      "step": 452
+    },
+    {
+      "epoch": 2.445344129554656,
+      "grad_norm": 0.6977039302218991,
+      "learning_rate": 2.419762204129695e-05,
+      "loss": 0.1342,
+      "step": 453
+    },
+    {
+      "epoch": 2.4507422402159245,
+      "grad_norm": 0.5622313512053422,
+      "learning_rate": 2.412375528036051e-05,
+      "loss": 0.1082,
+      "step": 454
+    },
+    {
+      "epoch": 2.456140350877193,
+      "grad_norm": 0.686695291821724,
+      "learning_rate": 2.4049829723738127e-05,
+      "loss": 0.1294,
+      "step": 455
+    },
+    {
+      "epoch": 2.4615384615384617,
+      "grad_norm": 0.5834700201658436,
+      "learning_rate": 2.3975846425445745e-05,
+      "loss": 0.1067,
+      "step": 456
+    },
+    {
+      "epoch": 2.4669365721997303,
+      "grad_norm": 0.6058986941820341,
+      "learning_rate": 2.390180644032257e-05,
+      "loss": 0.1168,
+      "step": 457
+    },
+    {
+      "epoch": 2.472334682860999,
+      "grad_norm": 0.6116066930817056,
+      "learning_rate": 2.382771082401605e-05,
+      "loss": 0.1131,
+      "step": 458
+    },
+    {
+      "epoch": 2.477732793522267,
+      "grad_norm": 0.6143580414867089,
+      "learning_rate": 2.3753560632966803e-05,
+      "loss": 0.2727,
+      "step": 459
+    },
+    {
+      "epoch": 2.4831309041835357,
+      "grad_norm": 0.6842933583990936,
+      "learning_rate": 2.367935692439357e-05,
+      "loss": 0.1295,
+      "step": 460
+    },
+    {
+      "epoch": 2.4885290148448043,
+      "grad_norm": 0.6126900645614165,
+      "learning_rate": 2.360510075627812e-05,
+      "loss": 0.1244,
+      "step": 461
+    },
+    {
+      "epoch": 2.493927125506073,
+      "grad_norm": 0.6164405026646849,
+      "learning_rate": 2.3530793187350196e-05,
+      "loss": 0.1176,
+      "step": 462
+    },
+    {
+      "epoch": 2.4993252361673415,
+      "grad_norm": 0.5884638793025013,
+      "learning_rate": 2.3456435277072392e-05,
+      "loss": 0.1122,
+      "step": 463
+    },
+    {
+      "epoch": 2.50472334682861,
+      "grad_norm": 0.5861711395789297,
+      "learning_rate": 2.3382028085625074e-05,
+      "loss": 0.1157,
+      "step": 464
+    },
+    {
+      "epoch": 2.5101214574898787,
+      "grad_norm": 0.5674154026944289,
+      "learning_rate": 2.3307572673891227e-05,
+      "loss": 0.1137,
+      "step": 465
+    },
+    {
+      "epoch": 2.515519568151147,
+      "grad_norm": 0.6561990533529527,
+      "learning_rate": 2.3233070103441372e-05,
+      "loss": 0.1245,
+      "step": 466
+    },
+    {
+      "epoch": 2.520917678812416,
+      "grad_norm": 0.5974437314880152,
+      "learning_rate": 2.3158521436518395e-05,
+      "loss": 0.1324,
+      "step": 467
+    },
+    {
+      "epoch": 2.526315789473684,
+      "grad_norm": 0.595231195490965,
+      "learning_rate": 2.3083927736022423e-05,
+      "loss": 0.1143,
+      "step": 468
+    },
+    {
+      "epoch": 2.5317139001349527,
+      "grad_norm": 0.6233086956383334,
+      "learning_rate": 2.3009290065495663e-05,
+      "loss": 0.1152,
+      "step": 469
+    },
+    {
+      "epoch": 2.5371120107962213,
+      "grad_norm": 0.6114526380121206,
+      "learning_rate": 2.2934609489107236e-05,
+      "loss": 0.1121,
+      "step": 470
+    },
+    {
+      "epoch": 2.54251012145749,
+      "grad_norm": 0.6110443440735952,
+      "learning_rate": 2.285988707163801e-05,
+      "loss": 0.1077,
+      "step": 471
+    },
+    {
+      "epoch": 2.5479082321187585,
+      "grad_norm": 0.6275400248619701,
+      "learning_rate": 2.2785123878465414e-05,
+      "loss": 0.1135,
+      "step": 472
+    },
+    {
+      "epoch": 2.553306342780027,
+      "grad_norm": 0.6967325937259534,
+      "learning_rate": 2.2710320975548238e-05,
+      "loss": 0.1283,
+      "step": 473
+    },
+    {
+      "epoch": 2.5587044534412957,
+      "grad_norm": 0.6038681175132602,
+      "learning_rate": 2.2635479429411463e-05,
+      "loss": 0.1155,
+      "step": 474
+    },
+    {
+      "epoch": 2.564102564102564,
+      "grad_norm": 0.6769688106373297,
+      "learning_rate": 2.256060030713102e-05,
+      "loss": 0.1298,
+      "step": 475
+    },
+    {
+      "epoch": 2.569500674763833,
+      "grad_norm": 0.6423702990126002,
+      "learning_rate": 2.248568467631862e-05,
+      "loss": 0.2641,
+      "step": 476
+    },
+    {
+      "epoch": 2.574898785425101,
+      "grad_norm": 0.6622129767608923,
+      "learning_rate": 2.2410733605106462e-05,
+      "loss": 0.1266,
+      "step": 477
+    },
+    {
+      "epoch": 2.5802968960863697,
+      "grad_norm": 0.7600089016209558,
+      "learning_rate": 2.2335748162132093e-05,
+      "loss": 0.1422,
+      "step": 478
+    },
+    {
+      "epoch": 2.5856950067476383,
+      "grad_norm": 0.5830834947808062,
+      "learning_rate": 2.2260729416523103e-05,
+      "loss": 0.1094,
+      "step": 479
+    },
+    {
+      "epoch": 2.591093117408907,
+      "grad_norm": 0.6337156292721836,
+      "learning_rate": 2.2185678437881898e-05,
+      "loss": 0.1148,
+      "step": 480
+    },
+    {
+      "epoch": 2.5964912280701755,
+      "grad_norm": 0.6443855753805541,
+      "learning_rate": 2.2110596296270472e-05,
+      "loss": 0.1167,
+      "step": 481
+    },
+    {
+      "epoch": 2.601889338731444,
+      "grad_norm": 0.6586716116184356,
+      "learning_rate": 2.2035484062195124e-05,
+      "loss": 0.1261,
+      "step": 482
+    },
+    {
+      "epoch": 2.6072874493927127,
+      "grad_norm": 0.6106432005902749,
+      "learning_rate": 2.196034280659122e-05,
+      "loss": 0.1149,
+      "step": 483
+    },
+    {
+      "epoch": 2.612685560053981,
+      "grad_norm": 0.6669027767971515,
+      "learning_rate": 2.188517360080788e-05,
+      "loss": 0.1368,
+      "step": 484
+    },
+    {
+      "epoch": 2.6180836707152495,
+      "grad_norm": 0.5950915576214866,
+      "learning_rate": 2.180997751659276e-05,
+      "loss": 0.1117,
+      "step": 485
+    },
+    {
+      "epoch": 2.623481781376518,
+      "grad_norm": 0.6481155328467912,
+      "learning_rate": 2.1734755626076733e-05,
+      "loss": 0.1346,
+      "step": 486
+    },
+    {
+      "epoch": 2.6288798920377867,
+      "grad_norm": 0.6629611894312477,
+      "learning_rate": 2.1659509001758616e-05,
+      "loss": 0.1286,
+      "step": 487
+    },
+    {
+      "epoch": 2.6342780026990553,
+      "grad_norm": 0.6046987801511539,
+      "learning_rate": 2.158423871648988e-05,
+      "loss": 0.1084,
+      "step": 488
+    },
+    {
+      "epoch": 2.639676113360324,
+      "grad_norm": 0.6902589784360342,
+      "learning_rate": 2.1508945843459322e-05,
+      "loss": 0.1804,
+      "step": 489
+    },
+    {
+      "epoch": 2.6450742240215925,
+      "grad_norm": 0.6943416749369566,
+      "learning_rate": 2.1433631456177823e-05,
+      "loss": 0.3025,
+      "step": 490
+    },
+    {
+      "epoch": 2.650472334682861,
+      "grad_norm": 0.5995490469668333,
+      "learning_rate": 2.1358296628463008e-05,
+      "loss": 0.1139,
+      "step": 491
+    },
+    {
+      "epoch": 2.6558704453441297,
+      "grad_norm": 0.6846769621619672,
+      "learning_rate": 2.1282942434423917e-05,
+      "loss": 0.1434,
+      "step": 492
+    },
+    {
+      "epoch": 2.661268556005398,
+      "grad_norm": 0.5987834015527275,
+      "learning_rate": 2.1207569948445724e-05,
+      "loss": 0.2782,
+      "step": 493
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.6574038992983974,
+      "learning_rate": 2.11321802451744e-05,
+      "loss": 0.1228,
+      "step": 494
+    },
+    {
+      "epoch": 2.672064777327935,
+      "grad_norm": 0.6393995486745645,
+      "learning_rate": 2.1056774399501413e-05,
+      "loss": 0.118,
+      "step": 495
+    },
+    {
+      "epoch": 2.6774628879892037,
+      "grad_norm": 0.5975778741535159,
+      "learning_rate": 2.0981353486548363e-05,
+      "loss": 0.1035,
+      "step": 496
+    },
+    {
+      "epoch": 2.6828609986504723,
+      "grad_norm": 0.644545839847509,
+      "learning_rate": 2.0905918581651692e-05,
+      "loss": 0.1244,
+      "step": 497
+    },
+    {
+      "epoch": 2.688259109311741,
+      "grad_norm": 0.6922625801477104,
+      "learning_rate": 2.0830470760347326e-05,
+      "loss": 0.1406,
+      "step": 498
+    },
+    {
+      "epoch": 2.6936572199730096,
+      "grad_norm": 0.6386404177754778,
+      "learning_rate": 2.0755011098355366e-05,
+      "loss": 0.1246,
+      "step": 499
+    },
+    {
+      "epoch": 2.699055330634278,
+      "grad_norm": 0.6032485076875685,
+      "learning_rate": 2.067954067156472e-05,
+      "loss": 0.1196,
+      "step": 500
+    },
+    {
+      "epoch": 2.7044534412955468,
+      "grad_norm": 0.658988527360155,
+      "learning_rate": 2.060406055601778e-05,
+      "loss": 0.1377,
+      "step": 501
+    },
+    {
+      "epoch": 2.709851551956815,
+      "grad_norm": 0.5989352994619056,
+      "learning_rate": 2.052857182789509e-05,
+      "loss": 0.1056,
+      "step": 502
+    },
+    {
+      "epoch": 2.7152496626180835,
+      "grad_norm": 0.6367876108307421,
+      "learning_rate": 2.0453075563499975e-05,
+      "loss": 0.1106,
+      "step": 503
+    },
+    {
+      "epoch": 2.720647773279352,
+      "grad_norm": 0.7090103740670995,
+      "learning_rate": 2.0377572839243225e-05,
+      "loss": 0.2028,
+      "step": 504
+    },
+    {
+      "epoch": 2.7260458839406208,
+      "grad_norm": 0.583231171229185,
+      "learning_rate": 2.030206473162772e-05,
+      "loss": 0.2519,
+      "step": 505
+    },
+    {
+      "epoch": 2.7314439946018894,
+      "grad_norm": 0.5429732432961625,
+      "learning_rate": 2.022655231723311e-05,
+      "loss": 0.2273,
+      "step": 506
+    },
+    {
+      "epoch": 2.736842105263158,
+      "grad_norm": 0.6753077778506595,
+      "learning_rate": 2.0151036672700437e-05,
+      "loss": 0.1334,
+      "step": 507
+    },
+    {
+      "epoch": 2.7422402159244266,
+      "grad_norm": 0.6176589395883935,
+      "learning_rate": 2.0075518874716797e-05,
+      "loss": 0.1045,
+      "step": 508
+    },
+    {
+      "epoch": 2.7476383265856947,
+      "grad_norm": 0.69994730732916,
+      "learning_rate": 2e-05,
+      "loss": 0.121,
+      "step": 509
+    },
+    {
+      "epoch": 2.753036437246964,
+      "grad_norm": 0.670186877323024,
+      "learning_rate": 1.9924481125283203e-05,
+      "loss": 0.5136,
+      "step": 510
+    },
+    {
+      "epoch": 2.758434547908232,
+      "grad_norm": 0.7091457000671242,
+      "learning_rate": 1.984896332729957e-05,
+      "loss": 0.14,
+      "step": 511
+    },
+    {
+      "epoch": 2.7638326585695006,
+      "grad_norm": 0.6092774087849933,
