diff --git "a/ajaymin28/vl-sg/trainer_state.json" "b/ajaymin28/vl-sg/trainer_state.json"
--- "a/ajaymin28/vl-sg/trainer_state.json"
+++ "b/ajaymin28/vl-sg/trainer_state.json"
@@ -1,9 +1,9 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.9881422924901185,
+  "epoch": 2.0,
   "eval_steps": 500,
-  "global_step": 1000,
+  "global_step": 2024,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -7007,6 +7007,7183 @@
       "learning_rate": 5.339823642419174e-06,
       "loss": 0.0542,
       "step": 1000
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8943229287072868,
+      "learning_rate": 5.331839694520436e-06,
+      "loss": 0.0306,
+      "step": 1001
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.0061767635442918,
+      "learning_rate": 5.323854896683422e-06,
+      "loss": 0.0305,
+      "step": 1002
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.3117131384625127,
+      "learning_rate": 5.315869269359527e-06,
+      "loss": 0.0483,
+      "step": 1003
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.483692159272879,
+      "learning_rate": 5.307882833002271e-06,
+      "loss": 0.0306,
+      "step": 1004
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.4928013172540946,
+      "learning_rate": 5.299895608067253e-06,
+      "loss": 0.0272,
+      "step": 1005
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.9350981952793932,
+      "learning_rate": 5.291907615012081e-06,
+      "loss": 0.0257,
+      "step": 1006
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.1880304932342811,
+      "learning_rate": 5.283918874296338e-06,
+      "loss": 0.0478,
+      "step": 1007
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.567658791017144,
+      "learning_rate": 5.2759294063815205e-06,
+      "loss": 0.0207,
+      "step": 1008
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5878617256713423,
+      "learning_rate": 5.267939231730985e-06,
+      "loss": 0.0259,
+      "step": 1009
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6296036095600541,
+      "learning_rate": 5.259948370809902e-06,
+      "loss": 0.03,
+      "step": 1010
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.086582723137742,
+      "learning_rate": 5.251956844085198e-06,
+      "loss": 0.0315,
+      "step": 1011
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7493884416834921,
+      "learning_rate": 5.243964672025502e-06,
+      "loss": 0.027,
+      "step": 1012
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.5755763721876781,
+      "learning_rate": 5.235971875101101e-06,
+      "loss": 0.024,
+      "step": 1013
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.0336876559829822,
+      "learning_rate": 5.22797847378388e-06,
+      "loss": 0.0324,
+      "step": 1014
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.7680003907100028,
+      "learning_rate": 5.219984488547269e-06,
+      "loss": 0.0341,
+      "step": 1015
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.8226456680704546,
+      "learning_rate": 5.2119899398662e-06,
+      "loss": 0.0324,
+      "step": 1016
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.6982266202238494,
+      "learning_rate": 5.203994848217043e-06,
+      "loss": 0.0286,
+      "step": 1017
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.7139453038989596,
+      "learning_rate": 5.195999234077561e-06,
+      "loss": 0.0263,
+      "step": 1018
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.7446444911541764,
+      "learning_rate": 5.188003117926854e-06,
+      "loss": 0.0177,
+      "step": 1019
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.3514532648365165,
+      "learning_rate": 5.1800065202453095e-06,
+      "loss": 0.0361,
+      "step": 1020
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.9721865980606375,
+      "learning_rate": 5.1720094615145455e-06,
+      "loss": 0.0389,
+      "step": 1021
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.3703407080177544,
+      "learning_rate": 5.164011962217366e-06,
+      "loss": 0.019,
+      "step": 1022
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.9599534838359112,
+      "learning_rate": 5.156014042837696e-06,
+      "loss": 0.0285,
+      "step": 1023
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.6398162968662446,
+      "learning_rate": 5.148015723860543e-06,
+      "loss": 0.0251,
+      "step": 1024
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.6225810158973047,
+      "learning_rate": 5.140017025771936e-06,
+      "loss": 0.0332,
+      "step": 1025
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.7141483456674433,
+      "learning_rate": 5.1320179690588735e-06,
+      "loss": 0.0286,
+      "step": 1026
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.9907387551938683,
+      "learning_rate": 5.124018574209272e-06,
+      "loss": 0.0432,
+      "step": 1027
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.7904414734004622,
+      "learning_rate": 5.116018861711919e-06,
+      "loss": 0.0287,
+      "step": 1028
