diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6596 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9994672349493873,
+  "eval_steps": 500,
+  "global_step": 938,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.0,
+      "learning_rate": 0.0,
+      "loss": 4.4688,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.0,
+      "learning_rate": 0.0,
+      "loss": 5.7051,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.0,
+      "learning_rate": 0.0,
+      "loss": 4.6101,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.0,
+      "learning_rate": 0.0,
+      "loss": 5.1161,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.0,
+      "learning_rate": 0.0,
+      "loss": 4.7314,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.0,
+      "learning_rate": 0.0,
+      "loss": 5.4174,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.0,
+      "learning_rate": 0.0,
+      "loss": 5.3351,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 267.06263341170535,
+      "learning_rate": 2.1276595744680852e-07,
+      "loss": 4.3761,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 256.6399066161345,
+      "learning_rate": 4.2553191489361704e-07,
+      "loss": 4.911,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 259.8979037908876,
+      "learning_rate": 6.382978723404255e-07,
+      "loss": 4.5067,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 206.34210154966675,
+      "learning_rate": 8.510638297872341e-07,
+      "loss": 4.1617,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 206.34210154966675,
+      "learning_rate": 8.510638297872341e-07,
+      "loss": 4.7024,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 104.25967551689651,
+      "learning_rate": 1.0638297872340427e-06,
+      "loss": 2.8783,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 104.25967551689651,
+      "learning_rate": 1.0638297872340427e-06,
+      "loss": 3.6247,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 301.56418557125875,
+      "learning_rate": 1.276595744680851e-06,
+      "loss": 3.8274,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 158.0102677208041,
+      "learning_rate": 1.4893617021276596e-06,
+      "loss": 2.5054,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 144.94261605863426,
+      "learning_rate": 1.7021276595744682e-06,
+      "loss": 2.6479,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 77.83778037631348,
+      "learning_rate": 1.9148936170212767e-06,
+      "loss": 1.7461,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 110.09097679168819,
+      "learning_rate": 2.1276595744680853e-06,
+      "loss": 2.005,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 64.67189678957659,
+      "learning_rate": 2.340425531914894e-06,
+      "loss": 1.5011,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 85.70086135587184,
+      "learning_rate": 2.553191489361702e-06,
+      "loss": 1.4231,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 51.55499709016579,
+      "learning_rate": 2.765957446808511e-06,
+      "loss": 1.3443,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 39.458031493922036,
+      "learning_rate": 2.978723404255319e-06,
+      "loss": 1.1782,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 30.718044498953375,
+      "learning_rate": 3.191489361702128e-06,
+      "loss": 1.0572,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 54.03307642588333,
+      "learning_rate": 3.4042553191489363e-06,
+      "loss": 1.0557,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 45.15061216435745,
+      "learning_rate": 3.6170212765957453e-06,
+      "loss": 0.9468,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 85.20725035566016,
+      "learning_rate": 3.8297872340425535e-06,
+      "loss": 1.1611,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 61.03528945297965,
+      "learning_rate": 4.042553191489362e-06,
+      "loss": 0.9764,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 32.57984313596529,
+      "learning_rate": 4.255319148936171e-06,
+      "loss": 0.9542,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 46.71878018713139,
+      "learning_rate": 4.468085106382979e-06,
+      "loss": 1.0575,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 28.228382153829948,
+      "learning_rate": 4.680851063829788e-06,
+      "loss": 0.8927,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 32.305084533281175,
+      "learning_rate": 4.893617021276596e-06,
+      "loss": 0.9158,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 38.80822759339413,
+      "learning_rate": 5.106382978723404e-06,
+      "loss": 1.0127,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 16.765291089343936,
+      "learning_rate": 5.319148936170213e-06,
+      "loss": 0.8579,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 46.21887734855352,
+      "learning_rate": 5.531914893617022e-06,
+      "loss": 1.0595,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 33.51542422627954,
+      "learning_rate": 5.744680851063831e-06,
+      "loss": 0.7873,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 20.62520698421828,
+      "learning_rate": 5.957446808510638e-06,
+      "loss": 0.7463,
+      "step": 37
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 32.86540760673144,
+      "learning_rate": 6.170212765957447e-06,
+      "loss": 0.9393,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 16.78339589062748,
+      "learning_rate": 6.382978723404256e-06,
+      "loss": 0.7036,
+      "step": 39
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 20.638888081681362,
+      "learning_rate": 6.595744680851064e-06,
+      "loss": 0.8178,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 24.957942318493615,
+      "learning_rate": 6.808510638297873e-06,
+      "loss": 0.8192,
+      "step": 41
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 18.753926303171866,
+      "learning_rate": 7.021276595744682e-06,
+      "loss": 0.8053,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 20.380738488767815,
+      "learning_rate": 7.234042553191491e-06,
+      "loss": 0.7775,
+      "step": 43
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 15.358291004654351,
+      "learning_rate": 7.446808510638298e-06,
+      "loss": 0.768,
+      "step": 44
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 28.01853531392082,
+      "learning_rate": 7.659574468085107e-06,
+      "loss": 0.6495,
+      "step": 45
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 90.12246442870764,
+      "learning_rate": 7.872340425531916e-06,
+      "loss": 1.1254,
+      "step": 46
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 24.048650390369733,
+      "learning_rate": 8.085106382978723e-06,
+      "loss": 0.8769,
+      "step": 47
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 79.51846666706089,
+      "learning_rate": 8.297872340425532e-06,
+      "loss": 1.2554,
+      "step": 48
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 20.341370812622667,
+      "learning_rate": 8.510638297872341e-06,
+      "loss": 0.7153,
+      "step": 49
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 48.50233018085147,
+      "learning_rate": 8.72340425531915e-06,
+      "loss": 1.0358,
+      "step": 50
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 27.71805876780665,
+      "learning_rate": 8.936170212765958e-06,
+      "loss": 0.7539,
+      "step": 51
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 44.56567439721268,
+      "learning_rate": 9.148936170212767e-06,
+      "loss": 1.0363,
+      "step": 52
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 43.96614589042312,
+      "learning_rate": 9.361702127659576e-06,
+      "loss": 1.0823,
+      "step": 53
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 36.336438544851525,
+      "learning_rate": 9.574468085106385e-06,
+      "loss": 0.9189,
+      "step": 54
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 20.360154033813146,
+      "learning_rate": 9.787234042553192e-06,
+      "loss": 0.7138,
+      "step": 55
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 16.415639113466074,
+      "learning_rate": 1e-05,
+      "loss": 0.8165,
+      "step": 56
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 17.154150115372023,
+      "learning_rate": 1.0212765957446808e-05,
+      "loss": 0.8003,
+      "step": 57
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 23.485978008365016,
+      "learning_rate": 1.0425531914893619e-05,
+      "loss": 0.8017,
+      "step": 58
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 18.921319860871982,
+      "learning_rate": 1.0638297872340426e-05,
+      "loss": 0.6803,
+      "step": 59
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 28.458932361446607,
+      "learning_rate": 1.0851063829787233e-05,
+      "loss": 0.9205,
+      "step": 60
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 16.224593364238817,
+      "learning_rate": 1.1063829787234044e-05,
+      "loss": 0.7286,
+      "step": 61
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 30.78498419558807,
+      "learning_rate": 1.1276595744680851e-05,
+      "loss": 0.9564,
+      "step": 62
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 20.29389328745947,
+      "learning_rate": 1.1489361702127662e-05,
+      "loss": 0.8442,
+      "step": 63
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 18.23986171950765,
+      "learning_rate": 1.170212765957447e-05,
+      "loss": 0.7971,
+      "step": 64
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 13.607742168916245,
+      "learning_rate": 1.1914893617021277e-05,
+      "loss": 0.7479,
+      "step": 65
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 15.589517783052612,
+      "learning_rate": 1.2127659574468087e-05,
+      "loss": 0.6971,
+      "step": 66
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 11.730568689593209,
+      "learning_rate": 1.2340425531914895e-05,
+      "loss": 0.6136,
+      "step": 67
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 17.919191925814953,
+      "learning_rate": 1.2553191489361702e-05,
+      "loss": 0.6786,
+      "step": 68
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 46.93984753189794,
+      "learning_rate": 1.2765957446808513e-05,
+      "loss": 1.1316,
+      "step": 69
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 32.097199934080784,
+      "learning_rate": 1.297872340425532e-05,
+      "loss": 0.8888,
+      "step": 70
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 23.37535729938417,
+      "learning_rate": 1.3191489361702127e-05,
+      "loss": 0.7891,
+      "step": 71
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 16.127458600501658,
+      "learning_rate": 1.3404255319148938e-05,
+      "loss": 0.5976,
+      "step": 72
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 13.056670968499086,
+      "learning_rate": 1.3617021276595745e-05,
+      "loss": 0.6602,
+      "step": 73
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 18.80577219024569,
+      "learning_rate": 1.3829787234042556e-05,
+      "loss": 0.7343,
+      "step": 74
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 18.0606093200878,
+      "learning_rate": 1.4042553191489363e-05,
+      "loss": 0.7229,
+      "step": 75
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 13.62137898397275,
+      "learning_rate": 1.425531914893617e-05,
+      "loss": 0.7897,
+      "step": 76
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 10.783139067640777,
+      "learning_rate": 1.4468085106382981e-05,
+      "loss": 0.5573,
+      "step": 77
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 28.888686230080594,
+      "learning_rate": 1.4680851063829789e-05,
+      "loss": 0.7468,
+      "step": 78
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 24.696181606034965,
+      "learning_rate": 1.4893617021276596e-05,
+      "loss": 0.6625,
+      "step": 79
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 32.68489697451064,
+      "learning_rate": 1.5106382978723407e-05,
+      "loss": 1.0193,
+      "step": 80
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 20.801300816024295,
+      "learning_rate": 1.5319148936170214e-05,
+      "loss": 0.6352,
+      "step": 81
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 10.55513433584252,
+      "learning_rate": 1.5531914893617023e-05,
+      "loss": 0.7007,
+      "step": 82
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 15.716140126443527,
+      "learning_rate": 1.5744680851063832e-05,
+      "loss": 0.7187,
+      "step": 83
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 21.85726232694579,
+      "learning_rate": 1.595744680851064e-05,
+      "loss": 0.7349,
+      "step": 84
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 20.964587712026038,
+      "learning_rate": 1.6170212765957446e-05,
+      "loss": 0.7143,
+      "step": 85
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 8.250030396732411,
+      "learning_rate": 1.6382978723404255e-05,
+      "loss": 0.6892,
+      "step": 86
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 21.20006063299826,
+      "learning_rate": 1.6595744680851064e-05,
+      "loss": 0.8109,
+      "step": 87
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 19.09608912821583,
+      "learning_rate": 1.6808510638297873e-05,
+      "loss": 0.8082,
+      "step": 88
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 20.52502129407863,
+      "learning_rate": 1.7021276595744682e-05,
+      "loss": 0.6783,
+      "step": 89
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 39.20021747253875,
+      "learning_rate": 1.723404255319149e-05,
+      "loss": 0.9711,
+      "step": 90
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 31.095085040142717,
+      "learning_rate": 1.74468085106383e-05,
+      "loss": 0.8283,
+      "step": 91
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 37.26207899740689,
+      "learning_rate": 1.765957446808511e-05,
+      "loss": 1.0039,
+      "step": 92
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 16.273798667976465,
+      "learning_rate": 1.7872340425531915e-05,
+      "loss": 0.6814,
+      "step": 93
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 32.374520273410155,
+      "learning_rate": 1.8085106382978724e-05,
+      "loss": 0.9674,
+      "step": 94
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 34.748891951713134,
+      "learning_rate": 1.8297872340425533e-05,
+      "loss": 1.1996,
+      "step": 95
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 26.013510708052237,
+      "learning_rate": 1.8510638297872342e-05,
+      "loss": 0.9372,
+      "step": 96
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 19.99677186530767,
+      "learning_rate": 1.872340425531915e-05,
+      "loss": 0.8718,
+      "step": 97
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 16.62199424485097,
+      "learning_rate": 1.893617021276596e-05,
+      "loss": 0.9384,
+      "step": 98
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 20.111532243609883,
+      "learning_rate": 1.914893617021277e-05,
+      "loss": 0.9813,
+      "step": 99
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 15.191733826484391,
+      "learning_rate": 1.9361702127659575e-05,
+      "loss": 0.8069,
+      "step": 100
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 23.926673285223416,
+      "learning_rate": 1.9574468085106384e-05,
+      "loss": 0.8308,
+      "step": 101
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 23.218285916806728,
+      "learning_rate": 1.9787234042553193e-05,
+      "loss": 0.9754,
+      "step": 102
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 17.76104315857583,
+      "learning_rate": 2e-05,
+      "loss": 0.8006,
+      "step": 103
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 20.836820707421783,
+      "learning_rate": 1.9999930723752516e-05,
+      "loss": 0.8543,
+      "step": 104
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 17.0438780130028,
+      "learning_rate": 1.9999722895969904e-05,
+      "loss": 0.8443,
+      "step": 105
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 22.36760800612908,
+      "learning_rate": 1.9999376519531668e-05,
+      "loss": 0.8841,
+      "step": 106
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 11.874378681555248,
+      "learning_rate": 1.999889159923694e-05,
+      "loss": 0.6501,
+      "step": 107
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 17.245286092714633,
+      "learning_rate": 1.9998268141804412e-05,
+      "loss": 0.8119,
+      "step": 108
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 16.693402688517093,
+      "learning_rate": 1.9997506155872246e-05,
+      "loss": 0.7683,
+      "step": 109
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 18.100428165378997,
+      "learning_rate": 1.999660565199794e-05,
+      "loss": 0.8259,
+      "step": 110
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 10.547692312079072,
+      "learning_rate": 1.9995566642658208e-05,
+      "loss": 0.6651,
+      "step": 111
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 10.19017430503209,
+      "learning_rate": 1.9994389142248775e-05,
+      "loss": 0.7619,
+      "step": 112
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 13.012636488876366,
+      "learning_rate": 1.999307316708421e-05,
+      "loss": 0.6931,
+      "step": 113
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 13.672739554862506,
+      "learning_rate": 1.9991618735397674e-05,
+      "loss": 0.7014,
+      "step": 114
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 11.268690129927412,
+      "learning_rate": 1.9990025867340683e-05,
+      "loss": 0.7041,
+      "step": 115
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 13.835706135922516,
+      "learning_rate": 1.998829458498282e-05,
+      "loss": 0.7568,
+      "step": 116
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 12.580781013528279,
+      "learning_rate": 1.998642491231143e-05,
+      "loss": 0.7155,
+      "step": 117
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 11.88807897059113,
+      "learning_rate": 1.9984416875231303e-05,
+      "loss": 0.7374,
+      "step": 118
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 15.58619342416967,
+      "learning_rate": 1.9982270501564286e-05,
+      "loss": 0.8355,
+      "step": 119
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 14.670349037930592,
+      "learning_rate": 1.9979985821048925e-05,
+      "loss": 0.623,
+      "step": 120
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 17.596509727191997,
+      "learning_rate": 1.997756286534004e-05,
+      "loss": 0.7664,
+      "step": 121
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 16.388736277383142,
+      "learning_rate": 1.9975001668008282e-05,
+      "loss": 0.6749,
+      "step": 122
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 11.99481967344947,
