diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8652 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.989367088607595,
+  "eval_steps": 500,
+  "global_step": 1230,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004012036108324975,
+      "grad_norm": 6.612851998766499,
+      "learning_rate": 3.2e-07,
+      "loss": 1.0629,
+      "step": 1
+    },
+    {
+      "epoch": 0.00802407221664995,
+      "grad_norm": 6.573050487422954,
+      "learning_rate": 6.4e-07,
+      "loss": 1.0651,
+      "step": 2
+    },
+    {
+      "epoch": 0.012036108324974924,
+      "grad_norm": 6.725762103014051,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 1.0823,
+      "step": 3
+    },
+    {
+      "epoch": 0.0160481444332999,
+      "grad_norm": 6.397212782661277,
+      "learning_rate": 1.28e-06,
+      "loss": 1.0671,
+      "step": 4
+    },
+    {
+      "epoch": 0.020060180541624874,
+      "grad_norm": 6.099609580051789,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 1.0516,
+      "step": 5
+    },
+    {
+      "epoch": 0.024072216649949848,
+      "grad_norm": 5.981394290587972,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 1.0184,
+      "step": 6
+    },
+    {
+      "epoch": 0.028084252758274825,
+      "grad_norm": 4.861643103953028,
+      "learning_rate": 2.24e-06,
+      "loss": 1.001,
+      "step": 7
+    },
+    {
+      "epoch": 0.0320962888665998,
+      "grad_norm": 4.546889554265633,
+      "learning_rate": 2.56e-06,
+      "loss": 1.0014,
+      "step": 8
+    },
+    {
+      "epoch": 0.03610832497492478,
+      "grad_norm": 2.664216929534807,
+      "learning_rate": 2.88e-06,
+      "loss": 0.9665,
+      "step": 9
+    },
+    {
+      "epoch": 0.04012036108324975,
+      "grad_norm": 2.2958963287491323,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.9249,
+      "step": 10
+    },
+    {
+      "epoch": 0.044132397191574725,
+      "grad_norm": 2.05271269720836,
+      "learning_rate": 3.52e-06,
+      "loss": 0.9005,
+      "step": 11
+    },
+    {
+      "epoch": 0.048144433299899696,
+      "grad_norm": 4.512114254121433,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 0.9884,
+      "step": 12
+    },
+    {
+      "epoch": 0.05215646940822467,
+      "grad_norm": 4.475645355384806,
+      "learning_rate": 4.16e-06,
+      "loss": 0.9352,
+      "step": 13
+    },
+    {
+      "epoch": 0.05616850551654965,
+      "grad_norm": 4.391439743940724,
+      "learning_rate": 4.48e-06,
+      "loss": 0.886,
+      "step": 14
+    },
+    {
+      "epoch": 0.06018054162487462,
+      "grad_norm": 4.4330339838914785,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.9289,
+      "step": 15
+    },
+    {
+      "epoch": 0.0641925777331996,
+      "grad_norm": 3.2330424839883,
+      "learning_rate": 5.12e-06,
+      "loss": 0.894,
+      "step": 16
+    },
+    {
+      "epoch": 0.06820461384152457,
+      "grad_norm": 2.9825255370177093,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 0.8662,
+      "step": 17
+    },
+    {
+      "epoch": 0.07221664994984955,
+      "grad_norm": 2.841034044866774,
+      "learning_rate": 5.76e-06,
+      "loss": 0.8784,
+      "step": 18
+    },
+    {
+      "epoch": 0.07622868605817452,
+      "grad_norm": 2.459961183005845,
+      "learning_rate": 6.08e-06,
+      "loss": 0.8658,
+      "step": 19
+    },
+    {
+      "epoch": 0.0802407221664995,
+      "grad_norm": 2.184693142285669,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 0.837,
+      "step": 20
+    },
+    {
+      "epoch": 0.08425275827482448,
+      "grad_norm": 1.7237912612029918,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 0.8117,
+      "step": 21
+    },
+    {
+      "epoch": 0.08826479438314945,
+      "grad_norm": 1.7859872265706882,
+      "learning_rate": 7.04e-06,
+      "loss": 0.8189,
+      "step": 22
+    },
+    {
+      "epoch": 0.09227683049147442,
+      "grad_norm": 1.8897349914635586,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 0.7982,
+      "step": 23
+    },
+    {
+      "epoch": 0.09628886659979939,
+      "grad_norm": 1.8643147838314422,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 0.7923,
+      "step": 24
+    },
+    {
+      "epoch": 0.10030090270812438,
+      "grad_norm": 1.5253907659878532,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.8331,
+      "step": 25
+    },
+    {
+      "epoch": 0.10431293881644935,
+      "grad_norm": 1.1040179596357251,
+      "learning_rate": 8.32e-06,
+      "loss": 0.7942,
+      "step": 26
+    },
+    {
+      "epoch": 0.10832497492477432,
+      "grad_norm": 1.2648400093933025,
+      "learning_rate": 8.64e-06,
+      "loss": 0.7794,
+      "step": 27
+    },
+    {
+      "epoch": 0.1123370110330993,
+      "grad_norm": 1.2860859750010947,
+      "learning_rate": 8.96e-06,
+      "loss": 0.77,
+      "step": 28
+    },
+    {
+      "epoch": 0.11634904714142427,
+      "grad_norm": 1.0854009473006938,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 0.7667,
+      "step": 29
+    },
+    {
+      "epoch": 0.12036108324974924,
+      "grad_norm": 0.8521397512225974,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.7497,
+      "step": 30
+    },
+    {
+      "epoch": 0.12437311935807423,
+      "grad_norm": 1.0904725375786488,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 0.7536,
+      "step": 31
+    },
+    {
+      "epoch": 0.1283851554663992,
+      "grad_norm": 0.99872778798552,
+      "learning_rate": 1.024e-05,
+      "loss": 0.7612,
+      "step": 32
+    },
+    {
+      "epoch": 0.13239719157472418,
+      "grad_norm": 0.8153604252790577,
+      "learning_rate": 1.056e-05,
+      "loss": 0.7479,
+      "step": 33
+    },
+    {
+      "epoch": 0.13640922768304914,
+      "grad_norm": 0.7824028296699954,
+      "learning_rate": 1.0880000000000001e-05,
+      "loss": 0.7779,
+      "step": 34
+    },
+    {
+      "epoch": 0.14042126379137412,
+      "grad_norm": 0.9019613313335676,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 0.7669,
+      "step": 35
+    },
+    {
+      "epoch": 0.1444332998996991,
+      "grad_norm": 0.8495364535168003,
+      "learning_rate": 1.152e-05,
+      "loss": 0.7551,
+      "step": 36
+    },
+    {
+      "epoch": 0.14844533600802406,
+      "grad_norm": 0.6988266031561519,
+      "learning_rate": 1.184e-05,
+      "loss": 0.7517,
+      "step": 37
+    },
+    {
+      "epoch": 0.15245737211634905,
+      "grad_norm": 0.7297808622075487,
+      "learning_rate": 1.216e-05,
+      "loss": 0.7477,
+      "step": 38
+    },
+    {
+      "epoch": 0.15646940822467403,
+      "grad_norm": 0.7777427228388275,
+      "learning_rate": 1.248e-05,
+      "loss": 0.7515,
+      "step": 39
+    },
+    {
+      "epoch": 0.160481444332999,
+      "grad_norm": 0.7350633464498506,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 0.7393,
+      "step": 40
+    },
+    {
+      "epoch": 0.16449348044132397,
+      "grad_norm": 0.6473988796558481,
+      "learning_rate": 1.3120000000000001e-05,
+      "loss": 0.7195,
+      "step": 41
+    },
+    {
+      "epoch": 0.16850551654964896,
+      "grad_norm": 0.7513149493332046,
+      "learning_rate": 1.3440000000000002e-05,
+      "loss": 0.7409,
+      "step": 42
+    },
+    {
+      "epoch": 0.17251755265797392,
+      "grad_norm": 0.6765737631202615,
+      "learning_rate": 1.376e-05,
+      "loss": 0.729,
+      "step": 43
+    },
+    {
+      "epoch": 0.1765295887662989,
+      "grad_norm": 0.7954448530656657,
+      "learning_rate": 1.408e-05,
+      "loss": 0.7094,
+      "step": 44
+    },
+    {
+      "epoch": 0.18054162487462388,
+      "grad_norm": 0.5845307808118603,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 0.726,
+      "step": 45
+    },
+    {
+      "epoch": 0.18455366098294884,
+      "grad_norm": 0.7388167700697004,
+      "learning_rate": 1.4720000000000001e-05,
+      "loss": 0.7311,
+      "step": 46
+    },
+    {
+      "epoch": 0.18856569709127383,
+      "grad_norm": 0.7418097882167217,
+      "learning_rate": 1.5040000000000002e-05,
+      "loss": 0.7381,
+      "step": 47
+    },
+    {
+      "epoch": 0.19257773319959878,
+      "grad_norm": 0.6242473102288691,
+      "learning_rate": 1.5360000000000002e-05,
+      "loss": 0.7479,
+      "step": 48
+    },
+    {
+      "epoch": 0.19658976930792377,
+      "grad_norm": 0.8457801328139297,
+      "learning_rate": 1.5680000000000002e-05,
+      "loss": 0.726,
+      "step": 49
+    },
+    {
+      "epoch": 0.20060180541624875,
+      "grad_norm": 0.6962550850297534,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.7232,
+      "step": 50
+    },
+    {
+      "epoch": 0.2046138415245737,
+      "grad_norm": 0.7449908064653729,
+      "learning_rate": 1.632e-05,
+      "loss": 0.6954,
+      "step": 51
+    },
+    {
+      "epoch": 0.2086258776328987,
+      "grad_norm": 0.767778352232827,
+      "learning_rate": 1.664e-05,
+      "loss": 0.6998,
+      "step": 52
+    },
+    {
+      "epoch": 0.21263791374122368,
+      "grad_norm": 0.7688669555117182,
+      "learning_rate": 1.696e-05,
+      "loss": 0.7304,
+      "step": 53
+    },
+    {
+      "epoch": 0.21664994984954863,
+      "grad_norm": 0.6485837848159103,
+      "learning_rate": 1.728e-05,
+      "loss": 0.7249,
+      "step": 54
+    },
+    {
+      "epoch": 0.22066198595787362,
+      "grad_norm": 0.7786995782041959,
+      "learning_rate": 1.76e-05,
+      "loss": 0.7208,
+      "step": 55
+    },
+    {
+      "epoch": 0.2246740220661986,
+      "grad_norm": 0.6752645347449635,
+      "learning_rate": 1.792e-05,
+      "loss": 0.7264,
+      "step": 56
+    },
+    {
+      "epoch": 0.22868605817452356,
+      "grad_norm": 0.7496398046674226,
+      "learning_rate": 1.824e-05,
+      "loss": 0.7036,
+      "step": 57
+    },
+    {
+      "epoch": 0.23269809428284854,
+      "grad_norm": 0.7623985720679161,
+      "learning_rate": 1.8560000000000002e-05,
+      "loss": 0.7252,
+      "step": 58
+    },
+    {
+      "epoch": 0.23671013039117353,
+      "grad_norm": 0.7861458035762849,
+      "learning_rate": 1.8880000000000002e-05,
+      "loss": 0.7033,
+      "step": 59
+    },
+    {
+      "epoch": 0.24072216649949849,
+      "grad_norm": 0.6621454126360683,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 0.7038,
+      "step": 60
+    },
+    {
+      "epoch": 0.24473420260782347,
+      "grad_norm": 0.8781061932563883,
+      "learning_rate": 1.9520000000000003e-05,
+      "loss": 0.7009,
+      "step": 61
+    },
+    {
+      "epoch": 0.24874623871614845,
+      "grad_norm": 0.5952684605240373,
+      "learning_rate": 1.9840000000000003e-05,
+      "loss": 0.6995,
+      "step": 62
+    },
+    {
+      "epoch": 0.25275827482447344,
+      "grad_norm": 0.7434218969525137,
+      "learning_rate": 2.016e-05,
+      "loss": 0.6912,
+      "step": 63
+    },
+    {
+      "epoch": 0.2567703109327984,
+      "grad_norm": 0.8515137173174638,
+      "learning_rate": 2.048e-05,
+      "loss": 0.7159,
+      "step": 64
+    },
+    {
+      "epoch": 0.26078234704112335,
+      "grad_norm": 0.6439348438506264,
+      "learning_rate": 2.08e-05,
+      "loss": 0.6906,
+      "step": 65
+    },
+    {
+      "epoch": 0.26479438314944836,
+      "grad_norm": 0.7105102518250962,
+      "learning_rate": 2.112e-05,
+      "loss": 0.6703,
+      "step": 66
+    },
+    {
+      "epoch": 0.2688064192577733,
+      "grad_norm": 0.6253101948926647,
+      "learning_rate": 2.144e-05,
+      "loss": 0.713,
+      "step": 67
+    },
+    {
+      "epoch": 0.2728184553660983,
+      "grad_norm": 0.8039572423592197,
+      "learning_rate": 2.1760000000000002e-05,
+      "loss": 0.6858,
+      "step": 68
+    },
+    {
+      "epoch": 0.2768304914744233,
+      "grad_norm": 1.0479886353752794,
+      "learning_rate": 2.2080000000000002e-05,
+      "loss": 0.6847,
+      "step": 69
+    },
+    {
+      "epoch": 0.28084252758274825,
+      "grad_norm": 0.6635952287037824,
+      "learning_rate": 2.2400000000000002e-05,
+      "loss": 0.6812,
+      "step": 70
+    },
+    {
+      "epoch": 0.2848545636910732,
+      "grad_norm": 0.7975243045915003,
+      "learning_rate": 2.272e-05,
+      "loss": 0.7123,
+      "step": 71
+    },
+    {
+      "epoch": 0.2888665997993982,
+      "grad_norm": 0.9202985707625535,
+      "learning_rate": 2.304e-05,
+      "loss": 0.6809,
+      "step": 72
+    },
+    {
+      "epoch": 0.2928786359077232,
+      "grad_norm": 0.7634755379949932,
+      "learning_rate": 2.336e-05,
+      "loss": 0.693,
+      "step": 73
+    },
+    {
+      "epoch": 0.29689067201604813,
+      "grad_norm": 0.7729657056673682,
+      "learning_rate": 2.368e-05,
+      "loss": 0.6878,
+      "step": 74
+    },
+    {
+      "epoch": 0.30090270812437314,
+      "grad_norm": 0.7820964553693485,
+      "learning_rate": 2.4e-05,
+      "loss": 0.6485,
+      "step": 75
+    },
+    {
+      "epoch": 0.3049147442326981,
+      "grad_norm": 1.0561986101533776,
+      "learning_rate": 2.432e-05,
+      "loss": 0.6786,
+      "step": 76
+    },
+    {
+      "epoch": 0.30892678034102306,
+      "grad_norm": 0.6833847244851405,
+      "learning_rate": 2.464e-05,
+      "loss": 0.7066,
+      "step": 77
+    },
+    {
+      "epoch": 0.31293881644934807,
+      "grad_norm": 0.6610321170610101,
+      "learning_rate": 2.496e-05,
+      "loss": 0.6959,
+      "step": 78
+    },
+    {
+      "epoch": 0.316950852557673,
+      "grad_norm": 0.8571354788755239,
+      "learning_rate": 2.5280000000000002e-05,
+      "loss": 0.7013,
+      "step": 79
+    },
+    {
+      "epoch": 0.320962888665998,
+      "grad_norm": 0.6405947958045713,
+      "learning_rate": 2.5600000000000002e-05,
+      "loss": 0.6751,
+      "step": 80
+    },
+    {
+      "epoch": 0.324974924774323,
+      "grad_norm": 0.8921969130442118,
+      "learning_rate": 2.5920000000000003e-05,
+      "loss": 0.6508,
+      "step": 81
+    },
+    {
+      "epoch": 0.32898696088264795,
+      "grad_norm": 0.6343439758955924,
+      "learning_rate": 2.6240000000000003e-05,
+      "loss": 0.699,
+      "step": 82
+    },
+    {
+      "epoch": 0.3329989969909729,
+      "grad_norm": 0.847936895363307,
+      "learning_rate": 2.6560000000000003e-05,
+      "loss": 0.6991,
+      "step": 83
+    },
+    {
+      "epoch": 0.3370110330992979,
+      "grad_norm": 0.7049415665018225,
+      "learning_rate": 2.6880000000000004e-05,
+      "loss": 0.6758,
+      "step": 84
+    },
+    {
+      "epoch": 0.3410230692076229,
+      "grad_norm": 0.8241773684410245,
+      "learning_rate": 2.7200000000000004e-05,
+      "loss": 0.7108,
+      "step": 85
+    },
+    {
+      "epoch": 0.34503510531594783,
+      "grad_norm": 0.8442973070422346,
+      "learning_rate": 2.752e-05,
+      "loss": 0.6832,
+      "step": 86
+    },
+    {
+      "epoch": 0.34904714142427284,
+      "grad_norm": 0.6786123797152089,
+      "learning_rate": 2.784e-05,
+      "loss": 0.6681,
+      "step": 87
+    },
+    {
+      "epoch": 0.3530591775325978,
+      "grad_norm": 0.6626369599112003,
+      "learning_rate": 2.816e-05,
+      "loss": 0.6683,
+      "step": 88
+    },
+    {
+      "epoch": 0.35707121364092276,
+      "grad_norm": 0.7240363876177163,
+      "learning_rate": 2.8480000000000002e-05,
+      "loss": 0.6953,
+      "step": 89
+    },
+    {
+      "epoch": 0.36108324974924777,
+      "grad_norm": 0.8065134586866273,
+      "learning_rate": 2.8800000000000002e-05,
+      "loss": 0.6926,
+      "step": 90
+    },
+    {
+      "epoch": 0.3650952858575727,
+      "grad_norm": 0.9211441638549288,
+      "learning_rate": 2.9120000000000002e-05,
+      "loss": 0.6601,
+      "step": 91
+    },
+    {
+      "epoch": 0.3691073219658977,
+      "grad_norm": 0.810127103419794,
+      "learning_rate": 2.9440000000000003e-05,
+      "loss": 0.6782,
+      "step": 92
+    },
+    {
+      "epoch": 0.3731193580742227,
+      "grad_norm": 0.8498838091341564,
+      "learning_rate": 2.9760000000000003e-05,
+      "loss": 0.6897,
+      "step": 93
+    },
+    {
+      "epoch": 0.37713139418254765,
+      "grad_norm": 1.0588407105675952,
+      "learning_rate": 3.0080000000000003e-05,
+      "loss": 0.6607,
+      "step": 94
+    },
+    {
+      "epoch": 0.3811434302908726,
+      "grad_norm": 0.7581183890713428,
+      "learning_rate": 3.0400000000000004e-05,
+      "loss": 0.6788,
+      "step": 95
+    },
+    {
+      "epoch": 0.38515546639919757,
+      "grad_norm": 0.7501196426567234,
+      "learning_rate": 3.0720000000000004e-05,
+      "loss": 0.6533,
+      "step": 96
+    },
+    {
+      "epoch": 0.3891675025075226,
+      "grad_norm": 0.8688700296185009,
+      "learning_rate": 3.104e-05,
+      "loss": 0.6761,
+      "step": 97
+    },
+    {
+      "epoch": 0.39317953861584753,
+      "grad_norm": 0.7001189235438992,
+      "learning_rate": 3.1360000000000005e-05,
+      "loss": 0.6499,
+      "step": 98
+    },
+    {
+      "epoch": 0.3971915747241725,
+      "grad_norm": 0.7650334955403152,
+      "learning_rate": 3.168e-05,
+      "loss": 0.6757,
+      "step": 99
+    },
+    {
+      "epoch": 0.4012036108324975,
+      "grad_norm": 0.7124778818597767,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.6669,
+      "step": 100
+    },
+    {
+      "epoch": 0.40521564694082246,
+      "grad_norm": 0.920263143749945,
+      "learning_rate": 3.232e-05,
+      "loss": 0.6747,
+      "step": 101
+    },
+    {
+      "epoch": 0.4092276830491474,
+      "grad_norm": 0.833199689092061,
+      "learning_rate": 3.264e-05,
+      "loss": 0.678,
+      "step": 102
+    },
+    {
+      "epoch": 0.41323971915747243,
+      "grad_norm": 0.9431084466687941,
+      "learning_rate": 3.296e-05,
+      "loss": 0.6733,
+      "step": 103
+    },
+    {
+      "epoch": 0.4172517552657974,
+      "grad_norm": 0.9433177194880746,
+      "learning_rate": 3.328e-05,
+      "loss": 0.6874,
+      "step": 104
+    },
+    {
+      "epoch": 0.42126379137412234,
+      "grad_norm": 0.7672992421574426,
+      "learning_rate": 3.3600000000000004e-05,
+      "loss": 0.6722,
+      "step": 105
+    },
+    {
+      "epoch": 0.42527582748244735,
+      "grad_norm": 0.7341459808362081,
+      "learning_rate": 3.392e-05,
+      "loss": 0.6738,
+      "step": 106
+    },
+    {
+      "epoch": 0.4292878635907723,
+      "grad_norm": 0.7420564048052237,
+      "learning_rate": 3.4240000000000004e-05,
+      "loss": 0.6769,
+      "step": 107
+    },
+    {
+      "epoch": 0.43329989969909727,
+      "grad_norm": 0.8067334516501767,
+      "learning_rate": 3.456e-05,
+      "loss": 0.6688,
+      "step": 108
+    },
+    {
+      "epoch": 0.4373119358074223,
+      "grad_norm": 0.6307015567061565,
+      "learning_rate": 3.4880000000000005e-05,
+      "loss": 0.6754,
+      "step": 109
+    },
+    {
+      "epoch": 0.44132397191574724,
+      "grad_norm": 0.8956635708314229,
+      "learning_rate": 3.52e-05,
+      "loss": 0.6732,
+      "step": 110
+    },
+    {
+      "epoch": 0.4453360080240722,
+      "grad_norm": 0.9824226256184038,
+      "learning_rate": 3.5520000000000006e-05,
+      "loss": 0.6878,
+      "step": 111
+    },
+    {
+      "epoch": 0.4493480441323972,
+      "grad_norm": 0.7165479439019142,
+      "learning_rate": 3.584e-05,
+      "loss": 0.6581,
+      "step": 112
+    },
+    {
+      "epoch": 0.45336008024072216,
+      "grad_norm": 0.9304865971468473,
+      "learning_rate": 3.6160000000000006e-05,
+      "loss": 0.6674,
+      "step": 113
+    },
+    {
+      "epoch": 0.4573721163490471,
+      "grad_norm": 1.0280561009386016,
+      "learning_rate": 3.648e-05,
+      "loss": 0.6605,
+      "step": 114
+    },
+    {
+      "epoch": 0.46138415245737213,
+      "grad_norm": 0.7931760305262244,
+      "learning_rate": 3.680000000000001e-05,
+      "loss": 0.6873,
+      "step": 115
+    },
+    {
+      "epoch": 0.4653961885656971,
+      "grad_norm": 0.7377542604792668,
+      "learning_rate": 3.7120000000000004e-05,
+      "loss": 0.6765,
+      "step": 116
+    },
+    {
+      "epoch": 0.46940822467402205,
+      "grad_norm": 0.7466448220048797,
+      "learning_rate": 3.744000000000001e-05,
+      "loss": 0.6567,
+      "step": 117
+    },
+    {
+      "epoch": 0.47342026078234706,
+      "grad_norm": 0.6822587637436976,
+      "learning_rate": 3.7760000000000004e-05,
+      "loss": 0.6605,
+      "step": 118
+    },
+    {
+      "epoch": 0.477432296890672,
+      "grad_norm": 0.7602451103276817,
+      "learning_rate": 3.808e-05,
+      "loss": 0.6759,
+      "step": 119
+    },
+    {
+      "epoch": 0.48144433299899697,
+      "grad_norm": 0.8200012173574417,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 0.6779,
+      "step": 120
+    },
+    {
+      "epoch": 0.485456369107322,
+      "grad_norm": 0.9753695075147062,
+      "learning_rate": 3.872e-05,
+      "loss": 0.6813,
+      "step": 121
+    },
+    {
+      "epoch": 0.48946840521564694,
+      "grad_norm": 0.803261210370899,
+      "learning_rate": 3.9040000000000006e-05,
+      "loss": 0.6598,
+      "step": 122
+    },
+    {
+      "epoch": 0.4934804413239719,
+      "grad_norm": 0.7109658147808896,
+      "learning_rate": 3.936e-05,
+      "loss": 0.6447,
+      "step": 123
+    },
+    {
+      "epoch": 0.4974924774322969,
+      "grad_norm": 0.8880356087467619,
+      "learning_rate": 3.9680000000000006e-05,
+      "loss": 0.6604,
+      "step": 124
+    },
+    {
+      "epoch": 0.5015045135406219,
+      "grad_norm": 0.6711601892318234,
+      "learning_rate": 4e-05,
+      "loss": 0.687,
+      "step": 125
+    },
+    {
+      "epoch": 0.5055165496489469,
+      "grad_norm": 0.7795220266844257,
+      "learning_rate": 3.9999921320169564e-05,
+      "loss": 0.6736,
+      "step": 126
+    },
+    {
+      "epoch": 0.5095285857572718,
+      "grad_norm": 0.9065638215596524,
+      "learning_rate": 3.999968528129731e-05,
+      "loss": 0.6772,
+      "step": 127
+    },
+    {
+      "epoch": 0.5135406218655968,
+      "grad_norm": 0.7943979987322751,
+      "learning_rate": 3.9999291885240385e-05,
+      "loss": 0.6711,
+      "step": 128
+    },
+    {
+      "epoch": 0.5175526579739218,
+      "grad_norm": 0.8044221440131943,
+      "learning_rate": 3.9998741135094016e-05,
+      "loss": 0.6761,
+      "step": 129
+    },
+    {
+      "epoch": 0.5215646940822467,
+      "grad_norm": 0.8971731886906648,
+      "learning_rate": 3.9998033035191504e-05,
+      "loss": 0.6848,
+      "step": 130
+    },
+    {
+      "epoch": 0.5255767301905717,
+      "grad_norm": 0.8538880223433852,
+      "learning_rate": 3.999716759110417e-05,
+      "loss": 0.6813,
+      "step": 131
+    },
+    {
+      "epoch": 0.5295887662988967,
+      "grad_norm": 0.7718951813970398,
+      "learning_rate": 3.9996144809641296e-05,
+      "loss": 0.6836,
+      "step": 132
+    },
+    {
+      "epoch": 0.5336008024072216,
+      "grad_norm": 0.8844893474448337,
+      "learning_rate": 3.999496469885014e-05,
+      "loss": 0.6821,
+      "step": 133
+    },
+    {
+      "epoch": 0.5376128385155466,
+      "grad_norm": 0.728840567886365,
+      "learning_rate": 3.999362726801576e-05,
+      "loss": 0.6731,
+      "step": 134
+    },
+    {
+      "epoch": 0.5416248746238717,
+      "grad_norm": 0.9551374593295036,
+      "learning_rate": 3.9992132527661065e-05,
+      "loss": 0.6993,
+      "step": 135
+    },
+    {
+      "epoch": 0.5456369107321966,
+      "grad_norm": 0.9090635222051794,
+      "learning_rate": 3.999048048954663e-05,
+      "loss": 0.6848,
+      "step": 136
+    },
+    {
+      "epoch": 0.5496489468405216,
+      "grad_norm": 0.6112577716524418,
+      "learning_rate": 3.998867116667067e-05,
+      "loss": 0.6748,
+      "step": 137
+    },
+    {
+      "epoch": 0.5536609829488466,
+      "grad_norm": 0.9293523389477538,
+      "learning_rate": 3.998670457326891e-05,
+      "loss": 0.6781,
+      "step": 138
+    },
+    {
+      "epoch": 0.5576730190571715,
+      "grad_norm": 0.7486580083516655,
+      "learning_rate": 3.9984580724814464e-05,
+      "loss": 0.6371,
+      "step": 139
+    },
+    {
+      "epoch": 0.5616850551654965,
+      "grad_norm": 0.7551550330691811,
+      "learning_rate": 3.998229963801774e-05,
+      "loss": 0.64,
+      "step": 140
+    },
+    {
+      "epoch": 0.5656970912738215,
+      "grad_norm": 0.7427182366440193,
+      "learning_rate": 3.9979861330826295e-05,
+      "loss": 0.6464,
+      "step": 141
+    },
+    {
+      "epoch": 0.5697091273821464,
+      "grad_norm": 0.692419918900504,
+      "learning_rate": 3.997726582242469e-05,
+      "loss": 0.673,
+      "step": 142
+    },
+    {
+      "epoch": 0.5737211634904714,
+      "grad_norm": 0.6587573618484559,
+      "learning_rate": 3.997451313323433e-05,
+      "loss": 0.6635,
+      "step": 143
+    },
+    {
+      "epoch": 0.5777331995987964,
+      "grad_norm": 0.8286910062445207,
+      "learning_rate": 3.997160328491333e-05,
+      "loss": 0.6691,
+      "step": 144
+    },
+    {
+      "epoch": 0.5817452357071213,
+      "grad_norm": 0.7316534837132789,
+      "learning_rate": 3.996853630035634e-05,
+      "loss": 0.6731,
+      "step": 145
+    },
+    {
+      "epoch": 0.5857572718154463,
+      "grad_norm": 0.6647027945565223,
+      "learning_rate": 3.996531220369432e-05,
+      "loss": 0.6614,
+      "step": 146
+    },
+    {
+      "epoch": 0.5897693079237714,
+      "grad_norm": 0.7819215606955189,
+      "learning_rate": 3.996193102029443e-05,
+      "loss": 0.6434,
+      "step": 147
+    },
+    {
+      "epoch": 0.5937813440320963,
+      "grad_norm": 0.5435529121383756,
+      "learning_rate": 3.995839277675975e-05,
+      "loss": 0.672,
+      "step": 148
+    },
+    {
+      "epoch": 0.5977933801404213,
+      "grad_norm": 0.6269522264831056,
+      "learning_rate": 3.995469750092912e-05,
+      "loss": 0.6565,
+      "step": 149
+    },
+    {
+      "epoch": 0.6018054162487463,
+      "grad_norm": 0.6458032413303357,
+      "learning_rate": 3.9950845221876926e-05,
+      "loss": 0.6663,
+      "step": 150
+    },
+    {
+      "epoch": 0.6058174523570712,
+      "grad_norm": 0.6401681643628693,
+      "learning_rate": 3.994683596991281e-05,
+      "loss": 0.6762,
+      "step": 151
+    },
+    {
+      "epoch": 0.6098294884653962,
+      "grad_norm": 0.781280971315514,
+      "learning_rate": 3.994266977658151e-05,
+      "loss": 0.6737,
+      "step": 152
+    },
+    {
+      "epoch": 0.6138415245737212,
+      "grad_norm": 1.0559529611747647,
+      "learning_rate": 3.9938346674662565e-05,
+      "loss": 0.6813,
+      "step": 153
+    },
+    {
+      "epoch": 0.6178535606820461,
+      "grad_norm": 0.6195586660543734,
+      "learning_rate": 3.9933866698170065e-05,
+      "loss": 0.6382,
+      "step": 154
+    },
+    {
+      "epoch": 0.6218655967903711,
+      "grad_norm": 0.7703511660846133,
+      "learning_rate": 3.992922988235239e-05,
+      "loss": 0.6438,
+      "step": 155
+    },
+    {
+      "epoch": 0.6258776328986961,
+      "grad_norm": 0.6832785415006298,
+      "learning_rate": 3.992443626369192e-05,
+      "loss": 0.6669,
+      "step": 156
+    },
+    {
+      "epoch": 0.629889669007021,
+      "grad_norm": 0.5896774850755083,
+      "learning_rate": 3.991948587990479e-05,
+      "loss": 0.655,
+      "step": 157
+    },
+    {
+      "epoch": 0.633901705115346,
+      "grad_norm": 0.676143021757847,
+      "learning_rate": 3.9914378769940506e-05,
+      "loss": 0.6531,
+      "step": 158
+    },
+    {