+      "learning_rate": 1.9773447682766894e-05,
+      "loss": 0.117,
+      "step": 512
+    },
+    {
+      "epoch": 2.769230769230769,
+      "grad_norm": 0.69499655788575,
+      "learning_rate": 1.9697935268372285e-05,
+      "loss": 0.1416,
+      "step": 513
+    },
+    {
+      "epoch": 2.7746288798920378,
+      "grad_norm": 0.611429420887083,
+      "learning_rate": 1.9622427160756778e-05,
+      "loss": 0.1135,
+      "step": 514
+    },
+    {
+      "epoch": 2.7800269905533064,
+      "grad_norm": 0.6303797016311802,
+      "learning_rate": 1.9546924436500028e-05,
+      "loss": 0.1168,
+      "step": 515
+    },
+    {
+      "epoch": 2.785425101214575,
+      "grad_norm": 0.62365852324568,
+      "learning_rate": 1.9471428172104916e-05,
+      "loss": 0.128,
+      "step": 516
+    },
+    {
+      "epoch": 2.7908232118758436,
+      "grad_norm": 0.6404183020004793,
+      "learning_rate": 1.9395939443982228e-05,
+      "loss": 0.2501,
+      "step": 517
+    },
+    {
+      "epoch": 2.7962213225371118,
+      "grad_norm": 0.6292426346258508,
+      "learning_rate": 1.9320459328435287e-05,
+      "loss": 0.1351,
+      "step": 518
+    },
+    {
+      "epoch": 2.801619433198381,
+      "grad_norm": 0.6385941431943223,
+      "learning_rate": 1.924498890164464e-05,
+      "loss": 0.2533,
+      "step": 519
+    },
+    {
+      "epoch": 2.807017543859649,
+      "grad_norm": 0.6593390043611441,
+      "learning_rate": 1.9169529239652678e-05,
+      "loss": 0.2079,
+      "step": 520
+    },
+    {
+      "epoch": 2.8124156545209176,
+      "grad_norm": 0.6208041690966352,
+      "learning_rate": 1.9094081418348318e-05,
+      "loss": 0.1157,
+      "step": 521
+    },
+    {
+      "epoch": 2.817813765182186,
+      "grad_norm": 0.6440545678890086,
+      "learning_rate": 1.901864651345164e-05,
+      "loss": 0.1339,
+      "step": 522
+    },
+    {
+      "epoch": 2.823211875843455,
+      "grad_norm": 0.5936846079230754,
+      "learning_rate": 1.894322560049859e-05,
+      "loss": 0.1242,
+      "step": 523
+    },
+    {
+      "epoch": 2.8286099865047234,
+      "grad_norm": 0.8776608622078234,
+      "learning_rate": 1.8867819754825602e-05,
+      "loss": 0.124,
+      "step": 524
+    },
+    {
+      "epoch": 2.834008097165992,
+      "grad_norm": 0.619866262203856,
+      "learning_rate": 1.879243005155428e-05,
+      "loss": 0.1228,
+      "step": 525
+    },
+    {
+      "epoch": 2.8394062078272606,
+      "grad_norm": 0.6472157396567825,
+      "learning_rate": 1.871705756557609e-05,
+      "loss": 0.2505,
+      "step": 526
+    },
+    {
+      "epoch": 2.8448043184885288,
+      "grad_norm": 0.6097332067194058,
+      "learning_rate": 1.8641703371536995e-05,
+      "loss": 0.1208,
+      "step": 527
+    },
+    {
+      "epoch": 2.850202429149798,
+      "grad_norm": 0.6737703505863861,
+      "learning_rate": 1.856636854382218e-05,
+      "loss": 0.2788,
+      "step": 528
+    },
+    {
+      "epoch": 2.855600539811066,
+      "grad_norm": 0.6219309424006997,
+      "learning_rate": 1.849105415654068e-05,
+      "loss": 0.1149,
+      "step": 529
+    },
+    {
+      "epoch": 2.8609986504723346,
+      "grad_norm": 0.6645437129953545,
+      "learning_rate": 1.841576128351013e-05,
+      "loss": 0.1397,
+      "step": 530
+    },
+    {
+      "epoch": 2.866396761133603,
+      "grad_norm": 0.6352886672687073,
+      "learning_rate": 1.8340490998241387e-05,
+      "loss": 0.1195,
+      "step": 531
+    },
+    {
+      "epoch": 2.871794871794872,
+      "grad_norm": 0.5667013233035931,
+      "learning_rate": 1.8265244373923274e-05,
+      "loss": 0.1046,
+      "step": 532
+    },
+    {
+      "epoch": 2.8771929824561404,
+      "grad_norm": 0.609701874237383,
+      "learning_rate": 1.8190022483407246e-05,
+      "loss": 0.1095,
+      "step": 533
+    },
+    {
+      "epoch": 2.882591093117409,
+      "grad_norm": 0.6667574173600572,
+      "learning_rate": 1.8114826399192133e-05,
+      "loss": 0.1188,
+      "step": 534
+    },
+    {
+      "epoch": 2.8879892037786776,
+      "grad_norm": 0.6261633444258052,
+      "learning_rate": 1.8039657193408788e-05,
+      "loss": 0.1208,
+      "step": 535
+    },
+    {
+      "epoch": 2.893387314439946,
+      "grad_norm": 0.557002161441068,
+      "learning_rate": 1.7964515937804875e-05,
+      "loss": 0.1051,
+      "step": 536
+    },
+    {
+      "epoch": 2.898785425101215,
+      "grad_norm": 0.5880503166312527,
+      "learning_rate": 1.7889403703729535e-05,
+      "loss": 0.1034,
+      "step": 537
+    },
+    {
+      "epoch": 2.904183535762483,
+      "grad_norm": 0.6386062087365171,
+      "learning_rate": 1.7814321562118105e-05,
+      "loss": 0.1207,
+      "step": 538
+    },
+    {
+      "epoch": 2.9095816464237516,
+      "grad_norm": 0.6281649321187761,
+      "learning_rate": 1.7739270583476904e-05,
+      "loss": 0.1127,
+      "step": 539
+    },
+    {
+      "epoch": 2.91497975708502,
+      "grad_norm": 0.6903441120949023,
+      "learning_rate": 1.766425183786791e-05,
+      "loss": 0.133,
+      "step": 540
+    },
+    {
+      "epoch": 2.920377867746289,
+      "grad_norm": 0.594699443176023,
+      "learning_rate": 1.758926639489354e-05,
+      "loss": 0.1126,
+      "step": 541
+    },
+    {
+      "epoch": 2.9257759784075574,
+      "grad_norm": 0.6534217265809632,
+      "learning_rate": 1.7514315323681386e-05,
+      "loss": 0.1399,
+      "step": 542
+    },
+    {
+      "epoch": 2.931174089068826,
+      "grad_norm": 0.5933497171590056,
+      "learning_rate": 1.7439399692868985e-05,
+      "loss": 0.1067,
+      "step": 543
+    },
+    {
+      "epoch": 2.9365721997300946,
+      "grad_norm": 0.7129140116302266,
+      "learning_rate": 1.7364520570588543e-05,
+      "loss": 0.1454,
+      "step": 544
+    },
+    {
+      "epoch": 2.941970310391363,
+      "grad_norm": 0.6040208976460203,
+      "learning_rate": 1.7289679024451772e-05,
+      "loss": 0.1051,
+      "step": 545
+    },
+    {
+      "epoch": 2.9473684210526314,
+      "grad_norm": 0.6415659724473479,
+      "learning_rate": 1.7214876121534592e-05,
+      "loss": 0.1002,
+      "step": 546
+    },
+    {
+      "epoch": 2.9527665317139,
+      "grad_norm": 0.6811604589833739,
+      "learning_rate": 1.7140112928361996e-05,
+      "loss": 0.1174,
+      "step": 547
+    },
+    {
+      "epoch": 2.9581646423751686,
+      "grad_norm": 0.6251248979220851,
+      "learning_rate": 1.7065390510892767e-05,
+      "loss": 0.1107,
+      "step": 548
+    },
+    {
+      "epoch": 2.9635627530364372,
+      "grad_norm": 0.6152395267761969,
+      "learning_rate": 1.699070993450434e-05,
+      "loss": 0.0994,
+      "step": 549
+    },
+    {
+      "epoch": 2.968960863697706,
+      "grad_norm": 0.6852482877150339,
+      "learning_rate": 1.6916072263977583e-05,
+      "loss": 0.1739,
+      "step": 550
+    },
+    {
+      "epoch": 2.9743589743589745,
+      "grad_norm": 0.6043771940281845,
+      "learning_rate": 1.6841478563481612e-05,
+      "loss": 0.1007,
+      "step": 551
+    },
+    {
+      "epoch": 2.979757085020243,
+      "grad_norm": 0.7251294070776313,
+      "learning_rate": 1.6766929896558638e-05,
+      "loss": 0.2776,
+      "step": 552
+    },
+    {
+      "epoch": 2.9851551956815117,
+      "grad_norm": 0.7218834500320331,
+      "learning_rate": 1.6692427326108776e-05,
+      "loss": 0.1519,
+      "step": 553
+    },
+    {
+      "epoch": 2.99055330634278,
+      "grad_norm": 0.6277344728773911,
+      "learning_rate": 1.6617971914374933e-05,
+      "loss": 0.2613,
+      "step": 554
+    },
+    {
+      "epoch": 2.9959514170040484,
+      "grad_norm": 0.6071469083911657,
+      "learning_rate": 1.654356472292761e-05,
+      "loss": 0.1127,
+      "step": 555
+    },
+    {
+      "epoch": 3.001349527665317,
+      "grad_norm": 1.7833210073385368,
+      "learning_rate": 1.6469206812649818e-05,
+      "loss": 0.2129,
+      "step": 556
+    },
+    {
+      "epoch": 3.0067476383265856,
+      "grad_norm": 0.32064299423776677,
+      "learning_rate": 1.6394899243721887e-05,
+      "loss": 0.0326,
+      "step": 557
+    },
+    {
+      "epoch": 3.0121457489878543,
+      "grad_norm": 0.35456048045323174,
+      "learning_rate": 1.632064307560644e-05,
+      "loss": 0.0343,
+      "step": 558
+    },
+    {
+      "epoch": 3.017543859649123,
+      "grad_norm": 0.3460859489223203,
+      "learning_rate": 1.6246439367033207e-05,
+      "loss": 0.0386,
+      "step": 559
+    },
+    {
+      "epoch": 3.0229419703103915,
+      "grad_norm": 0.34350308372675437,
+      "learning_rate": 1.617228917598396e-05,
+      "loss": 0.0337,
+      "step": 560
+    },
+    {
+      "epoch": 3.02834008097166,
+      "grad_norm": 0.33249338844321885,
+      "learning_rate": 1.609819355967744e-05,
+      "loss": 0.0325,
+      "step": 561
+    },
+    {
+      "epoch": 3.0337381916329287,
+      "grad_norm": 0.5631709206394763,
+      "learning_rate": 1.602415357455426e-05,
+      "loss": 0.054,
+      "step": 562
+    },
+    {
+      "epoch": 3.039136302294197,
+      "grad_norm": 0.40192112119976586,
+      "learning_rate": 1.5950170276261876e-05,
+      "loss": 0.0373,
+      "step": 563
+    },
+    {
+      "epoch": 3.0445344129554655,
+      "grad_norm": 0.4126502021158313,
+      "learning_rate": 1.587624471963949e-05,
+      "loss": 0.0373,
+      "step": 564
+    },
+    {
+      "epoch": 3.049932523616734,
+      "grad_norm": 0.43546597329447,
+      "learning_rate": 1.5802377958703054e-05,
+      "loss": 0.038,
+      "step": 565
+    },
+    {
+      "epoch": 3.0553306342780027,
+      "grad_norm": 0.3780585408034151,
+      "learning_rate": 1.5728571046630195e-05,
+      "loss": 0.0309,
+      "step": 566
+    },
+    {
+      "epoch": 3.0607287449392713,
+      "grad_norm": 0.5397727494180338,
+      "learning_rate": 1.5654825035745268e-05,
+      "loss": 0.0689,
+      "step": 567
+    },
+    {
+      "epoch": 3.06612685560054,
+      "grad_norm": 0.4007214251036579,
+      "learning_rate": 1.5581140977504273e-05,
+      "loss": 0.035,
+      "step": 568
+    },
+    {
+      "epoch": 3.0715249662618085,
+      "grad_norm": 0.5090009004527475,
+      "learning_rate": 1.550751992247993e-05,
+      "loss": 0.0368,
+      "step": 569
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 0.42083424924403967,
+      "learning_rate": 1.5433962920346638e-05,
+      "loss": 0.034,
+      "step": 570
+    },
+    {
+      "epoch": 3.0823211875843457,
+      "grad_norm": 0.37809287175104134,
+      "learning_rate": 1.5360471019865583e-05,
+      "loss": 0.0291,
+      "step": 571