+      "learning_rate": 1.9972302264539686e-05,
+      "loss": 0.6724,
+      "step": 123
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 20.53435074276929,
+      "learning_rate": 1.9969464692335153e-05,
+      "loss": 0.7242,
+      "step": 124
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 11.475488488305645,
+      "learning_rate": 1.996648899070996e-05,
+      "loss": 0.7357,
+      "step": 125
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 25.553564042295182,
+      "learning_rate": 1.996337520089319e-05,
+      "loss": 0.7927,
+      "step": 126
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 8.109394353226076,
+      "learning_rate": 1.9960123366027187e-05,
+      "loss": 0.7367,
+      "step": 127
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 8.254315086779002,
+      "learning_rate": 1.995673353116692e-05,
+      "loss": 0.5336,
+      "step": 128
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 28.75262073669196,
+      "learning_rate": 1.995320574327941e-05,
+      "loss": 0.9592,
+      "step": 129
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 22.09118124557931,
+      "learning_rate": 1.9949540051243032e-05,
+      "loss": 0.6935,
+      "step": 130
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 19.95406582271437,
+      "learning_rate": 1.9945736505846866e-05,
+      "loss": 0.8254,
+      "step": 131
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 13.525596071741782,
+      "learning_rate": 1.9941795159789983e-05,
+      "loss": 0.6531,
+      "step": 132
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 9.891408326838208,
+      "learning_rate": 1.9937716067680712e-05,
+      "loss": 0.7162,
+      "step": 133
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 23.11321488500328,
+      "learning_rate": 1.9933499286035896e-05,
+      "loss": 0.7962,
+      "step": 134
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 14.482317722843773,
+      "learning_rate": 1.9929144873280092e-05,
+      "loss": 0.7631,
+      "step": 135
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 19.49882212964946,
+      "learning_rate": 1.992465288974478e-05,
+      "loss": 0.7617,
+      "step": 136
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 14.59122368824527,
+      "learning_rate": 1.992002339766751e-05,
+      "loss": 0.6722,
+      "step": 137
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 14.253058587354259,
+      "learning_rate": 1.9915256461191054e-05,
+      "loss": 0.6869,
+      "step": 138
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 22.646528058916722,
+      "learning_rate": 1.99103521463625e-05,
+      "loss": 0.858,
+      "step": 139
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 19.599804799046947,
+      "learning_rate": 1.9905310521132353e-05,
+      "loss": 0.8475,
+      "step": 140
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 7.459796818444622,
+      "learning_rate": 1.9900131655353597e-05,
+      "loss": 0.6509,
+      "step": 141
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 14.795548081947178,
+      "learning_rate": 1.9894815620780705e-05,
+      "loss": 0.781,
+      "step": 142
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 19.287512528853632,
+      "learning_rate": 1.9889362491068658e-05,
+      "loss": 0.7946,
+      "step": 143
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 11.780467440280454,
+      "learning_rate": 1.9883772341771937e-05,
+      "loss": 0.6937,
+      "step": 144
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 17.590934140385546,
+      "learning_rate": 1.9878045250343445e-05,
+      "loss": 0.8808,
+      "step": 145
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 16.151381586194,
+      "learning_rate": 1.9872181296133473e-05,
+      "loss": 0.8076,
+      "step": 146
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 10.326349558310788,
+      "learning_rate": 1.986618056038856e-05,
+      "loss": 0.8052,
+      "step": 147
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 9.391393835716272,
+      "learning_rate": 1.9860043126250407e-05,
+      "loss": 0.7323,
+      "step": 148
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 13.654055394779414,
+      "learning_rate": 1.9853769078754685e-05,
+      "loss": 0.7861,
+      "step": 149
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 13.20685189778904,
+      "learning_rate": 1.9847358504829896e-05,
+      "loss": 0.7136,
+      "step": 150
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 8.948428090341855,
+      "learning_rate": 1.9840811493296134e-05,
+      "loss": 0.6935,
+      "step": 151
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 9.441488025541567,
+      "learning_rate": 1.9834128134863885e-05,
+      "loss": 0.7222,
+      "step": 152
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 12.322015322175998,
+      "learning_rate": 1.982730852213274e-05,
+      "loss": 0.7606,
+      "step": 153
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 6.325468800857368,
+      "learning_rate": 1.982035274959014e-05,
+      "loss": 0.5735,
+      "step": 154
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 6.406666981105762,
+      "learning_rate": 1.9813260913610048e-05,
+      "loss": 0.5348,
+      "step": 155
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 17.612437547183127,
+      "learning_rate": 1.9806033112451616e-05,
+      "loss": 0.6477,
+      "step": 156
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 18.78091352132794,
+      "learning_rate": 1.9798669446257844e-05,
+      "loss": 0.7878,
+      "step": 157
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 15.546105512429289,
+      "learning_rate": 1.979117001705415e-05,
+      "loss": 0.852,
+      "step": 158
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 16.923786294030265,
+      "learning_rate": 1.9783534928747006e-05,
+      "loss": 0.7754,
+      "step": 159
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 13.076565312797653,
+      "learning_rate": 1.9775764287122464e-05,
+      "loss": 0.7425,
+      "step": 160
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 14.51221649438122,
+      "learning_rate": 1.9767858199844697e-05,
+      "loss": 0.693,
+      "step": 161
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 16.76943181344085,
+      "learning_rate": 1.9759816776454527e-05,
+      "loss": 0.8376,
+      "step": 162
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 10.230125019828748,
+      "learning_rate": 1.9751640128367872e-05,
+      "loss": 0.7238,
+      "step": 163
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 24.10748192789658,
+      "learning_rate": 1.9743328368874237e-05,
+      "loss": 0.8927,
+      "step": 164
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 24.836153220553506,
+      "learning_rate": 1.973488161313512e-05,
+      "loss": 0.8796,
+      "step": 165
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 15.039263981577998,
+      "learning_rate": 1.972629997818243e-05,
+      "loss": 0.7177,
+      "step": 166
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 7.166311769976143,
+      "learning_rate": 1.9717583582916862e-05,
+      "loss": 0.7274,
+      "step": 167
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 13.36802727443076,
+      "learning_rate": 1.970873254810625e-05,
+      "loss": 0.7425,
+      "step": 168
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 12.26776202136434,
+      "learning_rate": 1.969974699638388e-05,
+      "loss": 0.7671,
+      "step": 169
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 14.871789262668347,
+      "learning_rate": 1.969062705224682e-05,
+      "loss": 0.8251,
+      "step": 170
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 13.01131507667653,
+      "learning_rate": 1.968137284205417e-05,
+      "loss": 0.8131,
+      "step": 171
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 8.692440461337334,
+      "learning_rate": 1.967198449402532e-05,
+      "loss": 0.7195,
+      "step": 172
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 15.307857384021617,
+      "learning_rate": 1.966246213823818e-05,
+      "loss": 0.7753,
+      "step": 173
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 22.444147986646815,
+      "learning_rate": 1.9652805906627356e-05,
+      "loss": 0.9884,
+      "step": 174
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 14.639613594215758,
+      "learning_rate": 1.9643015932982355e-05,
+      "loss": 0.7357,
+      "step": 175
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 10.34217870581945,
+      "learning_rate": 1.9633092352945698e-05,
+      "loss": 0.7147,
+      "step": 176
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 12.490262046586464,
+      "learning_rate": 1.9623035304011062e-05,
+      "loss": 0.7384,
+      "step": 177
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 15.433175693682523,
+      "learning_rate": 1.9612844925521375e-05,
+      "loss": 0.7652,
+      "step": 178
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 6.045080843254174,
+      "learning_rate": 1.960252135866687e-05,
+      "loss": 0.8282,
+      "step": 179
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 8.861252768013037,
+      "learning_rate": 1.9592064746483143e-05,
+      "loss": 0.7317,
+      "step": 180
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 7.8221459133117435,
+      "learning_rate": 1.9581475233849165e-05,
+      "loss": 0.5407,
+      "step": 181
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 12.879262754926978,
+      "learning_rate": 1.957075296748527e-05,
+      "loss": 0.6644,
+      "step": 182
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 5.635984955428702,
+      "learning_rate": 1.9559898095951137e-05,
+      "loss": 0.6606,
+      "step": 183
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 7.056126169759745,
+      "learning_rate": 1.9548910769643724e-05,
+      "loss": 0.5217,
+      "step": 184
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 5.817340227592049,
+      "learning_rate": 1.953779114079517e-05,
+      "loss": 0.694,
+      "step": 185
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 11.240599667838348,
+      "learning_rate": 1.9526539363470715e-05,
+      "loss": 0.6613,
+      "step": 186
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 6.377305729246939,
+      "learning_rate": 1.9515155593566536e-05,
+      "loss": 0.6127,
+      "step": 187
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 12.70631731046457,
+      "learning_rate": 1.9503639988807605e-05,
+      "loss": 0.648,
+      "step": 188
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 7.740556879132694,
+      "learning_rate": 1.9491992708745502e-05,
+      "loss": 0.6363,
+      "step": 189
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 9.838244865368074,
+      "learning_rate": 1.94802139147562e-05,
+      "loss": 0.5207,
+      "step": 190
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 12.292539349351499,
+      "learning_rate": 1.946830377003782e-05,
+      "loss": 0.7129,
+      "step": 191
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 9.434980765550153,
+      "learning_rate": 1.94562624396084e-05,
+      "loss": 0.623,
+      "step": 192
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 17.774682137642376,
+      "learning_rate": 1.9444090090303567e-05,
+      "loss": 0.8399,
+      "step": 193
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 9.97338739760112,
+      "learning_rate": 1.9431786890774265e-05,
+      "loss": 0.5893,
+      "step": 194
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 11.708748373782319,
+      "learning_rate": 1.941935301148439e-05,
+      "loss": 0.7569,
+      "step": 195
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 10.640144063322005,
+      "learning_rate": 1.9406788624708438e-05,
+      "loss": 0.694,
+      "step": 196
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 14.454447375580173,
+      "learning_rate": 1.939409390452913e-05,
+      "loss": 0.7807,
+      "step": 197
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 10.35715254572469,
+      "learning_rate": 1.9381269026834975e-05,
+      "loss": 0.6799,
+      "step": 198
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 6.919008432462478,
+      "learning_rate": 1.9368314169317858e-05,
+      "loss": 0.6883,
+      "step": 199
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 9.137624019639443,
+      "learning_rate": 1.935522951147056e-05,
+      "loss": 0.5919,
+      "step": 200
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 15.35731927734825,
+      "learning_rate": 1.9342015234584277e-05,
+      "loss": 0.5982,
+      "step": 201
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 28.1752923954513,
+      "learning_rate": 1.932867152174612e-05,
+      "loss": 0.6482,
+      "step": 202
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 28.1752923954513,
+      "learning_rate": 1.932867152174612e-05,
+      "loss": 1.748,
+      "step": 203
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1587.2702484305357,
+      "learning_rate": 1.9315198557836555e-05,
+      "loss": 1.6079,
+      "step": 204
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1587.2702484305357,
+      "learning_rate": 1.9315198557836555e-05,
+      "loss": 11.678,
+      "step": 205
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2434.8463463776825,
+      "learning_rate": 1.9301596529526856e-05,
+      "loss": 11.5182,
+      "step": 206
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2901.204727033937,
+      "learning_rate": 1.928786562527652e-05,
+      "loss": 9.7123,
+      "step": 207
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 490.65682411342476,
+      "learning_rate": 1.9274006035330654e-05,
+      "loss": 3.1282,
+      "step": 208
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 480.76138720661936,
+      "learning_rate": 1.9260017951717334e-05,
+      "loss": 2.4949,
+      "step": 209
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 143.8799125203908,
+      "learning_rate": 1.9245901568244945e-05,
+      "loss": 2.1616,
+      "step": 210
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 226.80553179811704,
+      "learning_rate": 1.9231657080499507e-05,
+      "loss": 1.9539,
+      "step": 211
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 210.9739319214792,
+      "learning_rate": 1.921728468584195e-05,
+      "loss": 1.7932,
+      "step": 212
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 102.69163082309667,
+      "learning_rate": 1.9202784583405386e-05,
+      "loss": 1.5288,
+      "step": 213
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 80.79475053684271,
+      "learning_rate": 1.9188156974092355e-05,
+      "loss": 1.4608,
+      "step": 214
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 131.37203972527402,
+      "learning_rate": 1.9173402060572028e-05,
+      "loss": 1.3191,
+      "step": 215
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 641.6313774584162,
+      "learning_rate": 1.915852004727742e-05,
+      "loss": 1.8877,
+      "step": 216
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 535.4790750541629,
+      "learning_rate": 1.9143511140402532e-05,
+      "loss": 1.4452,
+      "step": 217
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 548.3152023208836,
+      "learning_rate": 1.9128375547899518e-05,
+      "loss": 1.9607,
+      "step": 218
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 568.5307973204539,
+      "learning_rate": 1.9113113479475784e-05,
+      "loss": 1.8329,
+      "step": 219
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 247.07029020503853,
+      "learning_rate": 1.9097725146591103e-05,
+      "loss": 1.2675,
+      "step": 220
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 483.1919344273096,
+      "learning_rate": 1.908221076245466e-05,
+      "loss": 3.1382,
+      "step": 221
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 435.67666262712555,
+      "learning_rate": 1.9066570542022122e-05,
+      "loss": 2.9579,
+      "step": 222
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 106.36477945538998,
+      "learning_rate": 1.905080470199264e-05,
+      "loss": 1.9584,
+      "step": 223
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 150.21543919990003,
+      "learning_rate": 1.9034913460805868e-05,
+      "loss": 1.5164,
+      "step": 224
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 265.0211172938759,
+      "learning_rate": 1.901889703863891e-05,
+      "loss": 1.4234,
+      "step": 225
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 135.22687197035,
+      "learning_rate": 1.90027556574033e-05,
+      "loss": 1.3242,
+      "step": 226
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 69.09048458856181,
+      "learning_rate": 1.8986489540741895e-05,
+      "loss": 1.1573,
+      "step": 227
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 317.48370008905204,
+      "learning_rate": 1.89700989140258e-05,
+      "loss": 1.6469,
+      "step": 228
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 178.1220954711751,
+      "learning_rate": 1.8953584004351243e-05,
+      "loss": 1.3997,
+      "step": 229
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 58.594800768267845,
+      "learning_rate": 1.8936945040536413e-05,
+      "loss": 1.4203,
+      "step": 230
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 100.8540607851877,
+      "learning_rate": 1.892018225311831e-05,
+      "loss": 1.3959,
+      "step": 231
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 40.196456568889715,
+      "learning_rate": 1.8903295874349528e-05,
+      "loss": 1.3036,
+      "step": 232
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 49.160977991667785,
+      "learning_rate": 1.8886286138195063e-05,
+      "loss": 1.2698,
+      "step": 233
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 38.84069117366192,
+      "learning_rate": 1.8869153280329054e-05,
+      "loss": 1.1846,
+      "step": 234
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 43.81901383324617,
+      "learning_rate": 1.885189753813152e-05,
+      "loss": 1.2088,
+      "step": 235
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 19.950274945570303,
+      "learning_rate": 1.8834519150685074e-05,
+      "loss": 1.0781,
+      "step": 236
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 50.927777642645886,
+      "learning_rate": 1.8817018358771612e-05,
+      "loss": 1.207,
+      "step": 237
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 45.137241355795545,
+      "learning_rate": 1.879939540486897e-05,
+      "loss": 1.1345,
+      "step": 238
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 49.611501553831054,
+      "learning_rate": 1.8781650533147572e-05,