+      "epoch": 0.6379137412236711,
+      "grad_norm": 0.5881255193761082,
+      "learning_rate": 3.990911497398174e-05,
+      "loss": 0.6557,
+      "step": 159
+    },
+    {
+      "epoch": 0.641925777331996,
+      "grad_norm": 0.6081104446257762,
+      "learning_rate": 3.990369453344394e-05,
+      "loss": 0.6551,
+      "step": 160
+    },
+    {
+      "epoch": 0.645937813440321,
+      "grad_norm": 0.6248203292839142,
+      "learning_rate": 3.989811749097505e-05,
+      "loss": 0.6605,
+      "step": 161
+    },
+    {
+      "epoch": 0.649949849548646,
+      "grad_norm": 0.6035810871038287,
+      "learning_rate": 3.989238389045514e-05,
+      "loss": 0.6547,
+      "step": 162
+    },
+    {
+      "epoch": 0.6539618856569709,
+      "grad_norm": 0.7926731069152031,
+      "learning_rate": 3.988649377699608e-05,
+      "loss": 0.6226,
+      "step": 163
+    },
+    {
+      "epoch": 0.6579739217652959,
+      "grad_norm": 0.6959472818612854,
+      "learning_rate": 3.988044719694119e-05,
+      "loss": 0.6332,
+      "step": 164
+    },
+    {
+      "epoch": 0.6619859578736209,
+      "grad_norm": 0.766303643141116,
+      "learning_rate": 3.9874244197864856e-05,
+      "loss": 0.6665,
+      "step": 165
+    },
+    {
+      "epoch": 0.6659979939819458,
+      "grad_norm": 0.5995589318743519,
+      "learning_rate": 3.9867884828572163e-05,
+      "loss": 0.6721,
+      "step": 166
+    },
+    {
+      "epoch": 0.6700100300902708,
+      "grad_norm": 0.6108832789673108,
+      "learning_rate": 3.986136913909853e-05,
+      "loss": 0.6357,
+      "step": 167
+    },
+    {
+      "epoch": 0.6740220661985958,
+      "grad_norm": 0.5802664532414953,
+      "learning_rate": 3.9854697180709283e-05,
+      "loss": 0.625,
+      "step": 168
+    },
+    {
+      "epoch": 0.6780341023069207,
+      "grad_norm": 0.5988247998772378,
+      "learning_rate": 3.984786900589929e-05,
+      "loss": 0.6686,
+      "step": 169
+    },
+    {
+      "epoch": 0.6820461384152458,
+      "grad_norm": 0.566551446365298,
+      "learning_rate": 3.9840884668392495e-05,
+      "loss": 0.6442,
+      "step": 170
+    },
+    {
+      "epoch": 0.6860581745235708,
+      "grad_norm": 0.7722998678351436,
+      "learning_rate": 3.9833744223141565e-05,
+      "loss": 0.6595,
+      "step": 171
+    },
+    {
+      "epoch": 0.6900702106318957,
+      "grad_norm": 0.7810169467015965,
+      "learning_rate": 3.9826447726327395e-05,
+      "loss": 0.6528,
+      "step": 172
+    },
+    {
+      "epoch": 0.6940822467402207,
+      "grad_norm": 0.6395402458938686,
+      "learning_rate": 3.98189952353587e-05,
+      "loss": 0.6588,
+      "step": 173
+    },
+    {
+      "epoch": 0.6980942828485457,
+      "grad_norm": 0.7047599562391664,
+      "learning_rate": 3.981138680887154e-05,
+      "loss": 0.6624,
+      "step": 174
+    },
+    {
+      "epoch": 0.7021063189568706,
+      "grad_norm": 0.6273745924712153,
+      "learning_rate": 3.9803622506728916e-05,
+      "loss": 0.6452,
+      "step": 175
+    },
+    {
+      "epoch": 0.7061183550651956,
+      "grad_norm": 0.5548997802524247,
+      "learning_rate": 3.97957023900202e-05,
+      "loss": 0.6612,
+      "step": 176
+    },
+    {
+      "epoch": 0.7101303911735206,
+      "grad_norm": 0.6595797152990318,
+      "learning_rate": 3.9787626521060736e-05,
+      "loss": 0.6478,
+      "step": 177
+    },
+    {
+      "epoch": 0.7141424272818455,
+      "grad_norm": 0.6378873771790127,
+      "learning_rate": 3.977939496339134e-05,
+      "loss": 0.647,
+      "step": 178
+    },
+    {
+      "epoch": 0.7181544633901705,
+      "grad_norm": 0.5880209564871013,
+      "learning_rate": 3.977100778177775e-05,
+      "loss": 0.6322,
+      "step": 179
+    },
+    {
+      "epoch": 0.7221664994984955,
+      "grad_norm": 0.7887481327033323,
+      "learning_rate": 3.976246504221018e-05,
+      "loss": 0.6284,
+      "step": 180
+    },
+    {
+      "epoch": 0.7261785356068204,
+      "grad_norm": 0.6766759082879258,
+      "learning_rate": 3.9753766811902756e-05,
+      "loss": 0.6568,
+      "step": 181
+    },
+    {
+      "epoch": 0.7301905717151455,
+      "grad_norm": 0.675038232957483,
+      "learning_rate": 3.974491315929301e-05,
+      "loss": 0.6547,
+      "step": 182
+    },
+    {
+      "epoch": 0.7342026078234705,
+      "grad_norm": 0.8842048624887232,
+      "learning_rate": 3.9735904154041335e-05,
+      "loss": 0.6633,
+      "step": 183
+    },
+    {
+      "epoch": 0.7382146439317954,
+      "grad_norm": 0.6932980000439893,
+      "learning_rate": 3.972673986703042e-05,
+      "loss": 0.6546,
+      "step": 184
+    },
+    {
+      "epoch": 0.7422266800401204,
+      "grad_norm": 0.7287436247809296,
+      "learning_rate": 3.971742037036472e-05,
+      "loss": 0.6445,
+      "step": 185
+    },
+    {
+      "epoch": 0.7462387161484454,
+      "grad_norm": 0.7215005004556381,
+      "learning_rate": 3.9707945737369885e-05,
+      "loss": 0.6582,
+      "step": 186
+    },
+    {
+      "epoch": 0.7502507522567703,
+      "grad_norm": 0.6409669812642776,
+      "learning_rate": 3.969831604259217e-05,
+      "loss": 0.6335,
+      "step": 187
+    },
+    {
+      "epoch": 0.7542627883650953,
+      "grad_norm": 0.7477447726910106,
+      "learning_rate": 3.9688531361797834e-05,
+      "loss": 0.6553,
+      "step": 188
+    },
+    {
+      "epoch": 0.7582748244734202,
+      "grad_norm": 0.5822180467820733,
+      "learning_rate": 3.96785917719726e-05,
+      "loss": 0.6612,
+      "step": 189
+    },
+    {
+      "epoch": 0.7622868605817452,
+      "grad_norm": 0.5798332990978287,
+      "learning_rate": 3.966849735132097e-05,
+      "loss": 0.6446,
+      "step": 190
+    },
+    {
+      "epoch": 0.7662988966900702,
+      "grad_norm": 0.6724702260183473,
+      "learning_rate": 3.96582481792657e-05,
+      "loss": 0.6535,
+      "step": 191
+    },
+    {
+      "epoch": 0.7703109327983951,
+      "grad_norm": 0.5664978751141979,
+      "learning_rate": 3.9647844336447074e-05,
+      "loss": 0.665,
+      "step": 192
+    },
+    {
+      "epoch": 0.7743229689067201,
+      "grad_norm": 0.7511085017145228,
+      "learning_rate": 3.9637285904722376e-05,
+      "loss": 0.6689,
+      "step": 193
+    },
+    {
+      "epoch": 0.7783350050150452,
+      "grad_norm": 0.5838719100787517,
+      "learning_rate": 3.962657296716515e-05,
+      "loss": 0.6454,
+      "step": 194
+    },
+    {
+      "epoch": 0.7823470411233701,
+      "grad_norm": 0.6989096327760085,
+      "learning_rate": 3.961570560806461e-05,
+      "loss": 0.6565,
+      "step": 195
+    },
+    {
+      "epoch": 0.7863590772316951,
+      "grad_norm": 0.6200370813403686,
+      "learning_rate": 3.960468391292496e-05,
+      "loss": 0.6382,
+      "step": 196
+    },
+    {
+      "epoch": 0.7903711133400201,
+      "grad_norm": 0.5368998272962549,
+      "learning_rate": 3.9593507968464714e-05,
+      "loss": 0.6614,
+      "step": 197
+    },
+    {
+      "epoch": 0.794383149448345,
+      "grad_norm": 0.8231037556830179,
+      "learning_rate": 3.958217786261601e-05,
+      "loss": 0.6484,
+      "step": 198
+    },
+    {
+      "epoch": 0.79839518555667,
+      "grad_norm": 0.7018383893276623,
+      "learning_rate": 3.9570693684523914e-05,
+      "loss": 0.6315,
+      "step": 199
+    },
+    {
+      "epoch": 0.802407221664995,
+      "grad_norm": 0.6582831085438146,
+      "learning_rate": 3.9559055524545755e-05,
+      "loss": 0.6357,
+      "step": 200
+    },
+    {
+      "epoch": 0.8064192577733199,
+      "grad_norm": 0.7652381152451531,
+      "learning_rate": 3.9547263474250385e-05,
+      "loss": 0.6409,
+      "step": 201
+    },
+    {
+      "epoch": 0.8104312938816449,
+      "grad_norm": 0.6635501223411701,
+      "learning_rate": 3.953531762641745e-05,
+      "loss": 0.6481,
+      "step": 202
+    },
+    {
+      "epoch": 0.8144433299899699,
+      "grad_norm": 0.8063586594930956,
+      "learning_rate": 3.952321807503668e-05,
+      "loss": 0.6607,
+      "step": 203
+    },
+    {
+      "epoch": 0.8184553660982948,
+      "grad_norm": 0.5891022967170688,
+      "learning_rate": 3.951096491530714e-05,
+      "loss": 0.6522,
+      "step": 204
+    },
+    {
+      "epoch": 0.8224674022066198,
+      "grad_norm": 0.6081956410290555,
+      "learning_rate": 3.949855824363647e-05,
+      "loss": 0.6422,
+      "step": 205
+    },
+    {
+      "epoch": 0.8264794383149449,
+      "grad_norm": 0.5857933872264018,
+      "learning_rate": 3.948599815764018e-05,
+      "loss": 0.6635,
+      "step": 206
+    },
+    {
+      "epoch": 0.8304914744232698,
+      "grad_norm": 0.5771734519430035,
+      "learning_rate": 3.9473284756140786e-05,
+      "loss": 0.6295,
+      "step": 207
+    },
+    {
+      "epoch": 0.8345035105315948,
+      "grad_norm": 0.6167785949187584,
+      "learning_rate": 3.946041813916713e-05,
+      "loss": 0.6328,
+      "step": 208
+    },
+    {
+      "epoch": 0.8385155466399198,
+      "grad_norm": 0.7698250529133521,
+      "learning_rate": 3.9447398407953536e-05,
+      "loss": 0.6433,
+      "step": 209
+    },
+    {
+      "epoch": 0.8425275827482447,
+      "grad_norm": 0.6924784020925326,
+      "learning_rate": 3.943422566493902e-05,
+      "loss": 0.6832,
+      "step": 210
+    },
+    {
+      "epoch": 0.8465396188565697,
+      "grad_norm": 0.6835035108799093,
+      "learning_rate": 3.9420900013766515e-05,
+      "loss": 0.6315,
+      "step": 211
+    },
+    {
+      "epoch": 0.8505516549648947,
+      "grad_norm": 0.7976652155495385,
+      "learning_rate": 3.9407421559282e-05,
+      "loss": 0.6445,
+      "step": 212
+    },
+    {
+      "epoch": 0.8545636910732196,
+      "grad_norm": 0.5416316484500985,
+      "learning_rate": 3.939379040753374e-05,
+      "loss": 0.6357,
+      "step": 213
+    },
+    {
+      "epoch": 0.8585757271815446,
+      "grad_norm": 0.6137296730908093,
+      "learning_rate": 3.93800066657714e-05,
+      "loss": 0.6536,
+      "step": 214
+    },
+    {
+      "epoch": 0.8625877632898696,
+      "grad_norm": 0.5538516654519803,
+      "learning_rate": 3.9366070442445234e-05,
+      "loss": 0.6759,
+      "step": 215
+    },
+    {
+      "epoch": 0.8665997993981945,
+      "grad_norm": 0.5721813961498722,
+      "learning_rate": 3.93519818472052e-05,
+      "loss": 0.6356,
+      "step": 216
+    },
+    {
+      "epoch": 0.8706118355065195,
+      "grad_norm": 0.5971347060814602,
+      "learning_rate": 3.933774099090013e-05,
+      "loss": 0.6603,
+      "step": 217
+    },
+    {
+      "epoch": 0.8746238716148446,
+      "grad_norm": 0.63901252274,
+      "learning_rate": 3.9323347985576846e-05,
+      "loss": 0.6529,
+      "step": 218
+    },
+    {
+      "epoch": 0.8786359077231695,
+      "grad_norm": 0.6049910606878818,
+      "learning_rate": 3.930880294447926e-05,
+      "loss": 0.636,
+      "step": 219
+    },
+    {
+      "epoch": 0.8826479438314945,
+      "grad_norm": 0.5897564983679663,
+      "learning_rate": 3.929410598204752e-05,
+      "loss": 0.6072,
+      "step": 220
+    },
+    {
+      "epoch": 0.8866599799398195,
+      "grad_norm": 0.5982499288798169,
+      "learning_rate": 3.927925721391707e-05,
+      "loss": 0.6071,
+      "step": 221
+    },
+    {
+      "epoch": 0.8906720160481444,
+      "grad_norm": 0.5559270336468647,
+      "learning_rate": 3.9264256756917765e-05,
+      "loss": 0.6386,
+      "step": 222
+    },
+    {
+      "epoch": 0.8946840521564694,
+      "grad_norm": 0.7374141302463203,
+      "learning_rate": 3.9249104729072944e-05,
+      "loss": 0.6363,
+      "step": 223
+    },
+    {
+      "epoch": 0.8986960882647944,
+      "grad_norm": 0.5066034505521827,
+      "learning_rate": 3.923380124959852e-05,
+      "loss": 0.6422,
+      "step": 224
+    },
+    {
+      "epoch": 0.9027081243731193,
+      "grad_norm": 0.7503479926622877,
+      "learning_rate": 3.9218346438901996e-05,
+      "loss": 0.644,
+      "step": 225
+    },
+    {
+      "epoch": 0.9067201604814443,
+      "grad_norm": 0.5952925909453672,
+      "learning_rate": 3.920274041858156e-05,
+      "loss": 0.6252,
+      "step": 226
+    },
+    {
+      "epoch": 0.9107321965897693,
+      "grad_norm": 0.7276892127750862,
+      "learning_rate": 3.918698331142512e-05,
+      "loss": 0.646,
+      "step": 227
+    },
+    {
+      "epoch": 0.9147442326980942,
+      "grad_norm": 0.6908984131345813,
+      "learning_rate": 3.917107524140933e-05,
+      "loss": 0.6557,
+      "step": 228
+    },
+    {
+      "epoch": 0.9187562688064193,
+      "grad_norm": 0.559677574389918,
+      "learning_rate": 3.9155016333698615e-05,
+      "loss": 0.6233,
+      "step": 229
+    },
+    {
+      "epoch": 0.9227683049147443,
+      "grad_norm": 0.6878045501894693,
+      "learning_rate": 3.913880671464418e-05,
+      "loss": 0.614,
+      "step": 230
+    },
+    {
+      "epoch": 0.9267803410230692,
+      "grad_norm": 0.5935301645497979,
+      "learning_rate": 3.912244651178304e-05,
+      "loss": 0.6363,
+      "step": 231
+    },
+    {
+      "epoch": 0.9307923771313942,
+      "grad_norm": 0.5306757451540638,
+      "learning_rate": 3.9105935853837e-05,
+      "loss": 0.6404,
+      "step": 232
+    },
+    {
+      "epoch": 0.9348044132397192,
+      "grad_norm": 0.6424362936771318,
+      "learning_rate": 3.908927487071162e-05,
+      "loss": 0.6395,
+      "step": 233
+    },
+    {
+      "epoch": 0.9388164493480441,
+      "grad_norm": 0.5324914095322344,
+      "learning_rate": 3.9072463693495243e-05,
+      "loss": 0.637,
+      "step": 234
+    },
+    {
+      "epoch": 0.9428284854563691,
+      "grad_norm": 0.5776586353405196,
+      "learning_rate": 3.905550245445793e-05,
+      "loss": 0.6224,
+      "step": 235
+    },
+    {
+      "epoch": 0.9468405215646941,
+      "grad_norm": 0.8665176704584113,
+      "learning_rate": 3.903839128705041e-05,
+      "loss": 0.6511,
+      "step": 236
+    },
+    {
+      "epoch": 0.950852557673019,
+      "grad_norm": 0.4970414825832502,
+      "learning_rate": 3.9021130325903076e-05,
+      "loss": 0.639,
+      "step": 237
+    },
+    {
+      "epoch": 0.954864593781344,
+      "grad_norm": 0.8048858304633686,
+      "learning_rate": 3.9003719706824863e-05,
+      "loss": 0.6158,
+      "step": 238
+    },
+    {
+      "epoch": 0.958876629889669,
+      "grad_norm": 0.523301010342669,
+      "learning_rate": 3.898615956680223e-05,
+      "loss": 0.6577,
+      "step": 239
+    },
+    {
+      "epoch": 0.9628886659979939,
+      "grad_norm": 0.7246918499160878,
+      "learning_rate": 3.8968450043998064e-05,
+      "loss": 0.6416,
+      "step": 240
+    },
+    {
+      "epoch": 0.966900702106319,
+      "grad_norm": 0.6971017591771864,
+      "learning_rate": 3.895059127775058e-05,
+      "loss": 0.6399,
+      "step": 241
+    },
+    {
+      "epoch": 0.970912738214644,
+      "grad_norm": 0.5388402078755408,
+      "learning_rate": 3.893258340857226e-05,
+      "loss": 0.6502,
+      "step": 242
+    },
+    {
+      "epoch": 0.9749247743229689,
+      "grad_norm": 0.6989045563311478,
+      "learning_rate": 3.891442657814871e-05,
+      "loss": 0.6472,
+      "step": 243
+    },
+    {
+      "epoch": 0.9789368104312939,
+      "grad_norm": 0.5862412832585728,
+      "learning_rate": 3.889612092933756e-05,
+      "loss": 0.6337,
+      "step": 244
+    },
+    {
+      "epoch": 0.9829488465396189,
+      "grad_norm": 0.6047455986395117,
+      "learning_rate": 3.8877666606167354e-05,
+      "loss": 0.6025,
+      "step": 245
+    },
+    {
+      "epoch": 0.9869608826479438,
+      "grad_norm": 0.5767709502232039,
+      "learning_rate": 3.8859063753836386e-05,
+      "loss": 0.6444,
+      "step": 246
+    },
+    {
+      "epoch": 0.9909729187562688,
+      "grad_norm": 0.4914713768244347,
+      "learning_rate": 3.884031251871158e-05,
+      "loss": 0.6399,
+      "step": 247
+    },
+    {
+      "epoch": 0.9949849548645938,
+      "grad_norm": 0.5675619299043949,
+      "learning_rate": 3.8821413048327346e-05,
+      "loss": 0.6321,
+      "step": 248
+    },
+    {
+      "epoch": 0.9989969909729187,
+      "grad_norm": 0.4577095150754147,
+      "learning_rate": 3.880236549138438e-05,
+      "loss": 0.6277,
+      "step": 249
+    },
+    {
+      "epoch": 1.0030090270812437,
+      "grad_norm": 0.6240421175734373,
+      "learning_rate": 3.878316999774856e-05,
+      "loss": 0.5593,
+      "step": 250
+    },
+    {
+      "epoch": 1.0070210631895686,
+      "grad_norm": 0.5499503269040985,
+      "learning_rate": 3.876382671844969e-05,
+      "loss": 0.5183,
+      "step": 251
+    },
+    {
+      "epoch": 1.0110330992978938,
+      "grad_norm": 0.6748947666147845,
+      "learning_rate": 3.8744335805680355e-05,
+      "loss": 0.533,
+      "step": 252
+    },
+    {
+      "epoch": 1.0150451354062187,
+      "grad_norm": 0.6530565848983463,
+      "learning_rate": 3.872469741279475e-05,
+      "loss": 0.5436,
+      "step": 253
+    },
+    {
+      "epoch": 1.0190571715145436,
+      "grad_norm": 0.6512591328597944,
+      "learning_rate": 3.87049116943074e-05,
+      "loss": 0.5331,
+      "step": 254
+    },
+    {
+      "epoch": 1.0230692076228687,
+      "grad_norm": 0.7502793150045433,
+      "learning_rate": 3.8684978805891994e-05,
+      "loss": 0.4977,
+      "step": 255
+    },
+    {
+      "epoch": 1.0270812437311936,
+      "grad_norm": 0.6995250851734305,
+      "learning_rate": 3.866489890438019e-05,
+      "loss": 0.5232,
+      "step": 256
+    },
+    {
+      "epoch": 1.0310932798395185,
+      "grad_norm": 0.5876279053110866,
+      "learning_rate": 3.8644672147760286e-05,
+      "loss": 0.5132,
+      "step": 257
+    },
+    {
+      "epoch": 1.0351053159478436,
+      "grad_norm": 0.7727694327663975,
+      "learning_rate": 3.862429869517607e-05,
+      "loss": 0.5133,
+      "step": 258
+    },
+    {
+      "epoch": 1.0391173520561685,
+      "grad_norm": 0.6365228244356602,
+      "learning_rate": 3.860377870692553e-05,
+      "loss": 0.5062,
+      "step": 259
+    },
+    {
+      "epoch": 1.0431293881644934,
+      "grad_norm": 0.5994364878214419,
+      "learning_rate": 3.858311234445957e-05,
+      "loss": 0.4916,
+      "step": 260
+    },
+    {
+      "epoch": 1.0471414242728185,
+      "grad_norm": 0.7202028365412168,
+      "learning_rate": 3.856229977038078e-05,
+      "loss": 0.5126,
+      "step": 261
+    },
+    {
+      "epoch": 1.0511534603811434,
+      "grad_norm": 0.5112821253307894,
+      "learning_rate": 3.854134114844215e-05,
+      "loss": 0.5063,
+      "step": 262
+    },
+    {
+      "epoch": 1.0551654964894683,
+      "grad_norm": 0.7047621776238286,
+      "learning_rate": 3.852023664354576e-05,
+      "loss": 0.5211,
+      "step": 263
+    },
+    {
+      "epoch": 1.0591775325977935,
+      "grad_norm": 0.554261045200132,
+      "learning_rate": 3.849898642174149e-05,
+      "loss": 0.5248,
+      "step": 264
+    },
+    {
+      "epoch": 1.0631895687061184,
+      "grad_norm": 0.6507013946989241,
+      "learning_rate": 3.8477590650225735e-05,
+      "loss": 0.5295,
+      "step": 265
+    },
+    {
+      "epoch": 1.0672016048144433,
+      "grad_norm": 0.6455252212228699,
+      "learning_rate": 3.845604949734006e-05,
+      "loss": 0.5136,
+      "step": 266
+    },
+    {
+      "epoch": 1.0712136409227684,
+      "grad_norm": 0.5183928024507691,
+      "learning_rate": 3.8434363132569883e-05,
+      "loss": 0.5122,
+      "step": 267
+    },
+    {
+      "epoch": 1.0752256770310933,
+      "grad_norm": 0.7127263343217726,
+      "learning_rate": 3.841253172654316e-05,
+      "loss": 0.5323,
+      "step": 268
+    },
+    {
+      "epoch": 1.0792377131394182,
+      "grad_norm": 0.6016978071458462,
+      "learning_rate": 3.839055545102902e-05,
+      "loss": 0.5106,
+      "step": 269
+    },
+    {
+      "epoch": 1.0832497492477433,
+      "grad_norm": 0.6233802970317148,
+      "learning_rate": 3.8368434478936426e-05,
+      "loss": 0.4919,
+      "step": 270
+    },
+    {
+      "epoch": 1.0872617853560682,
+      "grad_norm": 0.6904819746302212,
+      "learning_rate": 3.834616898431281e-05,
+      "loss": 0.5296,
+      "step": 271
+    },
+    {
+      "epoch": 1.0912738214643931,
+      "grad_norm": 0.5765513601850859,
+      "learning_rate": 3.832375914234272e-05,
+      "loss": 0.5175,
+      "step": 272
+    },
+    {
+      "epoch": 1.0952858575727182,
+      "grad_norm": 0.7222361778784043,
+      "learning_rate": 3.83012051293464e-05,
+      "loss": 0.5147,
+      "step": 273
+    },
+    {
+      "epoch": 1.0992978936810431,
+      "grad_norm": 0.5714963016354521,
+      "learning_rate": 3.8278507122778444e-05,
+      "loss": 0.4999,
+      "step": 274
+    },
+    {
+      "epoch": 1.103309929789368,
+      "grad_norm": 0.6515220105080878,
+      "learning_rate": 3.825566530122638e-05,
+      "loss": 0.5293,
+      "step": 275
+    },
+    {
+      "epoch": 1.1073219658976932,
+      "grad_norm": 0.5887192619370362,
+      "learning_rate": 3.823267984440928e-05,
+      "loss": 0.5008,
+      "step": 276
+    },
+    {
+      "epoch": 1.111334002006018,
+      "grad_norm": 0.7258859632332559,
+      "learning_rate": 3.8209550933176324e-05,
+      "loss": 0.5278,
+      "step": 277
+    },
+    {
+      "epoch": 1.115346038114343,
+      "grad_norm": 0.4833701799609937,
+      "learning_rate": 3.8186278749505395e-05,
+      "loss": 0.5061,
+      "step": 278
+    },
+    {
+      "epoch": 1.119358074222668,
+      "grad_norm": 0.733309085218396,
+      "learning_rate": 3.816286347650163e-05,
+      "loss": 0.5105,
+      "step": 279
+    },
+    {
+      "epoch": 1.123370110330993,
+      "grad_norm": 0.5161738269248766,
+      "learning_rate": 3.813930529839601e-05,
+      "loss": 0.5171,
+      "step": 280
+    },
+    {
+      "epoch": 1.127382146439318,
+      "grad_norm": 0.5737752588500534,
+      "learning_rate": 3.8115604400543885e-05,
+      "loss": 0.5136,
+      "step": 281
+    },
+    {
+      "epoch": 1.131394182547643,
+      "grad_norm": 0.5729713316767231,
+      "learning_rate": 3.8091760969423507e-05,
+      "loss": 0.5389,
+      "step": 282
+    },
+    {
+      "epoch": 1.135406218655968,
+      "grad_norm": 0.49021578067638777,
+      "learning_rate": 3.80677751926346e-05,
+      "loss": 0.5361,
+      "step": 283
+    },
+    {
+      "epoch": 1.1394182547642928,
+      "grad_norm": 0.6269276418133901,
+      "learning_rate": 3.804364725889683e-05,
+      "loss": 0.5081,
+      "step": 284
+    },
+    {
+      "epoch": 1.143430290872618,
+      "grad_norm": 0.5467536115955154,
+      "learning_rate": 3.801937735804838e-05,
+      "loss": 0.539,
+      "step": 285
+    },
+    {
+      "epoch": 1.1474423269809428,
+      "grad_norm": 0.5452506420772878,
+      "learning_rate": 3.7994965681044436e-05,
+      "loss": 0.4983,
+      "step": 286
+    },
+    {
+      "epoch": 1.1514543630892677,
+      "grad_norm": 0.5350629106303321,
+      "learning_rate": 3.7970412419955626e-05,
+      "loss": 0.5259,
+      "step": 287
+    },
+    {
+      "epoch": 1.1554663991975929,
+      "grad_norm": 0.5785959077698662,
+      "learning_rate": 3.794571776796662e-05,
+      "loss": 0.5368,
+      "step": 288
+    },
+    {
+      "epoch": 1.1594784353059178,
+      "grad_norm": 0.552921741856963,
+      "learning_rate": 3.792088191937451e-05,
+      "loss": 0.5094,
+      "step": 289
+    },
+    {
+      "epoch": 1.1634904714142427,
+      "grad_norm": 0.5582365093353788,
+      "learning_rate": 3.7895905069587326e-05,
+      "loss": 0.5228,
+      "step": 290
+    },
+    {
+      "epoch": 1.1675025075225678,
+      "grad_norm": 0.5940730601204347,
+      "learning_rate": 3.78707874151225e-05,
+      "loss": 0.5245,
+      "step": 291
+    },
+    {
+      "epoch": 1.1715145436308927,
+      "grad_norm": 0.5633476584492644,
+      "learning_rate": 3.784552915360532e-05,
+      "loss": 0.5219,
+      "step": 292
+    },
+    {
+      "epoch": 1.1755265797392176,
+      "grad_norm": 0.6405199278521263,
+      "learning_rate": 3.782013048376736e-05,
+      "loss": 0.4938,
+      "step": 293
+    },
+    {
+      "epoch": 1.1795386158475427,
+      "grad_norm": 0.5445855457359872,
+      "learning_rate": 3.7794591605444915e-05,
+      "loss": 0.5216,
+      "step": 294
+    },
+    {
+      "epoch": 1.1835506519558676,
+      "grad_norm": 0.5967139769052741,
+      "learning_rate": 3.776891271957745e-05,
+      "loss": 0.5114,
+      "step": 295
+    },
+    {
+      "epoch": 1.1875626880641925,
+      "grad_norm": 0.5518457966022539,
+      "learning_rate": 3.7743094028206e-05,
+      "loss": 0.5208,
+      "step": 296
+    },
+    {
+      "epoch": 1.1915747241725176,
+      "grad_norm": 0.5929480182250662,
+      "learning_rate": 3.77171357344716e-05,
+      "loss": 0.5313,
+      "step": 297
+    },
+    {
+      "epoch": 1.1955867602808425,
+      "grad_norm": 0.5013934862418261,
+      "learning_rate": 3.769103804261366e-05,
+      "loss": 0.4958,
+      "step": 298
+    },
+    {
+      "epoch": 1.1995987963891674,
+      "grad_norm": 0.5731852757848859,
+      "learning_rate": 3.766480115796837e-05,
+      "loss": 0.5052,
+      "step": 299
+    },
+    {
+      "epoch": 1.2036108324974926,
+      "grad_norm": 0.5465409143516308,
+      "learning_rate": 3.76384252869671e-05,
+      "loss": 0.5048,
+      "step": 300
+    },
+    {
+      "epoch": 1.2076228686058175,
+      "grad_norm": 0.6264675925813784,
+      "learning_rate": 3.761191063713476e-05,
+      "loss": 0.5282,
+      "step": 301
+    },
+    {
+      "epoch": 1.2116349047141424,
+      "grad_norm": 0.5421724702563706,
+      "learning_rate": 3.758525741708816e-05,
+      "loss": 0.4951,
+      "step": 302
+    },
+    {
+      "epoch": 1.2156469408224675,
+      "grad_norm": 0.5672942863957189,
+      "learning_rate": 3.755846583653439e-05,
+      "loss": 0.4934,
+      "step": 303
+    },
+    {
+      "epoch": 1.2196589769307924,
+      "grad_norm": 0.47835864287775454,
+      "learning_rate": 3.753153610626914e-05,
+      "loss": 0.5151,
+      "step": 304
+    },
+    {
+      "epoch": 1.2236710130391173,
+      "grad_norm": 0.559173870089285,
+      "learning_rate": 3.7504468438175076e-05,
+      "loss": 0.5098,
+      "step": 305
+    },
+    {
+      "epoch": 1.2276830491474424,
+      "grad_norm": 0.5473671941059122,
+      "learning_rate": 3.7477263045220155e-05,
+      "loss": 0.5244,
+      "step": 306
+    },
+    {
+      "epoch": 1.2316950852557673,
+      "grad_norm": 0.5657658432266967,
+      "learning_rate": 3.744992014145595e-05,
+      "loss": 0.5129,
+      "step": 307
+    },
+    {
+      "epoch": 1.2357071213640922,
+      "grad_norm": 0.6609906230324927,
+      "learning_rate": 3.742243994201595e-05,
+      "loss": 0.5093,
+      "step": 308
+    },
+    {
+      "epoch": 1.2397191574724173,
+      "grad_norm": 0.6179077735754257,
+      "learning_rate": 3.7394822663113915e-05,
+      "loss": 0.5301,
+      "step": 309
+    },
+    {
+      "epoch": 1.2437311935807422,
+      "grad_norm": 0.5335680587422695,
+      "learning_rate": 3.7367068522042115e-05,
+      "loss": 0.5071,
+      "step": 310
+    },
+    {
+      "epoch": 1.2477432296890671,
+      "grad_norm": 0.5799819731594408,
+      "learning_rate": 3.7339177737169664e-05,
+      "loss": 0.512,
+      "step": 311
+    },
+    {
+      "epoch": 1.2517552657973923,
+      "grad_norm": 0.49769186597087733,
+      "learning_rate": 3.731115052794079e-05,
+      "loss": 0.5028,