+    },
+    {
+      "epoch": 3.087719298245614,
+      "grad_norm": 0.43804777109347015,
+      "learning_rate": 1.5287045268869712e-05,
+      "loss": 0.0353,
+      "step": 572
+    },
+    {
+      "epoch": 3.0931174089068825,
+      "grad_norm": 0.47296403917832164,
+      "learning_rate": 1.5213686714248852e-05,
+      "loss": 0.0361,
+      "step": 573
+    },
+    {
+      "epoch": 3.098515519568151,
+      "grad_norm": 0.4100358348674376,
+      "learning_rate": 1.5140396401934725e-05,
+      "loss": 0.0294,
+      "step": 574
+    },
+    {
+      "epoch": 3.1039136302294197,
+      "grad_norm": 0.3626521943432172,
+      "learning_rate": 1.5067175376886099e-05,
+      "loss": 0.0271,
+      "step": 575
+    },
+    {
+      "epoch": 3.1093117408906883,
+      "grad_norm": 0.4325512783668008,
+      "learning_rate": 1.4994024683073848e-05,
+      "loss": 0.042,
+      "step": 576
+    },
+    {
+      "epoch": 3.114709851551957,
+      "grad_norm": 0.38243345056118383,
+      "learning_rate": 1.4920945363466062e-05,
+      "loss": 0.0301,
+      "step": 577
+    },
+    {
+      "epoch": 3.1201079622132255,
+      "grad_norm": 0.47125765412019066,
+      "learning_rate": 1.4847938460013219e-05,
+      "loss": 0.2469,
+      "step": 578
+    },
+    {
+      "epoch": 3.125506072874494,
+      "grad_norm": 0.38199687663330195,
+      "learning_rate": 1.4775005013633277e-05,
+      "loss": 0.0315,
+      "step": 579
+    },
+    {
+      "epoch": 3.1309041835357623,
+      "grad_norm": 0.40607226917742406,
+      "learning_rate": 1.4702146064196875e-05,
+      "loss": 0.0348,
+      "step": 580
+    },
+    {
+      "epoch": 3.136302294197031,
+      "grad_norm": 0.3882898487153505,
+      "learning_rate": 1.4629362650512464e-05,
+      "loss": 0.0371,
+      "step": 581
+    },
+    {
+      "epoch": 3.1417004048582995,
+      "grad_norm": 0.39251002263194035,
+      "learning_rate": 1.4556655810311566e-05,
+      "loss": 0.1615,
+      "step": 582
+    },
+    {
+      "epoch": 3.147098515519568,
+      "grad_norm": 0.4403240904362223,
+      "learning_rate": 1.4484026580233888e-05,
+      "loss": 0.0323,
+      "step": 583
+    },
+    {
+      "epoch": 3.1524966261808367,
+      "grad_norm": 0.34157066422446297,
+      "learning_rate": 1.4411475995812613e-05,
+      "loss": 0.0275,
+      "step": 584
+    },
+    {
+      "epoch": 3.1578947368421053,
+      "grad_norm": 0.3706904233443787,
+      "learning_rate": 1.4339005091459595e-05,
+      "loss": 0.0839,
+      "step": 585
+    },
+    {
+      "epoch": 3.163292847503374,
+      "grad_norm": 0.3560304765714981,
+      "learning_rate": 1.4266614900450645e-05,
+      "loss": 0.0295,
+      "step": 586
+    },
+    {
+      "epoch": 3.1686909581646425,
+      "grad_norm": 0.40503905536132107,
+      "learning_rate": 1.4194306454910757e-05,
+      "loss": 0.0431,
+      "step": 587
+    },
+    {
+      "epoch": 3.174089068825911,
+      "grad_norm": 0.3763838962823063,
+      "learning_rate": 1.4122080785799423e-05,
+      "loss": 0.033,
+      "step": 588
+    },
+    {
+      "epoch": 3.1794871794871793,
+      "grad_norm": 0.5153439226411922,
+      "learning_rate": 1.4049938922895945e-05,
+      "loss": 0.1264,
+      "step": 589
+    },
+    {
+      "epoch": 3.184885290148448,
+      "grad_norm": 0.4386948021509455,
+      "learning_rate": 1.3977881894784689e-05,
+      "loss": 0.1795,
+      "step": 590
+    },
+    {
+      "epoch": 3.1902834008097165,
+      "grad_norm": 0.38978579887041104,
+      "learning_rate": 1.390591072884051e-05,
+      "loss": 0.0302,
+      "step": 591
+    },
+    {
+      "epoch": 3.195681511470985,
+      "grad_norm": 0.39057096403698416,
+      "learning_rate": 1.383402645121402e-05,
+      "loss": 0.0303,
+      "step": 592
+    },
+    {
+      "epoch": 3.2010796221322537,
+      "grad_norm": 0.38395107864965483,
+      "learning_rate": 1.3762230086817024e-05,
+      "loss": 0.0297,
+      "step": 593
+    },
+    {
+      "epoch": 3.2064777327935223,
+      "grad_norm": 0.4335413534746179,
+      "learning_rate": 1.3690522659307846e-05,
+      "loss": 0.1675,
+      "step": 594
+    },
+    {
+      "epoch": 3.211875843454791,
+      "grad_norm": 0.4402590248785823,
+      "learning_rate": 1.3618905191076806e-05,
+      "loss": 0.0433,
+      "step": 595
+    },
+    {
+      "epoch": 3.2172739541160595,
+      "grad_norm": 0.36150571708276924,
+      "learning_rate": 1.3547378703231559e-05,
+      "loss": 0.0271,
+      "step": 596
+    },
+    {
+      "epoch": 3.2226720647773277,
+      "grad_norm": 0.3979936122911017,
+      "learning_rate": 1.3475944215582619e-05,
+      "loss": 0.0367,
+      "step": 597
+    },
+    {
+      "epoch": 3.2280701754385963,
+      "grad_norm": 0.4494705221787137,
+      "learning_rate": 1.340460274662874e-05,
+      "loss": 0.0427,
+      "step": 598
+    },
+    {
+      "epoch": 3.233468286099865,
+      "grad_norm": 0.4143799952810919,
+      "learning_rate": 1.3333355313542478e-05,
+      "loss": 0.0309,
+      "step": 599
+    },
+    {
+      "epoch": 3.2388663967611335,
+      "grad_norm": 0.3759757858796237,
+      "learning_rate": 1.3262202932155602e-05,
+      "loss": 0.0353,
+      "step": 600
+    },
+    {
+      "epoch": 3.244264507422402,
+      "grad_norm": 0.37209149960115473,
+      "learning_rate": 1.3191146616944676e-05,
+      "loss": 0.029,
+      "step": 601
+    },
+    {
+      "epoch": 3.2496626180836707,
+      "grad_norm": 0.380931247404278,
+      "learning_rate": 1.312018738101658e-05,
+      "loss": 0.0329,
+      "step": 602
+    },
+    {
+      "epoch": 3.2550607287449393,
+      "grad_norm": 0.47402538355936785,
+      "learning_rate": 1.304932623609403e-05,
+      "loss": 0.1415,
+      "step": 603
+    },
+    {
+      "epoch": 3.260458839406208,
+      "grad_norm": 0.37952381899472815,
+      "learning_rate": 1.2978564192501203e-05,
+      "loss": 0.0393,
+      "step": 604
+    },
+    {
+      "epoch": 3.2658569500674766,
+      "grad_norm": 0.36605859717517814,
+      "learning_rate": 1.2907902259149287e-05,
+      "loss": 0.0306,
+      "step": 605
+    },
+    {
+      "epoch": 3.2712550607287447,
+      "grad_norm": 0.35158118339250216,
+      "learning_rate": 1.2837341443522147e-05,
+      "loss": 0.0285,
+      "step": 606
+    },
+    {
+      "epoch": 3.2766531713900133,
+      "grad_norm": 0.39607433924431906,
+      "learning_rate": 1.2766882751661905e-05,
+      "loss": 0.0337,
+      "step": 607
+    },
+    {
+      "epoch": 3.282051282051282,
+      "grad_norm": 0.3794346946511997,
+      "learning_rate": 1.2696527188154639e-05,
+      "loss": 0.0324,
+      "step": 608
+    },
+    {
+      "epoch": 3.2874493927125505,
+      "grad_norm": 0.3824042106190188,
+      "learning_rate": 1.2626275756116027e-05,
+      "loss": 0.0291,
+      "step": 609
+    },
+    {
+      "epoch": 3.292847503373819,
+      "grad_norm": 0.455391744225054,
+      "learning_rate": 1.2556129457177084e-05,
+      "loss": 0.036,
+      "step": 610
+    },
+    {
+      "epoch": 3.2982456140350878,
+      "grad_norm": 0.36107306790981053,
+      "learning_rate": 1.2486089291469835e-05,
+      "loss": 0.0309,
+      "step": 611
+    },
+    {
+      "epoch": 3.3036437246963564,
+      "grad_norm": 0.3706080718132542,
+      "learning_rate": 1.2416156257613098e-05,
+      "loss": 0.035,
+      "step": 612
+    },
+    {
+      "epoch": 3.309041835357625,
+      "grad_norm": 0.35100511614676194,
+      "learning_rate": 1.2346331352698206e-05,
+      "loss": 0.0314,
+      "step": 613
+    },
+    {
+      "epoch": 3.3144399460188936,
+      "grad_norm": 0.43435260721693547,
+      "learning_rate": 1.227661557227483e-05,
+      "loss": 0.0435,
+      "step": 614
+    },
+    {
+      "epoch": 3.3198380566801617,
+      "grad_norm": 0.3827631202590933,
+      "learning_rate": 1.2207009910336764e-05,
+      "loss": 0.0339,
+      "step": 615
+    },
+    {
+      "epoch": 3.3252361673414303,
+      "grad_norm": 0.38618564492650564,
+      "learning_rate": 1.2137515359307737e-05,
+      "loss": 0.0332,
+      "step": 616
+    },
+    {
+      "epoch": 3.330634278002699,
+      "grad_norm": 0.3881669840948988,
+      "learning_rate": 1.2068132910027299e-05,
+      "loss": 0.033,
+      "step": 617
+    },
+    {
+      "epoch": 3.3360323886639676,
+      "grad_norm": 0.342705426956321,
+      "learning_rate": 1.1998863551736659e-05,
+      "loss": 0.0267,
+      "step": 618
+    },
+    {
+      "epoch": 3.341430499325236,
+      "grad_norm": 0.3910606294035003,
+      "learning_rate": 1.192970827206462e-05,
+      "loss": 0.0311,
+      "step": 619
+    },
+    {
+      "epoch": 3.3468286099865048,
+      "grad_norm": 0.36847171002316065,
+      "learning_rate": 1.1860668057013451e-05,
+      "loss": 0.0323,
+      "step": 620
+    },
+    {
+      "epoch": 3.3522267206477734,
+      "grad_norm": 0.367842252656434,
+      "learning_rate": 1.1791743890944869e-05,
+      "loss": 0.035,
+      "step": 621
+    },
+    {
+      "epoch": 3.357624831309042,
+      "grad_norm": 0.4600393423512787,
+      "learning_rate": 1.1722936756565969e-05,
+      "loss": 0.2359,
+      "step": 622
+    },
+    {
+      "epoch": 3.3630229419703106,
+      "grad_norm": 0.3576984845035267,
+      "learning_rate": 1.1654247634915261e-05,
+      "loss": 0.032,
+      "step": 623
+    },
+    {
+      "epoch": 3.3684210526315788,
+      "grad_norm": 0.3832267766248835,
+      "learning_rate": 1.1585677505348618e-05,
+      "loss": 0.1769,
+      "step": 624
+    },
+    {
+      "epoch": 3.3738191632928474,
+      "grad_norm": 0.38537223823363365,
+      "learning_rate": 1.151722734552538e-05,
+      "loss": 0.1353,
+      "step": 625
+    },
+    {
+      "epoch": 3.379217273954116,
+      "grad_norm": 0.38423463599004537,
+      "learning_rate": 1.1448898131394364e-05,
+      "loss": 0.0374,
+      "step": 626
+    },
+    {
+      "epoch": 3.3846153846153846,
+      "grad_norm": 0.3507372400046207,
+      "learning_rate": 1.1380690837179955e-05,
+      "loss": 0.0276,
+      "step": 627
+    },
+    {
+      "epoch": 3.390013495276653,
+      "grad_norm": 0.35758020697729664,
+      "learning_rate": 1.1312606435368266e-05,
+      "loss": 0.0295,
+      "step": 628
+    },
+    {
+      "epoch": 3.395411605937922,
+      "grad_norm": 0.32629800377807394,
+      "learning_rate": 1.124464589669319e-05,
+      "loss": 0.0273,
+      "step": 629
+    },
+    {
+      "epoch": 3.4008097165991904,
+      "grad_norm": 0.38394595099713785,
+      "learning_rate": 1.1176810190122644e-05,
+      "loss": 0.0346,
+      "step": 630
+    },
+    {