+      "loss": 1.1675,
+      "step": 239
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 60.82753635007021,
+      "learning_rate": 1.876378398946704e-05,
+      "loss": 1.1098,
+      "step": 240
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 36.20738946945509,
+      "learning_rate": 1.87457960213728e-05,
+      "loss": 1.0201,
+      "step": 241
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 46.88562809493131,
+      "learning_rate": 1.8727686878092638e-05,
+      "loss": 1.0511,
+      "step": 242
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 38.81307588678313,
+      "learning_rate": 1.8709456810533248e-05,
+      "loss": 0.9694,
+      "step": 243
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 29.29930215775943,
+      "learning_rate": 1.8691106071276763e-05,
+      "loss": 0.9828,
+      "step": 244
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 54.42807589658664,
+      "learning_rate": 1.867263491457726e-05,
+      "loss": 1.1109,
+      "step": 245
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 29.031387361113243,
+      "learning_rate": 1.8654043596357215e-05,
+      "loss": 1.0194,
+      "step": 246
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 35.21872276907598,
+      "learning_rate": 1.8635332374203993e-05,
+      "loss": 1.0552,
+      "step": 247
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 22.43165286258535,
+      "learning_rate": 1.8616501507366233e-05,
+      "loss": 0.9364,
+      "step": 248
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 29.83402427558291,
+      "learning_rate": 1.85975512567503e-05,
+      "loss": 1.0535,
+      "step": 249
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 36.54199147143507,
+      "learning_rate": 1.8578481884916645e-05,
+      "loss": 0.9772,
+      "step": 250
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 16.06774723588221,
+      "learning_rate": 1.8559293656076167e-05,
+      "loss": 0.9044,
+      "step": 251
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 14.242434397235082,
+      "learning_rate": 1.8539986836086568e-05,
+      "loss": 0.8656,
+      "step": 252
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 34.06087523793793,
+      "learning_rate": 1.8520561692448655e-05,
+      "loss": 0.9776,
+      "step": 253
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 23.65421594444165,
+      "learning_rate": 1.850101849430264e-05,
+      "loss": 0.9739,
+      "step": 254
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 53.1622333489428,
+      "learning_rate": 1.848135751242441e-05,
+      "loss": 0.9659,
+      "step": 255
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 27.743193122343396,
+      "learning_rate": 1.8461579019221775e-05,
+      "loss": 0.9828,
+      "step": 256
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 36.32985964499645,
+      "learning_rate": 1.8441683288730686e-05,
+      "loss": 1.0724,
+      "step": 257
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 57.208457798230874,
+      "learning_rate": 1.8421670596611463e-05,
+      "loss": 1.1399,
+      "step": 258
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 36.265365003042525,
+      "learning_rate": 1.840154122014494e-05,
+      "loss": 1.061,
+      "step": 259
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 50.18323851957246,
+      "learning_rate": 1.8381295438228656e-05,
+      "loss": 0.9041,
+      "step": 260
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 23.117360915951355,
+      "learning_rate": 1.836093353137297e-05,
+      "loss": 0.974,
+      "step": 261
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 50.588104484420946,
+      "learning_rate": 1.8340455781697175e-05,
+      "loss": 0.9803,
+      "step": 262
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 43.25313730795733,
+      "learning_rate": 1.831986247292561e-05,
+      "loss": 0.8178,
+      "step": 263
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 25.22535296579869,
+      "learning_rate": 1.8299153890383705e-05,
+      "loss": 0.9212,
+      "step": 264
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 24.74441945494526,
+      "learning_rate": 1.8278330320994035e-05,
+      "loss": 0.9241,
+      "step": 265
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 24.98014366767532,
+      "learning_rate": 1.8257392053272345e-05,
+      "loss": 0.9996,
+      "step": 266
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 30.455334182511248,
+      "learning_rate": 1.823633937732357e-05,
+      "loss": 0.8491,
+      "step": 267
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 35.02803006376697,
+      "learning_rate": 1.8215172584837776e-05,
+      "loss": 0.9924,
+      "step": 268
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 21.742297963510474,
+      "learning_rate": 1.8193891969086164e-05,
+      "loss": 0.9068,
+      "step": 269
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 16.27634444074666,
+      "learning_rate": 1.817249782491697e-05,
+      "loss": 0.9532,
+      "step": 270
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 27.206570441621835,
+      "learning_rate": 1.8150990448751393e-05,
+      "loss": 0.9585,
+      "step": 271
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 9.683718717075713,
+      "learning_rate": 1.8129370138579507e-05,
+      "loss": 0.8514,
+      "step": 272
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 46.207488174236765,
+      "learning_rate": 1.8107637193956102e-05,
+      "loss": 0.9744,
+      "step": 273
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 24.195105770115998,
+      "learning_rate": 1.8085791915996538e-05,
+      "loss": 0.994,
+      "step": 274
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 33.17832980287206,
+      "learning_rate": 1.8063834607372603e-05,
+      "loss": 1.1351,
+      "step": 275
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 18.66629676451458,
+      "learning_rate": 1.804176557230828e-05,
+      "loss": 0.7503,
+      "step": 276
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 35.58783412196323,
+      "learning_rate": 1.8019585116575554e-05,
+      "loss": 0.8252,
+      "step": 277
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 98.83505787049795,
+      "learning_rate": 1.799729354749018e-05,
+      "loss": 0.772,
+      "step": 278
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 21.601957059226525,
+      "learning_rate": 1.7974891173907406e-05,
+      "loss": 0.7751,
+      "step": 279
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 25.682884674147235,
+      "learning_rate": 1.7952378306217705e-05,
+      "loss": 0.8812,
+      "step": 280
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 14.543550272731224,
+      "learning_rate": 1.792975525634248e-05,
+      "loss": 0.8405,
+      "step": 281
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 49.50597218889562,
+      "learning_rate": 1.790702233772973e-05,
+      "loss": 0.7983,
+      "step": 282
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 28.92796112130337,
+      "learning_rate": 1.7884179865349713e-05,
+      "loss": 0.8288,
+      "step": 283
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 31.306112233207404,
+      "learning_rate": 1.786122815569058e-05,
+      "loss": 1.0268,
+      "step": 284
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 23.72305170543349,
+      "learning_rate": 1.7838167526754002e-05,
+      "loss": 0.9805,
+      "step": 285
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 30.976885389109945,
+      "learning_rate": 1.7814998298050744e-05,
+      "loss": 0.9433,
+      "step": 286
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 9.017186183397232,
+      "learning_rate": 1.7791720790596242e-05,
+      "loss": 0.9042,
+      "step": 287
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 24.453663623228078,
+      "learning_rate": 1.7768335326906182e-05,
+      "loss": 0.916,
+      "step": 288
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 16.661532775610848,
+      "learning_rate": 1.774484223099199e-05,
+      "loss": 0.8476,
+      "step": 289
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 75.49348483174468,
+      "learning_rate": 1.7721241828356377e-05,
+      "loss": 0.9214,
+      "step": 290
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 28.038342380284316,
+      "learning_rate": 1.7697534445988804e-05,
+      "loss": 0.5674,
+      "step": 291
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 26.918490245230988,
+      "learning_rate": 1.7673720412360973e-05,
+      "loss": 0.7465,
+      "step": 292
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 64.73808049621005,
+      "learning_rate": 1.7649800057422256e-05,
+      "loss": 0.7832,
+      "step": 293
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 21.87842844322322,
+      "learning_rate": 1.7625773712595147e-05,
+      "loss": 0.5816,
+      "step": 294
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 15.934437709078225,
+      "learning_rate": 1.760164171077064e-05,
+      "loss": 0.7923,
+      "step": 295
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 50.69489297444858,
+      "learning_rate": 1.7577404386303646e-05,
+      "loss": 0.6258,
+      "step": 296
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 21.105815660081543,
+      "learning_rate": 1.755306207500834e-05,
+      "loss": 0.7464,
+      "step": 297
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 21.183565369421174,
+      "learning_rate": 1.7528615114153524e-05,
+      "loss": 0.6985,
+      "step": 298
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 15.244682840971725,
+      "learning_rate": 1.750406384245793e-05,
+      "loss": 0.767,
+      "step": 299
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 10.358767383759252,
+      "learning_rate": 1.747940860008556e-05,
+      "loss": 0.6928,
+      "step": 300
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 39.96772160224854,
+      "learning_rate": 1.7454649728640944e-05,
+      "loss": 0.6887,
+      "step": 301
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 16.221102491098176,
+      "learning_rate": 1.742978757116443e-05,
+      "loss": 0.6359,
+      "step": 302
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 36.830429589303506,
+      "learning_rate": 1.7404822472127406e-05,
+      "loss": 0.6717,
+      "step": 303
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 17.661289846973503,
+      "learning_rate": 1.7379754777427554e-05,
+      "loss": 0.6609,
+      "step": 304
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 8.91929875914757,
+      "learning_rate": 1.7354584834384036e-05,
+      "loss": 0.6851,
+      "step": 305
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 36.39317884774958,
+      "learning_rate": 1.7329312991732687e-05,
+      "loss": 0.8089,
+      "step": 306
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 20.66408093673374,
+      "learning_rate": 1.73039395996212e-05,
+      "loss": 0.6783,
+      "step": 307
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 13.03603460389346,
+      "learning_rate": 1.727846500960425e-05,
+      "loss": 0.7026,
+      "step": 308
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 17.971533560556523,
+      "learning_rate": 1.725288957463864e-05,
+      "loss": 0.7989,
+      "step": 309
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 9.851305189485732,
+      "learning_rate": 1.7227213649078395e-05,
+      "loss": 0.6135,
+      "step": 310
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 13.9093394062407,
+      "learning_rate": 1.720143758866988e-05,
+      "loss": 0.7099,
+      "step": 311
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 15.918012101752101,
+      "learning_rate": 1.7175561750546833e-05,
+      "loss": 0.6836,
+      "step": 312
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 9.424349721892472,
+      "learning_rate": 1.7149586493225453e-05,
+      "loss": 0.6301,
+      "step": 313
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 15.040725484530588,
+      "learning_rate": 1.7123512176599413e-05,
+      "loss": 0.5931,
+      "step": 314
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 10.404277489273346,
+      "learning_rate": 1.709733916193487e-05,
+      "loss": 0.5216,
+      "step": 315
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 10.351015057753036,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.6885,
+      "step": 316
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 12.500602158481879,
+      "learning_rate": 1.704469849038734e-05,
+      "loss": 0.66,
+      "step": 317
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 9.587643268920083,
+      "learning_rate": 1.7018231562853987e-05,
+      "loss": 0.6073,
+      "step": 318
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 13.14738192750954,
+      "learning_rate": 1.6991667395971306e-05,
+      "loss": 0.6194,
+      "step": 319
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 11.464802706452456,
+      "learning_rate": 1.6965006357792454e-05,
+      "loss": 0.4891,
+      "step": 320
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.153912186441177,
+      "learning_rate": 1.6938248817712767e-05,
+      "loss": 0.5314,
+      "step": 321
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 13.329406945514984,
+      "learning_rate": 1.691139514646464e-05,
+      "loss": 0.7992,
+      "step": 322
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 13.044412872566445,
+      "learning_rate": 1.6884445716112388e-05,
+      "loss": 0.7126,
+      "step": 323
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 8.05329073835129,
+      "learning_rate": 1.685740090004709e-05,
+      "loss": 0.5594,
+      "step": 324
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 12.513875641703635,
+      "learning_rate": 1.6830261072981423e-05,
+      "loss": 0.6085,
+      "step": 325
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 12.059036404622644,
+      "learning_rate": 1.680302661094446e-05,
+      "loss": 0.5629,
+      "step": 326
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 10.324831974232412,
+      "learning_rate": 1.677569789127647e-05,
+      "loss": 0.6268,
+      "step": 327
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 9.702271050156932,
+      "learning_rate": 1.6748275292623678e-05,
+      "loss": 0.621,
+      "step": 328
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 11.784449310834715,
+      "learning_rate": 1.6720759194933037e-05,
+      "loss": 0.6856,
+      "step": 329
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 8.598925774309697,
+      "learning_rate": 1.669314997944694e-05,
+      "loss": 0.6918,
+      "step": 330
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 11.857440160524549,
+      "learning_rate": 1.666544802869796e-05,
+      "loss": 0.6879,
+      "step": 331
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 8.032666301516855,
+      "learning_rate": 1.6637653726503533e-05,
+      "loss": 0.5792,
+      "step": 332
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 7.0937478718250295,
+      "learning_rate": 1.660976745796065e-05,
+      "loss": 0.6607,
+      "step": 333
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 9.465209749684965,
+      "learning_rate": 1.658178960944052e-05,
+      "loss": 0.6535,
+      "step": 334
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 15.131095507571468,
+      "learning_rate": 1.655372056858322e-05,
+      "loss": 0.6438,
+      "step": 335
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 17.304531600026,
+      "learning_rate": 1.6525560724292303e-05,
+      "loss": 0.7653,
+      "step": 336
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 8.695700770875131,
+      "learning_rate": 1.6497310466729448e-05,
+      "loss": 0.6045,
+      "step": 337
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 11.95231240235872,
+      "learning_rate": 1.646897018730902e-05,
+      "loss": 0.6791,
+      "step": 338
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 9.486589536350005,
+      "learning_rate": 1.6440540278692656e-05,
+      "loss": 0.5519,
+      "step": 339
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 10.961929493661737,
+      "learning_rate": 1.6412021134783835e-05,
+      "loss": 0.7537,
+      "step": 340
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 9.852616880372151,
+      "learning_rate": 1.6383413150722413e-05,
+      "loss": 0.5775,
+      "step": 341
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 4.770203374130839,
+      "learning_rate": 1.6354716722879152e-05,
+      "loss": 0.5684,
+      "step": 342
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 6.203239240182946,
+      "learning_rate": 1.6325932248850206e-05,
+      "loss": 0.5391,
+      "step": 343
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 14.714310109670022,
+      "learning_rate": 1.6297060127451656e-05,
+      "loss": 0.6426,
+      "step": 344
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 16.664110650870665,
+      "learning_rate": 1.626810075871394e-05,
+      "loss": 0.7493,
+      "step": 345
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 10.82617102662038,
+      "learning_rate": 1.6239054543876345e-05,
+      "loss": 0.6214,
+      "step": 346
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 13.189719768118387,
+      "learning_rate": 1.6209921885381418e-05,
+      "loss": 0.647,
+      "step": 347
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 6.638748272318463,
+      "learning_rate": 1.6180703186869414e-05,
+      "loss": 0.6062,
+      "step": 348
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 9.318868865822735,
+      "learning_rate": 1.615139885317269e-05,
+      "loss": 0.5877,
+      "step": 349
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 13.07963389547572,
+      "learning_rate": 1.6122009290310097e-05,
+      "loss": 0.697,
+      "step": 350
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 8.538286649824611,
+      "learning_rate": 1.6092534905481367e-05,
+      "loss": 0.605,
+      "step": 351
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 14.845192579391924,
+      "learning_rate": 1.6062976107061454e-05,
+      "loss": 0.5283,
+      "step": 352
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 11.398188334173918,
+      "learning_rate": 1.6033333304594886e-05,
+      "loss": 0.7799,
+      "step": 353
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 11.098935636927996,
+      "learning_rate": 1.6003606908790082e-05,
+      "loss": 0.6517,
+      "step": 354
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 9.295708766284237,
+      "learning_rate": 1.5973797331513674e-05,
+      "loss": 0.8253,
+      "step": 355
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 15.251818246107037,