+      "step": 312
+    },
+    {
+      "epoch": 1.2557673019057172,
+      "grad_norm": 0.5184235363792055,
+      "learning_rate": 3.72829871148731e-05,
+      "loss": 0.5163,
+      "step": 313
+    },
+    {
+      "epoch": 1.259779338014042,
+      "grad_norm": 0.5204750640180548,
+      "learning_rate": 3.725468771955584e-05,
+      "loss": 0.4973,
+      "step": 314
+    },
+    {
+      "epoch": 1.2637913741223672,
+      "grad_norm": 0.5438664337508811,
+      "learning_rate": 3.7226252564648183e-05,
+      "loss": 0.5268,
+      "step": 315
+    },
+    {
+      "epoch": 1.267803410230692,
+      "grad_norm": 0.5458345933059568,
+      "learning_rate": 3.719768187387744e-05,
+      "loss": 0.5323,
+      "step": 316
+    },
+    {
+      "epoch": 1.271815446339017,
+      "grad_norm": 0.49589316558519136,
+      "learning_rate": 3.716897587203733e-05,
+      "loss": 0.5238,
+      "step": 317
+    },
+    {
+      "epoch": 1.2758274824473421,
+      "grad_norm": 0.5390092508721251,
+      "learning_rate": 3.714013478498618e-05,
+      "loss": 0.5192,
+      "step": 318
+    },
+    {
+      "epoch": 1.279839518555667,
+      "grad_norm": 0.48971833076948756,
+      "learning_rate": 3.711115883964516e-05,
+      "loss": 0.5062,
+      "step": 319
+    },
+    {
+      "epoch": 1.283851554663992,
+      "grad_norm": 0.5819043934545967,
+      "learning_rate": 3.7082048263996554e-05,
+      "loss": 0.5334,
+      "step": 320
+    },
+    {
+      "epoch": 1.287863590772317,
+      "grad_norm": 0.4771007356272514,
+      "learning_rate": 3.705280328708185e-05,
+      "loss": 0.5307,
+      "step": 321
+    },
+    {
+      "epoch": 1.291875626880642,
+      "grad_norm": 0.6329460358725633,
+      "learning_rate": 3.702342413900003e-05,
+      "loss": 0.5283,
+      "step": 322
+    },
+    {
+      "epoch": 1.2958876629889668,
+      "grad_norm": 0.4769979001632641,
+      "learning_rate": 3.6993911050905754e-05,
+      "loss": 0.4959,
+      "step": 323
+    },
+    {
+      "epoch": 1.299899699097292,
+      "grad_norm": 0.5837831795374662,
+      "learning_rate": 3.696426425500747e-05,
+      "loss": 0.4886,
+      "step": 324
+    },
+    {
+      "epoch": 1.3039117352056169,
+      "grad_norm": 0.48888936364196284,
+      "learning_rate": 3.6934483984565684e-05,
+      "loss": 0.4892,
+      "step": 325
+    },
+    {
+      "epoch": 1.3079237713139418,
+      "grad_norm": 0.47591830100385957,
+      "learning_rate": 3.690457047389106e-05,
+      "loss": 0.5291,
+      "step": 326
+    },
+    {
+      "epoch": 1.311935807422267,
+      "grad_norm": 0.5177986417883397,
+      "learning_rate": 3.687452395834259e-05,
+      "loss": 0.5123,
+      "step": 327
+    },
+    {
+      "epoch": 1.3159478435305918,
+      "grad_norm": 0.45224301805863004,
+      "learning_rate": 3.684434467432573e-05,
+      "loss": 0.517,
+      "step": 328
+    },
+    {
+      "epoch": 1.3199598796389167,
+      "grad_norm": 0.47469175528412033,
+      "learning_rate": 3.681403285929061e-05,
+      "loss": 0.5128,
+      "step": 329
+    },
+    {
+      "epoch": 1.3239719157472418,
+      "grad_norm": 0.574902606746805,
+      "learning_rate": 3.678358875173005e-05,
+      "loss": 0.5173,
+      "step": 330
+    },
+    {
+      "epoch": 1.3279839518555667,
+      "grad_norm": 0.4690257877200028,
+      "learning_rate": 3.6753012591177774e-05,
+      "loss": 0.4942,
+      "step": 331
+    },
+    {
+      "epoch": 1.3319959879638916,
+      "grad_norm": 0.4814515988500195,
+      "learning_rate": 3.672230461820651e-05,
+      "loss": 0.5051,
+      "step": 332
+    },
+    {
+      "epoch": 1.3360080240722167,
+      "grad_norm": 0.47645613027457,
+      "learning_rate": 3.669146507442606e-05,
+      "loss": 0.5504,
+      "step": 333
+    },
+    {
+      "epoch": 1.3400200601805417,
+      "grad_norm": 0.4853479100265655,
+      "learning_rate": 3.6660494202481426e-05,
+      "loss": 0.5027,
+      "step": 334
+    },
+    {
+      "epoch": 1.3440320962888666,
+      "grad_norm": 0.466849007851617,
+      "learning_rate": 3.662939224605091e-05,
+      "loss": 0.5327,
+      "step": 335
+    },
+    {
+      "epoch": 1.3480441323971917,
+      "grad_norm": 0.4817374982775801,
+      "learning_rate": 3.6598159449844175e-05,
+      "loss": 0.5157,
+      "step": 336
+    },
+    {
+      "epoch": 1.3520561685055166,
+      "grad_norm": 0.525239200146859,
+      "learning_rate": 3.6566796059600334e-05,
+      "loss": 0.5071,
+      "step": 337
+    },
+    {
+      "epoch": 1.3560682046138415,
+      "grad_norm": 0.5171180475045565,
+      "learning_rate": 3.653530232208601e-05,
+      "loss": 0.5051,
+      "step": 338
+    },
+    {
+      "epoch": 1.3600802407221666,
+      "grad_norm": 0.4788776047079267,
+      "learning_rate": 3.650367848509339e-05,
+      "loss": 0.5026,
+      "step": 339
+    },
+    {
+      "epoch": 1.3640922768304915,
+      "grad_norm": 0.5058116985763004,
+      "learning_rate": 3.64719247974383e-05,
+      "loss": 0.512,
+      "step": 340
+    },
+    {
+      "epoch": 1.3681043129388164,
+      "grad_norm": 0.5424195696315768,
+      "learning_rate": 3.644004150895821e-05,
+      "loss": 0.4954,
+      "step": 341
+    },
+    {
+      "epoch": 1.3721163490471415,
+      "grad_norm": 0.5050479077032102,
+      "learning_rate": 3.640802887051027e-05,
+      "loss": 0.52,
+      "step": 342
+    },
+    {
+      "epoch": 1.3761283851554664,
+      "grad_norm": 0.577965801910359,
+      "learning_rate": 3.637588713396942e-05,
+      "loss": 0.5362,
+      "step": 343
+    },
+    {
+      "epoch": 1.3801404212637913,
+      "grad_norm": 0.47848391330176654,
+      "learning_rate": 3.6343616552226267e-05,
+      "loss": 0.5258,
+      "step": 344
+    },
+    {
+      "epoch": 1.3841524573721165,
+      "grad_norm": 0.5025164029290684,
+      "learning_rate": 3.631121737918521e-05,
+      "loss": 0.521,
+      "step": 345
+    },
+    {
+      "epoch": 1.3881644934804414,
+      "grad_norm": 0.4919085859522591,
+      "learning_rate": 3.627868986976239e-05,
+      "loss": 0.5233,
+      "step": 346
+    },
+    {
+      "epoch": 1.3921765295887663,
+      "grad_norm": 0.5605337346096146,
+      "learning_rate": 3.6246034279883706e-05,
+      "loss": 0.529,
+      "step": 347
+    },
+    {
+      "epoch": 1.3961885656970914,
+      "grad_norm": 0.5666097514025779,
+      "learning_rate": 3.621325086648278e-05,
+      "loss": 0.5165,
+      "step": 348
+    },
+    {
+      "epoch": 1.4002006018054163,
+      "grad_norm": 0.5976034688890988,
+      "learning_rate": 3.6180339887498953e-05,
+      "loss": 0.5353,
+      "step": 349
+    },
+    {
+      "epoch": 1.4042126379137412,
+      "grad_norm": 0.5524960337867185,
+      "learning_rate": 3.6147301601875254e-05,
+      "loss": 0.5214,
+      "step": 350
+    },
+    {
+      "epoch": 1.4082246740220663,
+      "grad_norm": 0.4864378934971289,
+      "learning_rate": 3.611413626955635e-05,
+      "loss": 0.521,
+      "step": 351
+    },
+    {
+      "epoch": 1.4122367101303912,
+      "grad_norm": 0.5245634127611527,
+      "learning_rate": 3.6080844151486505e-05,
+      "loss": 0.508,
+      "step": 352
+    },
+    {
+      "epoch": 1.416248746238716,
+      "grad_norm": 0.4623750972342063,
+      "learning_rate": 3.6047425509607566e-05,
+      "loss": 0.5432,
+      "step": 353
+    },
+    {
+      "epoch": 1.4202607823470412,
+      "grad_norm": 0.46685485469105226,
+      "learning_rate": 3.601388060685681e-05,
+      "loss": 0.5151,
+      "step": 354
+    },
+    {
+      "epoch": 1.4242728184553661,
+      "grad_norm": 0.4351160886184323,
+      "learning_rate": 3.598020970716499e-05,
+      "loss": 0.5319,
+      "step": 355
+    },
+    {
+      "epoch": 1.428284854563691,
+      "grad_norm": 0.4834251224005961,
+      "learning_rate": 3.594641307545414e-05,
+      "loss": 0.4964,
+      "step": 356
+    },
+    {
+      "epoch": 1.4322968906720162,
+      "grad_norm": 0.4549641962463708,
+      "learning_rate": 3.591249097763562e-05,
+      "loss": 0.5049,
+      "step": 357
+    },
+    {
+      "epoch": 1.436308926780341,
+      "grad_norm": 0.44935023069863617,
+      "learning_rate": 3.58784436806079e-05,
+      "loss": 0.5011,
+      "step": 358
+    },
+    {
+      "epoch": 1.440320962888666,
+      "grad_norm": 0.5172251710375654,
+      "learning_rate": 3.584427145225455e-05,
+      "loss": 0.5153,
+      "step": 359
+    },
+    {
+      "epoch": 1.444332998996991,
+      "grad_norm": 0.47748488270629474,
+      "learning_rate": 3.5809974561442074e-05,
+      "loss": 0.512,
+      "step": 360
+    },
+    {
+      "epoch": 1.448345035105316,
+      "grad_norm": 0.5359034955126112,
+      "learning_rate": 3.5775553278017824e-05,
+      "loss": 0.5248,
+      "step": 361
+    },
+    {
+      "epoch": 1.4523570712136409,
+      "grad_norm": 0.5248897446417429,
+      "learning_rate": 3.574100787280788e-05,
+      "loss": 0.5309,
+      "step": 362
+    },
+    {
+      "epoch": 1.456369107321966,
+      "grad_norm": 0.46516875268048596,
+      "learning_rate": 3.57063386176149e-05,
+      "loss": 0.4956,
+      "step": 363
+    },
+    {
+      "epoch": 1.460381143430291,
+      "grad_norm": 0.5374091530526378,
+      "learning_rate": 3.5671545785216004e-05,
+      "loss": 0.5152,
+      "step": 364
+    },
+    {
+      "epoch": 1.4643931795386158,
+      "grad_norm": 0.49483595462784746,
+      "learning_rate": 3.56366296493606e-05,
+      "loss": 0.5269,
+      "step": 365
+    },
+    {
+      "epoch": 1.468405215646941,
+      "grad_norm": 0.5114482415870087,
+      "learning_rate": 3.560159048476825e-05,
+      "loss": 0.5334,
+      "step": 366
+    },
+    {
+      "epoch": 1.4724172517552658,
+      "grad_norm": 0.5685405928412336,
+      "learning_rate": 3.556642856712652e-05,
+      "loss": 0.5287,
+      "step": 367
+    },
+    {
+      "epoch": 1.4764292878635907,
+      "grad_norm": 0.527932748183303,
+      "learning_rate": 3.5531144173088774e-05,
+      "loss": 0.5135,
+      "step": 368
+    },
+    {
+      "epoch": 1.4804413239719159,
+      "grad_norm": 0.6274479985770136,
+      "learning_rate": 3.5495737580272024e-05,
+      "loss": 0.5194,
+      "step": 369
+    },
+    {
+      "epoch": 1.4844533600802408,
+      "grad_norm": 0.5399770754204466,
+      "learning_rate": 3.546020906725474e-05,
+      "loss": 0.5233,
+      "step": 370
+    },
+    {
+      "epoch": 1.4884653961885657,
+      "grad_norm": 0.5441483055060222,
+      "learning_rate": 3.542455891357467e-05,
+      "loss": 0.5223,
+      "step": 371
+    },
+    {
+      "epoch": 1.4924774322968908,
+      "grad_norm": 0.6030836289332245,
+      "learning_rate": 3.5388787399726605e-05,
+      "loss": 0.5172,
+      "step": 372
+    },
+    {
+      "epoch": 1.4964894684052157,
+      "grad_norm": 0.46659983569674995,
+      "learning_rate": 3.535289480716023e-05,
+      "loss": 0.501,
+      "step": 373
+    },
+    {
+      "epoch": 1.5005015045135406,
+      "grad_norm": 0.6288052889727458,
+      "learning_rate": 3.531688141827783e-05,
+      "loss": 0.5233,
+      "step": 374
+    },
+    {
+      "epoch": 1.5045135406218657,
+      "grad_norm": 0.4962893207054877,
+      "learning_rate": 3.528074751643215e-05,
+      "loss": 0.5215,
+      "step": 375
+    },
+    {
+      "epoch": 1.5085255767301906,
+      "grad_norm": 0.589362448250224,
+      "learning_rate": 3.524449338592412e-05,
+      "loss": 0.5073,
+      "step": 376
+    },
+    {
+      "epoch": 1.5125376128385155,
+      "grad_norm": 0.5632513561886133,
+      "learning_rate": 3.520811931200063e-05,
+      "loss": 0.5305,
+      "step": 377
+    },
+    {
+      "epoch": 1.5165496489468406,
+      "grad_norm": 0.6298531924872256,
+      "learning_rate": 3.5171625580852256e-05,
+      "loss": 0.5327,
+      "step": 378
+    },
+    {
+      "epoch": 1.5205616850551655,
+      "grad_norm": 0.5683802795988507,
+      "learning_rate": 3.513501247961106e-05,
+      "loss": 0.5078,
+      "step": 379
+    },
+    {
+      "epoch": 1.5245737211634904,
+      "grad_norm": 0.582638474048695,
+      "learning_rate": 3.509828029634832e-05,
+      "loss": 0.5164,
+      "step": 380
+    },
+    {
+      "epoch": 1.5285857572718156,
+      "grad_norm": 0.6280914458540929,
+      "learning_rate": 3.5061429320072225e-05,
+      "loss": 0.5345,
+      "step": 381
+    },
+    {
+      "epoch": 1.5325977933801405,
+      "grad_norm": 0.5020889343880065,
+      "learning_rate": 3.502445984072562e-05,
+      "loss": 0.5061,
+      "step": 382
+    },
+    {
+      "epoch": 1.5366098294884654,
+      "grad_norm": 0.5393845763964079,
+      "learning_rate": 3.4987372149183745e-05,
+      "loss": 0.5227,
+      "step": 383
+    },
+    {
+      "epoch": 1.5406218655967905,
+      "grad_norm": 0.5345391390748299,
+      "learning_rate": 3.495016653725194e-05,
+      "loss": 0.521,
+      "step": 384
+    },
+    {
+      "epoch": 1.5446339017051154,
+      "grad_norm": 0.45235812230370975,
+      "learning_rate": 3.4912843297663315e-05,
+      "loss": 0.5168,
+      "step": 385
+    },
+    {
+      "epoch": 1.5486459378134403,
+      "grad_norm": 0.5335451921265258,
+      "learning_rate": 3.48754027240765e-05,
+      "loss": 0.5134,
+      "step": 386
+    },
+    {
+      "epoch": 1.5526579739217654,
+      "grad_norm": 0.4891630920829491,
+      "learning_rate": 3.483784511107328e-05,
+      "loss": 0.5028,
+      "step": 387
+    },
+    {
+      "epoch": 1.5566700100300903,
+      "grad_norm": 0.4884584180286181,
+      "learning_rate": 3.480017075415632e-05,
+      "loss": 0.5136,
+      "step": 388
+    },
+    {
+      "epoch": 1.5606820461384152,
+      "grad_norm": 0.6071050887255214,
+      "learning_rate": 3.476237994974682e-05,
+      "loss": 0.5282,
+      "step": 389
+    },
+    {
+      "epoch": 1.5646940822467403,
+      "grad_norm": 0.49696323776801105,
+      "learning_rate": 3.4724472995182194e-05,
+      "loss": 0.499,
+      "step": 390
+    },
+    {
+      "epoch": 1.5687061183550652,
+      "grad_norm": 0.5239271478542761,
+      "learning_rate": 3.468645018871371e-05,
+      "loss": 0.5083,
+      "step": 391
+    },
+    {
+      "epoch": 1.5727181544633901,
+      "grad_norm": 0.5446261555334742,
+      "learning_rate": 3.464831182950418e-05,
+      "loss": 0.5206,
+      "step": 392
+    },
+    {
+      "epoch": 1.5767301905717153,
+      "grad_norm": 0.5473388833292854,
+      "learning_rate": 3.4610058217625554e-05,
+      "loss": 0.5295,
+      "step": 393
+    },
+    {
+      "epoch": 1.5807422266800402,
+      "grad_norm": 0.45705961982015514,
+      "learning_rate": 3.4571689654056605e-05,
+      "loss": 0.504,
+      "step": 394
+    },
+    {
+      "epoch": 1.584754262788365,
+      "grad_norm": 0.5623482376792627,
+      "learning_rate": 3.4533206440680546e-05,
+      "loss": 0.5058,
+      "step": 395
+    },
+    {
+      "epoch": 1.5887662988966902,
+      "grad_norm": 0.4484630691192091,
+      "learning_rate": 3.449460888028264e-05,
+      "loss": 0.523,
+      "step": 396
+    },
+    {
+      "epoch": 1.592778335005015,
+      "grad_norm": 0.5631991047765686,
+      "learning_rate": 3.4455897276547836e-05,
+      "loss": 0.5244,
+      "step": 397
+    },
+    {
+      "epoch": 1.59679037111334,
+      "grad_norm": 0.4388414027276364,
+      "learning_rate": 3.441707193405838e-05,
+      "loss": 0.5284,
+      "step": 398
+    },
+    {
+      "epoch": 1.6008024072216651,
+      "grad_norm": 0.496959221645331,
+      "learning_rate": 3.437813315829141e-05,
+      "loss": 0.4915,
+      "step": 399
+    },
+    {
+      "epoch": 1.60481444332999,
+      "grad_norm": 0.42288799774924485,
+      "learning_rate": 3.433908125561655e-05,
+      "loss": 0.5259,
+      "step": 400
+    },
+    {
+      "epoch": 1.608826479438315,
+      "grad_norm": 0.49406953788215646,
+      "learning_rate": 3.429991653329351e-05,
+      "loss": 0.5272,
+      "step": 401
+    },
+    {
+      "epoch": 1.61283851554664,
+      "grad_norm": 0.5132165414817065,
+      "learning_rate": 3.426063929946966e-05,
+      "loss": 0.5429,
+      "step": 402
+    },
+    {
+      "epoch": 1.616850551654965,
+      "grad_norm": 0.49815515373337205,
+      "learning_rate": 3.422124986317761e-05,
+      "loss": 0.5181,
+      "step": 403
+    },
+    {
+      "epoch": 1.6208625877632898,
+      "grad_norm": 0.4916003127773752,
+      "learning_rate": 3.418174853433278e-05,
+      "loss": 0.5232,
+      "step": 404
+    },
+    {
+      "epoch": 1.624874623871615,
+      "grad_norm": 0.46466469965282964,
+      "learning_rate": 3.4142135623730954e-05,
+      "loss": 0.5093,
+      "step": 405
+    },
+    {
+      "epoch": 1.6288866599799399,
+      "grad_norm": 0.48969059155051037,
+      "learning_rate": 3.410241144304584e-05,
+      "loss": 0.5108,
+      "step": 406
+    },
+    {
+      "epoch": 1.6328986960882648,
+      "grad_norm": 0.5216176968436806,
+      "learning_rate": 3.406257630482661e-05,
+      "loss": 0.5474,
+      "step": 407
+    },
+    {
+      "epoch": 1.63691073219659,
+      "grad_norm": 0.5501025614746948,
+      "learning_rate": 3.402263052249547e-05,
+      "loss": 0.5448,
+      "step": 408
+    },
+    {
+      "epoch": 1.6409227683049148,
+      "grad_norm": 0.4708240815575044,
+      "learning_rate": 3.398257441034515e-05,
+      "loss": 0.5144,
+      "step": 409
+    },
+    {
+      "epoch": 1.6449348044132397,
+      "grad_norm": 0.5242617253035858,
+      "learning_rate": 3.394240828353647e-05,
+      "loss": 0.5216,
+      "step": 410
+    },
+    {
+      "epoch": 1.6489468405215648,
+      "grad_norm": 0.46080198084531887,
+      "learning_rate": 3.390213245809582e-05,
+      "loss": 0.5046,
+      "step": 411
+    },
+    {
+      "epoch": 1.6529588766298897,
+      "grad_norm": 0.47757666177832514,
+      "learning_rate": 3.3861747250912724e-05,
+      "loss": 0.5161,
+      "step": 412
+    },
+    {
+      "epoch": 1.6569709127382146,
+      "grad_norm": 0.4830970579799819,
+      "learning_rate": 3.38212529797373e-05,
+      "loss": 0.507,
+      "step": 413
+    },
+    {
+      "epoch": 1.6609829488465397,
+      "grad_norm": 0.4821420756910173,
+      "learning_rate": 3.378064996317779e-05,
+      "loss": 0.5012,
+      "step": 414
+    },
+    {
+      "epoch": 1.6649949849548646,
+      "grad_norm": 0.47478718965353794,
+      "learning_rate": 3.373993852069804e-05,
+      "loss": 0.5272,
+      "step": 415
+    },
+    {
+      "epoch": 1.6690070210631895,
+      "grad_norm": 0.44612705706004185,
+      "learning_rate": 3.369911897261499e-05,
+      "loss": 0.5068,
+      "step": 416
+    },
+    {
+      "epoch": 1.6730190571715147,
+      "grad_norm": 0.4079540194200061,
+      "learning_rate": 3.365819164009614e-05,
+      "loss": 0.5146,
+      "step": 417
+    },
+    {
+      "epoch": 1.6770310932798396,
+      "grad_norm": 0.4532854344185365,
+      "learning_rate": 3.361715684515707e-05,
+      "loss": 0.5127,
+      "step": 418
+    },
+    {
+      "epoch": 1.6810431293881645,
+      "grad_norm": 0.42800544505874866,
+      "learning_rate": 3.357601491065884e-05,
+      "loss": 0.5042,
+      "step": 419
+    },
+    {
+      "epoch": 1.6850551654964896,
+      "grad_norm": 0.4328095376538618,
+      "learning_rate": 3.353476616030549e-05,
+      "loss": 0.5166,
+      "step": 420
+    },
+    {
+      "epoch": 1.6890672016048145,
+      "grad_norm": 0.4611166739503156,
+      "learning_rate": 3.349341091864149e-05,
+      "loss": 0.5128,
+      "step": 421
+    },
+    {
+      "epoch": 1.6930792377131394,
+      "grad_norm": 0.42105910956535564,
+      "learning_rate": 3.345194951104918e-05,
+      "loss": 0.5093,
+      "step": 422
+    },
+    {
+      "epoch": 1.6970912738214645,
+      "grad_norm": 0.47138060968184065,
+      "learning_rate": 3.341038226374622e-05,
+      "loss": 0.533,
+      "step": 423
+    },
+    {
+      "epoch": 1.7011033099297894,
+      "grad_norm": 0.45956558720735347,
+      "learning_rate": 3.3368709503783e-05,
+      "loss": 0.4988,
+      "step": 424
+    },
+    {
+      "epoch": 1.7051153460381143,
+      "grad_norm": 0.4398144943712062,
+      "learning_rate": 3.3326931559040084e-05,
+      "loss": 0.5124,
+      "step": 425
+    },
+    {
+      "epoch": 1.7091273821464394,
+      "grad_norm": 0.46805086553126873,
+      "learning_rate": 3.328504875822564e-05,
+      "loss": 0.5082,
+      "step": 426
+    },
+    {
+      "epoch": 1.7131394182547643,
+      "grad_norm": 0.4310014021424791,
+      "learning_rate": 3.3243061430872826e-05,
+      "loss": 0.5249,
+      "step": 427
+    },
+    {
+      "epoch": 1.7171514543630892,
+      "grad_norm": 0.529288499792788,
+      "learning_rate": 3.320096990733724e-05,
+      "loss": 0.5136,
+      "step": 428
+    },
+    {
+      "epoch": 1.7211634904714144,
+      "grad_norm": 0.46941698367754414,
+      "learning_rate": 3.315877451879426e-05,
+      "loss": 0.5354,
+      "step": 429
+    },
+    {
+      "epoch": 1.7251755265797393,
+      "grad_norm": 0.5028649945200033,
+      "learning_rate": 3.3116475597236486e-05,
+      "loss": 0.5022,
+      "step": 430
+    },
+    {
+      "epoch": 1.7291875626880642,
+      "grad_norm": 0.531724208554603,
+      "learning_rate": 3.3074073475471125e-05,
+      "loss": 0.4845,
+      "step": 431
+    },
+    {
+      "epoch": 1.7331995987963893,
+      "grad_norm": 0.5143302705923445,
+      "learning_rate": 3.303156848711734e-05,
+      "loss": 0.5207,
+      "step": 432
+    },
+    {
+      "epoch": 1.7372116349047142,
+      "grad_norm": 0.6256879836540098,
+      "learning_rate": 3.298896096660367e-05,
+      "loss": 0.5256,
+      "step": 433
+    },
+    {
+      "epoch": 1.741223671013039,
+      "grad_norm": 0.49970193728116863,
+      "learning_rate": 3.2946251249165366e-05,
+      "loss": 0.4944,
+      "step": 434
+    },
+    {
+      "epoch": 1.7452357071213642,
+      "grad_norm": 0.47548535571588696,
+      "learning_rate": 3.290343967084176e-05,
+      "loss": 0.4948,
+      "step": 435
+    },
+    {
+      "epoch": 1.7492477432296891,
+      "grad_norm": 0.5668308917582725,
+      "learning_rate": 3.2860526568473605e-05,
+      "loss": 0.4979,
+      "step": 436
+    },
+    {
+      "epoch": 1.753259779338014,
+      "grad_norm": 0.49073432522355714,
+      "learning_rate": 3.2817512279700486e-05,
+      "loss": 0.5072,
+      "step": 437
+    },
+    {
+      "epoch": 1.7572718154463391,
+      "grad_norm": 0.5174487945113188,
+      "learning_rate": 3.277439714295809e-05,
+      "loss": 0.517,
+      "step": 438
+    },
+    {
+      "epoch": 1.7612838515546638,
+      "grad_norm": 0.54734331687402,
+      "learning_rate": 3.273118149747557e-05,
+      "loss": 0.5279,
+      "step": 439
+    },
+    {
+      "epoch": 1.765295887662989,
+      "grad_norm": 0.5862058280981682,
+      "learning_rate": 3.268786568327291e-05,
+      "loss": 0.4813,
+      "step": 440
+    },
+    {
+      "epoch": 1.769307923771314,
+      "grad_norm": 0.5175444188076624,
+      "learning_rate": 3.26444500411582e-05,
+      "loss": 0.5152,
+      "step": 441
+    },
+    {
+      "epoch": 1.7733199598796388,
+      "grad_norm": 0.4881308680456048,
+      "learning_rate": 3.260093491272496e-05,
+      "loss": 0.5077,
+      "step": 442
+    },
+    {
+      "epoch": 1.7773319959879639,
+      "grad_norm": 0.6611554232977016,
+      "learning_rate": 3.25573206403495e-05,
+      "loss": 0.5054,
+      "step": 443
+    },
+    {
+      "epoch": 1.781344032096289,
+      "grad_norm": 0.4823222366565848,
+      "learning_rate": 3.251360756718816e-05,
+      "loss": 0.5128,
+      "step": 444
+    },
+    {
+      "epoch": 1.7853560682046137,
+      "grad_norm": 0.6438403000704762,
+      "learning_rate": 3.246979603717467e-05,
+      "loss": 0.5059,
+      "step": 445
+    },
+    {
+      "epoch": 1.7893681043129388,
+      "grad_norm": 0.5146890978318049,
+      "learning_rate": 3.242588639501741e-05,
+      "loss": 0.5164,
+      "step": 446
+    },
+    {
+      "epoch": 1.793380140421264,
+      "grad_norm": 0.49533353832262833,
+      "learning_rate": 3.238187898619669e-05,
+      "loss": 0.5022,
+      "step": 447
+    },
+    {
+      "epoch": 1.7973921765295886,
+      "grad_norm": 0.42717724604663515,
+      "learning_rate": 3.2337774156962045e-05,
+      "loss": 0.4899,
+      "step": 448
+    },
+    {
+      "epoch": 1.8014042126379137,
+      "grad_norm": 0.44442546887725365,
+      "learning_rate": 3.2293572254329546e-05,
+      "loss": 0.5115,
+      "step": 449
+    },
+    {
+      "epoch": 1.8054162487462388,
+      "grad_norm": 0.44501833868029755,
+      "learning_rate": 3.2249273626079005e-05,
+      "loss": 0.4831,
+      "step": 450
+    },
+    {
+      "epoch": 1.8094282848545635,
+      "grad_norm": 0.4418776412918822,
+      "learning_rate": 3.2204878620751274e-05,
+      "loss": 0.504,
+      "step": 451
+    },
+    {
+      "epoch": 1.8134403209628887,
+      "grad_norm": 0.4256719154208649,
+      "learning_rate": 3.2160387587645515e-05,
+      "loss": 0.4905,
+      "step": 452
+    },
+    {
+      "epoch": 1.8174523570712138,
+      "grad_norm": 0.5052749912932102,
+      "learning_rate": 3.21158008768164e-05,
+      "loss": 0.5176,
+      "step": 453
+    },
+    {
+      "epoch": 1.8214643931795385,
+      "grad_norm": 0.4687856555281907,
+      "learning_rate": 3.207111883907143e-05,
+      "loss": 0.5126,
+      "step": 454
+    },
+    {
+      "epoch": 1.8254764292878636,
+      "grad_norm": 0.5473438379310454,
+      "learning_rate": 3.202634182596812e-05,
+      "loss": 0.4925,
+      "step": 455
+    },
+    {
+      "epoch": 1.8294884653961887,
+      "grad_norm": 0.5343196190452553,
+      "learning_rate": 3.1981470189811244e-05,
+      "loss": 0.5044,
+      "step": 456
+    },
+    {
+      "epoch": 1.8335005015045134,
+      "grad_norm": 0.5421306829483467,
+      "learning_rate": 3.1936504283650076e-05,
+      "loss": 0.517,
+      "step": 457
+    },
+    {
+      "epoch": 1.8375125376128385,
+      "grad_norm": 0.5743757330071741,
+      "learning_rate": 3.189144446127561e-05,
+      "loss": 0.4957,
+      "step": 458
+    },
+    {
+      "epoch": 1.8415245737211636,
+      "grad_norm": 0.47110459472470867,
+      "learning_rate": 3.184629107721775e-05,
+      "loss": 0.5239,
+      "step": 459
+    },
+    {
+      "epoch": 1.8455366098294883,
+      "grad_norm": 0.5339813882541631,
+      "learning_rate": 3.1801044486742567e-05,
+      "loss": 0.5049,
+      "step": 460
+    },
+    {
+      "epoch": 1.8495486459378134,
+      "grad_norm": 0.45841650781747373,
+      "learning_rate": 3.1755705045849465e-05,
+      "loss": 0.493,
+      "step": 461
+    },
+    {
+      "epoch": 1.8535606820461386,
+      "grad_norm": 0.49467906485797375,
+      "learning_rate": 3.1710273111268396e-05,
+      "loss": 0.5165,
+      "step": 462
+    },
+    {
+      "epoch": 1.8575727181544632,
+      "grad_norm": 0.4682656656976467,
+      "learning_rate": 3.1664749040457054e-05,
+      "loss": 0.5104,
+      "step": 463
+    },
+    {
+      "epoch": 1.8615847542627884,
+      "grad_norm": 0.497609444713446,
+      "learning_rate": 3.161913319159806e-05,
+      "loss": 0.5002,
+      "step": 464
+    },
+    {
+      "epoch": 1.8655967903711135,
+      "grad_norm": 0.4554751111920033,
+      "learning_rate": 3.157342592359612e-05,
+      "loss": 0.4981,
+      "step": 465
+    },
+    {
+      "epoch": 1.8696088264794382,
+      "grad_norm": 0.5434986094551325,