+      "epoch": 3.406207827260459,
+      "grad_norm": 0.37159451623978973,
+      "learning_rate": 1.110910028284468e-05,
+      "loss": 0.0274,
+      "step": 631
+    },
+    {
+      "epoch": 3.4116059379217276,
+      "grad_norm": 0.38103429546463063,
+      "learning_rate": 1.1041517140253761e-05,
+      "loss": 0.0532,
+      "step": 632
+    },
+    {
+      "epoch": 3.417004048582996,
+      "grad_norm": 0.3605161816760705,
+      "learning_rate": 1.0974061725936935e-05,
+      "loss": 0.0292,
+      "step": 633
+    },
+    {
+      "epoch": 3.4224021592442644,
+      "grad_norm": 0.3313781482523291,
+      "learning_rate": 1.0906735001660138e-05,
+      "loss": 0.0254,
+      "step": 634
+    },
+    {
+      "epoch": 3.427800269905533,
+      "grad_norm": 0.4194498324076769,
+      "learning_rate": 1.0839537927354466e-05,
+      "loss": 0.0386,
+      "step": 635
+    },
+    {
+      "epoch": 3.4331983805668016,
+      "grad_norm": 0.3759933524425867,
+      "learning_rate": 1.0772471461102505e-05,
+      "loss": 0.0256,
+      "step": 636
+    },
+    {
+      "epoch": 3.43859649122807,
+      "grad_norm": 0.349992896262736,
+      "learning_rate": 1.070553655912463e-05,
+      "loss": 0.0272,
+      "step": 637
+    },
+    {
+      "epoch": 3.443994601889339,
+      "grad_norm": 0.39741478649164097,
+      "learning_rate": 1.0638734175765433e-05,
+      "loss": 0.0333,
+      "step": 638
+    },
+    {
+      "epoch": 3.4493927125506074,
+      "grad_norm": 0.38940446539885276,
+      "learning_rate": 1.0572065263480046e-05,
+      "loss": 0.0322,
+      "step": 639
+    },
+    {
+      "epoch": 3.454790823211876,
+      "grad_norm": 0.44353228199656364,
+      "learning_rate": 1.0505530772820625e-05,
+      "loss": 0.0915,
+      "step": 640
+    },
+    {
+      "epoch": 3.4601889338731446,
+      "grad_norm": 0.4220014313707858,
+      "learning_rate": 1.0439131652422763e-05,
+      "loss": 0.2163,
+      "step": 641
+    },
+    {
+      "epoch": 3.465587044534413,
+      "grad_norm": 0.44668294522299473,
+      "learning_rate": 1.037286884899196e-05,
+      "loss": 0.1473,
+      "step": 642
+    },
+    {
+      "epoch": 3.4709851551956814,
+      "grad_norm": 0.43187627571552006,
+      "learning_rate": 1.030674330729014e-05,
+      "loss": 0.0341,
+      "step": 643
+    },
+    {
+      "epoch": 3.47638326585695,
+      "grad_norm": 0.3105938531461725,
+      "learning_rate": 1.0240755970122173e-05,
+      "loss": 0.0215,
+      "step": 644
+    },
+    {
+      "epoch": 3.4817813765182186,
+      "grad_norm": 0.3692698555406238,
+      "learning_rate": 1.0174907778322458e-05,
+      "loss": 0.0276,
+      "step": 645
+    },
+    {
+      "epoch": 3.4871794871794872,
+      "grad_norm": 0.38173730507671527,
+      "learning_rate": 1.0109199670741447e-05,
+      "loss": 0.0331,
+      "step": 646
+    },
+    {
+      "epoch": 3.492577597840756,
+      "grad_norm": 0.4016001432856542,
+      "learning_rate": 1.004363258423235e-05,
+      "loss": 0.2211,
+      "step": 647
+    },
+    {
+      "epoch": 3.4979757085020244,
+      "grad_norm": 0.3353430006657887,
+      "learning_rate": 9.978207453637671e-06,
+      "loss": 0.024,
+      "step": 648
+    },
+    {
+      "epoch": 3.5033738191632926,
+      "grad_norm": 0.38732646655722897,
+      "learning_rate": 9.912925211775989e-06,
+      "loss": 0.0292,
+      "step": 649
+    },
+    {
+      "epoch": 3.5087719298245617,
+      "grad_norm": 0.37360264408486105,
+      "learning_rate": 9.84778678942856e-06,
+      "loss": 0.0332,
+      "step": 650
+    },
+    {
+      "epoch": 3.51417004048583,
+      "grad_norm": 0.4098664770234494,
+      "learning_rate": 9.782793115326112e-06,
+      "loss": 0.0343,
+      "step": 651
+    },
+    {
+      "epoch": 3.5195681511470984,
+      "grad_norm": 0.383720673124196,
+      "learning_rate": 9.717945116135568e-06,
+      "loss": 0.0352,
+      "step": 652
+    },
+    {
+      "epoch": 3.524966261808367,
+      "grad_norm": 0.332454800557511,
+      "learning_rate": 9.653243716446862e-06,
+      "loss": 0.0264,
+      "step": 653
+    },
+    {
+      "epoch": 3.5303643724696356,
+      "grad_norm": 0.35974880920866587,
+      "learning_rate": 9.588689838759744e-06,
+      "loss": 0.0293,
+      "step": 654
+    },
+    {
+      "epoch": 3.5357624831309042,
+      "grad_norm": 0.3557557424993659,
+      "learning_rate": 9.524284403470598e-06,
+      "loss": 0.0271,
+      "step": 655
+    },
+    {
+      "epoch": 3.541160593792173,
+      "grad_norm": 0.34794388980211743,
+      "learning_rate": 9.460028328859379e-06,
+      "loss": 0.0236,
+      "step": 656
+    },
+    {
+      "epoch": 3.5465587044534415,
+      "grad_norm": 0.41538318570062105,
+      "learning_rate": 9.395922531076462e-06,
+      "loss": 0.032,
+      "step": 657
+    },
+    {
+      "epoch": 3.5519568151147096,
+      "grad_norm": 0.34986159330867506,
+      "learning_rate": 9.331967924129615e-06,
+      "loss": 0.025,
+      "step": 658
+    },
+    {
+      "epoch": 3.5573549257759787,
+      "grad_norm": 0.3856553620254245,
+      "learning_rate": 9.268165419870947e-06,
+      "loss": 0.0324,
+      "step": 659
+    },
+    {
+      "epoch": 3.562753036437247,
+      "grad_norm": 0.3969406379925748,
+      "learning_rate": 9.20451592798394e-06,
+      "loss": 0.0288,
+      "step": 660
+    },
+    {
+      "epoch": 3.5681511470985154,
+      "grad_norm": 0.4006671596368767,
+      "learning_rate": 9.141020355970427e-06,
+      "loss": 0.0355,
+      "step": 661
+    },
+    {
+      "epoch": 3.573549257759784,
+      "grad_norm": 0.3940255776794968,
+      "learning_rate": 9.077679609137714e-06,
+      "loss": 0.0299,
+      "step": 662
+    },
+    {
+      "epoch": 3.5789473684210527,
+      "grad_norm": 0.35204668879258955,
+      "learning_rate": 9.014494590585603e-06,
+      "loss": 0.0305,
+      "step": 663
+    },
+    {
+      "epoch": 3.5843454790823213,
+      "grad_norm": 0.40011854490942894,
+      "learning_rate": 8.95146620119359e-06,
+      "loss": 0.1591,
+      "step": 664
+    },
+    {
+      "epoch": 3.58974358974359,
+      "grad_norm": 0.40974210440307884,
+      "learning_rate": 8.888595339607961e-06,
+      "loss": 0.0327,
+      "step": 665
+    },
+    {
+      "epoch": 3.5951417004048585,
+      "grad_norm": 0.3767642702530189,
+      "learning_rate": 8.825882902228998e-06,
+      "loss": 0.0328,
+      "step": 666
+    },
+    {
+      "epoch": 3.6005398110661266,
+      "grad_norm": 0.3646739324385535,
+      "learning_rate": 8.763329783198222e-06,
+      "loss": 0.0304,
+      "step": 667
+    },
+    {
+      "epoch": 3.6059379217273952,
+      "grad_norm": 0.4058623960988426,
+      "learning_rate": 8.700936874385601e-06,
+      "loss": 0.0311,
+      "step": 668
+    },
+    {
+      "epoch": 3.611336032388664,
+      "grad_norm": 0.3881703853633181,
+      "learning_rate": 8.638705065376887e-06,
+      "loss": 0.0311,
+      "step": 669
+    },
+    {
+      "epoch": 3.6167341430499325,
+      "grad_norm": 0.3486816059151549,
+      "learning_rate": 8.576635243460868e-06,
+      "loss": 0.025,
+      "step": 670
+    },
+    {
+      "epoch": 3.622132253711201,
+      "grad_norm": 0.3630098032989897,
+      "learning_rate": 8.514728293616788e-06,
+      "loss": 0.0285,
+      "step": 671
+    },
+    {
+      "epoch": 3.6275303643724697,
+      "grad_norm": 0.3997818117655459,
+      "learning_rate": 8.452985098501659e-06,
+      "loss": 0.0558,
+      "step": 672
+    },
+    {
+      "epoch": 3.6329284750337383,
+      "grad_norm": 0.36268701848097207,
+      "learning_rate": 8.391406538437737e-06,
+      "loss": 0.0265,
+      "step": 673
+    },
+    {
+      "epoch": 3.638326585695007,
+      "grad_norm": 0.3580824378387524,
+      "learning_rate": 8.329993491399933e-06,
+      "loss": 0.0317,
+      "step": 674
+    },
+    {
+      "epoch": 3.6437246963562755,
+      "grad_norm": 0.382343050277055,
+      "learning_rate": 8.268746833003304e-06,
+      "loss": 0.0289,
+      "step": 675
+    },
+    {
+      "epoch": 3.6491228070175437,
+      "grad_norm": 0.4396148697909173,
+      "learning_rate": 8.207667436490564e-06,
+      "loss": 0.0313,
+      "step": 676
+    },
+    {
+      "epoch": 3.6545209176788123,
+      "grad_norm": 0.3721908379893672,
+      "learning_rate": 8.146756172719668e-06,
+      "loss": 0.0334,
+      "step": 677
+    },
+    {
+      "epoch": 3.659919028340081,
+      "grad_norm": 0.33458204606323144,
+      "learning_rate": 8.086013910151334e-06,
+      "loss": 0.0297,
+      "step": 678
+    },
+    {
+      "epoch": 3.6653171390013495,
+      "grad_norm": 0.38656021901009874,
+      "learning_rate": 8.025441514836715e-06,
+      "loss": 0.0308,
+      "step": 679
+    },
+    {
+      "epoch": 3.670715249662618,
+      "grad_norm": 0.36672406205294283,
+      "learning_rate": 7.965039850405037e-06,
+      "loss": 0.0259,
+      "step": 680
+    },
+    {
+      "epoch": 3.6761133603238867,
+      "grad_norm": 0.33162755478379324,
+      "learning_rate": 7.904809778051252e-06,
+      "loss": 0.028,
+      "step": 681
+    },
+    {
+      "epoch": 3.6815114709851553,
+      "grad_norm": 0.3896391348891471,
+      "learning_rate": 7.84475215652381e-06,
+      "loss": 0.1327,
+      "step": 682
+    },
+    {
+      "epoch": 3.686909581646424,
+      "grad_norm": 0.4053652984242644,
+      "learning_rate": 7.784867842112367e-06,
+      "loss": 0.1707,
+      "step": 683
+    },
+    {
+      "epoch": 3.6923076923076925,
+      "grad_norm": 0.34864989460275414,
+      "learning_rate": 7.725157688635634e-06,
+      "loss": 0.0314,
+      "step": 684
+    },
+    {
+      "epoch": 3.6977058029689607,
+      "grad_norm": 0.3336341760052777,
+      "learning_rate": 7.665622547429139e-06,
+      "loss": 0.0271,
+      "step": 685
+    },
+    {
+      "epoch": 3.7031039136302293,
+      "grad_norm": 0.3449977266605391,
+      "learning_rate": 7.606263267333145e-06,
+      "loss": 0.0254,
+      "step": 686
+    },
+    {
+      "epoch": 3.708502024291498,
+      "grad_norm": 0.35748579139380365,
+      "learning_rate": 7.547080694680495e-06,
+      "loss": 0.0234,
+      "step": 687
+    },
+    {
+      "epoch": 3.7139001349527665,
+      "grad_norm": 0.3902324680292894,
+      "learning_rate": 7.4880756732846e-06,
+      "loss": 0.0328,
+      "step": 688
+    },
+    {
+      "epoch": 3.719298245614035,
+      "grad_norm": 0.36776270410444617,
+      "learning_rate": 7.429249044427362e-06,
+      "loss": 0.0355,
+      "step": 689
+    },
+    {
+      "epoch": 3.7246963562753037,
+      "grad_norm": 0.34229324573694525,