+      "learning_rate": 1.5943904985784797e-05,
+      "loss": 0.729,
+      "step": 356
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 5.329126700593149,
+      "learning_rate": 1.5913930285769356e-05,
+      "loss": 0.5526,
+      "step": 357
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 12.120816580430638,
+      "learning_rate": 1.5883873646774296e-05,
+      "loss": 0.699,
+      "step": 358
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 10.153946398501363,
+      "learning_rate": 1.5853735485241858e-05,
+      "loss": 0.6955,
+      "step": 359
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 6.094377398479003,
+      "learning_rate": 1.582351621874378e-05,
+      "loss": 0.6237,
+      "step": 360
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 12.749739058129057,
+      "learning_rate": 1.579321626597554e-05,
+      "loss": 0.7169,
+      "step": 361
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 8.376136137589857,
+      "learning_rate": 1.5762836046750546e-05,
+      "loss": 0.6543,
+      "step": 362
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 6.832733254275183,
+      "learning_rate": 1.573237598199432e-05,
+      "loss": 0.6495,
+      "step": 363
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 10.567138540731836,
+      "learning_rate": 1.570183649373865e-05,
+      "loss": 0.6424,
+      "step": 364
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 4.242489418395466,
+      "learning_rate": 1.5671218005115767e-05,
+      "loss": 0.5227,
+      "step": 365
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 10.186721290405504,
+      "learning_rate": 1.5640520940352476e-05,
+      "loss": 0.6831,
+      "step": 366
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 5.2392014967068405,
+      "learning_rate": 1.5609745724764264e-05,
+      "loss": 0.5989,
+      "step": 367
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 7.53125873380665,
+      "learning_rate": 1.5578892784749414e-05,
+      "loss": 0.5461,
+      "step": 368
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 11.308114294485758,
+      "learning_rate": 1.5547962547783126e-05,
+      "loss": 0.5895,
+      "step": 369
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 9.200894044889376,
+      "learning_rate": 1.5516955442411533e-05,
+      "loss": 0.7269,
+      "step": 370
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 7.3282431817133,
+      "learning_rate": 1.5485871898245824e-05,
+      "loss": 0.5479,
+      "step": 371
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 14.868640801063636,
+      "learning_rate": 1.5454712345956254e-05,
+      "loss": 0.6124,
+      "step": 372
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 18.938959881407193,
+      "learning_rate": 1.54234772172662e-05,
+      "loss": 0.6996,
+      "step": 373
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 9.41122569113249,
+      "learning_rate": 1.539216694494616e-05,
+      "loss": 0.7736,
+      "step": 374
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 13.076645506243146,
+      "learning_rate": 1.536078196280777e-05,
+      "loss": 0.6184,
+      "step": 375
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 14.603964846269042,
+      "learning_rate": 1.532932270569778e-05,
+      "loss": 0.7387,
+      "step": 376
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 7.409964871939044,
+      "learning_rate": 1.5297789609492062e-05,
+      "loss": 0.6705,
+      "step": 377
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 10.532417778256553,
+      "learning_rate": 1.526618311108952e-05,
+      "loss": 0.6359,
+      "step": 378
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 7.918617892191258,
+      "learning_rate": 1.5234503648406075e-05,
+      "loss": 0.5597,
+      "step": 379
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 8.984191116234184,
+      "learning_rate": 1.5202751660368594e-05,
+      "loss": 0.5886,
+      "step": 380
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 8.875342349884049,
+      "learning_rate": 1.5170927586908787e-05,
+      "loss": 0.579,
+      "step": 381
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 12.368851666890444,
+      "learning_rate": 1.513903186895713e-05,
+      "loss": 0.684,
+      "step": 382
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 8.674893219974978,
+      "learning_rate": 1.5107064948436758e-05,
+      "loss": 0.5412,
+      "step": 383
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 11.644842239089794,
+      "learning_rate": 1.5075027268257328e-05,
+      "loss": 0.6165,
+      "step": 384
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 24.313226821351616,
+      "learning_rate": 1.5042919272308895e-05,
+      "loss": 1.0199,
+      "step": 385
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 12.280306305433697,
+      "learning_rate": 1.5010741405455751e-05,
+      "loss": 0.5976,
+      "step": 386
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 5.830472782170163,
+      "learning_rate": 1.4978494113530268e-05,
+      "loss": 0.5179,
+      "step": 387
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 14.70235378916328,
+      "learning_rate": 1.4946177843326726e-05,
+      "loss": 0.7092,
+      "step": 388
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 17.543216455121254,
+      "learning_rate": 1.4913793042595109e-05,
+      "loss": 0.8279,
+      "step": 389
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 12.859056834147074,
+      "learning_rate": 1.4881340160034909e-05,
+      "loss": 0.7344,
+      "step": 390
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 9.382814693311708,
+      "learning_rate": 1.4848819645288915e-05,
+      "loss": 0.5823,
+      "step": 391
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 10.602701451692775,
+      "learning_rate": 1.4816231948936967e-05,
+      "loss": 0.706,
+      "step": 392
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 9.699931667366167,
+      "learning_rate": 1.4783577522489733e-05,
+      "loss": 0.6573,
+      "step": 393
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 8.280832106689648,
+      "learning_rate": 1.475085681838244e-05,
+      "loss": 0.6462,
+      "step": 394
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 10.316666897706078,
+      "learning_rate": 1.4718070289968602e-05,
+      "loss": 0.6581,
+      "step": 395
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 7.552089443149507,
+      "learning_rate": 1.4685218391513752e-05,
+      "loss": 0.6516,
+      "step": 396
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 10.041760330765845,
+      "learning_rate": 1.4652301578189141e-05,
+      "loss": 0.7184,
+      "step": 397
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 11.3264716915839,
+      "learning_rate": 1.4619320306065432e-05,
+      "loss": 0.6023,
+      "step": 398
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 9.31395109143123,
+      "learning_rate": 1.4586275032106373e-05,
+      "loss": 0.6248,
+      "step": 399
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 16.084796485739954,
+      "learning_rate": 1.4553166214162486e-05,
+      "loss": 0.5707,
+      "step": 400
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 5.559676215156368,
+      "learning_rate": 1.4519994310964697e-05,
+      "loss": 0.545,
+      "step": 401
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 7.5228362636067425,
+      "learning_rate": 1.4486759782118012e-05,
+      "loss": 0.634,
+      "step": 402
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 10.426847253583567,
+      "learning_rate": 1.4453463088095108e-05,
+      "loss": 0.613,
+      "step": 403
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 6.094049952534055,
+      "learning_rate": 1.4420104690229992e-05,
+      "loss": 0.5793,
+      "step": 404
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 6.245104416676842,
+      "learning_rate": 1.4386685050711593e-05,
+      "loss": 0.5631,
+      "step": 405
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 7.453663730544318,
+      "learning_rate": 1.4353204632577354e-05,
+      "loss": 0.5591,
+      "step": 406
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 10.466741619757949,
+      "learning_rate": 1.4319663899706818e-05,
+      "loss": 0.7189,
+      "step": 407
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 6.220499980519979,
+      "learning_rate": 1.4286063316815209e-05,
+      "loss": 0.4383,
+      "step": 408
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 10.892786776066355,
+      "learning_rate": 1.4252403349446986e-05,
+      "loss": 0.5664,
+      "step": 409
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 6.174114043802528,
+      "learning_rate": 1.4218684463969396e-05,
+      "loss": 0.5716,
+      "step": 410
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 4.9332645630935605,
+      "learning_rate": 1.4184907127566006e-05,
+      "loss": 0.553,
+      "step": 411
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 7.331476371683697,
+      "learning_rate": 1.4151071808230246e-05,
+      "loss": 0.5706,
+      "step": 412
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 9.54440691864147,
+      "learning_rate": 1.4117178974758903e-05,
+      "loss": 0.6349,
+      "step": 413
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 9.51093808442514,
+      "learning_rate": 1.4083229096745644e-05,
+      "loss": 0.6076,
+      "step": 414
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 7.416880704206411,
+      "learning_rate": 1.404922264457449e-05,
+      "loss": 0.5238,
+      "step": 415
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 9.30112890680624,
+      "learning_rate": 1.4015160089413331e-05,
+      "loss": 0.6277,
+      "step": 416
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 9.964620600235138,
+      "learning_rate": 1.3981041903207364e-05,
+      "loss": 0.6501,
+      "step": 417
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 8.049380804954936,
+      "learning_rate": 1.3946868558672569e-05,
+      "loss": 0.5853,
+      "step": 418
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 9.941935672040696,
+      "learning_rate": 1.3912640529289163e-05,
+      "loss": 0.6632,
+      "step": 419
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.961145697377557,
+      "learning_rate": 1.3878358289295032e-05,
+      "loss": 0.5542,
+      "step": 420
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 4.788363887274814,
+      "learning_rate": 1.3844022313679167e-05,
+      "loss": 0.5203,
+      "step": 421
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 12.11801121511047,
+      "learning_rate": 1.380963307817507e-05,
+      "loss": 0.6661,
+      "step": 422
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 11.34675588679876,
+      "learning_rate": 1.3775191059254185e-05,
+      "loss": 0.6116,
+      "step": 423
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 6.678425818352266,
+      "learning_rate": 1.3740696734119279e-05,
+      "loss": 0.6401,
+      "step": 424
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 12.160178505852342,
+      "learning_rate": 1.3706150580697826e-05,
+      "loss": 0.6138,
+      "step": 425
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 6.094569373598361,
+      "learning_rate": 1.3671553077635404e-05,
+      "loss": 0.5602,
+      "step": 426
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 5.908165508182039,
+      "learning_rate": 1.3636904704289053e-05,
+      "loss": 0.5606,
+      "step": 427
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 12.500164793835593,
+      "learning_rate": 1.3602205940720628e-05,
+      "loss": 0.5601,
+      "step": 428
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 5.981711152222135,
+      "learning_rate": 1.3567457267690152e-05,
+      "loss": 0.5011,
+      "step": 429
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 6.390064988076557,
+      "learning_rate": 1.3532659166649165e-05,
+      "loss": 0.515,
+      "step": 430
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 11.203440074699836,
+      "learning_rate": 1.3497812119734037e-05,
+      "loss": 0.7016,
+      "step": 431
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 13.733785334526278,
+      "learning_rate": 1.3462916609759298e-05,
+      "loss": 0.6623,
+      "step": 432
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 15.73461768409232,
+      "learning_rate": 1.342797312021094e-05,
+      "loss": 0.7519,
+      "step": 433
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 8.804555804165531,
+      "learning_rate": 1.3392982135239736e-05,
+      "loss": 0.531,
+      "step": 434
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 10.63905280022859,
+      "learning_rate": 1.3357944139654508e-05,
+      "loss": 0.5914,
+      "step": 435
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 14.738317600036572,
+      "learning_rate": 1.3322859618915431e-05,
+      "loss": 0.6387,
+      "step": 436
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 9.315538586392796,
+      "learning_rate": 1.3287729059127288e-05,
+      "loss": 0.5793,
+      "step": 437
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 7.330889488527192,
+      "learning_rate": 1.3252552947032755e-05,
+      "loss": 0.5955,
+      "step": 438
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 5.492945122022015,
+      "learning_rate": 1.3217331770005639e-05,
+      "loss": 0.4936,
+      "step": 439
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 6.447759334210931,
+      "learning_rate": 1.3182066016044135e-05,
+      "loss": 0.6306,
+      "step": 440
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 10.58800825139178,
+      "learning_rate": 1.3146756173764061e-05,
+      "loss": 0.682,
+      "step": 441
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 11.358838931871405,
+      "learning_rate": 1.3111402732392098e-05,
+      "loss": 0.7864,
+      "step": 442
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 8.567212453461783,
+      "learning_rate": 1.3076006181758989e-05,
+      "loss": 0.5594,
+      "step": 443
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 8.546126016623386,
+      "learning_rate": 1.3040567012292779e-05,
+      "loss": 0.6419,
+      "step": 444
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 11.957735628493818,
+      "learning_rate": 1.3005085715012003e-05,
+      "loss": 0.632,
+      "step": 445
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 10.122315964153882,
+      "learning_rate": 1.2969562781518885e-05,
+      "loss": 0.6758,
+      "step": 446
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 10.274234862260917,
+      "learning_rate": 1.2933998703992531e-05,
+      "loss": 0.4632,
+      "step": 447
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 9.56291976312044,
+      "learning_rate": 1.2898393975182113e-05,
+      "loss": 0.7572,
+      "step": 448
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 11.309501354950024,
+      "learning_rate": 1.2862749088400026e-05,
+      "loss": 0.5429,
+      "step": 449
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.4111570184934195,
+      "learning_rate": 1.2827064537515075e-05,
+      "loss": 0.5896,
+      "step": 450
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 4.403745251090165,
+      "learning_rate": 1.279134081694561e-05,
+      "loss": 0.6544,
+      "step": 451
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.2380739150735565,
+      "learning_rate": 1.2755578421652696e-05,
+      "loss": 0.4995,
+      "step": 452
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 6.1969762293333135,
+      "learning_rate": 1.2719777847133241e-05,
+      "loss": 0.6669,
+      "step": 453
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 6.009545288953141,
+      "learning_rate": 1.2683939589413139e-05,
+      "loss": 0.4661,
+      "step": 454
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 11.862108381435066,
+      "learning_rate": 1.2648064145040392e-05,
+      "loss": 0.5745,
+      "step": 455
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 8.371435047081937,
+      "learning_rate": 1.2612152011078233e-05,
+      "loss": 0.581,
+      "step": 456
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 6.570729807144181,
+      "learning_rate": 1.2576203685098233e-05,
+      "loss": 0.5283,
+      "step": 457
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 15.891490498465696,
+      "learning_rate": 1.2540219665173423e-05,
+      "loss": 0.6488,
+      "step": 458
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 11.348604428897819,
+      "learning_rate": 1.2504200449871378e-05,
+      "loss": 0.6554,
+      "step": 459
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 10.1223452644342,
+      "learning_rate": 1.2468146538247308e-05,
+      "loss": 0.6271,
+      "step": 460
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 14.257981744049157,
+      "learning_rate": 1.2432058429837153e-05,
+      "loss": 0.694,
+      "step": 461
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 14.906690269642054,
+      "learning_rate": 1.2395936624650664e-05,
+      "loss": 0.6699,
+      "step": 462
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 7.166307639896277,
+      "learning_rate": 1.2359781623164465e-05,
+      "loss": 0.5572,
+      "step": 463
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 13.74157282336694,
+      "learning_rate": 1.2323593926315113e-05,
+      "loss": 0.7576,
+      "step": 464
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 9.280109663389997,
+      "learning_rate": 1.2287374035492184e-05,
+      "loss": 0.5046,
+      "step": 465
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 10.090941451576859,
+      "learning_rate": 1.2251122452531301e-05,
+      "loss": 0.5604,
+      "step": 466
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 5.903088605165492,
+      "learning_rate": 1.2214839679707193e-05,
+      "loss": 0.5624,
+      "step": 467
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 10.167630601617299,
+      "learning_rate": 1.2178526219726723e-05,
+      "loss": 0.5295,
+      "step": 468
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 8.331133592273517,
+      "learning_rate": 1.2142182575721946e-05,
+      "loss": 0.6144,
+      "step": 469
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 9.705907066152204,
+      "learning_rate": 1.2105809251243113e-05,
+      "loss": 0.6389,
+      "step": 470
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 11.02758523623926,
+      "learning_rate": 1.2069406750251713e-05,
+      "loss": 0.666,
+      "step": 471
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 7.447181404677012,
+      "learning_rate": 1.2032975577113474e-05,
+      "loss": 0.7216,