+      "learning_rate": 3.152762759607526e-05,
+      "loss": 0.5055,
+      "step": 466
+    },
+    {
+      "epoch": 1.8736208625877633,
+      "grad_norm": 0.5854791670003544,
+      "learning_rate": 3.148173856937594e-05,
+      "loss": 0.5226,
+      "step": 467
+    },
+    {
+      "epoch": 1.8776328986960884,
+      "grad_norm": 0.5615536864901692,
+      "learning_rate": 3.1435759204552246e-05,
+      "loss": 0.5305,
+      "step": 468
+    },
+    {
+      "epoch": 1.881644934804413,
+      "grad_norm": 0.48420939361370163,
+      "learning_rate": 3.138968986336904e-05,
+      "loss": 0.4969,
+      "step": 469
+    },
+    {
+      "epoch": 1.8856569709127382,
+      "grad_norm": 0.5088618795036021,
+      "learning_rate": 3.1343530908299116e-05,
+      "loss": 0.5265,
+      "step": 470
+    },
+    {
+      "epoch": 1.8896690070210633,
+      "grad_norm": 0.4646027468282,
+      "learning_rate": 3.129728270252036e-05,
+      "loss": 0.4925,
+      "step": 471
+    },
+    {
+      "epoch": 1.893681043129388,
+      "grad_norm": 0.4700295272966421,
+      "learning_rate": 3.125094560991285e-05,
+      "loss": 0.5333,
+      "step": 472
+    },
+    {
+      "epoch": 1.8976930792377131,
+      "grad_norm": 0.4534952754728935,
+      "learning_rate": 3.1204519995056056e-05,
+      "loss": 0.5265,
+      "step": 473
+    },
+    {
+      "epoch": 1.9017051153460383,
+      "grad_norm": 0.4142265066504554,
+      "learning_rate": 3.115800622322594e-05,
+      "loss": 0.5172,
+      "step": 474
+    },
+    {
+      "epoch": 1.905717151454363,
+      "grad_norm": 0.4205801242715312,
+      "learning_rate": 3.111140466039205e-05,
+      "loss": 0.4986,
+      "step": 475
+    },
+    {
+      "epoch": 1.909729187562688,
+      "grad_norm": 0.4742278740826261,
+      "learning_rate": 3.10647156732147e-05,
+      "loss": 0.5219,
+      "step": 476
+    },
+    {
+      "epoch": 1.9137412236710132,
+      "grad_norm": 0.4331077847654615,
+      "learning_rate": 3.101793962904205e-05,
+      "loss": 0.49,
+      "step": 477
+    },
+    {
+      "epoch": 1.9177532597793379,
+      "grad_norm": 0.4573340356240811,
+      "learning_rate": 3.097107689590723e-05,
+      "loss": 0.5286,
+      "step": 478
+    },
+    {
+      "epoch": 1.921765295887663,
+      "grad_norm": 0.4369495958876047,
+      "learning_rate": 3.0924127842525415e-05,
+      "loss": 0.499,
+      "step": 479
+    },
+    {
+      "epoch": 1.925777331995988,
+      "grad_norm": 0.5121574853263784,
+      "learning_rate": 3.087709283829098e-05,
+      "loss": 0.5078,
+      "step": 480
+    },
+    {
+      "epoch": 1.9297893681043128,
+      "grad_norm": 0.4508944158071729,
+      "learning_rate": 3.082997225327452e-05,
+      "loss": 0.5184,
+      "step": 481
+    },
+    {
+      "epoch": 1.933801404212638,
+      "grad_norm": 0.47198983610169853,
+      "learning_rate": 3.078276645822001e-05,
+      "loss": 0.5085,
+      "step": 482
+    },
+    {
+      "epoch": 1.937813440320963,
+      "grad_norm": 0.46409329431899354,
+      "learning_rate": 3.073547582454184e-05,
+      "loss": 0.5086,
+      "step": 483
+    },
+    {
+      "epoch": 1.9418254764292877,
+      "grad_norm": 0.4104927488146024,
+      "learning_rate": 3.0688100724321924e-05,
+      "loss": 0.5026,
+      "step": 484
+    },
+    {
+      "epoch": 1.9458375125376128,
+      "grad_norm": 0.38353129695229743,
+      "learning_rate": 3.064064153030673e-05,
+      "loss": 0.4971,
+      "step": 485
+    },
+    {
+      "epoch": 1.949849548645938,
+      "grad_norm": 0.4606268628529823,
+      "learning_rate": 3.059309861590441e-05,
+      "loss": 0.4921,
+      "step": 486
+    },
+    {
+      "epoch": 1.9538615847542626,
+      "grad_norm": 0.4662176457909344,
+      "learning_rate": 3.0545472355181804e-05,
+      "loss": 0.5213,
+      "step": 487
+    },
+    {
+      "epoch": 1.9578736208625878,
+      "grad_norm": 0.45061554991192976,
+      "learning_rate": 3.0497763122861516e-05,
+      "loss": 0.5103,
+      "step": 488
+    },
+    {
+      "epoch": 1.9618856569709129,
+      "grad_norm": 0.4620294707524686,
+      "learning_rate": 3.0449971294318977e-05,
+      "loss": 0.5357,
+      "step": 489
+    },
+    {
+      "epoch": 1.9658976930792376,
+      "grad_norm": 0.4924661385297439,
+      "learning_rate": 3.0402097245579493e-05,
+      "loss": 0.5342,
+      "step": 490
+    },
+    {
+      "epoch": 1.9699097291875627,
+      "grad_norm": 0.479911683768078,
+      "learning_rate": 3.0354141353315264e-05,
+      "loss": 0.5046,
+      "step": 491
+    },
+    {
+      "epoch": 1.9739217652958878,
+      "grad_norm": 0.4677785504849579,
+      "learning_rate": 3.0306103994842435e-05,
+      "loss": 0.5144,
+      "step": 492
+    },
+    {
+      "epoch": 1.9779338014042125,
+      "grad_norm": 0.4796863988821568,
+      "learning_rate": 3.0257985548118127e-05,
+      "loss": 0.514,
+      "step": 493
+    },
+    {
+      "epoch": 1.9819458375125376,
+      "grad_norm": 0.46975159900780417,
+      "learning_rate": 3.0209786391737466e-05,
+      "loss": 0.4904,
+      "step": 494
+    },
+    {
+      "epoch": 1.9859578736208627,
+      "grad_norm": 0.4588142493039961,
+      "learning_rate": 3.0161506904930592e-05,
+      "loss": 0.5057,
+      "step": 495
+    },
+    {
+      "epoch": 1.9899699097291874,
+      "grad_norm": 0.41160684125218905,
+      "learning_rate": 3.0113147467559697e-05,
+      "loss": 0.5025,
+      "step": 496
+    },
+    {
+      "epoch": 1.9939819458375125,
+      "grad_norm": 0.5326070545153468,
+      "learning_rate": 3.0064708460116007e-05,
+      "loss": 0.5145,
+      "step": 497
+    },
+    {
+      "epoch": 1.9979939819458377,
+      "grad_norm": 0.4322859173107025,
+      "learning_rate": 3.0016190263716815e-05,
+      "loss": 0.5107,
+      "step": 498
+    },
+    {
+      "epoch": 2.0283544303797467,
+      "grad_norm": 0.7704102834836188,
+      "learning_rate": 2.965396831793362e-05,
+      "loss": 0.3794,
+      "step": 499
+    },
+    {
+      "epoch": 2.0324050632911392,
+      "grad_norm": 0.5692071918217336,
+      "learning_rate": 2.9604221019627316e-05,
+      "loss": 0.3646,
+      "step": 500
+    },
+    {
+      "epoch": 2.0364556962025318,
+      "grad_norm": 0.9467962877740839,
+      "learning_rate": 2.955439637024526e-05,
+      "loss": 0.3712,
+      "step": 501
+    },
+    {
+      "epoch": 2.0405063291139243,
+      "grad_norm": 0.5902514758303117,
+      "learning_rate": 2.9504494771068334e-05,
+      "loss": 0.3769,
+      "step": 502
+    },
+    {
+      "epoch": 2.0445569620253163,
+      "grad_norm": 0.573735013004024,
+      "learning_rate": 2.9454516623997156e-05,
+      "loss": 0.3704,
+      "step": 503
+    },
+    {
+      "epoch": 2.048607594936709,
+      "grad_norm": 0.5968572553741621,
+      "learning_rate": 2.9404462331548847e-05,
+      "loss": 0.3664,
+      "step": 504
+    },
+    {
+      "epoch": 2.0526582278481014,
+      "grad_norm": 0.4880125183827262,
+      "learning_rate": 2.93543322968538e-05,
+      "loss": 0.386,
+      "step": 505
+    },
+    {
+      "epoch": 2.056708860759494,
+      "grad_norm": 0.5368570466304421,
+      "learning_rate": 2.9304126923652428e-05,
+      "loss": 0.365,
+      "step": 506
+    },
+    {
+      "epoch": 2.060759493670886,
+      "grad_norm": 0.5019850315103731,
+      "learning_rate": 2.9253846616291896e-05,
+      "loss": 0.375,
+      "step": 507
+    },
+    {
+      "epoch": 2.0648101265822785,
+      "grad_norm": 0.5838346932972964,
+      "learning_rate": 2.9203491779722896e-05,
+      "loss": 0.3548,
+      "step": 508
+    },
+    {
+      "epoch": 2.068860759493671,
+      "grad_norm": 0.46290970944033705,
+      "learning_rate": 2.9153062819496357e-05,
+      "loss": 0.3919,
+      "step": 509
+    },
+    {
+      "epoch": 2.0729113924050635,
+      "grad_norm": 0.6031546656017543,
+      "learning_rate": 2.9102560141760178e-05,
+      "loss": 0.37,
+      "step": 510
+    },
+    {
+      "epoch": 2.0769620253164556,
+      "grad_norm": 0.4885735509055873,
+      "learning_rate": 2.9051984153256004e-05,
+      "loss": 0.3547,
+      "step": 511
+    },
+    {
+      "epoch": 2.081012658227848,
+      "grad_norm": 0.5210464099451096,
+      "learning_rate": 2.900133526131588e-05,
+      "loss": 0.3675,
+      "step": 512
+    },
+    {
+      "epoch": 2.0850632911392406,
+      "grad_norm": 0.4886475589670207,
+      "learning_rate": 2.8950613873859025e-05,
+      "loss": 0.3631,
+      "step": 513
+    },
+    {
+      "epoch": 2.089113924050633,
+      "grad_norm": 0.45541465821259763,
+      "learning_rate": 2.8899820399388515e-05,
+      "loss": 0.3604,
+      "step": 514
+    },
+    {
+      "epoch": 2.093164556962025,
+      "grad_norm": 0.47907749012248535,
+      "learning_rate": 2.8848955246988012e-05,
+      "loss": 0.3765,
+      "step": 515
+    },
+    {
+      "epoch": 2.0972151898734177,
+      "grad_norm": 0.4409541200038464,
+      "learning_rate": 2.879801882631847e-05,
+      "loss": 0.3641,
+      "step": 516
+    },
+    {
+      "epoch": 2.1012658227848102,
+      "grad_norm": 0.44754717711895387,
+      "learning_rate": 2.8747011547614808e-05,
+      "loss": 0.3408,
+      "step": 517
+    },
+    {
+      "epoch": 2.1053164556962027,
+      "grad_norm": 0.4212410924336331,
+      "learning_rate": 2.8695933821682635e-05,
+      "loss": 0.3737,
+      "step": 518
+    },
+    {
+      "epoch": 2.109367088607595,
+      "grad_norm": 0.44571253063197375,
+      "learning_rate": 2.864478605989494e-05,
+      "loss": 0.3503,
+      "step": 519
+    },
+    {
+      "epoch": 2.1134177215189873,
+      "grad_norm": 0.45461481765613937,
+      "learning_rate": 2.8593568674188765e-05,
+      "loss": 0.3605,
+      "step": 520
+    },
+    {
+      "epoch": 2.11746835443038,
+      "grad_norm": 0.4213992807800835,
+      "learning_rate": 2.8542282077061892e-05,
+      "loss": 0.3562,
+      "step": 521
+    },
+    {
+      "epoch": 2.1215189873417724,
+      "grad_norm": 0.4476234967709086,
+      "learning_rate": 2.8490926681569523e-05,
+      "loss": 0.3549,
+      "step": 522
+    },
+    {
+      "epoch": 2.1255696202531644,
+      "grad_norm": 0.4170659428313066,
+      "learning_rate": 2.8439502901320956e-05,
+      "loss": 0.3404,
+      "step": 523
+    },
+    {
+      "epoch": 2.129620253164557,
+      "grad_norm": 0.6119410797877736,
+      "learning_rate": 2.8388011150476237e-05,
+      "loss": 0.3433,
+      "step": 524
+    },
+    {
+      "epoch": 2.1336708860759495,
+      "grad_norm": 0.46927242092206944,
+      "learning_rate": 2.8336451843742866e-05,
+      "loss": 0.3393,
+      "step": 525
+    },
+    {
+      "epoch": 2.137721518987342,
+      "grad_norm": 0.48696362186995024,
+      "learning_rate": 2.8284825396372387e-05,
+      "loss": 0.3677,
+      "step": 526
+    },
+    {
+      "epoch": 2.141772151898734,
+      "grad_norm": 0.44206685134782275,
+      "learning_rate": 2.8233132224157132e-05,
+      "loss": 0.3652,
+      "step": 527
+    },
+    {
+      "epoch": 2.1458227848101266,
+      "grad_norm": 0.47920441430079386,
+      "learning_rate": 2.8181372743426805e-05,
+      "loss": 0.3767,
+      "step": 528
+    },
+    {
+      "epoch": 2.149873417721519,
+      "grad_norm": 0.496658988709585,
+      "learning_rate": 2.8129547371045128e-05,
+      "loss": 0.3597,
+      "step": 529
+    },
+    {
+      "epoch": 2.1539240506329116,
+      "grad_norm": 0.4286857490514013,
+      "learning_rate": 2.8077656524406534e-05,
+      "loss": 0.3496,
+      "step": 530
+    },
+    {
+      "epoch": 2.1579746835443037,
+      "grad_norm": 0.43225760971118893,
+      "learning_rate": 2.802570062143278e-05,
+      "loss": 0.3377,
+      "step": 531
+    },
+    {
+      "epoch": 2.162025316455696,
+      "grad_norm": 0.4755823097342794,
+      "learning_rate": 2.7973680080569555e-05,
+      "loss": 0.3658,
+      "step": 532
+    },
+    {
+      "epoch": 2.1660759493670887,
+      "grad_norm": 0.412381427080124,
+      "learning_rate": 2.792159532078314e-05,
+      "loss": 0.3665,
+      "step": 533
+    },
+    {
+      "epoch": 2.170126582278481,
+      "grad_norm": 0.4581030467450501,
+      "learning_rate": 2.7869446761557033e-05,
+      "loss": 0.374,
+      "step": 534
+    },
+    {
+      "epoch": 2.1741772151898733,
+      "grad_norm": 0.40945232049764396,
+      "learning_rate": 2.781723482288857e-05,
+      "loss": 0.3523,
+      "step": 535
+    },
+    {
+      "epoch": 2.178227848101266,
+      "grad_norm": 0.4432463171906232,
+      "learning_rate": 2.7764959925285517e-05,
+      "loss": 0.3435,
+      "step": 536
+    },
+    {
+      "epoch": 2.1822784810126583,
+      "grad_norm": 0.46449145228189204,
+      "learning_rate": 2.771262248976272e-05,
+      "loss": 0.3742,
+      "step": 537
+    },
+    {
+      "epoch": 2.186329113924051,
+      "grad_norm": 0.43514336264712855,
+      "learning_rate": 2.7660222937838677e-05,
+      "loss": 0.3561,
+      "step": 538
+    },
+    {
+      "epoch": 2.190379746835443,
+      "grad_norm": 0.40037743923643426,
+      "learning_rate": 2.7607761691532186e-05,
+      "loss": 0.363,
+      "step": 539
+    },
+    {
+      "epoch": 2.1944303797468354,
+      "grad_norm": 0.4115852899113939,
+      "learning_rate": 2.7555239173358916e-05,
+      "loss": 0.3592,
+      "step": 540
+    },
+    {
+      "epoch": 2.198481012658228,
+      "grad_norm": 0.414869014529127,
+      "learning_rate": 2.7502655806328e-05,
+      "loss": 0.3851,
+      "step": 541
+    },
+    {
+      "epoch": 2.2025316455696204,
+      "grad_norm": 0.463552713790135,
+      "learning_rate": 2.7450012013938648e-05,
+      "loss": 0.3535,
+      "step": 542
+    },
+    {
+      "epoch": 2.2065822784810125,
+      "grad_norm": 0.46202501120180556,
+      "learning_rate": 2.739730822017673e-05,
+      "loss": 0.3539,
+      "step": 543
+    },
+    {
+      "epoch": 2.210632911392405,
+      "grad_norm": 0.4438013704560041,
+      "learning_rate": 2.7344544849511355e-05,
+      "loss": 0.3581,
+      "step": 544
+    },
+    {
+      "epoch": 2.2146835443037975,
+      "grad_norm": 0.44089529085317986,
+      "learning_rate": 2.7291722326891456e-05,
+      "loss": 0.3545,
+      "step": 545
+    },
+    {
+      "epoch": 2.21873417721519,
+      "grad_norm": 0.43956053967384245,
+      "learning_rate": 2.723884107774236e-05,
+      "loss": 0.3682,
+      "step": 546
+    },
+    {
+      "epoch": 2.222784810126582,
+      "grad_norm": 0.3924480419751318,
+      "learning_rate": 2.718590152796239e-05,
+      "loss": 0.3649,
+      "step": 547
+    },
+    {
+      "epoch": 2.2268354430379746,
+      "grad_norm": 0.42940291054972923,
+      "learning_rate": 2.71329041039194e-05,
+      "loss": 0.3618,
+      "step": 548
+    },
+    {
+      "epoch": 2.230886075949367,
+      "grad_norm": 0.42669335957573,
+      "learning_rate": 2.7079849232447357e-05,
+      "loss": 0.3714,
+      "step": 549
+    },
+    {
+      "epoch": 2.2349367088607597,
+      "grad_norm": 0.41229494670367917,
+      "learning_rate": 2.7026737340842895e-05,
+      "loss": 0.3447,
+      "step": 550
+    },
+    {
+      "epoch": 2.2389873417721518,
+      "grad_norm": 0.43288057623576365,
+      "learning_rate": 2.697356885686189e-05,
+      "loss": 0.3589,
+      "step": 551
+    },
+    {
+      "epoch": 2.2430379746835443,
+      "grad_norm": 0.42050349288428895,
+      "learning_rate": 2.6920344208716014e-05,
+      "loss": 0.3541,
+      "step": 552
+    },
+    {
+      "epoch": 2.247088607594937,
+      "grad_norm": 0.4199965692836358,
+      "learning_rate": 2.6867063825069252e-05,
+      "loss": 0.3829,
+      "step": 553
+    },
+    {
+      "epoch": 2.2511392405063293,
+      "grad_norm": 0.4381940626567847,
+      "learning_rate": 2.6813728135034494e-05,
+      "loss": 0.3605,
+      "step": 554
+    },
+    {
+      "epoch": 2.2551898734177214,
+      "grad_norm": 0.40434878899532994,
+      "learning_rate": 2.6760337568170056e-05,
+      "loss": 0.3484,
+      "step": 555
+    },
+    {
+      "epoch": 2.259240506329114,
+      "grad_norm": 0.49184649978106343,
+      "learning_rate": 2.6706892554476226e-05,
+      "loss": 0.3765,
+      "step": 556
+    },
+    {
+      "epoch": 2.2632911392405064,
+      "grad_norm": 0.42436481232986467,
+      "learning_rate": 2.6653393524391795e-05,
+      "loss": 0.3809,
+      "step": 557
+    },
+    {
+      "epoch": 2.267341772151899,
+      "grad_norm": 0.4416660504352746,
+      "learning_rate": 2.6599840908790592e-05,
+      "loss": 0.367,
+      "step": 558
+    },
+    {
+      "epoch": 2.271392405063291,
+      "grad_norm": 0.4171173762818663,
+      "learning_rate": 2.6546235138978028e-05,
+      "loss": 0.3886,
+      "step": 559
+    },
+    {
+      "epoch": 2.2754430379746835,
+      "grad_norm": 0.41334081680662077,
+      "learning_rate": 2.6492576646687597e-05,
+      "loss": 0.3449,
+      "step": 560
+    },
+    {
+      "epoch": 2.279493670886076,
+      "grad_norm": 0.43922691670714553,
+      "learning_rate": 2.6438865864077425e-05,
+      "loss": 0.3582,
+      "step": 561
+    },
+    {
+      "epoch": 2.2835443037974685,
+      "grad_norm": 0.4089701676553786,
+      "learning_rate": 2.6385103223726766e-05,
+      "loss": 0.3657,
+      "step": 562
+    },
+    {
+      "epoch": 2.2875949367088606,
+      "grad_norm": 0.45745456788013816,
+      "learning_rate": 2.6331289158632537e-05,
+      "loss": 0.3776,
+      "step": 563
+    },
+    {
+      "epoch": 2.291645569620253,
+      "grad_norm": 0.44074386839421675,
+      "learning_rate": 2.6277424102205817e-05,
+      "loss": 0.3502,
+      "step": 564
+    },
+    {
+      "epoch": 2.2956962025316456,
+      "grad_norm": 0.4505872470828455,
+      "learning_rate": 2.6223508488268374e-05,
+      "loss": 0.3612,
+      "step": 565
+    },
+    {
+      "epoch": 2.299746835443038,
+      "grad_norm": 0.39153083200757155,
+      "learning_rate": 2.6169542751049148e-05,
+      "loss": 0.3801,
+      "step": 566
+    },
+    {
+      "epoch": 2.3037974683544302,
+      "grad_norm": 0.51078223648984,
+      "learning_rate": 2.6115527325180754e-05,
+      "loss": 0.3731,
+      "step": 567
+    },
+    {
+      "epoch": 2.3078481012658227,
+      "grad_norm": 0.4316468796140628,
+      "learning_rate": 2.606146264569603e-05,
+      "loss": 0.3623,
+      "step": 568
+    },
+    {
+      "epoch": 2.3118987341772153,
+      "grad_norm": 0.471562668318298,
+      "learning_rate": 2.6007349148024447e-05,
+      "loss": 0.353,
+      "step": 569
+    },
+    {
+      "epoch": 2.3159493670886078,
+      "grad_norm": 0.44798217381467365,
+      "learning_rate": 2.5953187267988694e-05,
+      "loss": 0.3658,
+      "step": 570
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.4600089309218735,
+      "learning_rate": 2.5898977441801097e-05,
+      "loss": 0.3624,
+      "step": 571
+    },
+    {
+      "epoch": 2.3240506329113924,
+      "grad_norm": 0.3948645778105983,
+      "learning_rate": 2.584472010606015e-05,
+      "loss": 0.3655,
+      "step": 572
+    },
+    {
+      "epoch": 2.328101265822785,
+      "grad_norm": 0.4292424465666896,
+      "learning_rate": 2.5790415697746976e-05,
+      "loss": 0.3738,
+      "step": 573
+    },
+    {
+      "epoch": 2.3321518987341774,
+      "grad_norm": 0.39445584105228393,
+      "learning_rate": 2.5736064654221808e-05,
+      "loss": 0.3592,
+      "step": 574
+    },
+    {
+      "epoch": 2.3362025316455695,
+      "grad_norm": 0.4093521241560843,
+      "learning_rate": 2.568166741322048e-05,
+      "loss": 0.3763,
+      "step": 575
+    },
+    {
+      "epoch": 2.340253164556962,
+      "grad_norm": 0.41521489739778356,
+      "learning_rate": 2.56272244128509e-05,
+      "loss": 0.3702,
+      "step": 576
+    },
+    {
+      "epoch": 2.3443037974683545,
+      "grad_norm": 0.40831554840288803,
+      "learning_rate": 2.55727360915895e-05,
+      "loss": 0.3636,
+      "step": 577
+    },
+    {
+      "epoch": 2.348354430379747,
+      "grad_norm": 0.4343431482592834,
+      "learning_rate": 2.5518202888277734e-05,
+      "loss": 0.3658,
+      "step": 578
+    },
+    {
+      "epoch": 2.352405063291139,
+      "grad_norm": 0.429398207221872,
+      "learning_rate": 2.5463625242118523e-05,
+      "loss": 0.3637,
+      "step": 579
+    },
+    {
+      "epoch": 2.3564556962025316,
+      "grad_norm": 0.45112294723104207,
+      "learning_rate": 2.5409003592672723e-05,
+      "loss": 0.3637,
+      "step": 580
+    },
+    {
+      "epoch": 2.360506329113924,
+      "grad_norm": 0.41844458161614667,
+      "learning_rate": 2.535433837985559e-05,
+      "loss": 0.3605,
+      "step": 581
+    },
+    {
+      "epoch": 2.3645569620253166,
+      "grad_norm": 0.41967689586804774,
+      "learning_rate": 2.529963004393324e-05,
+      "loss": 0.3651,
+      "step": 582
+    },
+    {
+      "epoch": 2.3686075949367087,
+      "grad_norm": 0.41343002788190514,
+      "learning_rate": 2.524487902551908e-05,
+      "loss": 0.3603,
+      "step": 583
+    },
+    {
+      "epoch": 2.372658227848101,
+      "grad_norm": 0.43691899882683644,
+      "learning_rate": 2.519008576557029e-05,
+      "loss": 0.3718,
+      "step": 584
+    },
+    {
+      "epoch": 2.3767088607594937,
+      "grad_norm": 0.4469001593197778,
+      "learning_rate": 2.5135250705384254e-05,
+      "loss": 0.3844,
+      "step": 585
+    },
+    {
+      "epoch": 2.3807594936708862,
+      "grad_norm": 0.41211751741659636,
+      "learning_rate": 2.5080374286595007e-05,
+      "loss": 0.3814,
+      "step": 586
+    },
+    {
+      "epoch": 2.3848101265822783,
+      "grad_norm": 0.42953146319588964,
+      "learning_rate": 2.5025456951169677e-05,
+      "loss": 0.3456,
+      "step": 587
+    },
+    {
+      "epoch": 2.388860759493671,
+      "grad_norm": 0.3968544047751466,
+      "learning_rate": 2.4970499141404942e-05,
+      "loss": 0.3701,
+      "step": 588
+    },
+    {
+      "epoch": 2.3929113924050633,
+      "grad_norm": 0.3891069749721624,
+      "learning_rate": 2.491550129992345e-05,
+      "loss": 0.341,
+      "step": 589
+    },
+    {
+      "epoch": 2.396962025316456,
+      "grad_norm": 0.4524471436596732,
+      "learning_rate": 2.486046386967024e-05,
+      "loss": 0.3457,
+      "step": 590
+    },
+    {
+      "epoch": 2.401012658227848,
+      "grad_norm": 0.40439914837082963,
+      "learning_rate": 2.4805387293909214e-05,
+      "loss": 0.3509,
+      "step": 591
+    },
+    {
+      "epoch": 2.4050632911392404,
+      "grad_norm": 0.40876198435319255,
+      "learning_rate": 2.4750272016219552e-05,
+      "loss": 0.3502,
+      "step": 592
+    },
+    {
+      "epoch": 2.409113924050633,
+      "grad_norm": 0.3985636350751559,
+      "learning_rate": 2.4695118480492114e-05,
+      "loss": 0.3773,
+      "step": 593
+    },
+    {
+      "epoch": 2.4131645569620255,
+      "grad_norm": 0.3945749368438278,
+      "learning_rate": 2.4639927130925898e-05,
+      "loss": 0.3505,
+      "step": 594
+    },
+    {
+      "epoch": 2.4172151898734175,
+      "grad_norm": 0.39126839711383726,
+      "learning_rate": 2.458469841202444e-05,
+      "loss": 0.3624,
+      "step": 595
+    },
+    {
+      "epoch": 2.42126582278481,
+      "grad_norm": 0.42944568520297577,
+      "learning_rate": 2.452943276859226e-05,
+      "loss": 0.3644,
+      "step": 596
+    },
+    {
+      "epoch": 2.4253164556962026,
+      "grad_norm": 0.41403389092669113,
+      "learning_rate": 2.447413064573125e-05,
+      "loss": 0.3773,
+      "step": 597
+    },
+    {
+      "epoch": 2.429367088607595,
+      "grad_norm": 0.4186640510797011,
+      "learning_rate": 2.4418792488837095e-05,
+      "loss": 0.3777,
+      "step": 598
+    },
+    {
+      "epoch": 2.433417721518987,
+      "grad_norm": 0.3761059786533923,
+      "learning_rate": 2.4363418743595713e-05,
+      "loss": 0.3791,
+      "step": 599
+    },
+    {
+      "epoch": 2.4374683544303797,
+      "grad_norm": 0.3921168408255149,
+      "learning_rate": 2.430800985597963e-05,
+      "loss": 0.3596,
+      "step": 600
+    },
+    {
+      "epoch": 2.441518987341772,
+      "grad_norm": 0.4148312028226884,
+      "learning_rate": 2.4252566272244415e-05,
+      "loss": 0.3639,
+      "step": 601
+    },
+    {
+      "epoch": 2.4455696202531647,
+      "grad_norm": 0.3622479237659724,
+      "learning_rate": 2.4197088438925063e-05,
+      "loss": 0.3685,
+      "step": 602
+    },
+    {
+      "epoch": 2.449620253164557,
+      "grad_norm": 0.40760978761155037,
+      "learning_rate": 2.4141576802832417e-05,
+      "loss": 0.3678,
+      "step": 603
+    },
+    {
+      "epoch": 2.4536708860759493,
+      "grad_norm": 0.4645057232049641,
+      "learning_rate": 2.408603181104957e-05,
+      "loss": 0.3552,
+      "step": 604
+    },
+    {
+      "epoch": 2.457721518987342,
+      "grad_norm": 0.4067832502561007,
+      "learning_rate": 2.4030453910928245e-05,
+      "loss": 0.3678,
+      "step": 605
+    },
+    {
+      "epoch": 2.4617721518987343,
+      "grad_norm": 0.4387041068174868,
+      "learning_rate": 2.397484355008521e-05,
+      "loss": 0.352,
+      "step": 606
+    },
+    {
+      "epoch": 2.4658227848101264,
+      "grad_norm": 0.44034979824713943,
+      "learning_rate": 2.3919201176398662e-05,
+      "loss": 0.3728,
+      "step": 607
+    },
+    {
+      "epoch": 2.469873417721519,
+      "grad_norm": 0.417640131279379,
+      "learning_rate": 2.3863527238004633e-05,
+      "loss": 0.3669,
+      "step": 608
+    },
+    {
+      "epoch": 2.4739240506329114,
+      "grad_norm": 0.4080812380838371,
+      "learning_rate": 2.380782218329337e-05,
+      "loss": 0.3678,
+      "step": 609
+    },
+    {
+      "epoch": 2.477974683544304,
+      "grad_norm": 0.39121926494077963,
+      "learning_rate": 2.3752086460905725e-05,
+      "loss": 0.3533,
+      "step": 610
+    },
+    {
+      "epoch": 2.482025316455696,
+      "grad_norm": 0.442021071175835,
+      "learning_rate": 2.3696320519729544e-05,
+      "loss": 0.3726,
+      "step": 611
+    },
+    {
+      "epoch": 2.4860759493670885,
+      "grad_norm": 0.4066892479397453,
+      "learning_rate": 2.3640524808896045e-05,
+      "loss": 0.3617,
+      "step": 612
+    },
+    {
+      "epoch": 2.490126582278481,
+      "grad_norm": 0.4037199656223477,
+      "learning_rate": 2.3584699777776222e-05,
+      "loss": 0.3765,
+      "step": 613
+    },
+    {
+      "epoch": 2.4941772151898736,
+      "grad_norm": 0.46153057791002744,
+      "learning_rate": 2.3528845875977195e-05,
+      "loss": 0.3733,
+      "step": 614
+    },
+    {
+      "epoch": 2.4982278481012656,
+      "grad_norm": 0.37847827544964036,
+      "learning_rate": 2.3472963553338614e-05,
+      "loss": 0.366,
+      "step": 615
+    },
+    {
+      "epoch": 2.502278481012658,
+      "grad_norm": 0.3917091673792605,
+      "learning_rate": 2.341705325992901e-05,
+      "loss": 0.3444,
+      "step": 616
+    },
+    {
+      "epoch": 2.5063291139240507,
+      "grad_norm": 0.40027915826139066,
+      "learning_rate": 2.336111544604222e-05,
+      "loss": 0.3782,
+      "step": 617
+    },
+    {
+      "epoch": 2.510379746835443,
+      "grad_norm": 0.3875336727702107,
+      "learning_rate": 2.33051505621937e-05,
+      "loss": 0.3595,
+      "step": 618
+    },
+    {
+      "epoch": 2.5144303797468357,
+      "grad_norm": 0.4024707268570018,
+      "learning_rate": 2.324915905911693e-05,
+      "loss": 0.3643,