+      "learning_rate": 7.3706016468472045e-06,
+      "loss": 0.0283,
+      "step": 690
+    },
+    {
+      "epoch": 3.7300944669365723,
+      "grad_norm": 0.3430143200128147,
+      "learning_rate": 7.312134316727093e-06,
+      "loss": 0.0277,
+      "step": 691
+    },
+    {
+      "epoch": 3.7354925775978405,
+      "grad_norm": 0.3755365478047152,
+      "learning_rate": 7.25384788768265e-06,
+      "loss": 0.1848,
+      "step": 692
+    },
+    {
+      "epoch": 3.7408906882591095,
+      "grad_norm": 0.40747375304504935,
+      "learning_rate": 7.195743190750241e-06,
+      "loss": 0.0316,
+      "step": 693
+    },
+    {
+      "epoch": 3.7462887989203777,
+      "grad_norm": 0.36320411850321793,
+      "learning_rate": 7.137821054375114e-06,
+      "loss": 0.061,
+      "step": 694
+    },
+    {
+      "epoch": 3.7516869095816463,
+      "grad_norm": 0.34677209449967655,
+      "learning_rate": 7.080082304399625e-06,
+      "loss": 0.0261,
+      "step": 695
+    },
+    {
+      "epoch": 3.757085020242915,
+      "grad_norm": 0.339929781006867,
+      "learning_rate": 7.022527764051423e-06,
+      "loss": 0.0279,
+      "step": 696
+    },
+    {
+      "epoch": 3.7624831309041835,
+      "grad_norm": 0.3391973928664608,
+      "learning_rate": 6.9651582539317565e-06,
+      "loss": 0.0267,
+      "step": 697
+    },
+    {
+      "epoch": 3.767881241565452,
+      "grad_norm": 0.3492234616457014,
+      "learning_rate": 6.907974592003732e-06,
+      "loss": 0.0278,
+      "step": 698
+    },
+    {
+      "epoch": 3.7732793522267207,
+      "grad_norm": 0.46416298428484226,
+      "learning_rate": 6.850977593580675e-06,
+      "loss": 0.0369,
+      "step": 699
+    },
+    {
+      "epoch": 3.7786774628879893,
+      "grad_norm": 0.33497380932471116,
+      "learning_rate": 6.794168071314486e-06,
+      "loss": 0.0218,
+      "step": 700
+    },
+    {
+      "epoch": 3.7840755735492575,
+      "grad_norm": 0.3840112538093901,
+      "learning_rate": 6.737546835184101e-06,
+      "loss": 0.0268,
+      "step": 701
+    },
+    {
+      "epoch": 3.7894736842105265,
+      "grad_norm": 0.43018231327354933,
+      "learning_rate": 6.6811146924838725e-06,
+      "loss": 0.0656,
+      "step": 702
+    },
+    {
+      "epoch": 3.7948717948717947,
+      "grad_norm": 0.34687755720402597,
+      "learning_rate": 6.6248724478121254e-06,
+      "loss": 0.0276,
+      "step": 703
+    },
+    {
+      "epoch": 3.8002699055330633,
+      "grad_norm": 0.3648563529378022,
+      "learning_rate": 6.568820903059632e-06,
+      "loss": 0.0363,
+      "step": 704
+    },
+    {
+      "epoch": 3.805668016194332,
+      "grad_norm": 0.4492418675592268,
+      "learning_rate": 6.512960857398227e-06,
+      "loss": 0.0376,
+      "step": 705
+    },
+    {
+      "epoch": 3.8110661268556005,
+      "grad_norm": 0.37428080603957575,
+      "learning_rate": 6.457293107269371e-06,
+      "loss": 0.0291,
+      "step": 706
+    },
+    {
+      "epoch": 3.816464237516869,
+      "grad_norm": 0.369499926635395,
+      "learning_rate": 6.401818446372809e-06,
+      "loss": 0.0482,
+      "step": 707
+    },
+    {
+      "epoch": 3.8218623481781377,
+      "grad_norm": 0.3415777063161995,
+      "learning_rate": 6.346537665655286e-06,
+      "loss": 0.1112,
+      "step": 708
+    },
+    {
+      "epoch": 3.8272604588394064,
+      "grad_norm": 0.287305312482991,
+      "learning_rate": 6.291451553299204e-06,
+      "loss": 0.021,
+      "step": 709
+    },
+    {
+      "epoch": 3.8326585695006745,
+      "grad_norm": 0.3780601777928885,
+      "learning_rate": 6.236560894711459e-06,
+      "loss": 0.031,
+      "step": 710
+    },
+    {
+      "epoch": 3.8380566801619436,
+      "grad_norm": 0.4695675829432281,
+      "learning_rate": 6.181866472512175e-06,
+      "loss": 0.137,
+      "step": 711
+    },
+    {
+      "epoch": 3.8434547908232117,
+      "grad_norm": 0.45331630879040147,
+      "learning_rate": 6.127369066523599e-06,
+      "loss": 0.3132,
+      "step": 712
+    },
+    {
+      "epoch": 3.8488529014844803,
+      "grad_norm": 0.34731060254756707,
+      "learning_rate": 6.073069453758946e-06,
+      "loss": 0.0242,
+      "step": 713
+    },
+    {
+      "epoch": 3.854251012145749,
+      "grad_norm": 0.3544578398852602,
+      "learning_rate": 6.018968408411341e-06,
+      "loss": 0.0259,
+      "step": 714
+    },
+    {
+      "epoch": 3.8596491228070176,
+      "grad_norm": 0.35168683563675,
+      "learning_rate": 5.965066701842766e-06,
+      "loss": 0.1441,
+      "step": 715
+    },
+    {
+      "epoch": 3.865047233468286,
+      "grad_norm": 0.357743732049281,
+      "learning_rate": 5.9113651025730835e-06,
+      "loss": 0.0292,
+      "step": 716
+    },
+    {
+      "epoch": 3.8704453441295548,
+      "grad_norm": 0.3677528896181293,
+      "learning_rate": 5.857864376269051e-06,
+      "loss": 0.0798,
+      "step": 717
+    },
+    {
+      "epoch": 3.8758434547908234,
+      "grad_norm": 0.35923867256807285,
+      "learning_rate": 5.804565285733432e-06,
+      "loss": 0.0328,
+      "step": 718
+    },
+    {
+      "epoch": 3.8812415654520915,
+      "grad_norm": 0.38030735474824356,
+      "learning_rate": 5.751468590894107e-06,
+      "loss": 0.1131,
+      "step": 719
+    },
+    {
+      "epoch": 3.8866396761133606,
+      "grad_norm": 0.36213732163193973,
+      "learning_rate": 5.698575048793223e-06,
+      "loss": 0.0285,
+      "step": 720
+    },
+    {
+      "epoch": 3.8920377867746287,
+      "grad_norm": 0.4427125231379987,
+      "learning_rate": 5.645885413576433e-06,
+      "loss": 0.1179,
+      "step": 721
+    },
+    {
+      "epoch": 3.8974358974358974,
+      "grad_norm": 0.37081258331408545,
+      "learning_rate": 5.593400436482119e-06,
+      "loss": 0.0416,
+      "step": 722
+    },
+    {
+      "epoch": 3.902834008097166,
+      "grad_norm": 0.4002983215995271,
+      "learning_rate": 5.541120865830687e-06,
+      "loss": 0.0312,
+      "step": 723
+    },
+    {
+      "epoch": 3.9082321187584346,
+      "grad_norm": 0.34949068915726794,
+      "learning_rate": 5.489047447013891e-06,
+      "loss": 0.0296,
+      "step": 724
+    },
+    {
+      "epoch": 3.913630229419703,
+      "grad_norm": 0.3464543170966297,
+      "learning_rate": 5.4371809224842354e-06,
+      "loss": 0.0249,
+      "step": 725
+    },
+    {
+      "epoch": 3.919028340080972,
+      "grad_norm": 0.3442707542539264,
+      "learning_rate": 5.3855220317443416e-06,
+      "loss": 0.0272,
+      "step": 726
+    },
+    {
+      "epoch": 3.9244264507422404,
+      "grad_norm": 0.3948593131237832,
+      "learning_rate": 5.334071511336449e-06,
+      "loss": 0.0263,
+      "step": 727
+    },
+    {
+      "epoch": 3.9298245614035086,
+      "grad_norm": 0.3574464081518333,
+      "learning_rate": 5.282830094831881e-06,
+      "loss": 0.0268,
+      "step": 728
+    },
+    {
+      "epoch": 3.9352226720647776,
+      "grad_norm": 0.3420025327246806,
+      "learning_rate": 5.231798512820612e-06,
+      "loss": 0.0249,
+      "step": 729
+    },
+    {
+      "epoch": 3.9406207827260458,
+      "grad_norm": 0.4021770089954796,
+      "learning_rate": 5.180977492900823e-06,
+      "loss": 0.1576,
+      "step": 730
+    },
+    {
+      "epoch": 3.9460188933873144,
+      "grad_norm": 0.399473122697206,
+      "learning_rate": 5.130367759668544e-06,
+      "loss": 0.1145,
+      "step": 731
+    },
+    {
+      "epoch": 3.951417004048583,
+      "grad_norm": 0.39532463873364626,
+      "learning_rate": 5.07997003470734e-06,
+      "loss": 0.0305,
+      "step": 732
+    },
+    {
+      "epoch": 3.9568151147098516,
+      "grad_norm": 0.3383915433366634,
+      "learning_rate": 5.029785036577976e-06,
+      "loss": 0.024,
+      "step": 733
+    },
+    {
+      "epoch": 3.96221322537112,
+      "grad_norm": 0.35374596721588925,
+      "learning_rate": 4.979813480808231e-06,
+      "loss": 0.027,
+      "step": 734
+    },
+    {
+      "epoch": 3.967611336032389,
+      "grad_norm": 0.3523879250836844,
+      "learning_rate": 4.930056079882632e-06,
+      "loss": 0.0299,
+      "step": 735
+    },
+    {
+      "epoch": 3.9730094466936574,
+      "grad_norm": 0.35312835882530846,
+      "learning_rate": 4.880513543232361e-06,
+      "loss": 0.0265,
+      "step": 736
+    },
+    {
+      "epoch": 3.9784075573549256,
+      "grad_norm": 0.3932291227653852,
+      "learning_rate": 4.831186577225082e-06,
+      "loss": 0.0345,
+      "step": 737
+    },
+    {
+      "epoch": 3.983805668016194,
+      "grad_norm": 0.4189705607669111,
+      "learning_rate": 4.782075885154909e-06,
+      "loss": 0.0301,
+      "step": 738
+    },
+    {
+      "epoch": 3.989203778677463,
+      "grad_norm": 0.43157943937613985,
+      "learning_rate": 4.733182167232356e-06,
+      "loss": 0.0362,
+      "step": 739
+    },
+    {
+      "epoch": 3.9946018893387314,
+      "grad_norm": 0.34724202073129157,
+      "learning_rate": 4.684506120574375e-06,
+      "loss": 0.0245,
+      "step": 740
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 1.7041116732425667,
+      "learning_rate": 4.636048439194392e-06,
+      "loss": 0.0627,
+      "step": 741
+    },
+    {
+      "epoch": 4.005398110661268,
+      "grad_norm": 0.15584224827038143,
+      "learning_rate": 4.587809813992437e-06,
+      "loss": 0.0116,
+      "step": 742
+    },
+    {
+      "epoch": 4.010796221322537,
+      "grad_norm": 0.1392348554326972,
+      "learning_rate": 4.53979093274526e-06,
+      "loss": 0.0084,
+      "step": 743
+    },
+    {
+      "epoch": 4.016194331983805,
+      "grad_norm": 0.1282245610540337,
+      "learning_rate": 4.4919924800965696e-06,
+      "loss": 0.0082,
+      "step": 744
+    },
+    {
+      "epoch": 4.021592442645074,
+      "grad_norm": 0.17084049043505972,
+      "learning_rate": 4.444415137547238e-06,
+      "loss": 0.0092,
+      "step": 745
+    },
+    {
+      "epoch": 4.026990553306343,
+      "grad_norm": 0.16361824667883376,
+      "learning_rate": 4.3970595834455845e-06,
+      "loss": 0.0087,
+      "step": 746
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 0.2446505920500862,
+      "learning_rate": 4.349926492977719e-06,
+      "loss": 0.0831,
+      "step": 747
+    },
+    {
+      "epoch": 4.03778677462888,
+      "grad_norm": 0.22089359546985632,
+      "learning_rate": 4.303016538157907e-06,
+      "loss": 0.0321,
+      "step": 748
+    },
+    {
+      "epoch": 4.043184885290149,
+      "grad_norm": 0.2468941327645301,
+      "learning_rate": 4.256330387818999e-06,
+      "loss": 0.1825,
+      "step": 749
+    },
+    {