+      "step": 472
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 7.658897343141703,
+      "learning_rate": 1.1996516236591398e-05,
+      "loss": 0.6087,
+      "step": 473
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 4.187819572120313,
+      "learning_rate": 1.1960029233838737e-05,
+      "loss": 0.5356,
+      "step": 474
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 5.7176090622427385,
+      "learning_rate": 1.1923515074392022e-05,
+      "loss": 0.619,
+      "step": 475
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 6.823057157296125,
+      "learning_rate": 1.1886974264164037e-05,
+      "loss": 0.6037,
+      "step": 476
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 11.059981117618824,
+      "learning_rate": 1.1850407309436831e-05,
+      "loss": 0.604,
+      "step": 477
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 9.504143330556396,
+      "learning_rate": 1.181381471685468e-05,
+      "loss": 0.647,
+      "step": 478
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 6.557755314159755,
+      "learning_rate": 1.1777196993417087e-05,
+      "loss": 0.4595,
+      "step": 479
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 6.9247206639927095,
+      "learning_rate": 1.1740554646471742e-05,
+      "loss": 0.5228,
+      "step": 480
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 10.232093839271188,
+      "learning_rate": 1.1703888183707513e-05,
+      "loss": 0.6062,
+      "step": 481
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 6.510487958239591,
+      "learning_rate": 1.166719811314738e-05,
+      "loss": 0.5219,
+      "step": 482
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 8.968572470896715,
+      "learning_rate": 1.1630484943141428e-05,
+      "loss": 0.5209,
+      "step": 483
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 5.0494617841107265,
+      "learning_rate": 1.1593749182359788e-05,
+      "loss": 0.5948,
+      "step": 484
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 11.340103876058729,
+      "learning_rate": 1.1556991339785595e-05,
+      "loss": 0.6203,
+      "step": 485
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 10.146196176770191,
+      "learning_rate": 1.1520211924707919e-05,
+      "loss": 0.7785,
+      "step": 486
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.43271626514912,
+      "learning_rate": 1.1483411446714744e-05,
+      "loss": 0.4862,
+      "step": 487
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 7.824102062875491,
+      "learning_rate": 1.1446590415685863e-05,
+      "loss": 0.6113,
+      "step": 488
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 6.336647655938098,
+      "learning_rate": 1.1409749341785859e-05,
+      "loss": 0.5343,
+      "step": 489
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 4.855528076878574,
+      "learning_rate": 1.1372888735456988e-05,
+      "loss": 0.5014,
+      "step": 490
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 8.244191689910387,
+      "learning_rate": 1.1336009107412162e-05,
+      "loss": 0.64,
+      "step": 491
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 9.72186897530395,
+      "learning_rate": 1.1299110968627822e-05,
+      "loss": 0.5919,
+      "step": 492
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 6.153546625651434,
+      "learning_rate": 1.1262194830336888e-05,
+      "loss": 0.4917,
+      "step": 493
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 8.261625036086883,
+      "learning_rate": 1.1225261204021662e-05,
+      "loss": 0.5592,
+      "step": 494
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 7.32183514687012,
+      "learning_rate": 1.118831060140676e-05,
+      "loss": 0.4822,
+      "step": 495
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 6.309347895522231,
+      "learning_rate": 1.1151343534451995e-05,
+      "loss": 0.5727,
+      "step": 496
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 7.8065524764950265,
+      "learning_rate": 1.1114360515345301e-05,
+      "loss": 0.6533,
+      "step": 497
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 9.553172955626218,
+      "learning_rate": 1.107736205649564e-05,
+      "loss": 0.4895,
+      "step": 498
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 8.353357382799153,
+      "learning_rate": 1.1040348670525889e-05,
+      "loss": 0.5254,
+      "step": 499
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 11.74491323882519,
+      "learning_rate": 1.1003320870265741e-05,
+      "loss": 0.6004,
+      "step": 500
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.4228605283872797,
+      "learning_rate": 1.096627916874461e-05,
+      "loss": 0.4572,
+      "step": 501
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 11.366613402895425,
+      "learning_rate": 1.0929224079184514e-05,
+      "loss": 0.648,
+      "step": 502
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.915082226142763,
+      "learning_rate": 1.0892156114992963e-05,
+      "loss": 0.5606,
+      "step": 503
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.481987693582871,
+      "learning_rate": 1.0855075789755845e-05,
+      "loss": 0.5845,
+      "step": 504
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 6.226052256005356,
+      "learning_rate": 1.0817983617230326e-05,
+      "loss": 0.5025,
+      "step": 505
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 8.531553272430466,
+      "learning_rate": 1.0780880111337704e-05,
+      "loss": 0.5657,
+      "step": 506
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 6.998952295190077,
+      "learning_rate": 1.0743765786156313e-05,
+      "loss": 0.7092,
+      "step": 507
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 6.513406245121362,
+      "learning_rate": 1.0706641155914384e-05,
+      "loss": 0.548,
+      "step": 508
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 8.48581353131954,
+      "learning_rate": 1.066950673498294e-05,
+      "loss": 0.5551,
+      "step": 509
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 5.34639942444931,
+      "learning_rate": 1.0632363037868638e-05,
+      "loss": 0.4334,
+      "step": 510
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 11.269718113412772,
+      "learning_rate": 1.0595210579206676e-05,
+      "loss": 0.5803,
+      "step": 511
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 11.41953037054744,
+      "learning_rate": 1.0558049873753637e-05,
+      "loss": 0.5186,
+      "step": 512
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 11.07638179235055,
+      "learning_rate": 1.0520881436380366e-05,
+      "loss": 0.6146,
+      "step": 513
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 11.357643871564994,
+      "learning_rate": 1.0483705782064835e-05,
+      "loss": 0.4825,
+      "step": 514
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 8.12832951837023,
+      "learning_rate": 1.0446523425885008e-05,
+      "loss": 0.5972,
+      "step": 515
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.573339318696573,
+      "learning_rate": 1.040933488301171e-05,
+      "loss": 0.4511,
+      "step": 516
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 7.488115795627891,
+      "learning_rate": 1.0372140668701483e-05,
+      "loss": 0.5612,
+      "step": 517
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 4.416399464656011,
+      "learning_rate": 1.0334941298289441e-05,
+      "loss": 0.5678,
+      "step": 518
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 8.472247115461284,
+      "learning_rate": 1.0297737287182144e-05,
+      "loss": 0.5397,
+      "step": 519
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 7.278212668194716,
+      "learning_rate": 1.026052915085045e-05,
+      "loss": 0.5597,
+      "step": 520
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 5.04716478079048,
+      "learning_rate": 1.022331740482237e-05,
+      "loss": 0.5014,
+      "step": 521
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 11.570053771414813,
+      "learning_rate": 1.0186102564675928e-05,
+      "loss": 0.661,
+      "step": 522
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 8.829505233768511,
+      "learning_rate": 1.014888514603202e-05,
+      "loss": 0.5232,
+      "step": 523
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 9.542821950950582,
+      "learning_rate": 1.0111665664547267e-05,
+      "loss": 0.5823,
+      "step": 524
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.63327242048301,
+      "learning_rate": 1.0074444635906875e-05,
+      "loss": 0.5625,
+      "step": 525
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 6.365741719759197,
+      "learning_rate": 1.0037222575817476e-05,
+      "loss": 0.6587,
+      "step": 526
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 8.488178372143695,
+      "learning_rate": 1e-05,
+      "loss": 0.5501,
+      "step": 527
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 9.897138663353587,
+      "learning_rate": 9.962777424182527e-06,
+      "loss": 0.6661,
+      "step": 528
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 10.569907059227898,
+      "learning_rate": 9.92555536409313e-06,
+      "loss": 0.5958,
+      "step": 529
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 3.736421534561604,
+      "learning_rate": 9.888334335452734e-06,
+      "loss": 0.5303,
+      "step": 530
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 8.836116584372025,
+      "learning_rate": 9.85111485396798e-06,
+      "loss": 0.6466,
+      "step": 531
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 7.181117148918435,
+      "learning_rate": 9.813897435324075e-06,
+      "loss": 0.54,
+      "step": 532
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 4.785898028898109,
+      "learning_rate": 9.776682595177633e-06,
+      "loss": 0.4985,
+      "step": 533
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 11.140585777357792,
+      "learning_rate": 9.739470849149554e-06,
+      "loss": 0.6496,
+      "step": 534
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 3.531792374263498,
+      "learning_rate": 9.702262712817857e-06,
+      "loss": 0.5364,
+      "step": 535
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 6.601039058908546,
+      "learning_rate": 9.66505870171056e-06,
+      "loss": 0.5116,
+      "step": 536
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 8.927040349981295,
+      "learning_rate": 9.627859331298522e-06,
+      "loss": 0.5219,
+      "step": 537
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 6.1651706502501264,
+      "learning_rate": 9.590665116988293e-06,
+      "loss": 0.5608,
+      "step": 538
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 9.745244839875141,
+      "learning_rate": 9.553476574114993e-06,
+      "loss": 0.5524,
+      "step": 539
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 7.008580981680822,
+      "learning_rate": 9.51629421793517e-06,
+      "loss": 0.5662,
+      "step": 540
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 7.687272495275157,
+      "learning_rate": 9.479118563619638e-06,
+      "loss": 0.5806,
+      "step": 541
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 5.593165759891498,
+      "learning_rate": 9.441950126246365e-06,
+      "loss": 0.5447,
+      "step": 542
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 4.518154487773691,
+      "learning_rate": 9.404789420793327e-06,
+      "loss": 0.5662,
+      "step": 543
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 5.979882247197896,
+      "learning_rate": 9.367636962131364e-06,
+      "loss": 0.5499,
+      "step": 544
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 10.248432776308547,
+      "learning_rate": 9.330493265017062e-06,
+      "loss": 0.6341,
+      "step": 545
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 7.4913799071398275,
+      "learning_rate": 9.29335884408562e-06,
+      "loss": 0.4995,
+      "step": 546
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 12.97254908941589,
+      "learning_rate": 9.25623421384369e-06,
+      "loss": 0.5575,
+      "step": 547
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 7.7639027583981095,
+      "learning_rate": 9.2191198886623e-06,
+      "loss": 0.5819,
+      "step": 548
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 7.6543085882238895,
+      "learning_rate": 9.182016382769678e-06,
+      "loss": 0.6647,
+      "step": 549
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.7656791466080355,
+      "learning_rate": 9.144924210244155e-06,
+      "loss": 0.6249,
+      "step": 550
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.2481708291921505,
+      "learning_rate": 9.107843885007042e-06,
+      "loss": 0.5478,
+      "step": 551
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 10.21885595901045,
+      "learning_rate": 9.070775920815488e-06,
+      "loss": 0.5803,
+      "step": 552
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 7.389626516947618,
+      "learning_rate": 9.033720831255391e-06,
+      "loss": 0.5334,
+      "step": 553
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 11.910444324111724,
+      "learning_rate": 8.996679129734266e-06,
+      "loss": 0.5788,
+      "step": 554
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 10.70692641976889,
+      "learning_rate": 8.959651329474115e-06,
+      "loss": 0.5155,
+      "step": 555
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 5.6372805638926,
+      "learning_rate": 8.922637943504362e-06,
+      "loss": 0.5881,
+      "step": 556
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 9.8644472116661,
+      "learning_rate": 8.8856394846547e-06,
+      "loss": 0.5072,
+      "step": 557
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 11.356856786195184,
+      "learning_rate": 8.848656465548007e-06,
+      "loss": 0.549,
+      "step": 558
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 11.998300599364015,
+      "learning_rate": 8.811689398593245e-06,
+      "loss": 0.5355,
+      "step": 559
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 5.815584273089378,
+      "learning_rate": 8.774738795978341e-06,
+      "loss": 0.4673,
+      "step": 560
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 9.210932501903986,
+      "learning_rate": 8.737805169663113e-06,
+      "loss": 0.5137,
+      "step": 561
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 10.604624719913076,
+      "learning_rate": 8.700889031372183e-06,
+      "loss": 0.6092,
+      "step": 562
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 6.733937738438429,
+      "learning_rate": 8.663990892587839e-06,
+      "loss": 0.5617,
+      "step": 563
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 6.687819879267525,
+      "learning_rate": 8.627111264543012e-06,
+      "loss": 0.4319,
+      "step": 564
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 11.209429725910061,
+      "learning_rate": 8.590250658214148e-06,
+      "loss": 0.5174,
+      "step": 565
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 5.337704061513362,
+      "learning_rate": 8.553409584314138e-06,
+      "loss": 0.4295,
+      "step": 566
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 9.126368763120784,
+      "learning_rate": 8.516588553285258e-06,
+      "loss": 0.4436,
+      "step": 567
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 10.856129898989279,
+      "learning_rate": 8.479788075292083e-06,
+      "loss": 0.6126,
+      "step": 568
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 5.163627248000245,
+      "learning_rate": 8.443008660214409e-06,
+      "loss": 0.5352,
+      "step": 569
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 6.372067670711978,
+      "learning_rate": 8.406250817640212e-06,
+      "loss": 0.5076,
+      "step": 570
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 6.426831912031028,
+      "learning_rate": 8.369515056858575e-06,
+      "loss": 0.6041,
+      "step": 571
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 6.985038214862641,
+      "learning_rate": 8.332801886852624e-06,
+      "loss": 0.6586,
+      "step": 572
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 6.593624344866879,
+      "learning_rate": 8.296111816292494e-06,
+      "loss": 0.6751,
+      "step": 573
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 7.554153671309798,
+      "learning_rate": 8.259445353528261e-06,
+      "loss": 0.5276,
+      "step": 574
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 9.224509289941837,
+      "learning_rate": 8.222803006582915e-06,
+      "loss": 0.5993,
+      "step": 575
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 16.50239364012574,
+      "learning_rate": 8.186185283145325e-06,
+      "loss": 0.6259,
+      "step": 576
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 15.120105685997352,
+      "learning_rate": 8.149592690563172e-06,
+      "loss": 0.6467,
+      "step": 577
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 7.220612802422961,
+      "learning_rate": 8.113025735835965e-06,
+      "loss": 0.5715,
+      "step": 578
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 8.562863959435413,
+      "learning_rate": 8.076484925607983e-06,
+      "loss": 0.5466,
+      "step": 579
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 12.490548007030851,
+      "learning_rate": 8.039970766161264e-06,
+      "loss": 0.5514,
+      "step": 580
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 6.5845388402784435,
+      "learning_rate": 8.003483763408604e-06,
+      "loss": 0.5687,
+      "step": 581
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 8.50284284365726,
+      "learning_rate": 7.967024422886528e-06,
+      "loss": 0.4796,
+      "step": 582
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 7.038271102438535,
+      "learning_rate": 7.930593249748289e-06,
+      "loss": 0.5228,
+      "step": 583
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 11.889834227941325,
+      "learning_rate": 7.894190748756892e-06,
+      "loss": 0.4577,
+      "step": 584
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 6.711588374541616,
+      "learning_rate": 7.857817424278056e-06,
+      "loss": 0.498,
+      "step": 585
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 7.966852766279544,
+      "learning_rate": 7.821473780273279e-06,
+      "loss": 0.5579,
+      "step": 586
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 8.417280128263915,
+      "learning_rate": 7.785160320292812e-06,
+      "loss": 0.4942,
+      "step": 587
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 4.894439384156987,
+      "learning_rate": 7.7488775474687e-06,
+      "loss": 0.5082,
+      "step": 588
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 7.960251253611578,
+      "learning_rate": 7.712625964507818e-06,
+      "loss": 0.4997,
+      "step": 589