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.3076855298975811,
+      "learning_rate": 5.108018852056411e-06,
+      "loss": 0.0428,
+      "step": 1029
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.47849407068874555,
+      "learning_rate": 5.100018565733106e-06,
+      "loss": 0.0245,
+      "step": 1030
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.2447035809415234,
+      "learning_rate": 5.092018023233072e-06,
+      "loss": 0.0332,
+      "step": 1031
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.8381997558957843,
+      "learning_rate": 5.084017245048034e-06,
+      "loss": 0.0321,
+      "step": 1032
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.7945475267321247,
+      "learning_rate": 5.0760162516703156e-06,
+      "loss": 0.0382,
+      "step": 1033
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.8671733382239873,
+      "learning_rate": 5.068015063592799e-06,
+      "loss": 0.0372,
+      "step": 1034
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.8604675588518298,
+      "learning_rate": 5.06001370130886e-06,
+      "loss": 0.0178,
+      "step": 1035
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.0987612417558899,
+      "learning_rate": 5.052012185312322e-06,
+      "loss": 0.0402,
+      "step": 1036
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.107296934615035,
+      "learning_rate": 5.044010536097402e-06,
+      "loss": 0.03,
+      "step": 1037
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.8614135705236582,
+      "learning_rate": 5.036008774158658e-06,
+      "loss": 0.034,
+      "step": 1038
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.7233307057992016,
+      "learning_rate": 5.028006919990936e-06,
+      "loss": 0.027,
+      "step": 1039
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.6366263075262523,
+      "learning_rate": 5.0200049940893225e-06,
+      "loss": 0.03,
+      "step": 1040
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5525916393083419,
+      "learning_rate": 5.012003016949082e-06,
+      "loss": 0.0271,
+      "step": 1041
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.8220356610097129,
+      "learning_rate": 5.004001009065611e-06,
+      "loss": 0.031,
+      "step": 1042
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.8725303219826496,
+      "learning_rate": 4.99599899093439e-06,
+      "loss": 0.0317,
+      "step": 1043
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.969863129435251,
+      "learning_rate": 4.98799698305092e-06,
+      "loss": 0.0257,
+      "step": 1044
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.7438173977745536,
+      "learning_rate": 4.979995005910679e-06,
+      "loss": 0.0189,
+      "step": 1045
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.3051006180712026,
+      "learning_rate": 4.971993080009065e-06,
+      "loss": 0.0304,
+      "step": 1046
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5562679937676483,
+      "learning_rate": 4.9639912258413435e-06,
+      "loss": 0.0237,
+      "step": 1047
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.8810611430495988,
+      "learning_rate": 4.955989463902599e-06,
+      "loss": 0.0297,
+      "step": 1048
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.7041819608996103,
+      "learning_rate": 4.94798781468768e-06,
+      "loss": 0.0306,
+      "step": 1049
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.850027647311031,
+      "learning_rate": 4.939986298691141e-06,
+      "loss": 0.0218,
+      "step": 1050
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 2.8282525428235044,
+      "learning_rate": 4.931984936407202e-06,
+      "loss": 0.0393,
+      "step": 1051
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 2.018508686889688,
+      "learning_rate": 4.923983748329685e-06,
+      "loss": 0.0312,
+      "step": 1052
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 2.6184868153222927,
+      "learning_rate": 4.9159827549519676e-06,
+      "loss": 0.0345,
+      "step": 1053
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.964887841469877,
+      "learning_rate": 4.907981976766928e-06,
+      "loss": 0.0469,
+      "step": 1054
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.7869754204093123,
+      "learning_rate": 4.899981434266895e-06,
+      "loss": 0.0314,
+      "step": 1055
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.514845941566159,
+      "learning_rate": 4.891981147943589e-06,
+      "loss": 0.0255,
+      "step": 1056
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.8788715161807736,
+      "learning_rate": 4.883981138288081e-06,
+      "loss": 0.0244,
+      "step": 1057
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.046570067083794,
+      "learning_rate": 4.8759814257907275e-06,
+      "loss": 0.0316,
+      "step": 1058
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.0154582399674494,
+      "learning_rate": 4.867982030941127e-06,
+      "loss": 0.0345,
+      "step": 1059