+      "step": 619
+    },
+    {
+      "epoch": 2.5184810126582278,
+      "grad_norm": 0.4124896669244531,
+      "learning_rate": 2.319314138775977e-05,
+      "loss": 0.384,
+      "step": 620
+    },
+    {
+      "epoch": 2.5225316455696203,
+      "grad_norm": 0.43889154148980547,
+      "learning_rate": 2.3137097999280856e-05,
+      "loss": 0.3735,
+      "step": 621
+    },
+    {
+      "epoch": 2.526582278481013,
+      "grad_norm": 0.4384150665220281,
+      "learning_rate": 2.308102934504593e-05,
+      "loss": 0.3589,
+      "step": 622
+    },
+    {
+      "epoch": 2.530632911392405,
+      "grad_norm": 0.6349940771678356,
+      "learning_rate": 2.3024935876624222e-05,
+      "loss": 0.3678,
+      "step": 623
+    },
+    {
+      "epoch": 2.5346835443037974,
+      "grad_norm": 0.4145927775483372,
+      "learning_rate": 2.2968818045784813e-05,
+      "loss": 0.3488,
+      "step": 624
+    },
+    {
+      "epoch": 2.53873417721519,
+      "grad_norm": 0.3926650061202865,
+      "learning_rate": 2.2912676304493006e-05,
+      "loss": 0.3906,
+      "step": 625
+    },
+    {
+      "epoch": 2.5427848101265824,
+      "grad_norm": 0.43931573429103166,
+      "learning_rate": 2.2856511104906668e-05,
+      "loss": 0.3561,
+      "step": 626
+    },
+    {
+      "epoch": 2.546835443037975,
+      "grad_norm": 0.3963999189137664,
+      "learning_rate": 2.2800322899372586e-05,
+      "loss": 0.3452,
+      "step": 627
+    },
+    {
+      "epoch": 2.550886075949367,
+      "grad_norm": 0.43128977890761244,
+      "learning_rate": 2.2744112140422844e-05,
+      "loss": 0.3711,
+      "step": 628
+    },
+    {
+      "epoch": 2.5549367088607595,
+      "grad_norm": 0.41309434363012687,
+      "learning_rate": 2.2687879280771177e-05,
+      "loss": 0.3586,
+      "step": 629
+    },
+    {
+      "epoch": 2.558987341772152,
+      "grad_norm": 0.40272860635209407,
+      "learning_rate": 2.26316247733093e-05,
+      "loss": 0.3549,
+      "step": 630
+    },
+    {
+      "epoch": 2.563037974683544,
+      "grad_norm": 0.4282174477234997,
+      "learning_rate": 2.257534907110328e-05,
+      "loss": 0.3654,
+      "step": 631
+    },
+    {
+      "epoch": 2.5670886075949366,
+      "grad_norm": 0.40400990199569786,
+      "learning_rate": 2.2519052627389882e-05,
+      "loss": 0.3278,
+      "step": 632
+    },
+    {
+      "epoch": 2.571139240506329,
+      "grad_norm": 0.4277612156239767,
+      "learning_rate": 2.246273589557294e-05,
+      "loss": 0.3612,
+      "step": 633
+    },
+    {
+      "epoch": 2.5751898734177217,
+      "grad_norm": 0.43267072375613386,
+      "learning_rate": 2.240639932921966e-05,
+      "loss": 0.365,
+      "step": 634
+    },
+    {
+      "epoch": 2.579240506329114,
+      "grad_norm": 0.4141213506664964,
+      "learning_rate": 2.2350043382056995e-05,
+      "loss": 0.369,
+      "step": 635
+    },
+    {
+      "epoch": 2.5832911392405062,
+      "grad_norm": 0.4096518643691402,
+      "learning_rate": 2.2293668507968015e-05,
+      "loss": 0.3836,
+      "step": 636
+    },
+    {
+      "epoch": 2.5873417721518988,
+      "grad_norm": 0.45435665741906817,
+      "learning_rate": 2.2237275160988186e-05,
+      "loss": 0.3633,
+      "step": 637
+    },
+    {
+      "epoch": 2.5913924050632913,
+      "grad_norm": 0.4112511128399866,
+      "learning_rate": 2.2180863795301787e-05,
+      "loss": 0.3484,
+      "step": 638
+    },
+    {
+      "epoch": 2.5954430379746833,
+      "grad_norm": 0.42711876119190056,
+      "learning_rate": 2.212443486523819e-05,
+      "loss": 0.3794,
+      "step": 639
+    },
+    {
+      "epoch": 2.599493670886076,
+      "grad_norm": 0.3982073769715853,
+      "learning_rate": 2.2067988825268243e-05,
+      "loss": 0.3538,
+      "step": 640
+    },
+    {
+      "epoch": 2.6035443037974684,
+      "grad_norm": 0.40838534652185576,
+      "learning_rate": 2.2011526130000596e-05,
+      "loss": 0.3532,
+      "step": 641
+    },
+    {
+      "epoch": 2.607594936708861,
+      "grad_norm": 0.45442443639091146,
+      "learning_rate": 2.1955047234178038e-05,
+      "loss": 0.3779,
+      "step": 642
+    },
+    {
+      "epoch": 2.6116455696202534,
+      "grad_norm": 0.42074538919250787,
+      "learning_rate": 2.1898552592673825e-05,
+      "loss": 0.3653,
+      "step": 643
+    },
+    {
+      "epoch": 2.6156962025316455,
+      "grad_norm": 0.3882370640712819,
+      "learning_rate": 2.184204266048803e-05,
+      "loss": 0.3589,
+      "step": 644
+    },
+    {
+      "epoch": 2.619746835443038,
+      "grad_norm": 0.4560052748068996,
+      "learning_rate": 2.1785517892743887e-05,
+      "loss": 0.3519,
+      "step": 645
+    },
+    {
+      "epoch": 2.6237974683544305,
+      "grad_norm": 0.42082089010228585,
+      "learning_rate": 2.17289787446841e-05,
+      "loss": 0.3844,
+      "step": 646
+    },
+    {
+      "epoch": 2.6278481012658226,
+      "grad_norm": 0.45739922014976686,
+      "learning_rate": 2.1672425671667198e-05,
+      "loss": 0.3621,
+      "step": 647
+    },
+    {
+      "epoch": 2.631898734177215,
+      "grad_norm": 0.4482335849089613,
+      "learning_rate": 2.161585912916385e-05,
+      "loss": 0.3393,
+      "step": 648
+    },
+    {
+      "epoch": 2.6359493670886076,
+      "grad_norm": 0.4190731571343439,
+      "learning_rate": 2.1559279572753214e-05,
+      "loss": 0.3594,
+      "step": 649
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.4323630375932559,
+      "learning_rate": 2.1502687458119268e-05,
+      "loss": 0.3629,
+      "step": 650
+    },
+    {
+      "epoch": 2.6440506329113926,
+      "grad_norm": 0.43519282474043997,
+      "learning_rate": 2.1446083241047116e-05,
+      "loss": 0.3572,
+      "step": 651
+    },
+    {
+      "epoch": 2.6481012658227847,
+      "grad_norm": 0.42807656969711144,
+      "learning_rate": 2.1389467377419333e-05,
+      "loss": 0.3624,
+      "step": 652
+    },
+    {
+      "epoch": 2.6521518987341772,
+      "grad_norm": 0.39184821355521987,
+      "learning_rate": 2.133284032321232e-05,
+      "loss": 0.3761,
+      "step": 653
+    },
+    {
+      "epoch": 2.6562025316455697,
+      "grad_norm": 0.4037555748486543,
+      "learning_rate": 2.1276202534492566e-05,
+      "loss": 0.3437,
+      "step": 654
+    },
+    {
+      "epoch": 2.660253164556962,
+      "grad_norm": 0.40049649479049837,
+      "learning_rate": 2.121955446741306e-05,
+      "loss": 0.361,
+      "step": 655
+    },
+    {
+      "epoch": 2.6643037974683543,
+      "grad_norm": 0.40635358306977337,
+      "learning_rate": 2.1162896578209517e-05,
+      "loss": 0.3756,
+      "step": 656
+    },
+    {
+      "epoch": 2.668354430379747,
+      "grad_norm": 0.400506329278798,
+      "learning_rate": 2.1106229323196813e-05,
+      "loss": 0.3502,
+      "step": 657
+    },
+    {
+      "epoch": 2.6724050632911394,
+      "grad_norm": 0.3960654787190359,
+      "learning_rate": 2.1049553158765214e-05,
+      "loss": 0.384,
+      "step": 658
+    },
+    {
+      "epoch": 2.676455696202532,
+      "grad_norm": 0.4094184939277511,
+      "learning_rate": 2.0992868541376764e-05,
+      "loss": 0.3645,
+      "step": 659
+    },
+    {
+      "epoch": 2.680506329113924,
+      "grad_norm": 0.3925198679588752,
+      "learning_rate": 2.093617592756158e-05,
+      "loss": 0.3976,
+      "step": 660
+    },
+    {
+      "epoch": 2.6845569620253165,
+      "grad_norm": 0.38804789374148585,
+      "learning_rate": 2.0879475773914167e-05,
+      "loss": 0.3856,
+      "step": 661
+    },
+    {
+      "epoch": 2.688607594936709,
+      "grad_norm": 0.42501207445400274,
+      "learning_rate": 2.082276853708978e-05,
+      "loss": 0.3747,
+      "step": 662
+    },
+    {
+      "epoch": 2.692658227848101,
+      "grad_norm": 0.37382414632280586,
+      "learning_rate": 2.076605467380071e-05,
+      "loss": 0.3715,
+      "step": 663
+    },
+    {
+      "epoch": 2.6967088607594936,
+      "grad_norm": 0.3961961619386375,
+      "learning_rate": 2.0709334640812613e-05,
+      "loss": 0.3744,
+      "step": 664
+    },
+    {
+      "epoch": 2.700759493670886,
+      "grad_norm": 0.4338197059648437,
+      "learning_rate": 2.0652608894940824e-05,
+      "loss": 0.3764,
+      "step": 665
+    },
+    {
+      "epoch": 2.7048101265822786,
+      "grad_norm": 0.4128711504983498,
+      "learning_rate": 2.0595877893046722e-05,
+      "loss": 0.3835,
+      "step": 666
+    },
+    {
+      "epoch": 2.708860759493671,
+      "grad_norm": 0.4246980338063562,
+      "learning_rate": 2.0539142092033985e-05,
+      "loss": 0.3717,
+      "step": 667
+    },
+    {
+      "epoch": 2.712911392405063,
+      "grad_norm": 0.4162255084171925,
+      "learning_rate": 2.048240194884496e-05,
+      "loss": 0.3794,
+      "step": 668
+    },
+    {
+      "epoch": 2.7169620253164557,
+      "grad_norm": 0.4084885581204382,
+      "learning_rate": 2.042565792045695e-05,
+      "loss": 0.3765,
+      "step": 669
+    },
+    {
+      "epoch": 2.721012658227848,
+      "grad_norm": 0.42031336510768375,
+      "learning_rate": 2.036891046387857e-05,
+      "loss": 0.348,
+      "step": 670
+    },
+    {
+      "epoch": 2.7250632911392403,
+      "grad_norm": 0.4301351351338328,
+      "learning_rate": 2.0312160036146036e-05,
+      "loss": 0.3549,
+      "step": 671
+    },
+    {
+      "epoch": 2.729113924050633,
+      "grad_norm": 0.37434436742401866,
+      "learning_rate": 2.025540709431948e-05,
+      "loss": 0.371,
+      "step": 672
+    },
+    {
+      "epoch": 2.7331645569620253,
+      "grad_norm": 0.4057242692209515,
+      "learning_rate": 2.0198652095479298e-05,
+      "loss": 0.3617,
+      "step": 673
+    },
+    {
+      "epoch": 2.737215189873418,
+      "grad_norm": 0.3954003210181357,
+      "learning_rate": 2.014189549672245e-05,
+      "loss": 0.3705,
+      "step": 674
+    },
+    {
+      "epoch": 2.7412658227848103,
+      "grad_norm": 0.39545646464547907,
+      "learning_rate": 2.0085137755158776e-05,
+      "loss": 0.3323,
+      "step": 675
+    },
+    {
+      "epoch": 2.7453164556962024,
+      "grad_norm": 0.38616264374660375,
+      "learning_rate": 2.0028379327907327e-05,
+      "loss": 0.3579,
+      "step": 676
+    },
+    {
+      "epoch": 2.749367088607595,
+      "grad_norm": 0.425911545893184,
+      "learning_rate": 1.9971620672092676e-05,
+      "loss": 0.3498,
+      "step": 677
+    },
+    {
+      "epoch": 2.7534177215189874,
+      "grad_norm": 0.3751945286360797,
+      "learning_rate": 1.991486224484123e-05,
+      "loss": 0.3581,
+      "step": 678
+    },
+    {
+      "epoch": 2.7574683544303795,
+      "grad_norm": 0.4133013785241136,
+      "learning_rate": 1.985810450327756e-05,
+      "loss": 0.3789,
+      "step": 679
+    },
+    {
+      "epoch": 2.761518987341772,
+      "grad_norm": 0.4058262176322736,
+      "learning_rate": 1.9801347904520706e-05,
+      "loss": 0.3737,
+      "step": 680
+    },
+    {
+      "epoch": 2.7655696202531646,
+      "grad_norm": 0.40883288281610924,
+      "learning_rate": 1.974459290568053e-05,
+      "loss": 0.3607,
+      "step": 681
+    },
+    {
+      "epoch": 2.769620253164557,
+      "grad_norm": 0.42675479082953005,
+      "learning_rate": 1.968783996385397e-05,
+      "loss": 0.3565,
+      "step": 682
+    },
+    {
+      "epoch": 2.7736708860759496,
+      "grad_norm": 0.3714362337910614,
+      "learning_rate": 1.963108953612143e-05,
+      "loss": 0.3688,
+      "step": 683
+    },
+    {
+      "epoch": 2.7777215189873417,
+      "grad_norm": 0.3977423232476304,
+      "learning_rate": 1.9574342079543056e-05,
+      "loss": 0.356,
+      "step": 684
+    },
+    {
+      "epoch": 2.781772151898734,
+      "grad_norm": 0.4025204092361647,
+      "learning_rate": 1.9517598051155046e-05,
+      "loss": 0.3637,
+      "step": 685
+    },
+    {
+      "epoch": 2.7858227848101267,
+      "grad_norm": 0.44143888915520335,
+      "learning_rate": 1.9460857907966025e-05,
+      "loss": 0.3746,
+      "step": 686
+    },
+    {
+      "epoch": 2.7898734177215188,
+      "grad_norm": 0.4191196145265337,
+      "learning_rate": 1.9404122106953285e-05,
+      "loss": 0.3697,
+      "step": 687
+    },
+    {
+      "epoch": 2.7939240506329113,
+      "grad_norm": 0.3647265205125135,
+      "learning_rate": 1.9347391105059176e-05,
+      "loss": 0.3559,
+      "step": 688
+    },
+    {
+      "epoch": 2.797974683544304,
+      "grad_norm": 0.5243395780650103,
+      "learning_rate": 1.92906653591874e-05,
+      "loss": 0.3775,
+      "step": 689
+    },
+    {
+      "epoch": 2.8020253164556963,
+      "grad_norm": 0.39064796745102004,
+      "learning_rate": 1.9233945326199295e-05,
+      "loss": 0.3662,
+      "step": 690
+    },
+    {
+      "epoch": 2.806075949367089,
+      "grad_norm": 0.4476868858780202,
+      "learning_rate": 1.917723146291022e-05,
+      "loss": 0.3757,
+      "step": 691
+    },
+    {
+      "epoch": 2.810126582278481,
+      "grad_norm": 0.45047805522010687,
+      "learning_rate": 1.912052422608584e-05,
+      "loss": 0.3567,
+      "step": 692
+    },
+    {
+      "epoch": 2.8141772151898734,
+      "grad_norm": 0.402323510891344,
+      "learning_rate": 1.9063824072438428e-05,
+      "loss": 0.3846,
+      "step": 693
+    },
+    {
+      "epoch": 2.818227848101266,
+      "grad_norm": 0.4096436505706435,
+      "learning_rate": 1.9007131458623246e-05,
+      "loss": 0.349,
+      "step": 694
+    },
+    {
+      "epoch": 2.822278481012658,
+      "grad_norm": 0.4372649374096489,
+      "learning_rate": 1.895044684123479e-05,
+      "loss": 0.378,
+      "step": 695
+    },
+    {
+      "epoch": 2.8263291139240505,
+      "grad_norm": 0.3890301487150277,
+      "learning_rate": 1.8893770676803194e-05,
+      "loss": 0.3559,
+      "step": 696
+    },
+    {
+      "epoch": 2.830379746835443,
+      "grad_norm": 0.38835321946941,
+      "learning_rate": 1.8837103421790486e-05,
+      "loss": 0.3481,
+      "step": 697
+    },
+    {
+      "epoch": 2.8344303797468355,
+      "grad_norm": 0.390000320232789,
+      "learning_rate": 1.8780445532586952e-05,
+      "loss": 0.3553,
+      "step": 698
+    },
+    {
+      "epoch": 2.838481012658228,
+      "grad_norm": 0.3630639968761546,
+      "learning_rate": 1.872379746550743e-05,
+      "loss": 0.3609,
+      "step": 699
+    },
+    {
+      "epoch": 2.84253164556962,
+      "grad_norm": 0.378084073016459,
+      "learning_rate": 1.866715967678769e-05,
+      "loss": 0.3456,
+      "step": 700
+    },
+    {
+      "epoch": 2.8465822784810126,
+      "grad_norm": 0.3787235930734076,
+      "learning_rate": 1.861053262258067e-05,
+      "loss": 0.366,
+      "step": 701
+    },
+    {
+      "epoch": 2.850632911392405,
+      "grad_norm": 0.3659583714420959,
+      "learning_rate": 1.8553916758952897e-05,
+      "loss": 0.3661,
+      "step": 702
+    },
+    {
+      "epoch": 2.8546835443037972,
+      "grad_norm": 0.3908175904905602,
+      "learning_rate": 1.8497312541880735e-05,
+      "loss": 0.3594,
+      "step": 703
+    },
+    {
+      "epoch": 2.8587341772151897,
+      "grad_norm": 0.39211978525036834,
+      "learning_rate": 1.8440720427246786e-05,
+      "loss": 0.3565,
+      "step": 704
+    },
+    {
+      "epoch": 2.8627848101265823,
+      "grad_norm": 0.3681433254667979,
+      "learning_rate": 1.8384140870836157e-05,
+      "loss": 0.382,
+      "step": 705
+    },
+    {
+      "epoch": 2.8668354430379748,
+      "grad_norm": 0.37892083695179746,
+      "learning_rate": 1.8327574328332806e-05,
+      "loss": 0.3685,
+      "step": 706
+    },
+    {
+      "epoch": 2.8708860759493673,
+      "grad_norm": 0.38293920343470145,
+      "learning_rate": 1.8271021255315906e-05,
+      "loss": 0.3444,
+      "step": 707
+    },
+    {
+      "epoch": 2.8749367088607594,
+      "grad_norm": 0.360437963783546,
+      "learning_rate": 1.8214482107256117e-05,
+      "loss": 0.3612,
+      "step": 708
+    },
+    {
+      "epoch": 2.878987341772152,
+      "grad_norm": 0.396378266156037,
+      "learning_rate": 1.8157957339511968e-05,
+      "loss": 0.3603,
+      "step": 709
+    },
+    {
+      "epoch": 2.8830379746835444,
+      "grad_norm": 0.37061211679324785,
+      "learning_rate": 1.8101447407326182e-05,
+      "loss": 0.3605,
+      "step": 710
+    },
+    {
+      "epoch": 2.8870886075949365,
+      "grad_norm": 0.37877055621923206,
+      "learning_rate": 1.8044952765821966e-05,
+      "loss": 0.3617,
+      "step": 711
+    },
+    {
+      "epoch": 2.891139240506329,
+      "grad_norm": 0.3955553744980108,
+      "learning_rate": 1.7988473869999407e-05,
+      "loss": 0.3694,
+      "step": 712
+    },
+    {
+      "epoch": 2.8951898734177215,
+      "grad_norm": 0.37585154840867857,
+      "learning_rate": 1.7932011174731764e-05,
+      "loss": 0.3517,
+      "step": 713
+    },
+    {
+      "epoch": 2.899240506329114,
+      "grad_norm": 0.4104028252317265,
+      "learning_rate": 1.7875565134761817e-05,
+      "loss": 0.352,
+      "step": 714
+    },
+    {
+      "epoch": 2.9032911392405065,
+      "grad_norm": 0.3666306369155531,
+      "learning_rate": 1.7819136204698226e-05,
+      "loss": 0.3711,
+      "step": 715
+    },
+    {
+      "epoch": 2.9073417721518986,
+      "grad_norm": 0.3879537132595155,
+      "learning_rate": 1.776272483901182e-05,
+      "loss": 0.3403,
+      "step": 716
+    },
+    {
+      "epoch": 2.911392405063291,
+      "grad_norm": 0.38170600735808985,
+      "learning_rate": 1.7706331492031995e-05,
+      "loss": 0.3657,
+      "step": 717
+    },
+    {
+      "epoch": 2.9154430379746836,
+      "grad_norm": 0.3786411139316161,
+      "learning_rate": 1.764995661794301e-05,
+      "loss": 0.3715,
+      "step": 718
+    },
+    {
+      "epoch": 2.9194936708860757,
+      "grad_norm": 0.4211524970645686,
+      "learning_rate": 1.759360067078035e-05,
+      "loss": 0.3556,
+      "step": 719
+    },
+    {
+      "epoch": 2.923544303797468,
+      "grad_norm": 0.4203795925606692,
+      "learning_rate": 1.7537264104427064e-05,
+      "loss": 0.3536,
+      "step": 720
+    },
+    {
+      "epoch": 2.9275949367088607,
+      "grad_norm": 0.40176669045936064,
+      "learning_rate": 1.748094737261012e-05,
+      "loss": 0.3579,
+      "step": 721
+    },
+    {
+      "epoch": 2.9316455696202532,
+      "grad_norm": 0.41496545175722493,
+      "learning_rate": 1.7424650928896726e-05,
+      "loss": 0.3681,
+      "step": 722
+    },
+    {
+      "epoch": 2.9356962025316458,
+      "grad_norm": 0.3607582194853792,
+      "learning_rate": 1.7368375226690712e-05,
+      "loss": 0.3778,
+      "step": 723
+    },
+    {
+      "epoch": 2.939746835443038,
+      "grad_norm": 0.4190746634094509,
+      "learning_rate": 1.731212071922883e-05,
+      "loss": 0.3525,
+      "step": 724
+    },
+    {
+      "epoch": 2.9437974683544303,
+      "grad_norm": 0.38839381616892416,
+      "learning_rate": 1.7255887859577156e-05,
+      "loss": 0.3531,
+      "step": 725
+    },
+    {
+      "epoch": 2.947848101265823,
+      "grad_norm": 0.36407062931466566,
+      "learning_rate": 1.7199677100627427e-05,
+      "loss": 0.3722,
+      "step": 726
+    },
+    {
+      "epoch": 2.951898734177215,
+      "grad_norm": 0.3804151499936817,
+      "learning_rate": 1.7143488895093343e-05,
+      "loss": 0.3611,
+      "step": 727
+    },
+    {
+      "epoch": 2.9559493670886074,
+      "grad_norm": 0.3875590720345679,
+      "learning_rate": 1.7087323695506994e-05,
+      "loss": 0.3554,
+      "step": 728
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.3862946460731437,
+      "learning_rate": 1.7031181954215194e-05,
+      "loss": 0.3544,
+      "step": 729
+    },
+    {
+      "epoch": 2.9640506329113925,
+      "grad_norm": 0.4097101839663008,
+      "learning_rate": 1.6975064123375788e-05,
+      "loss": 0.3648,
+      "step": 730
+    },
+    {
+      "epoch": 2.968101265822785,
+      "grad_norm": 0.37529235998203364,
+      "learning_rate": 1.6918970654954084e-05,
+      "loss": 0.3709,
+      "step": 731
+    },
+    {
+      "epoch": 2.972151898734177,
+      "grad_norm": 0.39568386223131,
+      "learning_rate": 1.686290200071915e-05,
+      "loss": 0.3347,
+      "step": 732
+    },
+    {
+      "epoch": 2.9762025316455696,
+      "grad_norm": 0.37545728946565077,
+      "learning_rate": 1.6806858612240234e-05,
+      "loss": 0.3947,
+      "step": 733
+    },
+    {
+      "epoch": 2.980253164556962,
+      "grad_norm": 0.3680992134024144,
+      "learning_rate": 1.6750840940883078e-05,
+      "loss": 0.359,
+      "step": 734
+    },
+    {
+      "epoch": 2.984303797468354,
+      "grad_norm": 0.3952515315569312,
+      "learning_rate": 1.6694849437806305e-05,
+      "loss": 0.3499,
+      "step": 735
+    },
+    {
+      "epoch": 2.9883544303797467,
+      "grad_norm": 1.0568810066385594,
+      "learning_rate": 1.663888455395778e-05,
+      "loss": 0.3588,
+      "step": 736
+    },
+    {
+      "epoch": 2.992405063291139,
+      "grad_norm": 0.3770502292069591,
+      "learning_rate": 1.6582946740070995e-05,
+      "loss": 0.3599,
+      "step": 737
+    },
+    {
+      "epoch": 2.9964556962025317,
+      "grad_norm": 0.3965899937182942,
+      "learning_rate": 1.6527036446661396e-05,
+      "loss": 0.3842,
+      "step": 738
+    },
+    {
+      "epoch": 3.0005063291139242,
+      "grad_norm": 0.3736633741682586,
+      "learning_rate": 1.6471154124022818e-05,
+      "loss": 0.3476,
+      "step": 739
+    },
+    {
+      "epoch": 3.0045569620253163,
+      "grad_norm": 0.594601155098279,
+      "learning_rate": 1.6415300222223788e-05,
+      "loss": 0.2679,
+      "step": 740
+    },
+    {
+      "epoch": 3.008607594936709,
+      "grad_norm": 0.34553848942197163,
+      "learning_rate": 1.6359475191103958e-05,
+      "loss": 0.2675,
+      "step": 741
+    },
+    {
+      "epoch": 3.0126582278481013,
+      "grad_norm": 0.5604158630845136,
+      "learning_rate": 1.6303679480270466e-05,
+      "loss": 0.2449,
+      "step": 742
+    },
+    {
+      "epoch": 3.016708860759494,
+      "grad_norm": 0.6696485750139916,
+      "learning_rate": 1.624791353909428e-05,
+      "loss": 0.275,
+      "step": 743
+    },
+    {
+      "epoch": 3.020759493670886,
+      "grad_norm": 0.4309685191341234,
+      "learning_rate": 1.619217781670663e-05,
+      "loss": 0.2396,
+      "step": 744
+    },
+    {
+      "epoch": 3.0248101265822784,
+      "grad_norm": 0.5240789398278343,
+      "learning_rate": 1.6136472761995373e-05,
+      "loss": 0.2573,
+      "step": 745
+    },
+    {
+      "epoch": 3.028860759493671,
+      "grad_norm": 0.5178327972694158,
+      "learning_rate": 1.608079882360134e-05,
+      "loss": 0.2709,
+      "step": 746
+    },
+    {
+      "epoch": 3.0329113924050635,
+      "grad_norm": 0.39026547276344614,
+      "learning_rate": 1.60251564499148e-05,
+      "loss": 0.2458,
+      "step": 747
+    },
+    {
+      "epoch": 3.0369620253164555,
+      "grad_norm": 0.7788041024723892,
+      "learning_rate": 1.596954608907176e-05,
+      "loss": 0.2794,
+      "step": 748
+    },
+    {
+      "epoch": 3.041012658227848,
+      "grad_norm": 0.5110646683650715,
+      "learning_rate": 1.591396818895043e-05,
+      "loss": 0.2466,
+      "step": 749
+    },
+    {
+      "epoch": 3.0450632911392406,
+      "grad_norm": 0.46998230764385973,
+      "learning_rate": 1.585842319716759e-05,
+      "loss": 0.2426,
+      "step": 750
+    },
+    {
+      "epoch": 3.049113924050633,
+      "grad_norm": 0.3959014114692123,
+      "learning_rate": 1.5802911561074944e-05,
+      "loss": 0.251,
+      "step": 751
+    },
+    {
+      "epoch": 3.053164556962025,
+      "grad_norm": 0.46254843607286866,
+      "learning_rate": 1.5747433727755595e-05,
+      "loss": 0.2426,
+      "step": 752
+    },
+    {
+      "epoch": 3.0572151898734177,
+      "grad_norm": 0.4360362045417157,
+      "learning_rate": 1.5691990144020376e-05,
+      "loss": 0.2409,
+      "step": 753
+    },
+    {
+      "epoch": 3.06126582278481,
+      "grad_norm": 0.36435836699200264,
+      "learning_rate": 1.5636581256404297e-05,
+      "loss": 0.2513,
+      "step": 754
+    },
+    {
+      "epoch": 3.0653164556962027,
+      "grad_norm": 0.445680259927462,
+      "learning_rate": 1.558120751116291e-05,
+      "loss": 0.2564,
+      "step": 755
+    },
+    {
+      "epoch": 3.0693670886075948,
+      "grad_norm": 0.4327554861609001,
+      "learning_rate": 1.552586935426876e-05,
+      "loss": 0.2567,
+      "step": 756
+    },
+    {
+      "epoch": 3.0734177215189873,
+      "grad_norm": 0.3586813947042321,
+      "learning_rate": 1.547056723140774e-05,
+      "loss": 0.2463,
+      "step": 757
+    },
+    {
+      "epoch": 3.07746835443038,
+      "grad_norm": 0.3857842109872023,
+      "learning_rate": 1.5415301587975565e-05,
+      "loss": 0.2373,
+      "step": 758
+    },
+    {
+      "epoch": 3.0815189873417723,
+      "grad_norm": 0.38225183095127296,
+      "learning_rate": 1.536007286907411e-05,
+      "loss": 0.2452,
+      "step": 759
+    },
+    {
+      "epoch": 3.0855696202531644,
+      "grad_norm": 0.3708899524323749,
+      "learning_rate": 1.5304881519507896e-05,
+      "loss": 0.2663,
+      "step": 760
+    },
+    {
+      "epoch": 3.089620253164557,
+      "grad_norm": 0.37733496863241633,
+      "learning_rate": 1.5249727983780453e-05,
+      "loss": 0.2479,
+      "step": 761
+    },
+    {
+      "epoch": 3.0936708860759494,
+      "grad_norm": 0.3597027042842214,
+      "learning_rate": 1.5194612706090786e-05,
+      "loss": 0.2647,
+      "step": 762
+    },
+    {
+      "epoch": 3.097721518987342,
+      "grad_norm": 0.35227136491319044,
+      "learning_rate": 1.5139536130329771e-05,
+      "loss": 0.2521,
+      "step": 763
+    },
+    {
+      "epoch": 3.101772151898734,
+      "grad_norm": 0.35885871258693447,
+      "learning_rate": 1.508449870007656e-05,
+      "loss": 0.2305,
+      "step": 764
+    },
+    {
+      "epoch": 3.1058227848101265,
+      "grad_norm": 0.35906296272638966,
+      "learning_rate": 1.5029500858595056e-05,
+      "loss": 0.2532,
+      "step": 765
+    },
+    {
+      "epoch": 3.109873417721519,
+      "grad_norm": 0.3415808836129023,
+      "learning_rate": 1.4974543048830328e-05,
+      "loss": 0.2491,
+      "step": 766
+    },
+    {
+      "epoch": 3.1139240506329116,
+      "grad_norm": 0.3658198454733206,
+      "learning_rate": 1.4919625713405e-05,
+      "loss": 0.2421,
+      "step": 767
+    },
+    {
+      "epoch": 3.1179746835443036,
+      "grad_norm": 0.37957534064456944,
+      "learning_rate": 1.4864749294615756e-05,
+      "loss": 0.2659,
+      "step": 768
+    },
+    {
+      "epoch": 3.122025316455696,
+      "grad_norm": 0.33458792632362844,
+      "learning_rate": 1.4809914234429716e-05,
+      "loss": 0.2597,
+      "step": 769
+    },
+    {
+      "epoch": 3.1260759493670887,
+      "grad_norm": 0.3525640711758802,
+      "learning_rate": 1.4755120974480923e-05,
+      "loss": 0.2524,
+      "step": 770
+    },
+    {
+      "epoch": 3.130126582278481,
+      "grad_norm": 0.3683911716372672,
+      "learning_rate": 1.4700369956066771e-05,
+      "loss": 0.2607,
+      "step": 771
+    },
+    {
+      "epoch": 3.1341772151898732,
+      "grad_norm": 0.3438032404044418,
+      "learning_rate": 1.4645661620144413e-05,
+      "loss": 0.2691,
+      "step": 772
+    },
+    {
+      "epoch": 3.1382278481012658,
+      "grad_norm": 0.3657343422220881,