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 8.640093158656311,
+      "learning_rate": 7.676406073684892e-06,
+      "loss": 0.5362,
+      "step": 590
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 11.82405754186769,
+      "learning_rate": 7.64021837683554e-06,
+      "loss": 0.5292,
+      "step": 591
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 5.258983750312472,
+      "learning_rate": 7.604063375349338e-06,
+      "loss": 0.547,
+      "step": 592
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 10.013781925100409,
+      "learning_rate": 7.567941570162849e-06,
+      "loss": 0.538,
+      "step": 593
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 7.8051606786973755,
+      "learning_rate": 7.531853461752696e-06,
+      "loss": 0.5122,
+      "step": 594
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 7.2586272669558145,
+      "learning_rate": 7.495799550128625e-06,
+      "loss": 0.4468,
+      "step": 595
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.774720107719125,
+      "learning_rate": 7.459780334826578e-06,
+      "loss": 0.5275,
+      "step": 596
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 7.511213422739932,
+      "learning_rate": 7.423796314901769e-06,
+      "loss": 0.5768,
+      "step": 597
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 13.06441655084729,
+      "learning_rate": 7.387847988921772e-06,
+      "loss": 0.6067,
+      "step": 598
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 9.818511674108413,
+      "learning_rate": 7.351935854959608e-06,
+      "loss": 0.5669,
+      "step": 599
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 12.085632705436097,
+      "learning_rate": 7.31606041058686e-06,
+      "loss": 0.6871,
+      "step": 600
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 6.460535141416251,
+      "learning_rate": 7.2802221528667604e-06,
+      "loss": 0.4964,
+      "step": 601
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 3.7229407170275715,
+      "learning_rate": 7.244421578347307e-06,
+      "loss": 0.4216,
+      "step": 602
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 8.837577427746787,
+      "learning_rate": 7.208659183054393e-06,
+      "loss": 0.5951,
+      "step": 603
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.340949956633562,
+      "learning_rate": 7.172935462484931e-06,
+      "loss": 0.3946,
+      "step": 604
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 5.8235051717682405,
+      "learning_rate": 7.137250911599978e-06,
+      "loss": 0.4795,
+      "step": 605
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.498783950116057,
+      "learning_rate": 7.101606024817889e-06,
+      "loss": 0.5192,
+      "step": 606
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.1814150008692534,
+      "learning_rate": 7.066001296007469e-06,
+      "loss": 0.3841,
+      "step": 607
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.5757391334979225,
+      "learning_rate": 7.030437218481116e-06,
+      "loss": 0.4353,
+      "step": 608
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 13.254243576587063,
+      "learning_rate": 6.9949142849880015e-06,
+      "loss": 0.5673,
+      "step": 609
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 4.1536903593774515,
+      "learning_rate": 6.959432987707223e-06,
+      "loss": 0.4192,
+      "step": 610
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 3.4786455475133287,
+      "learning_rate": 6.9239938182410126e-06,
+      "loss": 0.4249,
+      "step": 611
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 7.5677310862475915,
+      "learning_rate": 6.888597267607906e-06,
+      "loss": 0.4949,
+      "step": 612
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 8.368860704109643,
+      "learning_rate": 6.8532438262359404e-06,
+      "loss": 0.4154,
+      "step": 613
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 5.058418452049737,
+      "learning_rate": 6.817933983955867e-06,
+      "loss": 0.4951,
+      "step": 614
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 6.42488252810342,
+      "learning_rate": 6.7826682299943635e-06,
+      "loss": 0.5948,
+      "step": 615
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 7.090667184322379,
+      "learning_rate": 6.747447052967246e-06,
+      "loss": 0.564,
+      "step": 616
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 9.915670648118702,
+      "learning_rate": 6.712270940872713e-06,
+      "loss": 0.4821,
+      "step": 617
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 10.251562644163112,
+      "learning_rate": 6.677140381084573e-06,
+      "loss": 0.5258,
+      "step": 618
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.692081829738655,
+      "learning_rate": 6.642055860345494e-06,
+      "loss": 0.4832,
+      "step": 619
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 4.7513407845962545,
+      "learning_rate": 6.607017864760265e-06,
+      "loss": 0.4669,
+      "step": 620
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 7.302874981115006,
+      "learning_rate": 6.572026879789064e-06,
+      "loss": 0.4492,
+      "step": 621
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 13.730001115703725,
+      "learning_rate": 6.537083390240706e-06,
+      "loss": 0.5786,
+      "step": 622
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 5.826147970750949,
+      "learning_rate": 6.502187880265969e-06,
+      "loss": 0.3817,
+      "step": 623
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 8.854083264240913,
+      "learning_rate": 6.467340833350837e-06,
+      "loss": 0.4938,
+      "step": 624
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 5.711323962217592,
+      "learning_rate": 6.43254273230985e-06,
+      "loss": 0.4934,
+      "step": 625
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 6.916231992838763,
+      "learning_rate": 6.3977940592793766e-06,
+      "loss": 0.5225,
+      "step": 626
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 9.605613693440635,
+      "learning_rate": 6.36309529571095e-06,
+      "loss": 0.527,
+      "step": 627
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 7.226506391255898,
+      "learning_rate": 6.328446922364595e-06,
+      "loss": 0.4771,
+      "step": 628
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 5.211969056714009,
+      "learning_rate": 6.293849419302179e-06,
+      "loss": 0.5099,
+      "step": 629
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 11.660847177274526,
+      "learning_rate": 6.259303265880725e-06,
+      "loss": 0.5479,
+      "step": 630
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 8.043560392139083,
+      "learning_rate": 6.224808940745814e-06,
+      "loss": 0.5182,
+      "step": 631
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 8.052748988432281,
+      "learning_rate": 6.190366921824933e-06,
+      "loss": 0.6176,
+      "step": 632
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 7.108683500222242,
+      "learning_rate": 6.155977686320837e-06,
+      "loss": 0.5878,
+      "step": 633
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 11.823467597819771,
+      "learning_rate": 6.121641710704972e-06,
+      "loss": 0.4971,
+      "step": 634
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 10.40651530100889,
+      "learning_rate": 6.087359470710841e-06,
+      "loss": 0.4427,
+      "step": 635
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 11.976145453386437,
+      "learning_rate": 6.053131441327431e-06,
+      "loss": 0.5098,
+      "step": 636
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 7.971563751964657,
+      "learning_rate": 6.018958096792642e-06,
+      "loss": 0.5695,
+      "step": 637
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 6.129677385908451,
+      "learning_rate": 5.98483991058667e-06,
+      "loss": 0.4808,
+      "step": 638
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.479658413986526,
+      "learning_rate": 5.950777355425511e-06,
+      "loss": 0.4775,
+      "step": 639
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 4.309077539414612,
+      "learning_rate": 5.916770903254363e-06,
+      "loss": 0.4718,
+      "step": 640
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 5.726658232065661,
+      "learning_rate": 5.8828210252411e-06,
+      "loss": 0.5147,
+      "step": 641
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 6.432897045805533,
+      "learning_rate": 5.848928191769753e-06,
+      "loss": 0.5047,
+      "step": 642
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 6.06082150623531,
+      "learning_rate": 5.815092872433994e-06,
+      "loss": 0.5296,
+      "step": 643
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 7.760932921939526,
+      "learning_rate": 5.781315536030607e-06,
+      "loss": 0.5339,
+      "step": 644
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 6.0611679377908825,
+      "learning_rate": 5.74759665055302e-06,
+      "loss": 0.4212,
+      "step": 645
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 6.4875374667111565,
+      "learning_rate": 5.713936683184795e-06,
+      "loss": 0.5394,
+      "step": 646
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.850168689232539,
+      "learning_rate": 5.680336100293182e-06,
+      "loss": 0.4512,
+      "step": 647
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 9.444171703118819,
+      "learning_rate": 5.6467953674226505e-06,
+      "loss": 0.3942,
+      "step": 648
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 8.582206095093797,
+      "learning_rate": 5.613314949288409e-06,
+      "loss": 0.4582,
+      "step": 649
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 11.757390924671101,
+      "learning_rate": 5.579895309770009e-06,
+      "loss": 0.5435,
+      "step": 650
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 5.02260927644731,
+      "learning_rate": 5.546536911904896e-06,
+      "loss": 0.4719,
+      "step": 651
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 9.22536236771837,
+      "learning_rate": 5.513240217881992e-06,
+      "loss": 0.4668,
+      "step": 652
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 10.157555843180285,
+      "learning_rate": 5.4800056890353025e-06,
+      "loss": 0.5409,
+      "step": 653
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 12.323174404868091,
+      "learning_rate": 5.44683378583752e-06,
+      "loss": 0.5631,
+      "step": 654
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 8.185618668543789,
+      "learning_rate": 5.4137249678936265e-06,
+      "loss": 0.4284,
+      "step": 655
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 8.516855530899237,
+      "learning_rate": 5.380679693934569e-06,
+      "loss": 0.4221,
+      "step": 656
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 11.008555199751408,
+      "learning_rate": 5.347698421810861e-06,
+      "loss": 0.5924,
+      "step": 657
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 15.331344015670625,
+      "learning_rate": 5.31478160848625e-06,
+      "loss": 0.642,
+      "step": 658
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 10.747880055046071,
+      "learning_rate": 5.2819297100314e-06,
+      "loss": 0.5417,
+      "step": 659
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 15.482764407878273,
+      "learning_rate": 5.249143181617562e-06,
+      "loss": 0.63,
+      "step": 660
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 4.174929188215974,
+      "learning_rate": 5.216422477510267e-06,
+      "loss": 0.4989,
+      "step": 661
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 3.810136037355045,
+      "learning_rate": 5.183768051063036e-06,
+      "loss": 0.4893,
+      "step": 662
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 6.8676100081227744,
+      "learning_rate": 5.151180354711087e-06,
+      "loss": 0.5768,
+      "step": 663
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 10.044174931284044,
+      "learning_rate": 5.11865983996509e-06,
+      "loss": 0.4937,
+      "step": 664
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 5.716263993029701,
+      "learning_rate": 5.0862069574048956e-06,
+      "loss": 0.4918,
+      "step": 665
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 5.327730475446738,
+      "learning_rate": 5.053822156673276e-06,
+      "loss": 0.5112,
+      "step": 666
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 9.70768865377853,
+      "learning_rate": 5.021505886469733e-06,
+      "loss": 0.5682,
+      "step": 667
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 7.449826876587748,
+      "learning_rate": 4.989258594544252e-06,
+      "loss": 0.5914,
+      "step": 668
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 6.246979693998428,
+      "learning_rate": 4.957080727691107e-06,
+      "loss": 0.5022,
+      "step": 669
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 6.234169934241622,
+      "learning_rate": 4.9249727317426765e-06,
+      "loss": 0.5086,
+      "step": 670
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 8.795959544866905,
+      "learning_rate": 4.892935051563243e-06,
+      "loss": 0.5631,
+      "step": 671
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 6.585824557633543,
+      "learning_rate": 4.86096813104287e-06,
+      "loss": 0.5387,
+      "step": 672
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 9.412528445371356,
+      "learning_rate": 4.829072413091219e-06,
+      "loss": 0.5903,
+      "step": 673
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 4.934030658656449,
+      "learning_rate": 4.797248339631411e-06,
+      "loss": 0.5905,
+      "step": 674
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 7.225794451726558,
+      "learning_rate": 4.765496351593927e-06,
+      "loss": 0.595,
+      "step": 675
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 10.387771035955092,
+      "learning_rate": 4.733816888910483e-06,
+      "loss": 0.4901,
+      "step": 676
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 5.585297554337063,
+      "learning_rate": 4.7022103905079405e-06,
+      "loss": 0.3997,
+      "step": 677
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 5.2187286296388375,
+      "learning_rate": 4.67067729430222e-06,
+      "loss": 0.4856,
+      "step": 678
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.4821810592954296,
+      "learning_rate": 4.639218037192235e-06,
+      "loss": 0.4361,
+      "step": 679
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 3.959673908934417,
+      "learning_rate": 4.607833055053843e-06,
+      "loss": 0.3775,
+      "step": 680
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.129251200576683,
+      "learning_rate": 4.576522782733802e-06,
+      "loss": 0.4947,
+      "step": 681
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.8501294404892,
+      "learning_rate": 4.545287654043751e-06,
+      "loss": 0.3936,
+      "step": 682
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.318966235176163,
+      "learning_rate": 4.514128101754183e-06,
+      "loss": 0.4649,
+      "step": 683
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 7.407733161742192,
+      "learning_rate": 4.483044557588473e-06,
+      "loss": 0.5102,
+      "step": 684
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 4.965311089759532,
+      "learning_rate": 4.45203745221688e-06,
+      "loss": 0.457,
+      "step": 685
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 7.5517122869625135,
+      "learning_rate": 4.4211072152505866e-06,
+      "loss": 0.436,
+      "step": 686
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.082245336395077,
+      "learning_rate": 4.3902542752357415e-06,
+      "loss": 0.4936,
+      "step": 687
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 9.843843848674208,
+      "learning_rate": 4.359479059647527e-06,
+      "loss": 0.4336,
+      "step": 688
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 5.231213013499434,
+      "learning_rate": 4.3287819948842334e-06,
+      "loss": 0.4247,
+      "step": 689
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 6.305006534689214,
+      "learning_rate": 4.2981635062613556e-06,
+      "loss": 0.4826,
+      "step": 690
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 5.533371229509814,
+      "learning_rate": 4.267624018005686e-06,
+      "loss": 0.4353,
+      "step": 691
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 7.108534233713205,
+      "learning_rate": 4.237163953249453e-06,
+      "loss": 0.5991,
+      "step": 692
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.76045650818351,
+      "learning_rate": 4.206783734024463e-06,
+      "loss": 0.4514,
+      "step": 693
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 8.886091309499616,
+      "learning_rate": 4.176483781256224e-06,
+      "loss": 0.4023,
+      "step": 694
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 7.524320382507302,
+      "learning_rate": 4.1462645147581456e-06,
+      "loss": 0.4514,
+      "step": 695
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 7.482440196178911,
+      "learning_rate": 4.116126353225703e-06,
+      "loss": 0.5668,
+      "step": 696
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 9.764506225206542,
+      "learning_rate": 4.086069714230646e-06,
+      "loss": 0.539,
+      "step": 697
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 8.336719846814763,
+      "learning_rate": 4.056095014215208e-06,
+      "loss": 0.6426,
+      "step": 698
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 4.148296296213888,
+      "learning_rate": 4.0262026684863295e-06,
+      "loss": 0.3376,
+      "step": 699
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 9.203116559209336,
+      "learning_rate": 3.99639309120992e-06,
+      "loss": 0.5199,
+      "step": 700
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 4.576118300820756,
+      "learning_rate": 3.96666669540512e-06,
+      "loss": 0.3731,
+      "step": 701
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 9.971893856018951,
+      "learning_rate": 3.937023892938549e-06,
+      "loss": 0.591,
+      "step": 702
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 5.809308796580302,
+      "learning_rate": 3.907465094518636e-06,
+      "loss": 0.5063,
+      "step": 703
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 5.798443674221982,
+      "learning_rate": 3.877990709689905e-06,
+      "loss": 0.462,
+      "step": 704
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 9.114741488810273,
+      "learning_rate": 3.8486011468273145e-06,
+      "loss": 0.5972,
+      "step": 705
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 10.35491156365141,
+      "learning_rate": 3.819296813130588e-06,
+      "loss": 0.46,
+      "step": 706
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 8.13256163586952,