+      "epoch": 4.048582995951417,
+      "grad_norm": 0.2685001788895823,
+      "learning_rate": 4.209868707602871e-06,
+      "loss": 0.1063,
+      "step": 750
+    },
+    {
+      "epoch": 4.053981106612685,
+      "grad_norm": 0.1581251039928878,
+      "learning_rate": 4.163632159950965e-06,
+      "loss": 0.0074,
+      "step": 751
+    },
+    {
+      "epoch": 4.059379217273954,
+      "grad_norm": 0.19573771181593203,
+      "learning_rate": 4.117621404094811e-06,
+      "loss": 0.0085,
+      "step": 752
+    },
+    {
+      "epoch": 4.064777327935222,
+      "grad_norm": 0.1461438989441116,
+      "learning_rate": 4.07183709604666e-06,
+      "loss": 0.0082,
+      "step": 753
+    },
+    {
+      "epoch": 4.0701754385964914,
+      "grad_norm": 0.16302088011377536,
+      "learning_rate": 4.026279888590101e-06,
+      "loss": 0.0096,
+      "step": 754
+    },
+    {
+      "epoch": 4.07557354925776,
+      "grad_norm": 0.17495547254676413,
+      "learning_rate": 3.980950431270776e-06,
+      "loss": 0.0081,
+      "step": 755
+    },
+    {
+      "epoch": 4.080971659919029,
+      "grad_norm": 0.16147617638359743,
+      "learning_rate": 3.935849370387104e-06,
+      "loss": 0.0075,
+      "step": 756
+    },
+    {
+      "epoch": 4.086369770580297,
+      "grad_norm": 0.13531640776000314,
+      "learning_rate": 3.89097734898108e-06,
+      "loss": 0.0075,
+      "step": 757
+    },
+    {
+      "epoch": 4.091767881241566,
+      "grad_norm": 0.15833156615383603,
+      "learning_rate": 3.846335006829103e-06,
+      "loss": 0.0078,
+      "step": 758
+    },
+    {
+      "epoch": 4.097165991902834,
+      "grad_norm": 0.1502536180322697,
+      "learning_rate": 3.801922980432835e-06,
+      "loss": 0.0087,
+      "step": 759
+    },
+    {
+      "epoch": 4.102564102564102,
+      "grad_norm": 0.16439612264602832,
+      "learning_rate": 3.7577419030101615e-06,
+      "loss": 0.0085,
+      "step": 760
+    },
+    {
+      "epoch": 4.107962213225371,
+      "grad_norm": 0.26115821750394347,
+      "learning_rate": 3.7137924044861273e-06,
+      "loss": 0.2572,
+      "step": 761
+    },
+    {
+      "epoch": 4.113360323886639,
+      "grad_norm": 0.1956707726653882,
+      "learning_rate": 3.670075111483975e-06,
+      "loss": 0.0277,
+      "step": 762
+    },
+    {
+      "epoch": 4.1187584345479085,
+      "grad_norm": 0.172397433087837,
+      "learning_rate": 3.626590647316206e-06,
+      "loss": 0.009,
+      "step": 763
+    },
+    {
+      "epoch": 4.124156545209177,
+      "grad_norm": 0.16093660362400164,
+      "learning_rate": 3.583339631975704e-06,
+      "loss": 0.0084,
+      "step": 764
+    },
+    {
+      "epoch": 4.129554655870446,
+      "grad_norm": 0.13992297523457856,
+      "learning_rate": 3.5403226821268734e-06,
+      "loss": 0.0067,
+      "step": 765
+    },
+    {
+      "epoch": 4.134952766531714,
+      "grad_norm": 0.13111008475760771,
+      "learning_rate": 3.497540411096869e-06,
+      "loss": 0.0066,
+      "step": 766
+    },
+    {
+      "epoch": 4.140350877192983,
+      "grad_norm": 0.15452619417986485,
+      "learning_rate": 3.454993428866831e-06,
+      "loss": 0.0076,
+      "step": 767
+    },
+    {
+      "epoch": 4.145748987854251,
+      "grad_norm": 0.15886764259712663,
+      "learning_rate": 3.4126823420632095e-06,
+      "loss": 0.0081,
+      "step": 768
+    },
+    {
+      "epoch": 4.151147098515519,
+      "grad_norm": 0.18177047320434667,
+      "learning_rate": 3.3706077539490933e-06,
+      "loss": 0.0489,
+      "step": 769
+    },
+    {
+      "epoch": 4.156545209176788,
+      "grad_norm": 0.47899552021459674,
+      "learning_rate": 3.328770264415635e-06,
+      "loss": 0.0278,
+      "step": 770
+    },
+    {
+      "epoch": 4.161943319838056,
+      "grad_norm": 0.15628284095946662,
+      "learning_rate": 3.287170469973466e-06,
+      "loss": 0.007,
+      "step": 771
+    },
+    {
+      "epoch": 4.1673414304993255,
+      "grad_norm": 0.14899346110998693,
+      "learning_rate": 3.2458089637442124e-06,
+      "loss": 0.0071,
+      "step": 772
+    },
+    {
+      "epoch": 4.172739541160594,
+      "grad_norm": 0.17853919488702116,
+      "learning_rate": 3.204686335452043e-06,
+      "loss": 0.0086,
+      "step": 773
+    },
+    {
+      "epoch": 4.178137651821863,
+      "grad_norm": 0.155584081195197,
+      "learning_rate": 3.163803171415243e-06,
+      "loss": 0.0066,
+      "step": 774
+    },
+    {
+      "epoch": 4.183535762483131,
+      "grad_norm": 0.1686004890478719,
+      "learning_rate": 3.1231600545378703e-06,
+      "loss": 0.007,
+      "step": 775
+    },
+    {
+      "epoch": 4.1889338731444,
+      "grad_norm": 0.17950162185181553,
+      "learning_rate": 3.0827575643014283e-06,
+      "loss": 0.0076,
+      "step": 776
+    },
+    {
+      "epoch": 4.194331983805668,
+      "grad_norm": 0.19071738162381693,
+      "learning_rate": 3.0425962767566307e-06,
+      "loss": 0.0071,
+      "step": 777
+    },
+    {
+      "epoch": 4.199730094466936,
+      "grad_norm": 0.14933428520520017,
+      "learning_rate": 3.0026767645151532e-06,
+      "loss": 0.0084,
+      "step": 778
+    },
+    {
+      "epoch": 4.205128205128205,
+      "grad_norm": 0.15852536035642434,
+      "learning_rate": 2.9629995967414914e-06,
+      "loss": 0.0058,
+      "step": 779
+    },
+    {
+      "epoch": 4.2105263157894735,
+      "grad_norm": 0.18187853868086776,
+      "learning_rate": 2.9235653391448448e-06,
+      "loss": 0.0075,
+      "step": 780
+    },
+    {
+      "epoch": 4.2159244264507425,
+      "grad_norm": 0.1572003998731716,
+      "learning_rate": 2.8843745539710523e-06,
+      "loss": 0.008,
+      "step": 781
+    },
+    {
+      "epoch": 4.221322537112011,
+      "grad_norm": 0.17395896577365463,
+      "learning_rate": 2.8454277999945603e-06,
+      "loss": 0.0077,
+      "step": 782
+    },
+    {
+      "epoch": 4.22672064777328,
+      "grad_norm": 0.35041441469504137,
+      "learning_rate": 2.806725632510472e-06,
+      "loss": 0.2508,
+      "step": 783
+    },
+    {
+      "epoch": 4.232118758434548,
+      "grad_norm": 0.19085015890709575,
+      "learning_rate": 2.7682686033266337e-06,
+      "loss": 0.007,
+      "step": 784
+    },
+    {
+      "epoch": 4.237516869095816,
+      "grad_norm": 0.17434262637335643,
+      "learning_rate": 2.730057260755743e-06,
+      "loss": 0.0071,
+      "step": 785
+    },
+    {
+      "epoch": 4.242914979757085,
+      "grad_norm": 0.27692426702257456,
+      "learning_rate": 2.6920921496075545e-06,
+      "loss": 0.1894,
+      "step": 786
+    },
+    {
+      "epoch": 4.248313090418353,
+      "grad_norm": 0.18631641241017277,
+      "learning_rate": 2.654373811181099e-06,
+      "loss": 0.0072,
+      "step": 787
+    },
+    {
+      "epoch": 4.253711201079622,
+      "grad_norm": 0.2506854137435988,
+      "learning_rate": 2.6169027832569825e-06,
+      "loss": 0.137,
+      "step": 788
+    },
+    {
+      "epoch": 4.2591093117408905,
+      "grad_norm": 0.13306514956818488,
+      "learning_rate": 2.5796796000896882e-06,
+      "loss": 0.0057,
+      "step": 789
+    },
+    {
+      "epoch": 4.2645074224021595,
+      "grad_norm": 0.14198881177330058,
+      "learning_rate": 2.5427047923999993e-06,
+      "loss": 0.0067,
+      "step": 790
+    },
+    {
+      "epoch": 4.269905533063428,
+      "grad_norm": 0.17095322268675192,
+      "learning_rate": 2.505978887367393e-06,
+      "loss": 0.0092,
+      "step": 791
+    },
+    {
+      "epoch": 4.275303643724697,
+      "grad_norm": 0.2240504562461258,
+      "learning_rate": 2.4695024086225507e-06,
+      "loss": 0.1948,
+      "step": 792
+    },
+    {
+      "epoch": 4.280701754385965,
+      "grad_norm": 0.19767331740609034,
+      "learning_rate": 2.4332758762398846e-06,
+      "loss": 0.0084,
+      "step": 793
+    },
+    {
+      "epoch": 4.286099865047233,
+      "grad_norm": 0.1897898564392771,
+      "learning_rate": 2.3972998067301113e-06,
+      "loss": 0.0091,
+      "step": 794
+    },
+    {
+      "epoch": 4.291497975708502,
+      "grad_norm": 0.24557583754489332,
+      "learning_rate": 2.3615747130329013e-06,
+      "loss": 0.0645,
+      "step": 795
+    },
+    {
+      "epoch": 4.29689608636977,
+      "grad_norm": 0.11474220149767227,
+      "learning_rate": 2.3261011045095706e-06,
+      "loss": 0.0052,
+      "step": 796
+    },
+    {
+      "epoch": 4.302294197031039,
+      "grad_norm": 0.17934384465987277,
+      "learning_rate": 2.2908794869358044e-06,
+      "loss": 0.0078,
+      "step": 797
+    },
+    {
+      "epoch": 4.3076923076923075,
+      "grad_norm": 0.17105025868804596,
+      "learning_rate": 2.25591036249444e-06,
+      "loss": 0.0074,
+      "step": 798
+    },
+    {
+      "epoch": 4.3130904183535765,
+      "grad_norm": 0.1733830020298948,
+      "learning_rate": 2.22119422976834e-06,
+      "loss": 0.0078,
+      "step": 799
+    },
+    {
+      "epoch": 4.318488529014845,
+      "grad_norm": 0.14992868331209838,
+      "learning_rate": 2.186731583733235e-06,
+      "loss": 0.0063,
+      "step": 800
+    },
+    {
+      "epoch": 4.323886639676114,
+      "grad_norm": 0.15115286692517083,
+      "learning_rate": 2.1525229157507142e-06,
+      "loss": 0.0068,
+      "step": 801
+    },
+    {
+      "epoch": 4.329284750337382,
+      "grad_norm": 0.17561669168571467,
+      "learning_rate": 2.118568713561189e-06,
+      "loss": 0.0106,
+      "step": 802
+    },
+    {
+      "epoch": 4.33468286099865,
+      "grad_norm": 0.16726960898237694,
+      "learning_rate": 2.084869461276946e-06,
+      "loss": 0.0075,
+      "step": 803
+    },
+    {
+      "epoch": 4.340080971659919,
+      "grad_norm": 0.1519651965593108,
+      "learning_rate": 2.0514256393752465e-06,
+      "loss": 0.0059,
+      "step": 804
+    },
+    {
+      "epoch": 4.345479082321187,
+      "grad_norm": 0.173856541251779,
+      "learning_rate": 2.018237724691483e-06,
+      "loss": 0.0087,
+      "step": 805
+    },
+    {
+      "epoch": 4.350877192982456,
+      "grad_norm": 0.1613693859888312,
+      "learning_rate": 1.9853061904123662e-06,
+      "loss": 0.0069,
+      "step": 806
+    },
+    {
+      "epoch": 4.3562753036437245,
+      "grad_norm": 0.1381367748106028,
+      "learning_rate": 1.9526315060691915e-06,
+      "loss": 0.0063,
+      "step": 807
+    },
+    {
+      "epoch": 4.361673414304994,
+      "grad_norm": 0.16367559270740642,
+      "learning_rate": 1.9202141375311335e-06,
+      "loss": 0.0073,
+      "step": 808
+    },
+    {
+      "epoch": 4.367071524966262,
+      "grad_norm": 0.14051555571715985,