+      "learning_rate": 1.4590996407327284e-05,
+      "loss": 0.2581,
+      "step": 773
+    },
+    {
+      "epoch": 3.1422784810126583,
+      "grad_norm": 0.3262036603856629,
+      "learning_rate": 1.4536374757881487e-05,
+      "loss": 0.2461,
+      "step": 774
+    },
+    {
+      "epoch": 3.146329113924051,
+      "grad_norm": 0.35226041398460184,
+      "learning_rate": 1.4481797111722271e-05,
+      "loss": 0.2519,
+      "step": 775
+    },
+    {
+      "epoch": 3.150379746835443,
+      "grad_norm": 0.3399399760543513,
+      "learning_rate": 1.4427263908410507e-05,
+      "loss": 0.2483,
+      "step": 776
+    },
+    {
+      "epoch": 3.1544303797468354,
+      "grad_norm": 0.35752194982013435,
+      "learning_rate": 1.4372775587149108e-05,
+      "loss": 0.2555,
+      "step": 777
+    },
+    {
+      "epoch": 3.158481012658228,
+      "grad_norm": 0.33946549410908844,
+      "learning_rate": 1.4318332586779522e-05,
+      "loss": 0.2605,
+      "step": 778
+    },
+    {
+      "epoch": 3.1625316455696204,
+      "grad_norm": 0.34699952705382603,
+      "learning_rate": 1.4263935345778202e-05,
+      "loss": 0.2375,
+      "step": 779
+    },
+    {
+      "epoch": 3.1665822784810125,
+      "grad_norm": 0.3527624510621365,
+      "learning_rate": 1.420958430225303e-05,
+      "loss": 0.2574,
+      "step": 780
+    },
+    {
+      "epoch": 3.170632911392405,
+      "grad_norm": 0.33178906489813753,
+      "learning_rate": 1.415527989393985e-05,
+      "loss": 0.2545,
+      "step": 781
+    },
+    {
+      "epoch": 3.1746835443037975,
+      "grad_norm": 0.3446385794939468,
+      "learning_rate": 1.410102255819891e-05,
+      "loss": 0.25,
+      "step": 782
+    },
+    {
+      "epoch": 3.17873417721519,
+      "grad_norm": 0.3360680758842884,
+      "learning_rate": 1.404681273201131e-05,
+      "loss": 0.2652,
+      "step": 783
+    },
+    {
+      "epoch": 3.182784810126582,
+      "grad_norm": 0.3462033775655195,
+      "learning_rate": 1.399265085197556e-05,
+      "loss": 0.2618,
+      "step": 784
+    },
+    {
+      "epoch": 3.1868354430379746,
+      "grad_norm": 0.33410707065348416,
+      "learning_rate": 1.393853735430398e-05,
+      "loss": 0.2529,
+      "step": 785
+    },
+    {
+      "epoch": 3.190886075949367,
+      "grad_norm": 0.3494356870663623,
+      "learning_rate": 1.3884472674819246e-05,
+      "loss": 0.233,
+      "step": 786
+    },
+    {
+      "epoch": 3.1949367088607596,
+      "grad_norm": 0.35729890630395916,
+      "learning_rate": 1.3830457248950864e-05,
+      "loss": 0.2545,
+      "step": 787
+    },
+    {
+      "epoch": 3.1989873417721517,
+      "grad_norm": 0.3283632821927642,
+      "learning_rate": 1.377649151173163e-05,
+      "loss": 0.2474,
+      "step": 788
+    },
+    {
+      "epoch": 3.2030379746835442,
+      "grad_norm": 0.3321590604248325,
+      "learning_rate": 1.3722575897794181e-05,
+      "loss": 0.2521,
+      "step": 789
+    },
+    {
+      "epoch": 3.2070886075949367,
+      "grad_norm": 0.3410133662417123,
+      "learning_rate": 1.3668710841367472e-05,
+      "loss": 0.243,
+      "step": 790
+    },
+    {
+      "epoch": 3.2111392405063293,
+      "grad_norm": 0.31868615653736754,
+      "learning_rate": 1.361489677627324e-05,
+      "loss": 0.2569,
+      "step": 791
+    },
+    {
+      "epoch": 3.2151898734177213,
+      "grad_norm": 0.3505932933487357,
+      "learning_rate": 1.3561134135922585e-05,
+      "loss": 0.2452,
+      "step": 792
+    },
+    {
+      "epoch": 3.219240506329114,
+      "grad_norm": 0.34214408385215267,
+      "learning_rate": 1.350742335331241e-05,
+      "loss": 0.2528,
+      "step": 793
+    },
+    {
+      "epoch": 3.2232911392405064,
+      "grad_norm": 0.33926275534991907,
+      "learning_rate": 1.345376486102198e-05,
+      "loss": 0.2562,
+      "step": 794
+    },
+    {
+      "epoch": 3.227341772151899,
+      "grad_norm": 0.3661094172759477,
+      "learning_rate": 1.3400159091209414e-05,
+      "loss": 0.2623,
+      "step": 795
+    },
+    {
+      "epoch": 3.231392405063291,
+      "grad_norm": 0.3500075389336499,
+      "learning_rate": 1.3346606475608216e-05,
+      "loss": 0.2496,
+      "step": 796
+    },
+    {
+      "epoch": 3.2354430379746835,
+      "grad_norm": 0.3379131063045504,
+      "learning_rate": 1.3293107445523781e-05,
+      "loss": 0.2266,
+      "step": 797
+    },
+    {
+      "epoch": 3.239493670886076,
+      "grad_norm": 0.3492023029720776,
+      "learning_rate": 1.3239662431829949e-05,
+      "loss": 0.2456,
+      "step": 798
+    },
+    {
+      "epoch": 3.2435443037974685,
+      "grad_norm": 0.3468536478180755,
+      "learning_rate": 1.3186271864965509e-05,
+      "loss": 0.2742,
+      "step": 799
+    },
+    {
+      "epoch": 3.2475949367088606,
+      "grad_norm": 0.33138899255388177,
+      "learning_rate": 1.3132936174930756e-05,
+      "loss": 0.254,
+      "step": 800
+    },
+    {
+      "epoch": 3.251645569620253,
+      "grad_norm": 0.3519473495113768,
+      "learning_rate": 1.3079655791283995e-05,
+      "loss": 0.2401,
+      "step": 801
+    },
+    {
+      "epoch": 3.2556962025316456,
+      "grad_norm": 0.33901884792499853,
+      "learning_rate": 1.3026431143138108e-05,
+      "loss": 0.2294,
+      "step": 802
+    },
+    {
+      "epoch": 3.259746835443038,
+      "grad_norm": 0.30752747973059347,
+      "learning_rate": 1.2973262659157114e-05,
+      "loss": 0.2513,
+      "step": 803
+    },
+    {
+      "epoch": 3.26379746835443,
+      "grad_norm": 0.35977928573712314,
+      "learning_rate": 1.2920150767552651e-05,
+      "loss": 0.2572,
+      "step": 804
+    },
+    {
+      "epoch": 3.2678481012658227,
+      "grad_norm": 0.3246299823956867,
+      "learning_rate": 1.2867095896080607e-05,
+      "loss": 0.242,
+      "step": 805
+    },
+    {
+      "epoch": 3.271898734177215,
+      "grad_norm": 0.336404085627576,
+      "learning_rate": 1.2814098472037612e-05,
+      "loss": 0.2421,
+      "step": 806
+    },
+    {
+      "epoch": 3.2759493670886077,
+      "grad_norm": 0.32860257533958664,
+      "learning_rate": 1.276115892225764e-05,
+      "loss": 0.2392,
+      "step": 807
+    },
+    {
+      "epoch": 3.2800000000000002,
+      "grad_norm": 0.31751784716157533,
+      "learning_rate": 1.2708277673108555e-05,
+      "loss": 0.2299,
+      "step": 808
+    },
+    {
+      "epoch": 3.2840506329113923,
+      "grad_norm": 0.33412344864481963,
+      "learning_rate": 1.2655455150488649e-05,
+      "loss": 0.2482,
+      "step": 809
+    },
+    {
+      "epoch": 3.288101265822785,
+      "grad_norm": 0.3144511736899087,
+      "learning_rate": 1.2602691779823272e-05,
+      "loss": 0.266,
+      "step": 810
+    },
+    {
+      "epoch": 3.2921518987341774,
+      "grad_norm": 0.33203611872354655,
+      "learning_rate": 1.2549987986061355e-05,
+      "loss": 0.2379,
+      "step": 811
+    },
+    {
+      "epoch": 3.2962025316455694,
+      "grad_norm": 0.3407916940670116,
+      "learning_rate": 1.2497344193672005e-05,
+      "loss": 0.2344,
+      "step": 812
+    },
+    {
+      "epoch": 3.300253164556962,
+      "grad_norm": 0.3313704916585959,
+      "learning_rate": 1.2444760826641092e-05,
+      "loss": 0.275,
+      "step": 813
+    },
+    {
+      "epoch": 3.3043037974683545,
+      "grad_norm": 0.3247983709800989,
+      "learning_rate": 1.2392238308467817e-05,
+      "loss": 0.2691,
+      "step": 814
+    },
+    {
+      "epoch": 3.308354430379747,
+      "grad_norm": 0.32647475095491263,
+      "learning_rate": 1.2339777062161326e-05,
+      "loss": 0.2367,
+      "step": 815
+    },
+    {
+      "epoch": 3.3124050632911395,
+      "grad_norm": 0.3168057105821444,
+      "learning_rate": 1.2287377510237293e-05,
+      "loss": 0.2665,
+      "step": 816
+    },
+    {
+      "epoch": 3.3164556962025316,
+      "grad_norm": 0.32490515712812496,
+      "learning_rate": 1.2235040074714488e-05,
+      "loss": 0.2379,
+      "step": 817
+    },
+    {
+      "epoch": 3.320506329113924,
+      "grad_norm": 0.33369779279361417,
+      "learning_rate": 1.2182765177111434e-05,
+      "loss": 0.2423,
+      "step": 818
+    },
+    {
+      "epoch": 3.3245569620253166,
+      "grad_norm": 0.3433452806379234,
+      "learning_rate": 1.213055323844297e-05,
+      "loss": 0.2435,
+      "step": 819
+    },
+    {
+      "epoch": 3.3286075949367087,
+      "grad_norm": 0.34808474216714785,
+      "learning_rate": 1.2078404679216864e-05,
+      "loss": 0.2606,
+      "step": 820
+    },
+    {
+      "epoch": 3.332658227848101,
+      "grad_norm": 0.3414614638076738,
+      "learning_rate": 1.2026319919430458e-05,
+      "loss": 0.2556,
+      "step": 821
+    },
+    {
+      "epoch": 3.3367088607594937,
+      "grad_norm": 0.32985605918041583,
+      "learning_rate": 1.1974299378567227e-05,
+      "loss": 0.2491,
+      "step": 822
+    },
+    {
+      "epoch": 3.340759493670886,
+      "grad_norm": 0.32448394070685327,
+      "learning_rate": 1.1922343475593462e-05,
+      "loss": 0.2465,
+      "step": 823
+    },
+    {
+      "epoch": 3.3448101265822787,
+      "grad_norm": 0.3438363073053231,
+      "learning_rate": 1.187045262895488e-05,
+      "loss": 0.2327,
+      "step": 824
+    },
+    {
+      "epoch": 3.348860759493671,
+      "grad_norm": 0.3379459491655878,
+      "learning_rate": 1.1818627256573203e-05,
+      "loss": 0.2444,
+      "step": 825
+    },
+    {
+      "epoch": 3.3529113924050633,
+      "grad_norm": 0.33680430322989385,
+      "learning_rate": 1.1766867775842864e-05,
+      "loss": 0.2562,
+      "step": 826
+    },
+    {
+      "epoch": 3.356962025316456,
+      "grad_norm": 0.32770022840615487,
+      "learning_rate": 1.1715174603627615e-05,
+      "loss": 0.2469,
+      "step": 827
+    },
+    {
+      "epoch": 3.361012658227848,
+      "grad_norm": 0.3179200146828703,
+      "learning_rate": 1.1663548156257147e-05,
+      "loss": 0.2388,
+      "step": 828
+    },
+    {
+      "epoch": 3.3650632911392404,
+      "grad_norm": 0.3294408027295764,
+      "learning_rate": 1.161198884952377e-05,
+      "loss": 0.242,
+      "step": 829
+    },
+    {
+      "epoch": 3.369113924050633,
+      "grad_norm": 0.33753952340554444,
+      "learning_rate": 1.1560497098679056e-05,
+      "loss": 0.244,
+      "step": 830
+    },
+    {
+      "epoch": 3.3731645569620254,
+      "grad_norm": 0.3319540284083433,
+      "learning_rate": 1.1509073318430479e-05,
+      "loss": 0.2424,
+      "step": 831
+    },
+    {
+      "epoch": 3.377215189873418,
+      "grad_norm": 0.34277748775352074,
+      "learning_rate": 1.1457717922938116e-05,
+      "loss": 0.2481,
+      "step": 832
+    },
+    {
+      "epoch": 3.38126582278481,
+      "grad_norm": 0.32612093932613845,
+      "learning_rate": 1.1406431325811233e-05,
+      "loss": 0.2511,
+      "step": 833
+    },
+    {
+      "epoch": 3.3853164556962025,
+      "grad_norm": 0.3407639159950644,
+      "learning_rate": 1.135521394010506e-05,
+      "loss": 0.2532,
+      "step": 834
+    },
+    {
+      "epoch": 3.389367088607595,
+      "grad_norm": 0.3199119618300343,
+      "learning_rate": 1.1304066178317367e-05,
+      "loss": 0.2408,
+      "step": 835
+    },
+    {
+      "epoch": 3.393417721518987,
+      "grad_norm": 0.34232571631510955,
+      "learning_rate": 1.1252988452385199e-05,
+      "loss": 0.2377,
+      "step": 836
+    },
+    {
+      "epoch": 3.3974683544303796,
+      "grad_norm": 0.3254903995798074,
+      "learning_rate": 1.1201981173681536e-05,
+      "loss": 0.2626,
+      "step": 837
+    },
+    {
+      "epoch": 3.401518987341772,
+      "grad_norm": 0.325750734281776,
+      "learning_rate": 1.1151044753011991e-05,
+      "loss": 0.2536,
+      "step": 838
+    },
+    {
+      "epoch": 3.4055696202531647,
+      "grad_norm": 0.3338115661397887,
+      "learning_rate": 1.1100179600611491e-05,
+      "loss": 0.2375,
+      "step": 839
+    },
+    {
+      "epoch": 3.409620253164557,
+      "grad_norm": 0.3469070594946913,
+      "learning_rate": 1.1049386126140985e-05,
+      "loss": 0.2556,
+      "step": 840
+    },
+    {
+      "epoch": 3.4136708860759493,
+      "grad_norm": 0.32565315932531425,
+      "learning_rate": 1.0998664738684128e-05,
+      "loss": 0.2527,
+      "step": 841
+    },
+    {
+      "epoch": 3.4177215189873418,
+      "grad_norm": 0.31251893537995945,
+      "learning_rate": 1.0948015846744e-05,
+      "loss": 0.272,
+      "step": 842
+    },
+    {
+      "epoch": 3.4217721518987343,
+      "grad_norm": 0.336430932230224,
+      "learning_rate": 1.0897439858239832e-05,
+      "loss": 0.2412,
+      "step": 843
+    },
+    {
+      "epoch": 3.4258227848101264,
+      "grad_norm": 0.3250850808007104,
+      "learning_rate": 1.0846937180503652e-05,
+      "loss": 0.2431,
+      "step": 844
+    },
+    {
+      "epoch": 3.429873417721519,
+      "grad_norm": 0.33355070385040186,
+      "learning_rate": 1.0796508220277117e-05,
+      "loss": 0.2534,
+      "step": 845
+    },
+    {
+      "epoch": 3.4339240506329114,
+      "grad_norm": 0.32809870129345403,
+      "learning_rate": 1.0746153383708107e-05,
+      "loss": 0.2428,
+      "step": 846
+    },
+    {
+      "epoch": 3.437974683544304,
+      "grad_norm": 0.32944473952247233,
+      "learning_rate": 1.0695873076347579e-05,
+      "loss": 0.2658,
+      "step": 847
+    },
+    {
+      "epoch": 3.4420253164556964,
+      "grad_norm": 0.32694778251635953,
+      "learning_rate": 1.0645667703146205e-05,
+      "loss": 0.2504,
+      "step": 848
+    },
+    {
+      "epoch": 3.4460759493670885,
+      "grad_norm": 0.31626512776856436,
+      "learning_rate": 1.0595537668451161e-05,
+      "loss": 0.2738,
+      "step": 849
+    },
+    {
+      "epoch": 3.450126582278481,
+      "grad_norm": 0.3220964346450603,
+      "learning_rate": 1.0545483376002854e-05,
+      "loss": 0.2592,
+      "step": 850
+    },
+    {
+      "epoch": 3.4541772151898735,
+      "grad_norm": 0.3233608402814376,
+      "learning_rate": 1.0495505228931676e-05,
+      "loss": 0.2416,
+      "step": 851
+    },
+    {
+      "epoch": 3.4582278481012656,
+      "grad_norm": 0.32009190287457223,
+      "learning_rate": 1.044560362975474e-05,
+      "loss": 0.2526,
+      "step": 852
+    },
+    {
+      "epoch": 3.462278481012658,
+      "grad_norm": 0.3205866695867176,
+      "learning_rate": 1.0395778980372695e-05,
+      "loss": 0.2439,
+      "step": 853
+    },
+    {
+      "epoch": 3.4663291139240506,
+      "grad_norm": 0.3355730104002603,
+      "learning_rate": 1.0346031682066381e-05,
+      "loss": 0.2616,
+      "step": 854
+    },
+    {
+      "epoch": 3.470379746835443,
+      "grad_norm": 0.3261159893889133,
+      "learning_rate": 1.0296362135493724e-05,
+      "loss": 0.2594,
+      "step": 855
+    },
+    {
+      "epoch": 3.4744303797468357,
+      "grad_norm": 0.3367977427761997,
+      "learning_rate": 1.0246770740686422e-05,
+      "loss": 0.2454,
+      "step": 856
+    },
+    {
+      "epoch": 3.4784810126582277,
+      "grad_norm": 0.3407583645627075,
+      "learning_rate": 1.0197257897046743e-05,
+      "loss": 0.2389,
+      "step": 857
+    },
+    {
+      "epoch": 3.4825316455696202,
+      "grad_norm": 0.3159378093140294,
+      "learning_rate": 1.014782400334433e-05,
+      "loss": 0.2678,
+      "step": 858
+    },
+    {
+      "epoch": 3.4865822784810128,
+      "grad_norm": 0.3151786177072686,
+      "learning_rate": 1.009846945771296e-05,
+      "loss": 0.2457,
+      "step": 859
+    },
+    {
+      "epoch": 3.490632911392405,
+      "grad_norm": 0.32362852328787944,
+      "learning_rate": 1.0049194657647363e-05,
+      "loss": 0.2361,
+      "step": 860
+    },
+    {
+      "epoch": 3.4946835443037974,
+      "grad_norm": 0.33360120953189004,
+      "learning_rate": 1.0000000000000006e-05,
+      "loss": 0.242,
+      "step": 861
+    },
+    {
+      "epoch": 3.49873417721519,
+      "grad_norm": 0.2996474482581394,
+      "learning_rate": 9.950885880977891e-06,
+      "loss": 0.2564,
+      "step": 862
+    },
+    {
+      "epoch": 3.5027848101265824,
+      "grad_norm": 0.3262417210238,
+      "learning_rate": 9.901852696139382e-06,
+      "loss": 0.2302,
+      "step": 863
+    },
+    {
+      "epoch": 3.506835443037975,
+      "grad_norm": 0.3129822987394034,
+      "learning_rate": 9.852900840391027e-06,
+      "loss": 0.2619,
+      "step": 864
+    },
+    {
+      "epoch": 3.510886075949367,
+      "grad_norm": 0.3161323419362324,
+      "learning_rate": 9.804030707984313e-06,
+      "loss": 0.2518,
+      "step": 865
+    },
+    {
+      "epoch": 3.5149367088607595,
+      "grad_norm": 0.31798267882938225,
+      "learning_rate": 9.755242692512599e-06,
+      "loss": 0.2431,
+      "step": 866
+    },
+    {
+      "epoch": 3.518987341772152,
+      "grad_norm": 0.3119363362212112,
+      "learning_rate": 9.70653718690782e-06,
+      "loss": 0.2441,
+      "step": 867
+    },
+    {
+      "epoch": 3.523037974683544,
+      "grad_norm": 0.33278133127210513,
+      "learning_rate": 9.657914583437454e-06,
+      "loss": 0.247,
+      "step": 868
+    },
+    {
+      "epoch": 3.5270886075949366,
+      "grad_norm": 0.31537525449322373,
+      "learning_rate": 9.609375273701246e-06,
+      "loss": 0.2399,
+      "step": 869
+    },
+    {
+      "epoch": 3.531139240506329,
+      "grad_norm": 0.33035939275393,
+      "learning_rate": 9.560919648628133e-06,
+      "loss": 0.2263,
+      "step": 870
+    },
+    {
+      "epoch": 3.5351898734177216,
+      "grad_norm": 0.3362253570429956,
+      "learning_rate": 9.512548098473047e-06,
+      "loss": 0.2419,
+      "step": 871
+    },
+    {
+      "epoch": 3.539240506329114,
+      "grad_norm": 0.32574130494256637,
+      "learning_rate": 9.464261012813825e-06,
+      "loss": 0.2533,
+      "step": 872
+    },
+    {
+      "epoch": 3.543291139240506,
+      "grad_norm": 0.3106825019847393,
+      "learning_rate": 9.416058780547987e-06,
+      "loss": 0.2504,
+      "step": 873
+    },
+    {
+      "epoch": 3.5473417721518987,
+      "grad_norm": 0.32388623599921795,
+      "learning_rate": 9.367941789889714e-06,
+      "loss": 0.2388,
+      "step": 874
+    },
+    {
+      "epoch": 3.5513924050632912,
+      "grad_norm": 0.3144837556892047,
+      "learning_rate": 9.319910428366607e-06,
+      "loss": 0.2474,
+      "step": 875
+    },
+    {
+      "epoch": 3.5554430379746833,
+      "grad_norm": 0.3227513984445465,
+      "learning_rate": 9.271965082816667e-06,
+      "loss": 0.2446,
+      "step": 876
+    },
+    {
+      "epoch": 3.559493670886076,
+      "grad_norm": 0.32844324928442115,
+      "learning_rate": 9.224106139385111e-06,
+      "loss": 0.2494,
+      "step": 877
+    },
+    {
+      "epoch": 3.5635443037974683,
+      "grad_norm": 0.3329556954279331,
+      "learning_rate": 9.176333983521291e-06,
+      "loss": 0.2494,
+      "step": 878
+    },
+    {
+      "epoch": 3.567594936708861,
+      "grad_norm": 0.3276548329105692,
+      "learning_rate": 9.12864899997558e-06,
+      "loss": 0.2518,
+      "step": 879
+    },
+    {
+      "epoch": 3.5716455696202534,
+      "grad_norm": 0.31686533095754804,
+      "learning_rate": 9.08105157279628e-06,
+      "loss": 0.2503,
+      "step": 880
+    },
+    {
+      "epoch": 3.5756962025316454,
+      "grad_norm": 0.3272269065337202,
+      "learning_rate": 9.03354208532653e-06,
+      "loss": 0.2593,
+      "step": 881
+    },
+    {
+      "epoch": 3.579746835443038,
+      "grad_norm": 0.32923341537127604,
+      "learning_rate": 8.986120920201205e-06,
+      "loss": 0.2427,
+      "step": 882
+    },
+    {
+      "epoch": 3.5837974683544305,
+      "grad_norm": 0.32415189395384264,
+      "learning_rate": 8.938788459343852e-06,
+      "loss": 0.2484,
+      "step": 883
+    },
+    {
+      "epoch": 3.5878481012658225,
+      "grad_norm": 0.2944828337381627,
+      "learning_rate": 8.8915450839636e-06,
+      "loss": 0.2205,
+      "step": 884
+    },
+    {
+      "epoch": 3.591898734177215,
+      "grad_norm": 0.3376785701306564,
+      "learning_rate": 8.844391174552116e-06,
+      "loss": 0.2569,
+      "step": 885
+    },
+    {
+      "epoch": 3.5959493670886076,
+      "grad_norm": 0.33121067327959897,
+      "learning_rate": 8.797327110880479e-06,
+      "loss": 0.2412,
+      "step": 886
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.31590806095273927,
+      "learning_rate": 8.750353271996206e-06,
+      "loss": 0.2386,
+      "step": 887
+    },
+    {
+      "epoch": 3.6040506329113926,
+      "grad_norm": 0.30434819861079004,
+      "learning_rate": 8.703470036220132e-06,
+      "loss": 0.2707,
+      "step": 888
+    },
+    {
+      "epoch": 3.6081012658227847,
+      "grad_norm": 0.3074787538919148,
+      "learning_rate": 8.656677781143394e-06,
+      "loss": 0.2576,
+      "step": 889
+    },
+    {
+      "epoch": 3.612151898734177,
+      "grad_norm": 0.32283739864119976,
+      "learning_rate": 8.609976883624377e-06,
+      "loss": 0.2411,
+      "step": 890
+    },
+    {
+      "epoch": 3.6162025316455697,
+      "grad_norm": 0.3176628727695747,
+      "learning_rate": 8.563367719785698e-06,
+      "loss": 0.2628,
+      "step": 891
+    },
+    {
+      "epoch": 3.620253164556962,
+      "grad_norm": 0.30909932131848744,
+      "learning_rate": 8.516850665011138e-06,
+      "loss": 0.2344,
+      "step": 892
+    },
+    {
+      "epoch": 3.6243037974683543,
+      "grad_norm": 0.31399511881442044,
+      "learning_rate": 8.47042609394269e-06,
+      "loss": 0.2351,
+      "step": 893
+    },
+    {
+      "epoch": 3.628354430379747,
+      "grad_norm": 0.3329674257984442,
+      "learning_rate": 8.424094380477432e-06,
+      "loss": 0.2521,
+      "step": 894
+    },
+    {
+      "epoch": 3.6324050632911393,
+      "grad_norm": 0.33495655670692265,
+      "learning_rate": 8.37785589776465e-06,
+      "loss": 0.2537,
+      "step": 895
+    },
+    {
+      "epoch": 3.636455696202532,
+      "grad_norm": 0.32286194002261537,
+      "learning_rate": 8.331711018202694e-06,
+      "loss": 0.2568,
+      "step": 896
+    },
+    {
+      "epoch": 3.640506329113924,
+      "grad_norm": 0.3301342267362824,
+      "learning_rate": 8.285660113436104e-06,
+      "loss": 0.2368,
+      "step": 897
+    },
+    {
+      "epoch": 3.6445569620253164,
+      "grad_norm": 0.3210118704164727,
+      "learning_rate": 8.239703554352527e-06,
+      "loss": 0.2282,
+      "step": 898
+    },
+    {
+      "epoch": 3.648607594936709,
+      "grad_norm": 0.32135642996057173,
+      "learning_rate": 8.193841711079775e-06,
+      "loss": 0.2628,
+      "step": 899
+    },
+    {
+      "epoch": 3.652658227848101,
+      "grad_norm": 0.32961238223128847,
+      "learning_rate": 8.148074952982828e-06,
+      "loss": 0.2519,
+      "step": 900
+    },
+    {
+      "epoch": 3.6567088607594935,
+      "grad_norm": 0.3077724985266919,
+      "learning_rate": 8.102403648660859e-06,
+      "loss": 0.2238,
+      "step": 901
+    },
+    {
+      "epoch": 3.660759493670886,
+      "grad_norm": 0.3124234369324712,
+      "learning_rate": 8.056828165944282e-06,
+      "loss": 0.2631,
+      "step": 902
+    },
+    {
+      "epoch": 3.6648101265822786,
+      "grad_norm": 0.32516861556618154,
+      "learning_rate": 8.011348871891762e-06,
+      "loss": 0.2616,
+      "step": 903
+    },
+    {
+      "epoch": 3.668860759493671,
+      "grad_norm": 0.2958290015096066,
+      "learning_rate": 7.965966132787287e-06,
+      "loss": 0.2541,
+      "step": 904
+    },
+    {
+      "epoch": 3.672911392405063,
+      "grad_norm": 0.32857376072307354,
+      "learning_rate": 7.920680314137189e-06,
+      "loss": 0.2706,
+      "step": 905
+    },
+    {
+      "epoch": 3.6769620253164557,
+      "grad_norm": 0.3136535856706206,
+      "learning_rate": 7.875491780667246e-06,
+      "loss": 0.2718,
+      "step": 906
+    },
+    {
+      "epoch": 3.681012658227848,
+      "grad_norm": 0.3318030182625039,
+      "learning_rate": 7.830400896319667e-06,
+      "loss": 0.2242,
+      "step": 907
+    },
+    {
+      "epoch": 3.6850632911392402,
+      "grad_norm": 0.3145244472669007,
+      "learning_rate": 7.785408024250259e-06,
+      "loss": 0.2569,
+      "step": 908
+    },
+    {
+      "epoch": 3.6891139240506328,
+      "grad_norm": 0.3274704177771191,
+      "learning_rate": 7.74051352682542e-06,
+      "loss": 0.2274,
+      "step": 909
+    },
+    {
+      "epoch": 3.6931645569620253,
+      "grad_norm": 0.3260142166180878,
+      "learning_rate": 7.695717765619257e-06,
+      "loss": 0.2619,
+      "step": 910
+    },
+    {
+      "epoch": 3.697215189873418,
+      "grad_norm": 0.2969118478985658,
+      "learning_rate": 7.651021101410673e-06,
+      "loss": 0.2707,
+      "step": 911
+    },
+    {
+      "epoch": 3.7012658227848103,
+      "grad_norm": 0.3069590765447823,
+      "learning_rate": 7.606423894180464e-06,
+      "loss": 0.2705,
+      "step": 912
+    },
+    {
+      "epoch": 3.7053164556962024,
+      "grad_norm": 0.31143389485075146,
+      "learning_rate": 7.56192650310839e-06,
+      "loss": 0.2467,
+      "step": 913
+    },
+    {
+      "epoch": 3.709367088607595,
+      "grad_norm": 0.32703961769932827,
+      "learning_rate": 7.517529286570349e-06,
+      "loss": 0.2581,
+      "step": 914
+    },
+    {
+      "epoch": 3.7134177215189874,
+      "grad_norm": 0.3302235300357165,
+      "learning_rate": 7.473232602135387e-06,
+      "loss": 0.2734,
+      "step": 915
+    },
+    {
+      "epoch": 3.7174683544303795,
+      "grad_norm": 0.30058629613766713,
+      "learning_rate": 7.429036806562935e-06,
+      "loss": 0.2534,
+      "step": 916
+    },
+    {
+      "epoch": 3.721518987341772,
+      "grad_norm": 0.32258439341238676,
+      "learning_rate": 7.3849422557998455e-06,
+      "loss": 0.2919,
+      "step": 917
+    },
+    {
+      "epoch": 3.7255696202531645,
+      "grad_norm": 0.3347153864952782,
+      "learning_rate": 7.340949304977567e-06,
+      "loss": 0.2411,
+      "step": 918
+    },
+    {
+      "epoch": 3.729620253164557,
+      "grad_norm": 0.3241888478211681,
+      "learning_rate": 7.297058308409282e-06,
+      "loss": 0.2314,
+      "step": 919
+    },
+    {
+      "epoch": 3.7336708860759495,
+      "grad_norm": 0.29817555163994286,
+      "learning_rate": 7.25326961958704e-06,
+      "loss": 0.2694,
+      "step": 920
+    },
+    {
+      "epoch": 3.7377215189873416,
+      "grad_norm": 0.30884052521297295,
+      "learning_rate": 7.209583591178921e-06,
+      "loss": 0.2432,
+      "step": 921
+    },
+    {
+      "epoch": 3.741772151898734,
+      "grad_norm": 0.3323619083523225,
+      "learning_rate": 7.1660005750261925e-06,
+      "loss": 0.259,
+      "step": 922
+    },
+    {
+      "epoch": 3.7458227848101266,
+      "grad_norm": 0.32606565071803256,
+      "learning_rate": 7.1225209221404765e-06,
+      "loss": 0.2388,
+      "step": 923
+    },
+    {
+      "epoch": 3.7498734177215187,
+      "grad_norm": 0.29179000766146496,
+      "learning_rate": 7.079144982700909e-06,
+      "loss": 0.2436,
+      "step": 924
+    },
+    {
+      "epoch": 3.7539240506329112,
+      "grad_norm": 0.3183528591441658,
+      "learning_rate": 7.0358731060513695e-06,
+      "loss": 0.2531,
+      "step": 925
+    },
+    {
+      "epoch": 3.7579746835443038,
+      "grad_norm": 0.3217579115196145,
+      "learning_rate": 6.99270564069757e-06,
+      "loss": 0.2565,
+      "step": 926