+      "learning_rate": 3.790078114618586e-06,
+      "loss": 0.46,
+      "step": 707
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 8.953602041543364,
+      "learning_rate": 3.760945456123656e-06,
+      "loss": 0.5296,
+      "step": 708
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 6.830059498599811,
+      "learning_rate": 3.731899241286061e-06,
+      "loss": 0.4798,
+      "step": 709
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 8.333772044542284,
+      "learning_rate": 3.702939872548348e-06,
+      "loss": 0.5452,
+      "step": 710
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 7.7830518181331785,
+      "learning_rate": 3.6740677511497958e-06,
+      "loss": 0.52,
+      "step": 711
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 9.48244358766429,
+      "learning_rate": 3.6452832771208524e-06,
+      "loss": 0.4604,
+      "step": 712
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 5.325755301496669,
+      "learning_rate": 3.616586849277587e-06,
+      "loss": 0.4801,
+      "step": 713
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 7.883051056212922,
+      "learning_rate": 3.587978865216165e-06,
+      "loss": 0.4396,
+      "step": 714
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 7.231451053733015,
+      "learning_rate": 3.559459721307349e-06,
+      "loss": 0.5395,
+      "step": 715
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 7.736510298846521,
+      "learning_rate": 3.5310298126909814e-06,
+      "loss": 0.4596,
+      "step": 716
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 5.986545004035281,
+      "learning_rate": 3.5026895332705504e-06,
+      "loss": 0.4593,
+      "step": 717
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.438626475367621,
+      "learning_rate": 3.474439275707698e-06,
+      "loss": 0.397,
+      "step": 718
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 9.623477478063533,
+      "learning_rate": 3.4462794314167846e-06,
+      "loss": 0.509,
+      "step": 719
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 9.394298451398283,
+      "learning_rate": 3.418210390559481e-06,
+      "loss": 0.4663,
+      "step": 720
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 9.315469015310178,
+      "learning_rate": 3.390232542039352e-06,
+      "loss": 0.5065,
+      "step": 721
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 6.092259066777256,
+      "learning_rate": 3.3623462734964697e-06,
+      "loss": 0.4477,
+      "step": 722
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.319138234974373,
+      "learning_rate": 3.3345519713020445e-06,
+      "loss": 0.3925,
+      "step": 723
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 6.358450139297925,
+      "learning_rate": 3.3068500205530596e-06,
+      "loss": 0.4679,
+      "step": 724
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 7.701373209390687,
+      "learning_rate": 3.2792408050669634e-06,
+      "loss": 0.409,
+      "step": 725
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 9.55351492544124,
+      "learning_rate": 3.2517247073763237e-06,
+      "loss": 0.4771,
+      "step": 726
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 4.7780971794647025,
+      "learning_rate": 3.2243021087235336e-06,
+      "loss": 0.4689,
+      "step": 727
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 7.114890807561552,
+      "learning_rate": 3.196973389055542e-06,
+      "loss": 0.4797,
+      "step": 728
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 7.072041698936169,
+      "learning_rate": 3.16973892701858e-06,
+      "loss": 0.5768,
+      "step": 729
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.961275677943624,
+      "learning_rate": 3.142599099952912e-06,
+      "loss": 0.3881,
+      "step": 730
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.06251591074102,
+      "learning_rate": 3.115554283887614e-06,
+      "loss": 0.375,
+      "step": 731
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 7.326058092329529,
+      "learning_rate": 3.088604853535361e-06,
+      "loss": 0.5388,
+      "step": 732
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.274081659887408,
+      "learning_rate": 3.0617511822872337e-06,
+      "loss": 0.4742,
+      "step": 733
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 4.793423500693722,
+      "learning_rate": 3.03499364220755e-06,
+      "loss": 0.4056,
+      "step": 734
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 5.269835865291424,
+      "learning_rate": 3.0083326040286977e-06,
+      "loss": 0.4332,
+      "step": 735
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 6.413197977206429,
+      "learning_rate": 2.9817684371460155e-06,
+      "loss": 0.5452,
+      "step": 736
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 8.581131926197953,
+      "learning_rate": 2.9553015096126638e-06,
+      "loss": 0.4393,
+      "step": 737
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 4.5160115188573045,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.4465,
+      "step": 738
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 5.2592392037223235,
+      "learning_rate": 2.902660838065131e-06,
+      "loss": 0.5082,
+      "step": 739
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 7.05269405170704,
+      "learning_rate": 2.8764878234005876e-06,
+      "loss": 0.372,
+      "step": 740
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 5.274091473047936,
+      "learning_rate": 2.8504135067745463e-06,
+      "loss": 0.4121,
+      "step": 741
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 7.3781029800169655,
+      "learning_rate": 2.8244382494531675e-06,
+      "loss": 0.4067,
+      "step": 742
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 8.32474239381458,
+      "learning_rate": 2.798562411330126e-06,
+      "loss": 0.3679,
+      "step": 743
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 12.450031335391019,
+      "learning_rate": 2.772786350921608e-06,
+      "loss": 0.5521,
+      "step": 744
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 7.249666510251534,
+      "learning_rate": 2.7471104253613645e-06,
+      "loss": 0.3938,
+      "step": 745
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 7.2873098383670065,
+      "learning_rate": 2.721534990395752e-06,
+      "loss": 0.4884,
+      "step": 746
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 5.438711289922824,
+      "learning_rate": 2.6960604003788014e-06,
+      "loss": 0.364,
+      "step": 747
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 9.79198662580905,
+      "learning_rate": 2.670687008267313e-06,
+      "loss": 0.3911,
+      "step": 748
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 11.420021502949318,
+      "learning_rate": 2.6454151656159666e-06,
+      "loss": 0.4454,
+      "step": 749
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 8.204149460627178,
+      "learning_rate": 2.6202452225724463e-06,
+      "loss": 0.4276,
+      "step": 750
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 10.999756173661007,
+      "learning_rate": 2.5951775278725956e-06,
+      "loss": 0.5623,
+      "step": 751
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 8.568105116698392,
+      "learning_rate": 2.5702124288355723e-06,
+      "loss": 0.3292,
+      "step": 752
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 5.56439861064704,
+      "learning_rate": 2.545350271359055e-06,
+      "loss": 0.3879,
+      "step": 753
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 8.106162701050645,
+      "learning_rate": 2.520591399914444e-06,
+      "loss": 0.364,
+      "step": 754
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 10.29859188596712,
+      "learning_rate": 2.495936157542074e-06,
+      "loss": 0.4791,
+      "step": 755
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 7.657220905887133,
+      "learning_rate": 2.471384885846482e-06,
+      "loss": 0.4902,
+      "step": 756
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 6.191995160106613,
+      "learning_rate": 2.4469379249916614e-06,
+      "loss": 0.4585,
+      "step": 757
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 7.623947016373425,
+      "learning_rate": 2.422595613696356e-06,
+      "loss": 0.42,
+      "step": 758
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 5.456113345068306,
+      "learning_rate": 2.3983582892293642e-06,
+      "loss": 0.3625,
+      "step": 759
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 9.722994068985187,
+      "learning_rate": 2.3742262874048583e-06,
+      "loss": 0.5455,
+      "step": 760
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 10.210377059221097,
+      "learning_rate": 2.3501999425777433e-06,
+      "loss": 0.4177,
+      "step": 761
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 9.986453075138304,
+      "learning_rate": 2.326279587639031e-06,
+      "loss": 0.4519,
+      "step": 762
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 7.444568050178271,
+      "learning_rate": 2.3024655540111984e-06,
+      "loss": 0.4707,
+      "step": 763
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 6.882963377525352,
+      "learning_rate": 2.2787581716436257e-06,
+      "loss": 0.4142,
+      "step": 764
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 5.621837753619931,
+      "learning_rate": 2.255157769008011e-06,
+      "loss": 0.3837,
+      "step": 765
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 8.979013360736435,
+      "learning_rate": 2.2316646730938197e-06,
+      "loss": 0.4511,
+      "step": 766
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 9.387346843179332,
+      "learning_rate": 2.2082792094037585e-06,
+      "loss": 0.3837,
+      "step": 767
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 11.152697788037019,
+      "learning_rate": 2.185001701949263e-06,
+      "loss": 0.3815,
+      "step": 768
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 7.316279559405673,
+      "learning_rate": 2.1618324732459993e-06,
+      "loss": 0.5763,
+      "step": 769
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 4.60864460318581,
+      "learning_rate": 2.138771844309423e-06,
+      "loss": 0.3784,
+      "step": 770
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 8.893733826091104,
+      "learning_rate": 2.1158201346502927e-06,
+      "loss": 0.3993,
+      "step": 771
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 8.857140540604258,
+      "learning_rate": 2.092977662270275e-06,
+      "loss": 0.3334,
+      "step": 772
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 9.000582629375513,
+      "learning_rate": 2.0702447436575223e-06,
+      "loss": 0.5377,
+      "step": 773
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 11.407276883996886,
+      "learning_rate": 2.0476216937822967e-06,
+      "loss": 0.4657,
+      "step": 774
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 9.343576283040107,
+      "learning_rate": 2.0251088260925967e-06,
+      "loss": 0.5038,
+      "step": 775
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.742436296561597,
+      "learning_rate": 2.002706452509824e-06,
+      "loss": 0.462,
+      "step": 776
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 9.651440289066235,
+      "learning_rate": 1.9804148834244465e-06,
+      "loss": 0.4588,
+      "step": 777
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 5.795036174327087,
+      "learning_rate": 1.9582344276917233e-06,
+      "loss": 0.5565,
+      "step": 778
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 7.475417877764932,
+      "learning_rate": 1.9361653926274016e-06,
+      "loss": 0.3644,
+      "step": 779
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 7.2305893277419235,
+      "learning_rate": 1.9142080840034647e-06,
+      "loss": 0.451,
+      "step": 780
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 7.9485644539877365,
+      "learning_rate": 1.8923628060439037e-06,
+      "loss": 0.5415,
+      "step": 781
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 7.339019736495735,
+      "learning_rate": 1.8706298614204942e-06,
+      "loss": 0.4258,
+      "step": 782
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 6.229226290130147,
+      "learning_rate": 1.8490095512486072e-06,
+      "loss": 0.372,
+      "step": 783
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 8.318686997660855,
+      "learning_rate": 1.8275021750830368e-06,
+      "loss": 0.4337,
+      "step": 784
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 11.383779868869713,
+      "learning_rate": 1.8061080309138379e-06,
+      "loss": 0.5517,
+      "step": 785
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 4.319009051803344,
+      "learning_rate": 1.7848274151622236e-06,
+      "loss": 0.4032,
+      "step": 786
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 6.434125961202376,
+      "learning_rate": 1.7636606226764353e-06,
+      "loss": 0.4165,
+      "step": 787
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 5.041886706286037,
+      "learning_rate": 1.742607946727658e-06,
+      "loss": 0.4621,
+      "step": 788
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 6.183213056857235,
+      "learning_rate": 1.7216696790059718e-06,
+      "loss": 0.423,
+      "step": 789
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 6.373680744920821,
+      "learning_rate": 1.7008461096162999e-06,
+      "loss": 0.5464,
+      "step": 790
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 11.493069464165945,
+      "learning_rate": 1.6801375270743925e-06,
+      "loss": 0.4518,
+      "step": 791
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 9.805183898852514,
+      "learning_rate": 1.6595442183028266e-06,
+      "loss": 0.4836,
+      "step": 792
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 8.05887231146552,
+      "learning_rate": 1.6390664686270342e-06,
+      "loss": 0.3711,
+      "step": 793
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 6.975427069143952,
+      "learning_rate": 1.6187045617713459e-06,
+      "loss": 0.4149,
+      "step": 794
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 7.272454723310934,
+      "learning_rate": 1.5984587798550633e-06,
+      "loss": 0.4357,
+      "step": 795
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 6.941951052506783,
+      "learning_rate": 1.578329403388541e-06,
+      "loss": 0.3835,
+      "step": 796
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.800438948527679,
+      "learning_rate": 1.5583167112693153e-06,
+      "loss": 0.4408,
+      "step": 797
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.3480222782611175,
+      "learning_rate": 1.5384209807782291e-06,
+      "loss": 0.4553,
+      "step": 798
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 5.909917855292465,
+      "learning_rate": 1.518642487575591e-06,
+      "loss": 0.497,
+      "step": 799
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 7.683087880393552,
+      "learning_rate": 1.49898150569736e-06,
+      "loss": 0.4535,
+      "step": 800
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 7.446327856244201,
+      "learning_rate": 1.4794383075513453e-06,
+      "loss": 0.3873,
+      "step": 801
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 6.360170134685136,
+      "learning_rate": 1.460013163913433e-06,
+      "loss": 0.5037,
+      "step": 802
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.053653216562189,
+      "learning_rate": 1.4407063439238333e-06,
+      "loss": 0.4954,
+      "step": 803
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.2483525961998945,
+      "learning_rate": 1.421518115083359e-06,
+      "loss": 0.3751,
+      "step": 804
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 6.345969600489226,
+      "learning_rate": 1.4024487432497013e-06,
+      "loss": 0.4143,
+      "step": 805
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 5.646370329937249,
+      "learning_rate": 1.3834984926337657e-06,
+      "loss": 0.3376,
+      "step": 806
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 10.69680196625431,
+      "learning_rate": 1.36466762579601e-06,
+      "loss": 0.5439,
+      "step": 807
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 9.34187811581496,
+      "learning_rate": 1.3459564036427841e-06,
+      "loss": 0.4761,
+      "step": 808
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 8.322291418258063,
+      "learning_rate": 1.3273650854227438e-06,
+      "loss": 0.4699,
+      "step": 809
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 6.089668194777211,
+      "learning_rate": 1.3088939287232383e-06,
+      "loss": 0.4135,
+      "step": 810
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 6.349508519106518,
+      "learning_rate": 1.2905431894667552e-06,
+      "loss": 0.4429,
+      "step": 811
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 6.0067928128468955,
+      "learning_rate": 1.272313121907366e-06,
+      "loss": 0.5506,
+      "step": 812
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 5.057193603204914,
+      "learning_rate": 1.2542039786272008e-06,
+      "loss": 0.4029,
+      "step": 813
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 6.9264232438262,
+      "learning_rate": 1.236216010532959e-06,
+      "loss": 0.399,
+      "step": 814
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 8.111117092935682,
+      "learning_rate": 1.218349466852432e-06,
+      "loss": 0.3969,
+      "step": 815
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 12.617119783922481,
+      "learning_rate": 1.200604595131033e-06,
+      "loss": 0.3632,
+      "step": 816
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.997880390654607,
+      "learning_rate": 1.1829816412283912e-06,
+      "loss": 0.3727,
+      "step": 817
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 9.925641311100412,
+      "learning_rate": 1.1654808493149283e-06,
+      "loss": 0.4779,
+      "step": 818
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 10.259671670493303,
+      "learning_rate": 1.1481024618684821e-06,
+      "loss": 0.4792,
+      "step": 819
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 4.666467514007945,
+      "learning_rate": 1.1308467196709505e-06,
+      "loss": 0.3665,
+      "step": 820
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 5.8390766285836895,
+      "learning_rate": 1.1137138618049403e-06,
+      "loss": 0.4426,
+      "step": 821
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.079596026020399,
+      "learning_rate": 1.096704125650474e-06,
+      "loss": 0.3858,
+      "step": 822
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 6.355034616703483,
+      "learning_rate": 1.079817746881696e-06,
+      "loss": 0.4322,
+      "step": 823
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 4.964202739872203,