+      "learning_rate": 1.8880545469986144e-06,
+      "loss": 0.0073,
+      "step": 809
+    },
+    {
+      "epoch": 4.372469635627531,
+      "grad_norm": 0.23064834815909274,
+      "learning_rate": 1.8561531929967057e-06,
+      "loss": 0.0085,
+      "step": 810
+    },
+    {
+      "epoch": 4.377867746288799,
+      "grad_norm": 0.24358676131518123,
+      "learning_rate": 1.8245105303685861e-06,
+      "loss": 0.1195,
+      "step": 811
+    },
+    {
+      "epoch": 4.383265856950067,
+      "grad_norm": 0.19179647085523155,
+      "learning_rate": 1.7931270102690822e-06,
+      "loss": 0.0081,
+      "step": 812
+    },
+    {
+      "epoch": 4.388663967611336,
+      "grad_norm": 0.18571189286382062,
+      "learning_rate": 1.7620030801581988e-06,
+      "loss": 0.0074,
+      "step": 813
+    },
+    {
+      "epoch": 4.394062078272604,
+      "grad_norm": 0.16089829965288302,
+      "learning_rate": 1.7311391837947723e-06,
+      "loss": 0.0073,
+      "step": 814
+    },
+    {
+      "epoch": 4.399460188933873,
+      "grad_norm": 0.14409744033623018,
+      "learning_rate": 1.7005357612301176e-06,
+      "loss": 0.0066,
+      "step": 815
+    },
+    {
+      "epoch": 4.4048582995951415,
+      "grad_norm": 0.16665498770890472,
+      "learning_rate": 1.6701932488017791e-06,
+      "loss": 0.0099,
+      "step": 816
+    },
+    {
+      "epoch": 4.410256410256411,
+      "grad_norm": 0.16528493853060244,
+      "learning_rate": 1.6401120791272829e-06,
+      "loss": 0.007,
+      "step": 817
+    },
+    {
+      "epoch": 4.415654520917679,
+      "grad_norm": 0.22208885872699052,
+      "learning_rate": 1.6102926810979912e-06,
+      "loss": 0.0099,
+      "step": 818
+    },
+    {
+      "epoch": 4.421052631578947,
+      "grad_norm": 0.16655728113519622,
+      "learning_rate": 1.5807354798729678e-06,
+      "loss": 0.0081,
+      "step": 819
+    },
+    {
+      "epoch": 4.426450742240216,
+      "grad_norm": 0.21501198575920613,
+      "learning_rate": 1.5514408968729399e-06,
+      "loss": 0.0751,
+      "step": 820
+    },
+    {
+      "epoch": 4.431848852901484,
+      "grad_norm": 0.15438234919555596,
+      "learning_rate": 1.5224093497742654e-06,
+      "loss": 0.0062,
+      "step": 821
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 0.1411495215593871,
+      "learning_rate": 1.493641252502993e-06,
+      "loss": 0.0067,
+      "step": 822
+    },
+    {
+      "epoch": 4.442645074224021,
+      "grad_norm": 0.22091395786785076,
+      "learning_rate": 1.465137015228959e-06,
+      "loss": 0.0633,
+      "step": 823
+    },
+    {
+      "epoch": 4.44804318488529,
+      "grad_norm": 0.14056343619292191,
+      "learning_rate": 1.4368970443599261e-06,
+      "loss": 0.0067,
+      "step": 824
+    },
+    {
+      "epoch": 4.4534412955465585,
+      "grad_norm": 0.13346197367284868,
+      "learning_rate": 1.4089217425358115e-06,
+      "loss": 0.0048,
+      "step": 825
+    },
+    {
+      "epoch": 4.458839406207828,
+      "grad_norm": 0.148850613047919,
+      "learning_rate": 1.381211508622924e-06,
+      "loss": 0.0067,
+      "step": 826
+    },
+    {
+      "epoch": 4.464237516869096,
+      "grad_norm": 0.1740364263154185,
+      "learning_rate": 1.3537667377082885e-06,
+      "loss": 0.0083,
+      "step": 827
+    },
+    {
+      "epoch": 4.469635627530364,
+      "grad_norm": 0.17645888735792817,
+      "learning_rate": 1.3265878210940098e-06,
+      "loss": 0.0079,
+      "step": 828
+    },
+    {
+      "epoch": 4.475033738191633,
+      "grad_norm": 0.16562347185792808,
+      "learning_rate": 1.2996751462917057e-06,
+      "loss": 0.0069,
+      "step": 829
+    },
+    {
+      "epoch": 4.480431848852901,
+      "grad_norm": 0.1609564758936411,
+      "learning_rate": 1.2730290970169535e-06,
+      "loss": 0.0067,
+      "step": 830
+    },
+    {
+      "epoch": 4.48582995951417,
+      "grad_norm": 0.16735348748988518,
+      "learning_rate": 1.2466500531838532e-06,
+      "loss": 0.0063,
+      "step": 831
+    },
+    {
+      "epoch": 4.491228070175438,
+      "grad_norm": 0.2048968721784967,
+      "learning_rate": 1.2205383908995815e-06,
+      "loss": 0.2056,
+      "step": 832
+    },
+    {
+      "epoch": 4.496626180836707,
+      "grad_norm": 0.14903906992957436,
+      "learning_rate": 1.1946944824590513e-06,
+      "loss": 0.0067,
+      "step": 833
+    },
+    {
+      "epoch": 4.502024291497976,
+      "grad_norm": 0.16322359086669616,
+      "learning_rate": 1.1691186963395861e-06,
+      "loss": 0.0055,
+      "step": 834
+    },
+    {
+      "epoch": 4.507422402159245,
+      "grad_norm": 0.16278200834250284,
+      "learning_rate": 1.1438113971956776e-06,
+      "loss": 0.0079,
+      "step": 835
+    },
+    {
+      "epoch": 4.512820512820513,
+      "grad_norm": 0.17418333666535243,
+      "learning_rate": 1.11877294585379e-06,
+      "loss": 0.0077,
+      "step": 836
+    },
+    {
+      "epoch": 4.518218623481781,
+      "grad_norm": 0.1419711861910052,
+      "learning_rate": 1.0940036993071934e-06,
+      "loss": 0.0078,
+      "step": 837
+    },
+    {
+      "epoch": 4.52361673414305,
+      "grad_norm": 0.1845233797922832,
+      "learning_rate": 1.0695040107109089e-06,
+      "loss": 0.0067,
+      "step": 838
+    },
+    {
+      "epoch": 4.529014844804318,
+      "grad_norm": 0.14933552106266268,
+      "learning_rate": 1.045274229376636e-06,
+      "loss": 0.0062,
+      "step": 839
+    },
+    {
+      "epoch": 4.534412955465587,
+      "grad_norm": 0.22406063579380403,
+      "learning_rate": 1.0213147007678036e-06,
+      "loss": 0.008,
+      "step": 840
+    },
+    {
+      "epoch": 4.539811066126855,
+      "grad_norm": 0.14192764956365347,
+      "learning_rate": 9.97625766494621e-07,
+      "loss": 0.0073,
+      "step": 841
+    },
+    {
+      "epoch": 4.545209176788124,
+      "grad_norm": 0.21149852476408423,
+      "learning_rate": 9.742077643092207e-07,
+      "loss": 0.0094,
+      "step": 842
+    },
+    {
+      "epoch": 4.550607287449393,
+      "grad_norm": 0.2466190339001213,
+      "learning_rate": 9.51061028100837e-07,
+      "loss": 0.0064,
+      "step": 843
+    },
+    {
+      "epoch": 4.556005398110662,
+      "grad_norm": 0.21652228857071856,
+      "learning_rate": 9.281858878910555e-07,
+      "loss": 0.0086,
+      "step": 844
+    },
+    {
+      "epoch": 4.56140350877193,
+      "grad_norm": 0.21949546715476373,
+      "learning_rate": 9.055826698290881e-07,
+      "loss": 0.0073,
+      "step": 845
+    },
+    {
+      "epoch": 4.566801619433198,
+      "grad_norm": 0.3097045351458495,
+      "learning_rate": 8.832516961871462e-07,
+      "loss": 0.0057,
+      "step": 846
+    },
+    {
+      "epoch": 4.572199730094467,
+      "grad_norm": 0.2948294424140672,
+      "learning_rate": 8.611932853558236e-07,
+      "loss": 0.0087,
+      "step": 847
+    },
+    {
+      "epoch": 4.577597840755735,
+      "grad_norm": 0.1520476053391632,
+      "learning_rate": 8.394077518395738e-07,
+      "loss": 0.0069,
+      "step": 848
+    },
+    {
+      "epoch": 4.582995951417004,
+      "grad_norm": 0.17234756128502748,
+      "learning_rate": 8.178954062522226e-07,
+      "loss": 0.0084,
+      "step": 849
+    },
+    {
+      "epoch": 4.588394062078272,
+      "grad_norm": 0.1573022595895341,
+      "learning_rate": 7.966565553125294e-07,
+      "loss": 0.0059,
+      "step": 850
+    },
+    {
+      "epoch": 4.593792172739541,
+      "grad_norm": 0.16461969032896173,
+      "learning_rate": 7.756915018398214e-07,
+      "loss": 0.0068,
+      "step": 851
+    },
+    {
+      "epoch": 4.59919028340081,
+      "grad_norm": 0.14068127798869795,
+      "learning_rate": 7.5500054474968e-07,
+      "loss": 0.006,
+      "step": 852
+    },
+    {
+      "epoch": 4.604588394062079,
+      "grad_norm": 0.16850155343511367,
+      "learning_rate": 7.345839790496745e-07,
+      "loss": 0.0064,
+      "step": 853
+    },
+    {
+      "epoch": 4.609986504723347,
+      "grad_norm": 0.16763855234451672,
+      "learning_rate": 7.144420958351506e-07,
+      "loss": 0.0072,
+      "step": 854
+    },
+    {
+      "epoch": 4.615384615384615,
+      "grad_norm": 0.1592524289261433,
+      "learning_rate": 6.945751822850955e-07,
+      "loss": 0.0081,
+      "step": 855
+    },
+    {
+      "epoch": 4.620782726045884,
+      "grad_norm": 0.18735419355531763,
+      "learning_rate": 6.749835216580214e-07,
+      "loss": 0.0096,
+      "step": 856
+    },
+    {
+      "epoch": 4.626180836707152,
+      "grad_norm": 0.20754621796892195,
+      "learning_rate": 6.556673932879488e-07,
+      "loss": 0.0084,
+      "step": 857
+    },
+    {
+      "epoch": 4.631578947368421,
+      "grad_norm": 0.14315438531961788,
+      "learning_rate": 6.366270725804069e-07,
+      "loss": 0.0062,
+      "step": 858
+    },
+    {
+      "epoch": 4.636977058029689,
+      "grad_norm": 0.16421607941126087,
+      "learning_rate": 6.178628310085178e-07,
+      "loss": 0.0074,
+      "step": 859
+    },
+    {
+      "epoch": 4.6423751686909585,
+      "grad_norm": 0.21208690378634537,
+      "learning_rate": 5.993749361091206e-07,
+      "loss": 0.1303,
+      "step": 860
+    },
+    {
+      "epoch": 4.647773279352227,
+      "grad_norm": 0.14180171705428649,
+      "learning_rate": 5.811636514789598e-07,
+      "loss": 0.0059,
+      "step": 861
+    },
+    {
+      "epoch": 4.653171390013496,
+      "grad_norm": 0.206784771609866,
+      "learning_rate": 5.632292367709236e-07,
+      "loss": 0.1098,
+      "step": 862
+    },
+    {
+      "epoch": 4.658569500674764,
+      "grad_norm": 0.1529621394315393,
+      "learning_rate": 5.455719476903442e-07,
+      "loss": 0.0065,
+      "step": 863
+    },
+    {
+      "epoch": 4.663967611336032,
+      "grad_norm": 0.1683098142151735,
+      "learning_rate": 5.281920359913528e-07,
+      "loss": 0.0173,
+      "step": 864
+    },
+    {
+      "epoch": 4.669365721997301,
+      "grad_norm": 0.1519444172909866,
+      "learning_rate": 5.110897494732881e-07,
+      "loss": 0.0064,
+      "step": 865
+    },
+    {
+      "epoch": 4.674763832658569,
+      "grad_norm": 0.15996483183482557,
+      "learning_rate": 4.942653319771618e-07,
+      "loss": 0.0076,
+      "step": 866
+    },
+    {
+      "epoch": 4.680161943319838,
+      "grad_norm": 0.1319666927607246,
+      "learning_rate": 4.777190233821838e-07,
+      "loss": 0.0058,
+      "step": 867
+    },
+    {
+      "epoch": 4.685560053981106,
+      "grad_norm": 0.16475241129528057,
+      "learning_rate": 4.6145105960235136e-07,
+      "loss": 0.0074,