+    },
+    {
+      "epoch": 3.7620253164556963,
+      "grad_norm": 0.32060764097850236,
+      "learning_rate": 6.949642934304375e-06,
+      "loss": 0.2505,
+      "step": 927
+    },
+    {
+      "epoch": 3.766075949367089,
+      "grad_norm": 0.3176778857787001,
+      "learning_rate": 6.906685333692871e-06,
+      "loss": 0.2475,
+      "step": 928
+    },
+    {
+      "epoch": 3.770126582278481,
+      "grad_norm": 0.3149572026418284,
+      "learning_rate": 6.86383318483769e-06,
+      "loss": 0.2501,
+      "step": 929
+    },
+    {
+      "epoch": 3.7741772151898734,
+      "grad_norm": 0.33762877281928194,
+      "learning_rate": 6.821086832864139e-06,
+      "loss": 0.2468,
+      "step": 930
+    },
+    {
+      "epoch": 3.778227848101266,
+      "grad_norm": 0.3299009328738016,
+      "learning_rate": 6.77844662204546e-06,
+      "loss": 0.2648,
+      "step": 931
+    },
+    {
+      "epoch": 3.782278481012658,
+      "grad_norm": 0.3118251730471459,
+      "learning_rate": 6.7359128958000455e-06,
+      "loss": 0.2609,
+      "step": 932
+    },
+    {
+      "epoch": 3.7863291139240505,
+      "grad_norm": 0.3121524516530458,
+      "learning_rate": 6.693485996688695e-06,
+      "loss": 0.2296,
+      "step": 933
+    },
+    {
+      "epoch": 3.790379746835443,
+      "grad_norm": 0.3099671707014266,
+      "learning_rate": 6.651166266411801e-06,
+      "loss": 0.2538,
+      "step": 934
+    },
+    {
+      "epoch": 3.7944303797468355,
+      "grad_norm": 0.3320043681325413,
+      "learning_rate": 6.6089540458066725e-06,
+      "loss": 0.2418,
+      "step": 935
+    },
+    {
+      "epoch": 3.798481012658228,
+      "grad_norm": 0.5693800823107728,
+      "learning_rate": 6.566849674844711e-06,
+      "loss": 0.2475,
+      "step": 936
+    },
+    {
+      "epoch": 3.80253164556962,
+      "grad_norm": 0.308259170219992,
+      "learning_rate": 6.524853492628747e-06,
+      "loss": 0.2651,
+      "step": 937
+    },
+    {
+      "epoch": 3.8065822784810126,
+      "grad_norm": 0.3212609458548449,
+      "learning_rate": 6.4829658373902536e-06,
+      "loss": 0.2584,
+      "step": 938
+    },
+    {
+      "epoch": 3.810632911392405,
+      "grad_norm": 0.3196495421835934,
+      "learning_rate": 6.441187046486648e-06,
+      "loss": 0.241,
+      "step": 939
+    },
+    {
+      "epoch": 3.814683544303797,
+      "grad_norm": 0.3284541819591291,
+      "learning_rate": 6.399517456398567e-06,
+      "loss": 0.2399,
+      "step": 940
+    },
+    {
+      "epoch": 3.81873417721519,
+      "grad_norm": 0.31138954196715074,
+      "learning_rate": 6.357957402727164e-06,
+      "loss": 0.2432,
+      "step": 941
+    },
+    {
+      "epoch": 3.8227848101265822,
+      "grad_norm": 0.3180372884240123,
+      "learning_rate": 6.316507220191395e-06,
+      "loss": 0.2511,
+      "step": 942
+    },
+    {
+      "epoch": 3.8268354430379747,
+      "grad_norm": 0.3179101630832076,
+      "learning_rate": 6.275167242625331e-06,
+      "loss": 0.2567,
+      "step": 943
+    },
+    {
+      "epoch": 3.8308860759493673,
+      "grad_norm": 0.31681529367269207,
+      "learning_rate": 6.233937802975471e-06,
+      "loss": 0.2411,
+      "step": 944
+    },
+    {
+      "epoch": 3.8349367088607593,
+      "grad_norm": 0.3308996898179361,
+      "learning_rate": 6.192819233298046e-06,
+      "loss": 0.2433,
+      "step": 945
+    },
+    {
+      "epoch": 3.838987341772152,
+      "grad_norm": 0.30526514684013983,
+      "learning_rate": 6.151811864756383e-06,
+      "loss": 0.261,
+      "step": 946
+    },
+    {
+      "epoch": 3.8430379746835444,
+      "grad_norm": 0.32092624734078834,
+      "learning_rate": 6.1109160276181655e-06,
+      "loss": 0.2305,
+      "step": 947
+    },
+    {
+      "epoch": 3.847088607594937,
+      "grad_norm": 0.3117538108512426,
+      "learning_rate": 6.070132051252868e-06,
+      "loss": 0.2435,
+      "step": 948
+    },
+    {
+      "epoch": 3.8511392405063294,
+      "grad_norm": 0.3158161460794247,
+      "learning_rate": 6.0294602641290034e-06,
+      "loss": 0.2463,
+      "step": 949
+    },
+    {
+      "epoch": 3.8551898734177215,
+      "grad_norm": 0.30924145300830586,
+      "learning_rate": 5.988900993811575e-06,
+      "loss": 0.2488,
+      "step": 950
+    },
+    {
+      "epoch": 3.859240506329114,
+      "grad_norm": 0.30188046761588316,
+      "learning_rate": 5.948454566959363e-06,
+      "loss": 0.2365,
+      "step": 951
+    },
+    {
+      "epoch": 3.8632911392405065,
+      "grad_norm": 0.3142071059413291,
+      "learning_rate": 5.908121309322328e-06,
+      "loss": 0.2371,
+      "step": 952
+    },
+    {
+      "epoch": 3.8673417721518986,
+      "grad_norm": 0.3118930931968126,
+      "learning_rate": 5.867901545738976e-06,
+      "loss": 0.2603,
+      "step": 953
+    },
+    {
+      "epoch": 3.871392405063291,
+      "grad_norm": 0.4748112058468381,
+      "learning_rate": 5.827795600133774e-06,
+      "loss": 0.2599,
+      "step": 954
+    },
+    {
+      "epoch": 3.8754430379746836,
+      "grad_norm": 0.3070171697617278,
+      "learning_rate": 5.787803795514466e-06,
+      "loss": 0.2442,
+      "step": 955
+    },
+    {
+      "epoch": 3.879493670886076,
+      "grad_norm": 0.3141624306927876,
+      "learning_rate": 5.747926453969576e-06,
+      "loss": 0.2436,
+      "step": 956
+    },
+    {
+      "epoch": 3.8835443037974686,
+      "grad_norm": 0.31804599438924813,
+      "learning_rate": 5.708163896665708e-06,
+      "loss": 0.2486,
+      "step": 957
+    },
+    {
+      "epoch": 3.8875949367088607,
+      "grad_norm": 0.3016273595548192,
+      "learning_rate": 5.668516443845047e-06,
+      "loss": 0.2433,
+      "step": 958
+    },
+    {
+      "epoch": 3.891645569620253,
+      "grad_norm": 0.31300992578351555,
+      "learning_rate": 5.6289844148227225e-06,
+      "loss": 0.2127,
+      "step": 959
+    },
+    {
+      "epoch": 3.8956962025316457,
+      "grad_norm": 0.3095832358750485,
+      "learning_rate": 5.5895681279842615e-06,
+      "loss": 0.2447,
+      "step": 960
+    },
+    {
+      "epoch": 3.899746835443038,
+      "grad_norm": 0.3112784598498588,
+      "learning_rate": 5.550267900783019e-06,
+      "loss": 0.2588,
+      "step": 961
+    },
+    {
+      "epoch": 3.9037974683544303,
+      "grad_norm": 0.3163780658029571,
+      "learning_rate": 5.511084049737623e-06,
+      "loss": 0.2455,
+      "step": 962
+    },
+    {
+      "epoch": 3.907848101265823,
+      "grad_norm": 0.30899321664446533,
+      "learning_rate": 5.4720168904294215e-06,
+      "loss": 0.2291,
+      "step": 963
+    },
+    {
+      "epoch": 3.9118987341772153,
+      "grad_norm": 0.3194741867156702,
+      "learning_rate": 5.433066737499948e-06,
+      "loss": 0.235,
+      "step": 964
+    },
+    {
+      "epoch": 3.915949367088608,
+      "grad_norm": 0.31209134818422485,
+      "learning_rate": 5.394233904648376e-06,
+      "loss": 0.2409,
+      "step": 965
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 0.3080999413541567,
+      "learning_rate": 5.355518704628997e-06,
+      "loss": 0.2452,
+      "step": 966
+    },
+    {
+      "epoch": 3.9240506329113924,
+      "grad_norm": 0.2941213299205916,
+      "learning_rate": 5.316921449248731e-06,
+      "loss": 0.2618,
+      "step": 967
+    },
+    {
+      "epoch": 3.928101265822785,
+      "grad_norm": 0.31185993723036765,
+      "learning_rate": 5.278442449364538e-06,
+      "loss": 0.2443,
+      "step": 968
+    },
+    {
+      "epoch": 3.932151898734177,
+      "grad_norm": 0.3066755622094371,
+      "learning_rate": 5.240082014881016e-06,
+      "loss": 0.253,
+      "step": 969
+    },
+    {
+      "epoch": 3.9362025316455695,
+      "grad_norm": 0.4016112415826915,
+      "learning_rate": 5.201840454747822e-06,
+      "loss": 0.2713,
+      "step": 970
+    },
+    {
+      "epoch": 3.940253164556962,
+      "grad_norm": 0.29800873649686294,
+      "learning_rate": 5.163718076957223e-06,
+      "loss": 0.238,
+      "step": 971
+    },
+    {
+      "epoch": 3.9443037974683546,
+      "grad_norm": 0.30843545770579317,
+      "learning_rate": 5.125715188541609e-06,
+      "loss": 0.2405,
+      "step": 972
+    },
+    {
+      "epoch": 3.948354430379747,
+      "grad_norm": 0.30461082619991187,
+      "learning_rate": 5.087832095571021e-06,
+      "loss": 0.256,
+      "step": 973
+    },
+    {
+      "epoch": 3.952405063291139,
+      "grad_norm": 0.31561080971732375,
+      "learning_rate": 5.0500691031506766e-06,
+      "loss": 0.2684,
+      "step": 974
+    },
+    {
+      "epoch": 3.9564556962025317,
+      "grad_norm": 0.3306107198334396,
+      "learning_rate": 5.01242651541854e-06,
+      "loss": 0.2244,
+      "step": 975
+    },
+    {
+      "epoch": 3.960506329113924,
+      "grad_norm": 0.31256278348144045,
+      "learning_rate": 4.974904635542815e-06,
+      "loss": 0.2429,
+      "step": 976
+    },
+    {
+      "epoch": 3.9645569620253163,
+      "grad_norm": 0.3065392726329799,
+      "learning_rate": 4.937503765719582e-06,
+      "loss": 0.2414,
+      "step": 977
+    },
+    {
+      "epoch": 3.968607594936709,
+      "grad_norm": 0.3239486684163372,
+      "learning_rate": 4.900224207170299e-06,
+      "loss": 0.2591,
+      "step": 978
+    },
+    {
+      "epoch": 3.9726582278481013,
+      "grad_norm": 0.3234460084609529,
+      "learning_rate": 4.8630662601394065e-06,
+      "loss": 0.2363,
+      "step": 979
+    },
+    {
+      "epoch": 3.976708860759494,
+      "grad_norm": 0.3066608133311389,
+      "learning_rate": 4.8260302238918995e-06,
+      "loss": 0.2474,
+      "step": 980
+    },
+    {
+      "epoch": 3.9807594936708863,
+      "grad_norm": 0.2932063187612167,
+      "learning_rate": 4.789116396710924e-06,
+      "loss": 0.2446,
+      "step": 981
+    },
+    {
+      "epoch": 3.9848101265822784,
+      "grad_norm": 0.2972865884990358,
+      "learning_rate": 4.752325075895368e-06,
+      "loss": 0.2425,
+      "step": 982
+    },
+    {
+      "epoch": 3.988860759493671,
+      "grad_norm": 0.30939835553538647,
+      "learning_rate": 4.715656557757473e-06,
+      "loss": 0.2551,
+      "step": 983
+    },
+    {
+      "epoch": 3.9929113924050634,
+      "grad_norm": 0.3203601897933848,
+      "learning_rate": 4.679111137620442e-06,
+      "loss": 0.2457,
+      "step": 984
+    },
+    {
+      "epoch": 3.9969620253164555,
+      "grad_norm": 0.30741265489528163,
+      "learning_rate": 4.6426891098160585e-06,
+      "loss": 0.2758,
+      "step": 985
+    },
+    {
+      "epoch": 4.0010126582278485,
+      "grad_norm": 0.3031081685397107,
+      "learning_rate": 4.6063907676823474e-06,
+      "loss": 0.2357,
+      "step": 986
+    },
+    {
+      "epoch": 4.0050632911392405,
+      "grad_norm": 0.4595194209042673,
+      "learning_rate": 4.570216403561141e-06,
+      "loss": 0.1895,
+      "step": 987
+    },
+    {
+      "epoch": 4.009113924050633,
+      "grad_norm": 0.34026846188447457,
+      "learning_rate": 4.534166308795815e-06,
+      "loss": 0.1746,
+      "step": 988
+    },
+    {
+      "epoch": 4.013164556962026,
+      "grad_norm": 0.27506052740701026,
+      "learning_rate": 4.498240773728859e-06,
+      "loss": 0.1749,
+      "step": 989
+    },
+    {
+      "epoch": 4.017215189873418,
+      "grad_norm": 0.2966209517270352,
+      "learning_rate": 4.462440087699609e-06,
+      "loss": 0.1777,
+      "step": 990
+    },
+    {
+      "epoch": 4.02126582278481,
+      "grad_norm": 0.3845615381281242,
+      "learning_rate": 4.426764539041861e-06,
+      "loss": 0.1946,
+      "step": 991
+    },
+    {
+      "epoch": 4.025316455696203,
+      "grad_norm": 0.39380857240634104,
+      "learning_rate": 4.391214415081582e-06,
+      "loss": 0.1522,
+      "step": 992
+    },
+    {
+      "epoch": 4.029367088607595,
+      "grad_norm": 0.44506827856957437,
+      "learning_rate": 4.355790002134579e-06,
+      "loss": 0.1766,
+      "step": 993
+    },
+    {
+      "epoch": 4.033417721518988,
+      "grad_norm": 0.35028374800016493,
+      "learning_rate": 4.320491585504207e-06,
+      "loss": 0.1991,
+      "step": 994
+    },
+    {
+      "epoch": 4.03746835443038,
+      "grad_norm": 0.31749889319566177,
+      "learning_rate": 4.2853194494790615e-06,
+      "loss": 0.2052,
+      "step": 995
+    },
+    {
+      "epoch": 4.041518987341772,
+      "grad_norm": 0.3353447060163369,
+      "learning_rate": 4.250273877330691e-06,
+      "loss": 0.1694,
+      "step": 996
+    },
+    {
+      "epoch": 4.045569620253165,
+      "grad_norm": 0.38610441662796613,
+      "learning_rate": 4.215355151311313e-06,
+      "loss": 0.1766,
+      "step": 997
+    },
+    {
+      "epoch": 4.049620253164557,
+      "grad_norm": 0.34668046464631475,
+      "learning_rate": 4.180563552651542e-06,
+      "loss": 0.208,
+      "step": 998
+    },
+    {
+      "epoch": 4.053670886075949,
+      "grad_norm": 0.29743133340496036,
+      "learning_rate": 4.145899361558147e-06,
+      "loss": 0.1805,
+      "step": 999
+    },
+    {
+      "epoch": 4.057721518987342,
+      "grad_norm": 0.29714018854423874,
+      "learning_rate": 4.111362857211738e-06,
+      "loss": 0.2012,
+      "step": 1000
+    },
+    {
+      "epoch": 4.061772151898734,
+      "grad_norm": 0.32577451290083903,
+      "learning_rate": 4.076954317764592e-06,
+      "loss": 0.2199,
+      "step": 1001
+    },
+    {
+      "epoch": 4.065822784810127,
+      "grad_norm": 0.3344673125807758,
+      "learning_rate": 4.042674020338335e-06,
+      "loss": 0.1896,
+      "step": 1002
+    },
+    {
+      "epoch": 4.069873417721519,
+      "grad_norm": 0.3207409983411464,
+      "learning_rate": 4.0085222410217835e-06,
+      "loss": 0.1825,
+      "step": 1003
+    },
+    {
+      "epoch": 4.073924050632911,
+      "grad_norm": 0.2901937956933271,
+      "learning_rate": 3.974499254868674e-06,
+      "loss": 0.1837,
+      "step": 1004
+    },
+    {
+      "epoch": 4.077974683544304,
+      "grad_norm": 0.2753815379834354,
+      "learning_rate": 3.940605335895451e-06,
+      "loss": 0.1982,
+      "step": 1005
+    },
+    {
+      "epoch": 4.082025316455696,
+      "grad_norm": 0.29066822710769097,
+      "learning_rate": 3.90684075707908e-06,
+      "loss": 0.1936,
+      "step": 1006
+    },
+    {
+      "epoch": 4.086075949367088,
+      "grad_norm": 0.2928284788136984,
+      "learning_rate": 3.8732057903548505e-06,
+      "loss": 0.1931,
+      "step": 1007
+    },
+    {
+      "epoch": 4.090126582278481,
+      "grad_norm": 0.2910558789711315,
+      "learning_rate": 3.8397007066141375e-06,
+      "loss": 0.207,
+      "step": 1008
+    },
+    {
+      "epoch": 4.094177215189873,
+      "grad_norm": 0.2885502762471026,
+      "learning_rate": 3.806325775702304e-06,
+      "loss": 0.1776,
+      "step": 1009
+    },
+    {
+      "epoch": 4.098227848101266,
+      "grad_norm": 0.2703134423388916,
+      "learning_rate": 3.773081266416434e-06,
+      "loss": 0.1901,
+      "step": 1010
+    },
+    {
+      "epoch": 4.102278481012658,
+      "grad_norm": 0.2737721045111255,
+      "learning_rate": 3.739967446503245e-06,
+      "loss": 0.2147,
+      "step": 1011
+    },
+    {
+      "epoch": 4.10632911392405,
+      "grad_norm": 0.26493075872900396,
+      "learning_rate": 3.706984582656894e-06,
+      "loss": 0.2075,
+      "step": 1012
+    },
+    {
+      "epoch": 4.110379746835443,
+      "grad_norm": 0.27366110439042873,
+      "learning_rate": 3.6741329405168237e-06,
+      "loss": 0.1745,
+      "step": 1013
+    },
+    {
+      "epoch": 4.114430379746835,
+      "grad_norm": 0.2835211717604849,
+      "learning_rate": 3.641412784665648e-06,
+      "loss": 0.1861,
+      "step": 1014
+    },
+    {
+      "epoch": 4.118481012658227,
+      "grad_norm": 0.2665131281478132,
+      "learning_rate": 3.608824378627005e-06,
+      "loss": 0.1702,
+      "step": 1015
+    },
+    {
+      "epoch": 4.12253164556962,
+      "grad_norm": 0.25667898177733456,
+      "learning_rate": 3.5763679848634337e-06,
+      "loss": 0.1751,
+      "step": 1016
+    },
+    {
+      "epoch": 4.1265822784810124,
+      "grad_norm": 0.276806857359091,
+      "learning_rate": 3.544043864774269e-06,
+      "loss": 0.1806,
+      "step": 1017
+    },
+    {
+      "epoch": 4.130632911392405,
+      "grad_norm": 0.2684510182046335,
+      "learning_rate": 3.5118522786935282e-06,
+      "loss": 0.1885,
+      "step": 1018
+    },
+    {
+      "epoch": 4.1346835443037975,
+      "grad_norm": 0.27835672741268386,
+      "learning_rate": 3.479793485887819e-06,
+      "loss": 0.1895,
+      "step": 1019
+    },
+    {
+      "epoch": 4.1387341772151895,
+      "grad_norm": 0.2697235097026939,
+      "learning_rate": 3.4478677445542653e-06,
+      "loss": 0.2118,
+      "step": 1020
+    },
+    {
+      "epoch": 4.1427848101265825,
+      "grad_norm": 0.26468548594436625,
+      "learning_rate": 3.4160753118183767e-06,
+      "loss": 0.187,
+      "step": 1021
+    },
+    {
+      "epoch": 4.146835443037975,
+      "grad_norm": 0.2592722600736271,
+      "learning_rate": 3.3844164437320527e-06,
+      "loss": 0.178,
+      "step": 1022
+    },
+    {
+      "epoch": 4.1508860759493675,
+      "grad_norm": 0.24842089487877614,
+      "learning_rate": 3.3528913952714558e-06,
+      "loss": 0.2055,
+      "step": 1023
+    },
+    {
+      "epoch": 4.15493670886076,
+      "grad_norm": 0.24825048778269235,
+      "learning_rate": 3.321500420335e-06,
+      "loss": 0.1853,
+      "step": 1024
+    },
+    {
+      "epoch": 4.158987341772152,
+      "grad_norm": 0.26396083101876777,
+      "learning_rate": 3.290243771741275e-06,
+      "loss": 0.1822,
+      "step": 1025
+    },
+    {
+      "epoch": 4.163037974683545,
+      "grad_norm": 0.26604568909259646,
+      "learning_rate": 3.2591217012270325e-06,
+      "loss": 0.1891,
+      "step": 1026
+    },
+    {
+      "epoch": 4.167088607594937,
+      "grad_norm": 0.250122548695861,
+      "learning_rate": 3.228134459445149e-06,
+      "loss": 0.1745,
+      "step": 1027
+    },
+    {
+      "epoch": 4.171139240506329,
+      "grad_norm": 0.2536038780053431,
+      "learning_rate": 3.1972822959626205e-06,
+      "loss": 0.1736,
+      "step": 1028
+    },
+    {
+      "epoch": 4.175189873417722,
+      "grad_norm": 0.2517772782453091,
+      "learning_rate": 3.166565459258513e-06,
+      "loss": 0.1915,
+      "step": 1029
+    },
+    {
+      "epoch": 4.179240506329114,
+      "grad_norm": 0.24467213689456546,
+      "learning_rate": 3.1359841967220193e-06,
+      "loss": 0.1758,
+      "step": 1030
+    },
+    {
+      "epoch": 4.183291139240506,
+      "grad_norm": 0.26029553974622194,
+      "learning_rate": 3.105538754650419e-06,
+      "loss": 0.1924,
+      "step": 1031
+    },
+    {
+      "epoch": 4.187341772151899,
+      "grad_norm": 0.24885768562886795,
+      "learning_rate": 3.07522937824712e-06,
+      "loss": 0.1905,
+      "step": 1032
+    },
+    {
+      "epoch": 4.191392405063291,
+      "grad_norm": 0.2481575629449707,
+      "learning_rate": 3.0450563116196697e-06,
+      "loss": 0.1846,
+      "step": 1033
+    },
+    {
+      "epoch": 4.195443037974684,
+      "grad_norm": 0.24306143260101026,
+      "learning_rate": 3.0150197977778008e-06,
+      "loss": 0.191,
+      "step": 1034
+    },
+    {
+      "epoch": 4.199493670886076,
+      "grad_norm": 0.26025824968023703,
+      "learning_rate": 2.985120078631465e-06,
+      "loss": 0.2211,
+      "step": 1035
+    },
+    {
+      "epoch": 4.203544303797468,
+      "grad_norm": 0.26676426498579947,
+      "learning_rate": 2.9553573949888893e-06,
+      "loss": 0.1962,
+      "step": 1036
+    },
+    {
+      "epoch": 4.207594936708861,
+      "grad_norm": 0.25754587615203667,
+      "learning_rate": 2.9257319865546384e-06,
+      "loss": 0.2074,
+      "step": 1037
+    },
+    {
+      "epoch": 4.211645569620253,
+      "grad_norm": 0.24933700898524577,
+      "learning_rate": 2.896244091927678e-06,
+      "loss": 0.1921,
+      "step": 1038
+    },
+    {
+      "epoch": 4.215696202531646,
+      "grad_norm": 0.25989508912664366,
+      "learning_rate": 2.8668939485994584e-06,
+      "loss": 0.1694,
+      "step": 1039
+    },
+    {
+      "epoch": 4.219746835443038,
+      "grad_norm": 0.26247102457766713,
+      "learning_rate": 2.837681792951994e-06,
+      "loss": 0.165,
+      "step": 1040
+    },
+    {
+      "epoch": 4.22379746835443,
+      "grad_norm": 0.26704699650510094,
+      "learning_rate": 2.808607860255981e-06,
+      "loss": 0.1991,
+      "step": 1041
+    },
+    {
+      "epoch": 4.227848101265823,
+      "grad_norm": 0.2596733100102948,
+      "learning_rate": 2.7796723846688634e-06,
+      "loss": 0.1968,
+      "step": 1042
+    },
+    {
+      "epoch": 4.231898734177215,
+      "grad_norm": 0.2569159306745644,
+      "learning_rate": 2.7508755992329937e-06,
+      "loss": 0.1907,
+      "step": 1043
+    },
+    {
+      "epoch": 4.235949367088607,
+      "grad_norm": 0.2586006778938439,
+      "learning_rate": 2.722217735873718e-06,
+      "loss": 0.2073,
+      "step": 1044
+    },
+    {
+      "epoch": 4.24,
+      "grad_norm": 0.2577894995963308,
+      "learning_rate": 2.6936990253975315e-06,
+      "loss": 0.2065,
+      "step": 1045
+    },
+    {
+      "epoch": 4.244050632911392,
+      "grad_norm": 0.253324930538423,
+      "learning_rate": 2.665319697490205e-06,
+      "loss": 0.1691,
+      "step": 1046
+    },
+    {
+      "epoch": 4.248101265822784,
+      "grad_norm": 0.2554187175758191,
+      "learning_rate": 2.637079980714945e-06,
+      "loss": 0.1797,
+      "step": 1047
+    },
+    {
+      "epoch": 4.252151898734177,
+      "grad_norm": 0.26943404886828576,
+      "learning_rate": 2.6089801025105453e-06,
+      "loss": 0.1679,
+      "step": 1048
+    },
+    {
+      "epoch": 4.256202531645569,
+      "grad_norm": 0.254299399136655,
+      "learning_rate": 2.581020289189571e-06,
+      "loss": 0.2064,
+      "step": 1049
+    },
+    {
+      "epoch": 4.260253164556962,
+      "grad_norm": 0.2551047456232592,
+      "learning_rate": 2.553200765936501e-06,
+      "loss": 0.181,
+      "step": 1050
+    },
+    {
+      "epoch": 4.264303797468354,
+      "grad_norm": 0.26092461290843916,
+      "learning_rate": 2.525521756805962e-06,
+      "loss": 0.2011,
+      "step": 1051
+    },
+    {
+      "epoch": 4.2683544303797465,
+      "grad_norm": 0.25673809689490135,
+      "learning_rate": 2.497983484720885e-06,
+      "loss": 0.1953,
+      "step": 1052
+    },
+    {
+      "epoch": 4.2724050632911394,
+      "grad_norm": 0.27431324088168263,
+      "learning_rate": 2.470586171470728e-06,
+      "loss": 0.2115,
+      "step": 1053
+    },
+    {
+      "epoch": 4.2764556962025315,
+      "grad_norm": 0.2599292905303033,
+      "learning_rate": 2.4433300377096836e-06,
+      "loss": 0.1833,
+      "step": 1054
+    },
+    {
+      "epoch": 4.2805063291139245,
+      "grad_norm": 0.25324256718884447,
+      "learning_rate": 2.4162153029549073e-06,
+      "loss": 0.1837,
+      "step": 1055
+    },
+    {
+      "epoch": 4.2845569620253166,
+      "grad_norm": 0.25739286194374233,
+      "learning_rate": 2.3892421855847458e-06,
+      "loss": 0.1755,
+      "step": 1056
+    },
+    {
+      "epoch": 4.288607594936709,
+      "grad_norm": 0.2534584911923379,
+      "learning_rate": 2.362410902836978e-06,
+      "loss": 0.1805,
+      "step": 1057
+    },
+    {
+      "epoch": 4.292658227848102,
+      "grad_norm": 0.245901157617782,
+      "learning_rate": 2.3357216708070653e-06,
+      "loss": 0.1855,
+      "step": 1058
+    },
+    {
+      "epoch": 4.296708860759494,
+      "grad_norm": 0.2728682510337038,
+      "learning_rate": 2.309174704446411e-06,
+      "loss": 0.2009,
+      "step": 1059
+    },
+    {
+      "epoch": 4.300759493670886,
+      "grad_norm": 0.24960296534738396,
+      "learning_rate": 2.2827702175606437e-06,
+      "loss": 0.1813,
+      "step": 1060
+    },
+    {
+      "epoch": 4.304810126582279,
+      "grad_norm": 0.26157821361934414,
+      "learning_rate": 2.256508422807855e-06,
+      "loss": 0.174,
+      "step": 1061
+    },
+    {
+      "epoch": 4.308860759493671,
+      "grad_norm": 0.25507021937312874,
+      "learning_rate": 2.230389531696946e-06,
+      "loss": 0.1841,
+      "step": 1062
+    },
+    {
+      "epoch": 4.312911392405063,
+      "grad_norm": 0.25071092046622717,
+      "learning_rate": 2.204413754585857e-06,
+      "loss": 0.1814,
+      "step": 1063
+    },
+    {
+      "epoch": 4.316962025316456,
+      "grad_norm": 0.25756035628451834,
+      "learning_rate": 2.1785813006799406e-06,
+      "loss": 0.203,
+      "step": 1064
+    },
+    {
+      "epoch": 4.321012658227848,
+      "grad_norm": 0.253927267856713,
+      "learning_rate": 2.1528923780302224e-06,
+      "loss": 0.1918,
+      "step": 1065
+    },
+    {
+      "epoch": 4.325063291139241,
+      "grad_norm": 0.2562389143879186,
+      "learning_rate": 2.127347193531757e-06,
+      "loss": 0.1877,
+      "step": 1066
+    },
+    {
+      "epoch": 4.329113924050633,
+      "grad_norm": 0.2598052408390054,
+      "learning_rate": 2.101945952921942e-06,
+      "loss": 0.199,
+      "step": 1067
+    },
+    {
+      "epoch": 4.333164556962025,
+      "grad_norm": 0.25042471894562157,
+      "learning_rate": 2.0766888607788906e-06,
+      "loss": 0.1932,
+      "step": 1068
+    },
+    {
+      "epoch": 4.337215189873418,
+      "grad_norm": 0.2533891425006932,
+      "learning_rate": 2.0515761205197337e-06,
+      "loss": 0.1725,
+      "step": 1069
+    },
+    {
+      "epoch": 4.34126582278481,
+      "grad_norm": 0.2627583042566202,
+      "learning_rate": 2.0266079343990453e-06,
+      "loss": 0.1946,
+      "step": 1070
+    },
+    {
+      "epoch": 4.345316455696203,
+      "grad_norm": 0.24911748766449662,
+      "learning_rate": 2.0017845035071494e-06,
+      "loss": 0.1897,
+      "step": 1071
+    },
+    {
+      "epoch": 4.349367088607595,
+      "grad_norm": 0.26558405556056064,
+      "learning_rate": 1.9771060277685537e-06,
+      "loss": 0.1947,
+      "step": 1072
+    },
+    {
+      "epoch": 4.353417721518987,
+      "grad_norm": 0.25620696351136424,
+      "learning_rate": 1.95257270594031e-06,
+      "loss": 0.1805,
+      "step": 1073
+    },
+    {
+      "epoch": 4.35746835443038,
+      "grad_norm": 0.2529006218789878,
+      "learning_rate": 1.9281847356104188e-06,
+      "loss": 0.207,
+      "step": 1074
+    },
+    {
+      "epoch": 4.361518987341772,
+      "grad_norm": 0.26307518401242747,
+      "learning_rate": 1.9039423131962365e-06,
+      "loss": 0.16,
+      "step": 1075
+    },
+    {
+      "epoch": 4.365569620253164,
+      "grad_norm": 0.2540266779153604,
+      "learning_rate": 1.8798456339429027e-06,
+      "loss": 0.1819,
+      "step": 1076
+    },
+    {
+      "epoch": 4.369620253164557,
+      "grad_norm": 0.25357195949491973,
+      "learning_rate": 1.8558948919217612e-06,
+      "loss": 0.2057,
+      "step": 1077
+    },
+    {
+      "epoch": 4.373670886075949,
+      "grad_norm": 0.25115912610222035,
+      "learning_rate": 1.8320902800287954e-06,
+      "loss": 0.1926,
+      "step": 1078
+    },
+    {
+      "epoch": 4.377721518987342,
+      "grad_norm": 0.24573396432380099,
+      "learning_rate": 1.8084319899830726e-06,
+      "loss": 0.1878,
+      "step": 1079
+    },
+    {
+      "epoch": 4.381772151898734,
+      "grad_norm": 0.2506267882434238,