+      "learning_rate": 1.0630549594635896e-06,
+      "loss": 0.4246,
+      "step": 824
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 9.981500108734732,
+      "learning_rate": 1.0464159956487596e-06,
+      "loss": 0.4048,
+      "step": 825
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 5.314444267142196,
+      "learning_rate": 1.0299010859742009e-06,
+      "loss": 0.4226,
+      "step": 826
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 6.711068748283136,
+      "learning_rate": 1.013510459258108e-06,
+      "loss": 0.4626,
+      "step": 827
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 6.87746096070976,
+      "learning_rate": 9.972443425967026e-07,
+      "loss": 0.4762,
+      "step": 828
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 5.584995566992966,
+      "learning_rate": 9.811029613610913e-07,
+      "loss": 0.4638,
+      "step": 829
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 5.476555647757321,
+      "learning_rate": 9.650865391941345e-07,
+      "loss": 0.4174,
+      "step": 830
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.334567697843729,
+      "learning_rate": 9.491952980073604e-07,
+      "loss": 0.3264,
+      "step": 831
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 7.864733082511921,
+      "learning_rate": 9.334294579778813e-07,
+      "loss": 0.3613,
+      "step": 832
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 8.345283693272092,
+      "learning_rate": 9.177892375453413e-07,
+      "loss": 0.5469,
+      "step": 833
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 8.885940166363653,
+      "learning_rate": 9.022748534088988e-07,
+      "loss": 0.4455,
+      "step": 834
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 7.55591370496932,
+      "learning_rate": 8.86886520524216e-07,
+      "loss": 0.4356,
+      "step": 835
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 7.5174900881299465,
+      "learning_rate": 8.716244521004846e-07,
+      "loss": 0.4074,
+      "step": 836
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 7.226452727128449,
+      "learning_rate": 8.564888595974718e-07,
+      "loss": 0.473,
+      "step": 837
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 7.586424458860782,
+      "learning_rate": 8.414799527225826e-07,
+      "loss": 0.4757,
+      "step": 838
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 5.51882334743589,
+      "learning_rate": 8.265979394279732e-07,
+      "loss": 0.4272,
+      "step": 839
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 6.945131528688248,
+      "learning_rate": 8.118430259076493e-07,
+      "loss": 0.3003,
+      "step": 840
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 7.957372619224033,
+      "learning_rate": 7.972154165946155e-07,
+      "loss": 0.4485,
+      "step": 841
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 7.203632243202649,
+      "learning_rate": 7.827153141580501e-07,
+      "loss": 0.384,
+      "step": 842
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 5.973567270845765,
+      "learning_rate": 7.683429195004932e-07,
+      "loss": 0.415,
+      "step": 843
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 7.742846323788447,
+      "learning_rate": 7.540984317550548e-07,
+      "loss": 0.4261,
+      "step": 844
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 4.971511012508995,
+      "learning_rate": 7.399820482826692e-07,
+      "loss": 0.3073,
+      "step": 845
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 7.417662373162329,
+      "learning_rate": 7.25993964669347e-07,
+      "loss": 0.4293,
+      "step": 846
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 8.474162845843916,
+      "learning_rate": 7.12134374723481e-07,
+      "loss": 0.5288,
+      "step": 847
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 6.74629233498178,
+      "learning_rate": 6.984034704731479e-07,
+      "loss": 0.379,
+      "step": 848
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 6.51287661591934,
+      "learning_rate": 6.848014421634497e-07,
+      "loss": 0.4254,
+      "step": 849
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 8.203119397161476,
+      "learning_rate": 6.71328478253881e-07,
+      "loss": 0.4818,
+      "step": 850
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 7.091765460672564,
+      "learning_rate": 6.579847654157234e-07,
+      "loss": 0.3937,
+      "step": 851
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.848337659332669,
+      "learning_rate": 6.447704885294437e-07,
+      "loss": 0.3223,
+      "step": 852
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 11.861790191542282,
+      "learning_rate": 6.316858306821449e-07,
+      "loss": 0.3955,
+      "step": 853
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.89823677782504,
+      "learning_rate": 6.18730973165027e-07,
+      "loss": 0.52,
+      "step": 854
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 7.297372429370729,
+      "learning_rate": 6.05906095470874e-07,
+      "loss": 0.5625,
+      "step": 855
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 8.018266432717773,
+      "learning_rate": 5.932113752915659e-07,
+      "loss": 0.5943,
+      "step": 856
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 5.441022930966068,
+      "learning_rate": 5.806469885156163e-07,
+      "loss": 0.3584,
+      "step": 857
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 7.923651656498159,
+      "learning_rate": 5.682131092257381e-07,
+      "loss": 0.4374,
+      "step": 858
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.401381730215806,
+      "learning_rate": 5.55909909696436e-07,
+      "loss": 0.4548,
+      "step": 859
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 9.323436877274784,
+      "learning_rate": 5.437375603916039e-07,
+      "loss": 0.3949,
+      "step": 860
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 7.6852951141865455,
+      "learning_rate": 5.316962299621808e-07,
+      "loss": 0.4357,
+      "step": 861
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 5.468965570424503,
+      "learning_rate": 5.19786085243803e-07,
+      "loss": 0.4446,
+      "step": 862
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 6.3812841292130855,
+      "learning_rate": 5.080072912544987e-07,
+      "loss": 0.4428,
+      "step": 863
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 10.852851737536733,
+      "learning_rate": 4.96360011192396e-07,
+      "loss": 0.4991,
+      "step": 864
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 5.7584503904522775,
+      "learning_rate": 4.848444064334679e-07,
+      "loss": 0.5114,
+      "step": 865
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 6.376038086977472,
+      "learning_rate": 4.734606365292871e-07,
+      "loss": 0.3782,
+      "step": 866
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 6.224991564333229,
+      "learning_rate": 4.6220885920483014e-07,
+      "loss": 0.3527,
+      "step": 867
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 5.203231071738429,
+      "learning_rate": 4.510892303562797e-07,
+      "loss": 0.3518,
+      "step": 868
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.884197350261294,
+      "learning_rate": 4.401019040488652e-07,
+      "loss": 0.4493,
+      "step": 869
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 8.438339641746209,
+      "learning_rate": 4.2924703251473355e-07,
+      "loss": 0.5192,
+      "step": 870
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 9.929683884140545,
+      "learning_rate": 4.1852476615083957e-07,
+      "loss": 0.4769,
+      "step": 871
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 7.444778653156325,
+      "learning_rate": 4.0793525351685905e-07,
+      "loss": 0.3915,
+      "step": 872
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.55575801178513,
+      "learning_rate": 3.974786413331311e-07,
+      "loss": 0.3861,
+      "step": 873
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.154553375207641,
+      "learning_rate": 3.8715507447862455e-07,
+      "loss": 0.3713,
+      "step": 874
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 9.250721341718682,
+      "learning_rate": 3.7696469598893727e-07,
+      "loss": 0.4575,
+      "step": 875
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.654967363493775,
+      "learning_rate": 3.669076470543054e-07,
+      "loss": 0.4686,
+      "step": 876
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 5.710919914401177,
+      "learning_rate": 3.569840670176483e-07,
+      "loss": 0.4339,
+      "step": 877
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 6.7079973379018085,
+      "learning_rate": 3.4719409337264473e-07,
+      "loss": 0.4238,
+      "step": 878
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 10.924750798758199,
+      "learning_rate": 3.3753786176182303e-07,
+      "loss": 0.4977,
+      "step": 879
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 4.9608766209441315,
+      "learning_rate": 3.2801550597468033e-07,
+      "loss": 0.2732,
+      "step": 880
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 7.448404330000091,
+      "learning_rate": 3.186271579458333e-07,
+      "loss": 0.4483,
+      "step": 881
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.022846446335248,
+      "learning_rate": 3.093729477531826e-07,
+      "loss": 0.4363,
+      "step": 882
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 7.140771043711367,
+      "learning_rate": 3.002530036161222e-07,
+      "loss": 0.4566,
+      "step": 883
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 5.57307898414647,
+      "learning_rate": 2.912674518937553e-07,
+      "loss": 0.3565,
+      "step": 884
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 7.503093403901029,
+      "learning_rate": 2.824164170831389e-07,
+      "loss": 0.3495,
+      "step": 885
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 8.266577699990709,
+      "learning_rate": 2.737000218175712e-07,
+      "loss": 0.458,
+      "step": 886
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 8.007643327554275,
+      "learning_rate": 2.651183868648821e-07,
+      "loss": 0.3738,
+      "step": 887
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 6.8532106769801215,
+      "learning_rate": 2.566716311257655e-07,
+      "loss": 0.4602,
+      "step": 888
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 8.226940714871725,
+      "learning_rate": 2.483598716321289e-07,
+      "loss": 0.3939,
+      "step": 889
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 7.35412581775661,
+      "learning_rate": 2.4018322354547483e-07,
+      "loss": 0.4912,
+      "step": 890
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.189789718764145,
+      "learning_rate": 2.321418001553022e-07,
+      "loss": 0.3516,
+      "step": 891
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.643929108838239,
+      "learning_rate": 2.2423571287753853e-07,
+      "loss": 0.3811,
+      "step": 892
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.764646780403799,
+      "learning_rate": 2.1646507125299587e-07,
+      "loss": 0.4442,
+      "step": 893
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.42774429745915,
+      "learning_rate": 2.0882998294585177e-07,
+      "loss": 0.3746,
+      "step": 894
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 6.171794557047173,
+      "learning_rate": 2.013305537421606e-07,
+      "loss": 0.447,
+      "step": 895
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 5.130410450658874,
+      "learning_rate": 1.9396688754838355e-07,
+      "loss": 0.3469,
+      "step": 896
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 7.39720002220979,
+      "learning_rate": 1.867390863899543e-07,
+      "loss": 0.3783,
+      "step": 897
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 7.90865739350042,
+      "learning_rate": 1.796472504098612e-07,
+      "loss": 0.4422,
+      "step": 898
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.815959678060295,
+      "learning_rate": 1.726914778672606e-07,
+      "loss": 0.378,
+      "step": 899
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.081919878256831,
+      "learning_rate": 1.6587186513611685e-07,
+      "loss": 0.3205,
+      "step": 900
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 6.567585188289477,
+      "learning_rate": 1.5918850670386677e-07,
+      "loss": 0.44,
+      "step": 901
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 7.533656800692689,
+      "learning_rate": 1.5264149517010718e-07,
+      "loss": 0.4362,
+      "step": 902
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.4628885270524865,
+      "learning_rate": 1.4623092124531613e-07,
+      "loss": 0.4497,
+      "step": 903
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 5.858826602545033,
+      "learning_rate": 1.399568737495971e-07,
+      "loss": 0.43,
+      "step": 904
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 8.572231557747521,
+      "learning_rate": 1.3381943961144118e-07,
+      "loss": 0.3925,
+      "step": 905
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 5.873954819161939,
+      "learning_rate": 1.2781870386653016e-07,
+      "loss": 0.3683,
+      "step": 906
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 4.581416916538669,
+      "learning_rate": 1.2195474965655652e-07,
+      "loss": 0.2704,
+      "step": 907
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 6.965146521509025,
+      "learning_rate": 1.1622765822806636e-07,
+      "loss": 0.3669,
+      "step": 908
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 5.699569364212127,
+      "learning_rate": 1.1063750893134273e-07,
+      "loss": 0.4125,
+      "step": 909
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 7.0422218435015855,
+      "learning_rate": 1.0518437921929747e-07,
+      "loss": 0.3156,
+      "step": 910
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 9.329828931384606,
+      "learning_rate": 9.986834464640328e-08,
+      "loss": 0.4834,
+      "step": 911
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 5.211053129225831,
+      "learning_rate": 9.468947886764557e-08,
+      "loss": 0.4188,
+      "step": 912
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 5.70280316894895,
+      "learning_rate": 8.964785363750228e-08,
+      "loss": 0.3474,
+      "step": 913
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 5.824204755244337,
+      "learning_rate": 8.474353880894904e-08,
+      "loss": 0.2739,
+      "step": 914
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 7.610004827598348,
+      "learning_rate": 7.997660233249105e-08,
+      "loss": 0.4407,
+      "step": 915
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 6.558806283300632,
+      "learning_rate": 7.534711025522168e-08,
+      "loss": 0.3602,
+      "step": 916
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 6.692057122360635,
+      "learning_rate": 7.08551267199098e-08,
+      "loss": 0.4439,
+      "step": 917
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 6.4300501195085555,
+      "learning_rate": 6.650071396410718e-08,
+      "loss": 0.428,
+      "step": 918
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 6.994677903035368,
+      "learning_rate": 6.22839323192892e-08,
+      "loss": 0.517,
+      "step": 919
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 7.2674316715963325,
+      "learning_rate": 5.820484021001993e-08,
+      "loss": 0.3041,
+      "step": 920
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 8.22191067382583,
+      "learning_rate": 5.426349415313503e-08,
+      "loss": 0.4481,
+      "step": 921
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 5.065088639546076,
+      "learning_rate": 5.0459948756967914e-08,
+      "loss": 0.3333,
+      "step": 922
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 8.073376899592944,
+      "learning_rate": 4.679425672059035e-08,
+      "loss": 0.4493,
+      "step": 923
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 6.784518754930513,
+      "learning_rate": 4.326646883307972e-08,
+      "loss": 0.4454,
+      "step": 924
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 8.57495143340707,
+      "learning_rate": 3.987663397281627e-08,
+      "loss": 0.4892,
+      "step": 925
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.72551957891776,
+      "learning_rate": 3.662479910681027e-08,
+      "loss": 0.4259,
+      "step": 926
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 4.571968724650214,
+      "learning_rate": 3.3511009290042585e-08,
+      "loss": 0.364,
+      "step": 927
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 7.563468099974791,
+      "learning_rate": 3.0535307664848466e-08,
+      "loss": 0.4774,
+      "step": 928
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 6.8428676776213555,
+      "learning_rate": 2.7697735460316954e-08,
+      "loss": 0.4322,
+      "step": 929
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 3.9770591980462786,
+      "learning_rate": 2.499833199171908e-08,
+      "loss": 0.3225,
+      "step": 930
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.666940466332362,
+      "learning_rate": 2.2437134659962777e-08,
+      "loss": 0.4234,
+      "step": 931
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.153046102603702,
+      "learning_rate": 2.0014178951076602e-08,
+      "loss": 0.471,
+      "step": 932
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 5.709388029240762,
+      "learning_rate": 1.7729498435716808e-08,
+      "loss": 0.4391,
+      "step": 933
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 5.983125443983794,
+      "learning_rate": 1.5583124768701052e-08,
+      "loss": 0.483,
+      "step": 934
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 6.445238846878224,
+      "learning_rate": 1.3575087688570965e-08,
+      "loss": 0.3558,
+      "step": 935
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 6.816452865113282,
+      "learning_rate": 1.1705415017183585e-08,
+      "loss": 0.413,
+      "step": 936
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 5.457722190574727,
+      "learning_rate": 9.974132659319457e-09,
+      "loss": 0.3282,
+      "step": 937
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 6.969886154201284,
+      "learning_rate": 8.381264602327354e-09,
+      "loss": 0.438,
+      "step": 938
+    },
+    {
+      "epoch": 1.0,
+      "step": 938,
+      "total_flos": 373627158528000.0,
+      "train_loss": 0.7472518730773601,
+      "train_runtime": 19046.2174,
+      "train_samples_per_second": 6.306,
+      "train_steps_per_second": 0.049
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 938,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 200,
+  "total_flos": 373627158528000.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}