+      "step": 868
+    },
+    {
+      "epoch": 4.6909581646423755,
+      "grad_norm": 0.15587959822024125,
+      "learning_rate": 4.4546167258306296e-07,
+      "loss": 0.007,
+      "step": 869
+    },
+    {
+      "epoch": 4.696356275303644,
+      "grad_norm": 0.170013831426874,
+      "learning_rate": 4.297510902978341e-07,
+      "loss": 0.009,
+      "step": 870
+    },
+    {
+      "epoch": 4.701754385964913,
+      "grad_norm": 0.1490640310501832,
+      "learning_rate": 4.1431953674503586e-07,
+      "loss": 0.0066,
+      "step": 871
+    },
+    {
+      "epoch": 4.707152496626181,
+      "grad_norm": 0.14254928891164198,
+      "learning_rate": 3.991672319447015e-07,
+      "loss": 0.0066,
+      "step": 872
+    },
+    {
+      "epoch": 4.712550607287449,
+      "grad_norm": 0.16741315702690107,
+      "learning_rate": 3.842943919353914e-07,
+      "loss": 0.0067,
+      "step": 873
+    },
+    {
+      "epoch": 4.717948717948718,
+      "grad_norm": 0.234116846451384,
+      "learning_rate": 3.6970122877111104e-07,
+      "loss": 0.2032,
+      "step": 874
+    },
+    {
+      "epoch": 4.723346828609986,
+      "grad_norm": 0.15611337359536132,
+      "learning_rate": 3.5538795051829336e-07,
+      "loss": 0.0067,
+      "step": 875
+    },
+    {
+      "epoch": 4.728744939271255,
+      "grad_norm": 0.174776067692724,
+      "learning_rate": 3.413547612528212e-07,
+      "loss": 0.0066,
+      "step": 876
+    },
+    {
+      "epoch": 4.734143049932523,
+      "grad_norm": 0.17963649433676093,
+      "learning_rate": 3.2760186105712964e-07,
+      "loss": 0.0086,
+      "step": 877
+    },
+    {
+      "epoch": 4.7395411605937925,
+      "grad_norm": 0.12521550946408017,
+      "learning_rate": 3.1412944601734165e-07,
+      "loss": 0.0055,
+      "step": 878
+    },
+    {
+      "epoch": 4.744939271255061,
+      "grad_norm": 0.17200194683832373,
+      "learning_rate": 3.0093770822048786e-07,
+      "loss": 0.0069,
+      "step": 879
+    },
+    {
+      "epoch": 4.75033738191633,
+      "grad_norm": 0.1611762123853898,
+      "learning_rate": 2.8802683575174907e-07,
+      "loss": 0.0077,
+      "step": 880
+    },
+    {
+      "epoch": 4.755735492577598,
+      "grad_norm": 0.12936360386404333,
+      "learning_rate": 2.753970126917871e-07,
+      "loss": 0.0056,
+      "step": 881
+    },
+    {
+      "epoch": 4.761133603238866,
+      "grad_norm": 0.30585009959853815,
+      "learning_rate": 2.6304841911412027e-07,
+      "loss": 0.0188,
+      "step": 882
+    },
+    {
+      "epoch": 4.766531713900135,
+      "grad_norm": 0.14471118390180884,
+      "learning_rate": 2.509812310825499e-07,
+      "loss": 0.0062,
+      "step": 883
+    },
+    {
+      "epoch": 4.771929824561403,
+      "grad_norm": 0.17285030047405645,
+      "learning_rate": 2.391956206486468e-07,
+      "loss": 0.0071,
+      "step": 884
+    },
+    {
+      "epoch": 4.777327935222672,
+      "grad_norm": 0.1607607574024115,
+      "learning_rate": 2.2769175584931746e-07,
+      "loss": 0.0149,
+      "step": 885
+    },
+    {
+      "epoch": 4.7827260458839405,
+      "grad_norm": 0.22128649039370432,
+      "learning_rate": 2.1646980070437973e-07,
+      "loss": 0.0083,
+      "step": 886
+    },
+    {
+      "epoch": 4.7881241565452095,
+      "grad_norm": 0.3319798411658218,
+      "learning_rate": 2.0552991521425537e-07,
+      "loss": 0.27,
+      "step": 887
+    },
+    {
+      "epoch": 4.793522267206478,
+      "grad_norm": 0.28255511017291,
+      "learning_rate": 1.9487225535765873e-07,
+      "loss": 0.0225,
+      "step": 888
+    },
+    {
+      "epoch": 4.798920377867747,
+      "grad_norm": 0.16071587770507176,
+      "learning_rate": 1.8449697308939863e-07,
+      "loss": 0.0056,
+      "step": 889
+    },
+    {
+      "epoch": 4.804318488529015,
+      "grad_norm": 0.16475073813051386,
+      "learning_rate": 1.7440421633818872e-07,
+      "loss": 0.006,
+      "step": 890
+    },
+    {
+      "epoch": 4.809716599190283,
+      "grad_norm": 0.18289700223178906,
+      "learning_rate": 1.6459412900455607e-07,
+      "loss": 0.0115,
+      "step": 891
+    },
+    {
+      "epoch": 4.815114709851552,
+      "grad_norm": 0.13638603960683227,
+      "learning_rate": 1.5506685095878715e-07,
+      "loss": 0.006,
+      "step": 892
+    },
+    {
+      "epoch": 4.82051282051282,
+      "grad_norm": 0.1567400736733609,
+      "learning_rate": 1.4582251803892055e-07,
+      "loss": 0.0062,
+      "step": 893
+    },
+    {
+      "epoch": 4.825910931174089,
+      "grad_norm": 0.13245763270893754,
+      "learning_rate": 1.3686126204882634e-07,
+      "loss": 0.0063,
+      "step": 894
+    },
+    {
+      "epoch": 4.8313090418353575,
+      "grad_norm": 0.14266344369563544,
+      "learning_rate": 1.2818321075631635e-07,
+      "loss": 0.0056,
+      "step": 895
+    },
+    {
+      "epoch": 4.8367071524966265,
+      "grad_norm": 0.15296568107274008,
+      "learning_rate": 1.19788487891328e-07,
+      "loss": 0.0072,
+      "step": 896
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 0.17802279161940993,
+      "learning_rate": 1.1167721314415681e-07,
+      "loss": 0.0079,
+      "step": 897
+    },
+    {
+      "epoch": 4.847503373819164,
+      "grad_norm": 0.17793550727981747,
+      "learning_rate": 1.038495021637509e-07,
+      "loss": 0.01,
+      "step": 898
+    },
+    {
+      "epoch": 4.852901484480432,
+      "grad_norm": 0.17730496605693405,
+      "learning_rate": 9.630546655606365e-08,
+      "loss": 0.0086,
+      "step": 899
+    },
+    {
+      "epoch": 4.8582995951417,
+      "grad_norm": 0.17063989794920645,
+      "learning_rate": 8.904521388245713e-08,
+      "loss": 0.007,
+      "step": 900
+    },
+    {
+      "epoch": 4.863697705802969,
+      "grad_norm": 0.14312385412531,
+      "learning_rate": 8.206884765818102e-08,
+      "loss": 0.0068,
+      "step": 901
+    },
+    {
+      "epoch": 4.869095816464237,
+      "grad_norm": 0.25375269225697994,
+      "learning_rate": 7.537646735087833e-08,
+      "loss": 0.151,
+      "step": 902
+    },
+    {
+      "epoch": 4.874493927125506,
+      "grad_norm": 0.20399369941957426,
+      "learning_rate": 6.896816837918651e-08,
+      "loss": 0.0079,
+      "step": 903
+    },
+    {
+      "epoch": 4.8798920377867745,
+      "grad_norm": 0.15160002733191705,
+      "learning_rate": 6.284404211135853e-08,
+      "loss": 0.0056,
+      "step": 904
+    },
+    {
+      "epoch": 4.8852901484480435,
+      "grad_norm": 0.166046396519856,
+      "learning_rate": 5.700417586397722e-08,
+      "loss": 0.0062,
+      "step": 905
+    },
+    {
+      "epoch": 4.890688259109312,
+      "grad_norm": 0.13344237216852384,
+      "learning_rate": 5.144865290069634e-08,
+      "loss": 0.0081,
+      "step": 906
+    },
+    {
+      "epoch": 4.896086369770581,
+      "grad_norm": 0.14790764013135596,
+      "learning_rate": 4.61775524310637e-08,
+      "loss": 0.0081,
+      "step": 907
+    },
+    {
+      "epoch": 4.901484480431849,
+      "grad_norm": 0.19876666586618377,
+      "learning_rate": 4.119094960938208e-08,
+      "loss": 0.0139,
+      "step": 908
+    },
+    {
+      "epoch": 4.906882591093117,
+      "grad_norm": 0.2035196764075396,
+      "learning_rate": 3.648891553365008e-08,
+      "loss": 0.0079,
+      "step": 909
+    },
+    {
+      "epoch": 4.912280701754386,
+      "grad_norm": 0.22562390555546766,
+      "learning_rate": 3.20715172445385e-08,
+      "loss": 0.161,
+      "step": 910
+    },
+    {
+      "epoch": 4.917678812415654,
+      "grad_norm": 0.16996305915931423,
+      "learning_rate": 2.793881772443552e-08,
+      "loss": 0.0073,
+      "step": 911
+    },
+    {
+      "epoch": 4.923076923076923,
+      "grad_norm": 0.23973035958426467,
+      "learning_rate": 2.4090875896551903e-08,
+      "loss": 0.1603,
+      "step": 912
+    },
+    {
+      "epoch": 4.9284750337381915,
+      "grad_norm": 0.22254682144620083,
+      "learning_rate": 2.0527746624081635e-08,
+      "loss": 0.0073,
+      "step": 913
+    },
+    {
+      "epoch": 4.933873144399461,
+      "grad_norm": 0.16951767245965615,
+      "learning_rate": 1.72494807094159e-08,
+      "loss": 0.0061,
+      "step": 914
+    },
+    {
+      "epoch": 4.939271255060729,
+      "grad_norm": 0.17637730727963327,
+      "learning_rate": 1.425612489341921e-08,
+      "loss": 0.0087,
+      "step": 915
+    },
+    {
+      "epoch": 4.944669365721998,
+      "grad_norm": 0.15410690010074907,
+      "learning_rate": 1.1547721854767712e-08,
+      "loss": 0.0066,
+      "step": 916
+    },
+    {
+      "epoch": 4.950067476383266,
+      "grad_norm": 0.15498759379346,
+      "learning_rate": 9.12431020933191e-09,
+      "loss": 0.0076,
+      "step": 917
+    },
+    {
+      "epoch": 4.955465587044534,
+      "grad_norm": 0.3295876521751464,
+      "learning_rate": 6.985924509639308e-09,
+      "loss": 0.0056,
+      "step": 918
+    },
+    {
+      "epoch": 4.960863697705803,
+      "grad_norm": 0.20206463746530587,
+      "learning_rate": 5.1325952443681544e-09,
+      "loss": 0.0086,
+      "step": 919
+    },
+    {
+      "epoch": 4.966261808367071,
+      "grad_norm": 0.20582338460555613,
+      "learning_rate": 3.5643488379211167e-09,
+      "loss": 0.0061,
+      "step": 920
+    },
+    {
+      "epoch": 4.97165991902834,
+      "grad_norm": 0.2043474190499627,
+      "learning_rate": 2.2812076500411397e-09,
+      "loss": 0.1423,
+      "step": 921
+    },
+    {
+      "epoch": 4.9770580296896085,
+      "grad_norm": 0.19464105830490305,
+      "learning_rate": 1.283189975505028e-09,
+      "loss": 0.0704,
+      "step": 922
+    },
+    {
+      "epoch": 4.982456140350877,
+      "grad_norm": 0.1561265636121022,
+      "learning_rate": 5.703100438481102e-10,
+      "loss": 0.0065,
+      "step": 923
+    },
+    {
+      "epoch": 4.987854251012146,
+      "grad_norm": 0.16103815887383013,
+      "learning_rate": 1.4257801917549884e-10,
+      "loss": 0.0069,
+      "step": 924
+    },
+    {
+      "epoch": 4.993252361673415,
+      "grad_norm": 0.1456783005569135,
+      "learning_rate": 0.0,
+      "loss": 0.0078,
+      "step": 925
+    },
+    {
+      "epoch": 4.993252361673415,
+      "step": 925,
+      "total_flos": 1.7623108712815e+17,
+      "train_loss": 0.26654424302078583,
+      "train_runtime": 3168.6293,
+      "train_samples_per_second": 37.398,
+      "train_steps_per_second": 0.292
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 925,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.7623108712815e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}