+      "learning_rate": 1.7849202123252097e-06,
+      "loss": 0.1716,
+      "step": 1080
+    },
+    {
+      "epoch": 4.385822784810126,
+      "grad_norm": 0.26747727715938335,
+      "learning_rate": 1.7615551364158401e-06,
+      "loss": 0.187,
+      "step": 1081
+    },
+    {
+      "epoch": 4.389873417721519,
+      "grad_norm": 0.276050488286739,
+      "learning_rate": 1.738336950434061e-06,
+      "loss": 0.1712,
+      "step": 1082
+    },
+    {
+      "epoch": 4.393924050632911,
+      "grad_norm": 0.26757359813183224,
+      "learning_rate": 1.715265841375957e-06,
+      "loss": 0.1692,
+      "step": 1083
+    },
+    {
+      "epoch": 4.397974683544303,
+      "grad_norm": 0.25463222741321667,
+      "learning_rate": 1.6923419950530684e-06,
+      "loss": 0.1952,
+      "step": 1084
+    },
+    {
+      "epoch": 4.402025316455696,
+      "grad_norm": 0.25713393927024264,
+      "learning_rate": 1.6695655960909008e-06,
+      "loss": 0.1793,
+      "step": 1085
+    },
+    {
+      "epoch": 4.4060759493670885,
+      "grad_norm": 0.249161612259947,
+      "learning_rate": 1.646936827927441e-06,
+      "loss": 0.1801,
+      "step": 1086
+    },
+    {
+      "epoch": 4.410126582278481,
+      "grad_norm": 0.2577401471522469,
+      "learning_rate": 1.6244558728116766e-06,
+      "loss": 0.1788,
+      "step": 1087
+    },
+    {
+      "epoch": 4.4141772151898735,
+      "grad_norm": 0.26205374717138125,
+      "learning_rate": 1.6021229118021265e-06,
+      "loss": 0.1722,
+      "step": 1088
+    },
+    {
+      "epoch": 4.418227848101266,
+      "grad_norm": 0.2548238562567705,
+      "learning_rate": 1.5799381247653967e-06,
+      "loss": 0.1902,
+      "step": 1089
+    },
+    {
+      "epoch": 4.4222784810126585,
+      "grad_norm": 0.26574248204326734,
+      "learning_rate": 1.5579016903747013e-06,
+      "loss": 0.1991,
+      "step": 1090
+    },
+    {
+      "epoch": 4.426329113924051,
+      "grad_norm": 0.24877214049389834,
+      "learning_rate": 1.5360137861084656e-06,
+      "loss": 0.1958,
+      "step": 1091
+    },
+    {
+      "epoch": 4.430379746835443,
+      "grad_norm": 0.26152072916040553,
+      "learning_rate": 1.5142745882488475e-06,
+      "loss": 0.209,
+      "step": 1092
+    },
+    {
+      "epoch": 4.434430379746836,
+      "grad_norm": 0.2594646665598455,
+      "learning_rate": 1.4926842718803691e-06,
+      "loss": 0.1844,
+      "step": 1093
+    },
+    {
+      "epoch": 4.438481012658228,
+      "grad_norm": 0.2541742784581339,
+      "learning_rate": 1.4712430108884657e-06,
+      "loss": 0.1968,
+      "step": 1094
+    },
+    {
+      "epoch": 4.442531645569621,
+      "grad_norm": 0.25010020957311824,
+      "learning_rate": 1.4499509779581078e-06,
+      "loss": 0.1463,
+      "step": 1095
+    },
+    {
+      "epoch": 4.446582278481013,
+      "grad_norm": 0.4335725749332922,
+      "learning_rate": 1.4288083445723988e-06,
+      "loss": 0.1876,
+      "step": 1096
+    },
+    {
+      "epoch": 4.450632911392405,
+      "grad_norm": 0.25621744205099556,
+      "learning_rate": 1.4078152810112045e-06,
+      "loss": 0.1562,
+      "step": 1097
+    },
+    {
+      "epoch": 4.454683544303798,
+      "grad_norm": 0.2586805070319128,
+      "learning_rate": 1.3869719563497697e-06,
+      "loss": 0.1837,
+      "step": 1098
+    },
+    {
+      "epoch": 4.45873417721519,
+      "grad_norm": 0.26662135063873554,
+      "learning_rate": 1.3662785384573663e-06,
+      "loss": 0.2119,
+      "step": 1099
+    },
+    {
+      "epoch": 4.462784810126582,
+      "grad_norm": 0.2507422961018126,
+      "learning_rate": 1.3457351939959383e-06,
+      "loss": 0.1796,
+      "step": 1100
+    },
+    {
+      "epoch": 4.466835443037975,
+      "grad_norm": 0.2524930338555137,
+      "learning_rate": 1.3253420884187551e-06,
+      "loss": 0.1814,
+      "step": 1101
+    },
+    {
+      "epoch": 4.470886075949367,
+      "grad_norm": 0.25501970178141203,
+      "learning_rate": 1.3050993859690953e-06,
+      "loss": 0.1721,
+      "step": 1102
+    },
+    {
+      "epoch": 4.47493670886076,
+      "grad_norm": 0.24877526116289414,
+      "learning_rate": 1.2850072496788869e-06,
+      "loss": 0.1721,
+      "step": 1103
+    },
+    {
+      "epoch": 4.478987341772152,
+      "grad_norm": 0.2648997239177037,
+      "learning_rate": 1.2650658413674434e-06,
+      "loss": 0.1798,
+      "step": 1104
+    },
+    {
+      "epoch": 4.483037974683544,
+      "grad_norm": 0.24189624954783898,
+      "learning_rate": 1.2452753216401226e-06,
+      "loss": 0.1719,
+      "step": 1105
+    },
+    {
+      "epoch": 4.487088607594937,
+      "grad_norm": 0.25523778922270857,
+      "learning_rate": 1.2256358498870503e-06,
+      "loss": 0.1834,
+      "step": 1106
+    },
+    {
+      "epoch": 4.491139240506329,
+      "grad_norm": 0.2538635995147833,
+      "learning_rate": 1.2061475842818337e-06,
+      "loss": 0.199,
+      "step": 1107
+    },
+    {
+      "epoch": 4.495189873417721,
+      "grad_norm": 0.24865409986434014,
+      "learning_rate": 1.1868106817802816e-06,
+      "loss": 0.1782,
+      "step": 1108
+    },
+    {
+      "epoch": 4.499240506329114,
+      "grad_norm": 0.25006975330285586,
+      "learning_rate": 1.1676252981191482e-06,
+      "loss": 0.1795,
+      "step": 1109
+    },
+    {
+      "epoch": 4.503291139240506,
+      "grad_norm": 0.2596022867012277,
+      "learning_rate": 1.1485915878148823e-06,
+      "loss": 0.1887,
+      "step": 1110
+    },
+    {
+      "epoch": 4.507341772151898,
+      "grad_norm": 0.24363420746778314,
+      "learning_rate": 1.1297097041623584e-06,
+      "loss": 0.1799,
+      "step": 1111
+    },
+    {
+      "epoch": 4.511392405063291,
+      "grad_norm": 0.24752161616001797,
+      "learning_rate": 1.1109797992336847e-06,
+      "loss": 0.186,
+      "step": 1112
+    },
+    {
+      "epoch": 4.515443037974683,
+      "grad_norm": 0.26536162524424156,
+      "learning_rate": 1.092402023876933e-06,
+      "loss": 0.174,
+      "step": 1113
+    },
+    {
+      "epoch": 4.519493670886076,
+      "grad_norm": 0.24371501655363395,
+      "learning_rate": 1.0739765277149527e-06,
+      "loss": 0.1893,
+      "step": 1114
+    },
+    {
+      "epoch": 4.523544303797468,
+      "grad_norm": 0.26557361613879793,
+      "learning_rate": 1.0557034591441596e-06,
+      "loss": 0.1858,
+      "step": 1115
+    },
+    {
+      "epoch": 4.52759493670886,
+      "grad_norm": 0.24825292462621898,
+      "learning_rate": 1.0375829653333324e-06,
+      "loss": 0.1953,
+      "step": 1116
+    },
+    {
+      "epoch": 4.531645569620253,
+      "grad_norm": 0.24984810883025685,
+      "learning_rate": 1.0196151922224385e-06,
+      "loss": 0.1868,
+      "step": 1117
+    },
+    {
+      "epoch": 4.535696202531645,
+      "grad_norm": 0.2508884814620609,
+      "learning_rate": 1.0018002845214526e-06,
+      "loss": 0.1985,
+      "step": 1118
+    },
+    {
+      "epoch": 4.539746835443038,
+      "grad_norm": 0.25684781782910726,
+      "learning_rate": 9.841383857091947e-07,
+      "loss": 0.1995,
+      "step": 1119
+    },
+    {
+      "epoch": 4.54379746835443,
+      "grad_norm": 0.2453700174647778,
+      "learning_rate": 9.666296380321616e-07,
+      "loss": 0.1877,
+      "step": 1120
+    },
+    {
+      "epoch": 4.5478481012658225,
+      "grad_norm": 0.2517772344162403,
+      "learning_rate": 9.492741825034124e-07,
+      "loss": 0.1703,
+      "step": 1121
+    },
+    {
+      "epoch": 4.5518987341772155,
+      "grad_norm": 0.2669544400777196,
+      "learning_rate": 9.320721589013892e-07,
+      "loss": 0.1799,
+      "step": 1122
+    },
+    {
+      "epoch": 4.5559493670886075,
+      "grad_norm": 0.24774987328450715,
+      "learning_rate": 9.150237057688339e-07,
+      "loss": 0.1897,
+      "step": 1123
+    },
+    {
+      "epoch": 4.5600000000000005,
+      "grad_norm": 0.24767437398741987,
+      "learning_rate": 8.981289604116328e-07,
+      "loss": 0.1644,
+      "step": 1124
+    },
+    {
+      "epoch": 4.564050632911393,
+      "grad_norm": 0.2555203204815728,
+      "learning_rate": 8.813880588977542e-07,
+      "loss": 0.1694,
+      "step": 1125
+    },
+    {
+      "epoch": 4.568101265822785,
+      "grad_norm": 0.26753626313991863,
+      "learning_rate": 8.648011360561126e-07,
+      "loss": 0.2155,
+      "step": 1126
+    },
+    {
+      "epoch": 4.572151898734178,
+      "grad_norm": 0.24055867196203368,
+      "learning_rate": 8.483683254755037e-07,
+      "loss": 0.1876,
+      "step": 1127
+    },
+    {
+      "epoch": 4.57620253164557,
+      "grad_norm": 0.2531288880629054,
+      "learning_rate": 8.320897595035227e-07,
+      "loss": 0.1859,
+      "step": 1128
+    },
+    {
+      "epoch": 4.580253164556962,
+      "grad_norm": 0.26389023513668247,
+      "learning_rate": 8.159655692455093e-07,
+      "loss": 0.1688,
+      "step": 1129
+    },
+    {
+      "epoch": 4.584303797468355,
+      "grad_norm": 0.26100922533889986,
+      "learning_rate": 7.999958845634648e-07,
+      "loss": 0.1885,
+      "step": 1130
+    },
+    {
+      "epoch": 4.588354430379747,
+      "grad_norm": 0.25749472075493995,
+      "learning_rate": 7.841808340750478e-07,
+      "loss": 0.1831,
+      "step": 1131
+    },
+    {
+      "epoch": 4.592405063291139,
+      "grad_norm": 0.25796825622058955,
+      "learning_rate": 7.685205451524869e-07,
+      "loss": 0.16,
+      "step": 1132
+    },
+    {
+      "epoch": 4.596455696202532,
+      "grad_norm": 0.25197315212762045,
+      "learning_rate": 7.530151439216027e-07,
+      "loss": 0.1867,
+      "step": 1133
+    },
+    {
+      "epoch": 4.600506329113924,
+      "grad_norm": 0.26609241939914086,
+      "learning_rate": 7.376647552607675e-07,
+      "loss": 0.1691,
+      "step": 1134
+    },
+    {
+      "epoch": 4.604556962025317,
+      "grad_norm": 0.2539067371442286,
+      "learning_rate": 7.224695027998963e-07,
+      "loss": 0.1817,
+      "step": 1135
+    },
+    {
+      "epoch": 4.608607594936709,
+      "grad_norm": 0.2848910334500229,
+      "learning_rate": 7.07429508919466e-07,
+      "loss": 0.1796,
+      "step": 1136
+    },
+    {
+      "epoch": 4.612658227848101,
+      "grad_norm": 0.26430409798593085,
+      "learning_rate": 6.925448947495206e-07,
+      "loss": 0.1933,
+      "step": 1137
+    },
+    {
+      "epoch": 4.616708860759494,
+      "grad_norm": 0.2619366640191753,
+      "learning_rate": 6.778157801686936e-07,
+      "loss": 0.1737,
+      "step": 1138
+    },
+    {
+      "epoch": 4.620759493670886,
+      "grad_norm": 0.25217681025310673,
+      "learning_rate": 6.632422838032515e-07,
+      "loss": 0.2165,
+      "step": 1139
+    },
+    {
+      "epoch": 4.624810126582279,
+      "grad_norm": 0.2648493646604171,
+      "learning_rate": 6.488245230261281e-07,
+      "loss": 0.2036,
+      "step": 1140
+    },
+    {
+      "epoch": 4.628860759493671,
+      "grad_norm": 0.2495043936623337,
+      "learning_rate": 6.345626139559868e-07,
+      "loss": 0.2104,
+      "step": 1141
+    },
+    {
+      "epoch": 4.632911392405063,
+      "grad_norm": 0.24519087564415545,
+      "learning_rate": 6.204566714562866e-07,
+      "loss": 0.2194,
+      "step": 1142
+    },
+    {
+      "epoch": 4.636962025316456,
+      "grad_norm": 0.24861616338996975,
+      "learning_rate": 6.06506809134344e-07,
+      "loss": 0.172,
+      "step": 1143
+    },
+    {
+      "epoch": 4.641012658227848,
+      "grad_norm": 0.24151896211517,
+      "learning_rate": 5.927131393404373e-07,
+      "loss": 0.1819,
+      "step": 1144
+    },
+    {
+      "epoch": 4.64506329113924,
+      "grad_norm": 0.2570161636306086,
+      "learning_rate": 5.790757731668817e-07,
+      "loss": 0.195,
+      "step": 1145
+    },
+    {
+      "epoch": 4.649113924050633,
+      "grad_norm": 0.23675777977624773,
+      "learning_rate": 5.655948204471507e-07,
+      "loss": 0.1942,
+      "step": 1146
+    },
+    {
+      "epoch": 4.653164556962025,
+      "grad_norm": 0.2518173194197017,
+      "learning_rate": 5.522703897549875e-07,
+      "loss": 0.1849,
+      "step": 1147
+    },
+    {
+      "epoch": 4.657215189873417,
+      "grad_norm": 0.2551877202859815,
+      "learning_rate": 5.391025884035239e-07,
+      "loss": 0.207,
+      "step": 1148
+    },
+    {
+      "epoch": 4.66126582278481,
+      "grad_norm": 0.25492865702718365,
+      "learning_rate": 5.260915224444207e-07,
+      "loss": 0.1888,
+      "step": 1149
+    },
+    {
+      "epoch": 4.665316455696202,
+      "grad_norm": 0.2493315645583551,
+      "learning_rate": 5.132372966670129e-07,
+      "loss": 0.1964,
+      "step": 1150
+    },
+    {
+      "epoch": 4.669367088607595,
+      "grad_norm": 0.25557013935032535,
+      "learning_rate": 5.005400145974704e-07,
+      "loss": 0.1878,
+      "step": 1151
+    },
+    {
+      "epoch": 4.673417721518987,
+      "grad_norm": 0.2511065444790851,
+      "learning_rate": 4.879997784979562e-07,
+      "loss": 0.203,
+      "step": 1152
+    },
+    {
+      "epoch": 4.6774683544303794,
+      "grad_norm": 0.25003282865800974,
+      "learning_rate": 4.7561668936580984e-07,
+      "loss": 0.1875,
+      "step": 1153
+    },
+    {
+      "epoch": 4.681518987341772,
+      "grad_norm": 0.23908719228593767,
+      "learning_rate": 4.6339084693272306e-07,
+      "loss": 0.1859,
+      "step": 1154
+    },
+    {
+      "epoch": 4.6855696202531645,
+      "grad_norm": 0.25863818228177315,
+      "learning_rate": 4.5132234966395847e-07,
+      "loss": 0.2091,
+      "step": 1155
+    },
+    {
+      "epoch": 4.689620253164557,
+      "grad_norm": 0.2583831993961668,
+      "learning_rate": 4.3941129475752795e-07,
+      "loss": 0.1862,
+      "step": 1156
+    },
+    {
+      "epoch": 4.6936708860759495,
+      "grad_norm": 0.26410214765262197,
+      "learning_rate": 4.27657778143431e-07,
+      "loss": 0.1865,
+      "step": 1157
+    },
+    {
+      "epoch": 4.697721518987342,
+      "grad_norm": 0.246966951628436,
+      "learning_rate": 4.1606189448287757e-07,
+      "loss": 0.2111,
+      "step": 1158
+    },
+    {
+      "epoch": 4.7017721518987345,
+      "grad_norm": 0.25889157408497193,
+      "learning_rate": 4.046237371675177e-07,
+      "loss": 0.1571,
+      "step": 1159
+    },
+    {
+      "epoch": 4.705822784810127,
+      "grad_norm": 0.2554914263180254,
+      "learning_rate": 3.9334339831869963e-07,
+      "loss": 0.1863,
+      "step": 1160
+    },
+    {
+      "epoch": 4.709873417721519,
+      "grad_norm": 0.2678279985864465,
+      "learning_rate": 3.8222096878671955e-07,
+      "loss": 0.1897,
+      "step": 1161
+    },
+    {
+      "epoch": 4.713924050632912,
+      "grad_norm": 0.24866477085002517,
+      "learning_rate": 3.7125653815009545e-07,
+      "loss": 0.1602,
+      "step": 1162
+    },
+    {
+      "epoch": 4.717974683544304,
+      "grad_norm": 0.24080611906117488,
+      "learning_rate": 3.6045019471484974e-07,
+      "loss": 0.1982,
+      "step": 1163
+    },
+    {
+      "epoch": 4.722025316455696,
+      "grad_norm": 0.2536043606419714,
+      "learning_rate": 3.498020255137813e-07,
+      "loss": 0.1812,
+      "step": 1164
+    },
+    {
+      "epoch": 4.726075949367089,
+      "grad_norm": 0.2480059880786284,
+      "learning_rate": 3.393121163057811e-07,
+      "loss": 0.2043,
+      "step": 1165
+    },
+    {
+      "epoch": 4.730126582278481,
+      "grad_norm": 0.255932854382471,
+      "learning_rate": 3.289805515751399e-07,
+      "loss": 0.1905,
+      "step": 1166
+    },
+    {
+      "epoch": 4.734177215189874,
+      "grad_norm": 0.25122510934776665,
+      "learning_rate": 3.188074145308573e-07,
+      "loss": 0.1376,
+      "step": 1167
+    },
+    {
+      "epoch": 4.738227848101266,
+      "grad_norm": 0.24682429922038004,
+      "learning_rate": 3.087927871059804e-07,
+      "loss": 0.1892,
+      "step": 1168
+    },
+    {
+      "epoch": 4.742278481012658,
+      "grad_norm": 0.26055847310565466,
+      "learning_rate": 2.989367499569418e-07,
+      "loss": 0.1747,
+      "step": 1169
+    },
+    {
+      "epoch": 4.746329113924051,
+      "grad_norm": 0.2548218734780477,
+      "learning_rate": 2.8923938246290917e-07,
+      "loss": 0.1888,
+      "step": 1170
+    },
+    {
+      "epoch": 4.750379746835443,
+      "grad_norm": 0.26274280874008876,
+      "learning_rate": 2.7970076272514804e-07,
+      "loss": 0.1685,
+      "step": 1171
+    },
+    {
+      "epoch": 4.754430379746836,
+      "grad_norm": 0.2578746445374266,
+      "learning_rate": 2.703209675663887e-07,
+      "loss": 0.1704,
+      "step": 1172
+    },
+    {
+      "epoch": 4.758481012658228,
+      "grad_norm": 0.2538394236426901,
+      "learning_rate": 2.6110007253021374e-07,
+      "loss": 0.1785,
+      "step": 1173
+    },
+    {
+      "epoch": 4.76253164556962,
+      "grad_norm": 0.26403612045526087,
+      "learning_rate": 2.520381518804471e-07,
+      "loss": 0.1811,
+      "step": 1174
+    },
+    {
+      "epoch": 4.766582278481013,
+      "grad_norm": 0.25076280992008404,
+      "learning_rate": 2.4313527860054585e-07,
+      "loss": 0.1831,
+      "step": 1175
+    },
+    {
+      "epoch": 4.770632911392405,
+      "grad_norm": 0.24787297330435193,
+      "learning_rate": 2.343915243930317e-07,
+      "loss": 0.1668,
+      "step": 1176
+    },
+    {
+      "epoch": 4.774683544303797,
+      "grad_norm": 0.26151116573577937,
+      "learning_rate": 2.2580695967889367e-07,
+      "loss": 0.1985,
+      "step": 1177
+    },
+    {
+      "epoch": 4.77873417721519,
+      "grad_norm": 0.24860899008650683,
+      "learning_rate": 2.1738165359704189e-07,
+      "loss": 0.1752,
+      "step": 1178
+    },
+    {
+      "epoch": 4.782784810126582,
+      "grad_norm": 0.2617010342755883,
+      "learning_rate": 2.0911567400373257e-07,
+      "loss": 0.1656,
+      "step": 1179
+    },
+    {
+      "epoch": 4.786835443037974,
+      "grad_norm": 0.2554148324839167,
+      "learning_rate": 2.0100908747202607e-07,
+      "loss": 0.1969,
+      "step": 1180
+    },
+    {
+      "epoch": 4.790886075949367,
+      "grad_norm": 0.24137296760308233,
+      "learning_rate": 1.9306195929125638e-07,
+      "loss": 0.1956,
+      "step": 1181
+    },
+    {
+      "epoch": 4.794936708860759,
+      "grad_norm": 0.25322337941560996,
+      "learning_rate": 1.8527435346650247e-07,
+      "loss": 0.1875,
+      "step": 1182
+    },
+    {
+      "epoch": 4.798987341772152,
+      "grad_norm": 0.24088228386457156,
+      "learning_rate": 1.7764633271807108e-07,
+      "loss": 0.1762,
+      "step": 1183
+    },
+    {
+      "epoch": 4.803037974683544,
+      "grad_norm": 0.2574921411539647,
+      "learning_rate": 1.7017795848099262e-07,
+      "loss": 0.1948,
+      "step": 1184
+    },
+    {
+      "epoch": 4.807088607594936,
+      "grad_norm": 0.251522052765607,
+      "learning_rate": 1.6286929090452596e-07,
+      "loss": 0.2039,
+      "step": 1185
+    },
+    {
+      "epoch": 4.811139240506329,
+      "grad_norm": 0.2613452717462746,
+      "learning_rate": 1.557203888516745e-07,
+      "loss": 0.1689,
+      "step": 1186
+    },
+    {
+      "epoch": 4.815189873417721,
+      "grad_norm": 0.25355347156348534,
+      "learning_rate": 1.487313098987131e-07,
+      "loss": 0.1905,
+      "step": 1187
+    },
+    {
+      "epoch": 4.819240506329114,
+      "grad_norm": 0.2593328607682762,
+      "learning_rate": 1.4190211033472402e-07,
+      "loss": 0.1733,
+      "step": 1188
+    },
+    {
+      "epoch": 4.8232911392405065,
+      "grad_norm": 0.24551792742030615,
+      "learning_rate": 1.3523284516113955e-07,
+      "loss": 0.172,
+      "step": 1189
+    },
+    {
+      "epoch": 4.8273417721518985,
+      "grad_norm": 0.2478966788235146,
+      "learning_rate": 1.2872356809130682e-07,
+      "loss": 0.199,
+      "step": 1190
+    },
+    {
+      "epoch": 4.8313924050632915,
+      "grad_norm": 0.23717393286779834,
+      "learning_rate": 1.2237433155004807e-07,
+      "loss": 0.1863,
+      "step": 1191
+    },
+    {
+      "epoch": 4.8354430379746836,
+      "grad_norm": 0.2485479606272877,
+      "learning_rate": 1.1618518667323886e-07,
+      "loss": 0.183,
+      "step": 1192
+    },
+    {
+      "epoch": 4.839493670886076,
+      "grad_norm": 0.25866768681445945,
+      "learning_rate": 1.1015618330740385e-07,
+      "loss": 0.1709,
+      "step": 1193
+    },
+    {
+      "epoch": 4.843544303797469,
+      "grad_norm": 0.25526912748329317,
+      "learning_rate": 1.042873700093061e-07,
+      "loss": 0.1894,
+      "step": 1194
+    },
+    {
+      "epoch": 4.847594936708861,
+      "grad_norm": 0.2502979588581034,
+      "learning_rate": 9.857879404556291e-08,
+      "loss": 0.1857,
+      "step": 1195
+    },
+    {
+      "epoch": 4.851645569620253,
+      "grad_norm": 0.2501090636707632,
+      "learning_rate": 9.303050139225722e-08,
+      "loss": 0.1736,
+      "step": 1196
+    },
+    {
+      "epoch": 4.855696202531646,
+      "grad_norm": 0.2632300589615399,
+      "learning_rate": 8.76425367345779e-08,
+      "loss": 0.1823,
+      "step": 1197
+    },
+    {
+      "epoch": 4.859746835443038,
+      "grad_norm": 0.25904858168461176,
+      "learning_rate": 8.241494346644897e-08,
+      "loss": 0.1924,
+      "step": 1198
+    },
+    {
+      "epoch": 4.863797468354431,
+      "grad_norm": 0.24716213156936317,
+      "learning_rate": 7.734776369019204e-08,
+      "loss": 0.1713,
+      "step": 1199
+    },
+    {
+      "epoch": 4.867848101265823,
+      "grad_norm": 0.2507779728230523,
+      "learning_rate": 7.244103821617332e-08,
+      "loss": 0.2036,
+      "step": 1200
+    },
+    {
+      "epoch": 4.871898734177215,
+      "grad_norm": 0.24390293050725292,
+      "learning_rate": 6.769480656248606e-08,
+      "loss": 0.1728,
+      "step": 1201
+    },
+    {
+      "epoch": 4.875949367088608,
+      "grad_norm": 0.24940685973107296,
+      "learning_rate": 6.310910695462635e-08,
+      "loss": 0.1681,
+      "step": 1202
+    },
+    {
+      "epoch": 4.88,
+      "grad_norm": 0.25743270058759116,
+      "learning_rate": 5.8683976325191185e-08,
+      "loss": 0.1859,
+      "step": 1203
+    },
+    {
+      "epoch": 4.884050632911393,
+      "grad_norm": 0.2601780203600966,
+      "learning_rate": 5.4419450313571984e-08,
+      "loss": 0.2074,
+      "step": 1204
+    },
+    {
+      "epoch": 4.888101265822785,
+      "grad_norm": 0.24716918009439,
+      "learning_rate": 5.031556326567488e-08,
+      "loss": 0.1891,
+      "step": 1205
+    },
+    {
+      "epoch": 4.892151898734177,
+      "grad_norm": 0.2561112399948564,
+      "learning_rate": 4.637234823364312e-08,
+      "loss": 0.1989,
+      "step": 1206
+    },
+    {
+      "epoch": 4.89620253164557,
+      "grad_norm": 0.25525230452786235,
+      "learning_rate": 4.258983697558838e-08,
+      "loss": 0.1896,
+      "step": 1207
+    },
+    {
+      "epoch": 4.900253164556962,
+      "grad_norm": 0.24418081883190468,
+      "learning_rate": 3.896805995533548e-08,
+      "loss": 0.2029,
+      "step": 1208
+    },
+    {
+      "epoch": 4.904303797468354,
+      "grad_norm": 0.25634916109838896,
+      "learning_rate": 3.550704634218028e-08,
+      "loss": 0.1918,
+      "step": 1209
+    },
+    {
+      "epoch": 4.908354430379747,
+      "grad_norm": 0.24856887867033836,
+      "learning_rate": 3.2206824010647676e-08,
+      "loss": 0.1787,
+      "step": 1210
+    },
+    {
+      "epoch": 4.912405063291139,
+      "grad_norm": 0.2408410193979466,
+      "learning_rate": 2.9067419540278476e-08,
+      "loss": 0.1709,
+      "step": 1211
+    },
+    {
+      "epoch": 4.916455696202531,
+      "grad_norm": 0.2490300214282961,
+      "learning_rate": 2.6088858215400638e-08,
+      "loss": 0.2046,
+      "step": 1212
+    },
+    {
+      "epoch": 4.920506329113924,
+      "grad_norm": 0.2534107529794442,
+      "learning_rate": 2.3271164024940564e-08,
+      "loss": 0.1902,
+      "step": 1213
+    },
+    {
+      "epoch": 4.924556962025316,
+      "grad_norm": 0.25070039394281707,
+      "learning_rate": 2.061435966221881e-08,
+      "loss": 0.1995,
+      "step": 1214
+    },
+    {
+      "epoch": 4.928607594936709,
+      "grad_norm": 0.24684055096088142,
+      "learning_rate": 1.811846652477245e-08,
+      "loss": 0.1928,
+      "step": 1215
+    },
+    {
+      "epoch": 4.932658227848101,
+      "grad_norm": 0.26237927080501566,
+      "learning_rate": 1.5783504714184106e-08,
+      "loss": 0.186,
+      "step": 1216
+    },
+    {
+      "epoch": 4.936708860759493,
+      "grad_norm": 0.24108105635344573,
+      "learning_rate": 1.360949303591097e-08,
+      "loss": 0.2017,
+      "step": 1217
+    },
+    {
+      "epoch": 4.940759493670886,
+      "grad_norm": 0.25962399934554026,
+      "learning_rate": 1.1596448999144916e-08,
+      "loss": 0.175,
+      "step": 1218
+    },
+    {
+      "epoch": 4.944810126582278,
+      "grad_norm": 0.2586593898670324,
+      "learning_rate": 9.744388816668172e-09,
+      "loss": 0.1962,
+      "step": 1219
+    },
+    {
+      "epoch": 4.948860759493671,
+      "grad_norm": 0.24711722680678425,
+      "learning_rate": 8.05332740472009e-09,
+      "loss": 0.1753,
+      "step": 1220
+    },
+    {
+      "epoch": 4.952911392405063,
+      "grad_norm": 0.24705791620747586,
+      "learning_rate": 6.523278382872811e-09,
+      "loss": 0.1988,
+      "step": 1221
+    },
+    {
+      "epoch": 4.9569620253164555,
+      "grad_norm": 0.244786972942307,
+      "learning_rate": 5.15425407393133e-09,
+      "loss": 0.1677,
+      "step": 1222
+    },
+    {
+      "epoch": 4.961012658227848,
+      "grad_norm": 0.25264864878208926,
+      "learning_rate": 3.94626550383137e-09,
+      "loss": 0.2135,
+      "step": 1223
+    },
+    {
+      "epoch": 4.9650632911392405,
+      "grad_norm": 0.2625817204315561,
+      "learning_rate": 2.899322401546112e-09,
+      "loss": 0.1911,
+      "step": 1224
+    },
+    {
+      "epoch": 4.969113924050633,
+      "grad_norm": 0.25456742489230644,
+      "learning_rate": 2.013433199010706e-09,
+      "loss": 0.1765,
+      "step": 1225
+    },
+    {
+      "epoch": 4.9731645569620255,
+      "grad_norm": 0.24049151610391847,
+      "learning_rate": 1.2886050310556563e-09,
+      "loss": 0.1966,
+      "step": 1226
+    },
+    {
+      "epoch": 4.977215189873418,
+      "grad_norm": 0.25512404942263905,
+      "learning_rate": 7.248437353468695e-10,
+      "loss": 0.1823,
+      "step": 1227
+    },
+    {
+      "epoch": 4.98126582278481,
+      "grad_norm": 0.24909125986136094,
+      "learning_rate": 3.221538523412449e-10,
+      "loss": 0.1937,
+      "step": 1228
+    },
+    {
+      "epoch": 4.985316455696203,
+      "grad_norm": 0.32255868291040185,
+      "learning_rate": 8.053862524670663e-11,
+      "loss": 0.2062,
+      "step": 1229
+    },
+    {
+      "epoch": 4.989367088607595,
+      "grad_norm": 0.28072456729658973,
+      "learning_rate": 0.0,
+      "loss": 0.1805,
+      "step": 1230
+    },
+    {
+      "epoch": 4.989367088607595,
+      "step": 1230,
+      "total_flos": 2.53031023413307e+18,
+      "train_loss": 0.15847020327318007,
+      "train_runtime": 23887.8878,
+      "train_samples_per_second": 6.614,
+      "train_steps_per_second": 0.051
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1230,
+  "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": 2.53031023413307e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}