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.3103756035719953,
+      "learning_rate": 4.859982974228065e-06,
+      "loss": 0.0248,
+      "step": 1060
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 2.23368046234692,
+      "learning_rate": 4.851984276139458e-06,
+      "loss": 0.0395,
+      "step": 1061
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 2.1098419076483146,
+      "learning_rate": 4.843985957162304e-06,
+      "loss": 0.0276,
+      "step": 1062
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.942607118035288,
+      "learning_rate": 4.835988037782635e-06,
+      "loss": 0.0533,
+      "step": 1063
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.6421131635296107,
+      "learning_rate": 4.827990538485456e-06,
+      "loss": 0.0239,
+      "step": 1064
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.3053339694382833,
+      "learning_rate": 4.819993479754693e-06,
+      "loss": 0.0353,
+      "step": 1065
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.9986906730806447,
+      "learning_rate": 4.811996882073148e-06,
+      "loss": 0.0265,
+      "step": 1066
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.8181143387578583,
+      "learning_rate": 4.804000765922441e-06,
+      "loss": 0.0216,
+      "step": 1067
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.502546600151344,
+      "learning_rate": 4.796005151782958e-06,
+      "loss": 0.0445,
+      "step": 1068
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.7044947654375584,
+      "learning_rate": 4.788010060133802e-06,
+      "loss": 0.0251,
+      "step": 1069
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.6013429646716092,
+      "learning_rate": 4.780015511452732e-06,
+      "loss": 0.0256,
+      "step": 1070
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.5427439673220715,
+      "learning_rate": 4.772021526216123e-06,
+      "loss": 0.0263,
+      "step": 1071
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.769644548462023,
+      "learning_rate": 4.764028124898901e-06,
+      "loss": 0.0335,
+      "step": 1072
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.7447494663693941,
+      "learning_rate": 4.756035327974499e-06,
+      "loss": 0.0223,
+      "step": 1073
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.7204796166780826,
+      "learning_rate": 4.748043155914804e-06,
+      "loss": 0.03,
+      "step": 1074
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.0177867742746178,
+      "learning_rate": 4.740051629190099e-06,
+      "loss": 0.0341,
+      "step": 1075
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.7421615617814211,
+      "learning_rate": 4.732060768269016e-06,
+      "loss": 0.0273,
+      "step": 1076
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 3.8620381131051364,
+      "learning_rate": 4.724070593618482e-06,
+      "loss": 0.0263,
+      "step": 1077
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.20839082094382,
+      "learning_rate": 4.716081125703665e-06,
+      "loss": 0.0272,
+      "step": 1078
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.2494484460524444,
+      "learning_rate": 4.708092384987921e-06,
+      "loss": 0.0414,
+      "step": 1079
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.9688768305749882,
+      "learning_rate": 4.70010439193275e-06,
+      "loss": 0.0307,
+      "step": 1080
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.9176018744878512,
+      "learning_rate": 4.6921171669977304e-06,
+      "loss": 0.0333,
+      "step": 1081
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.4328823254667193,
+      "learning_rate": 4.684130730640475e-06,
+      "loss": 0.037,
+      "step": 1082
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.0825269206875192,
+      "learning_rate": 4.676145103316579e-06,
+      "loss": 0.0403,
+      "step": 1083
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.159048100788253,
+      "learning_rate": 4.6681603054795654e-06,
+      "loss": 0.0327,
+      "step": 1084
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.8399124550508421,
+      "learning_rate": 4.660176357580827e-06,
+      "loss": 0.0295,
+      "step": 1085
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.42909450654776243,
+      "learning_rate": 4.652193280069588e-06,
+      "loss": 0.0223,
+      "step": 1086
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5824237081174857,
+      "learning_rate": 4.644211093392837e-06,
+      "loss": 0.0277,
+      "step": 1087
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.7172546484982484,
+      "learning_rate": 4.636229817995281e-06,
+      "loss": 0.027,
+      "step": 1088
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.4487630330036912,
+      "learning_rate": 4.6282494743193e-06,
+      "loss": 0.0156,
+      "step": 1089
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.9578673232337767,
+      "learning_rate": 4.620270082804879e-06,
+      "loss": 0.035,
+      "step": 1090
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.8541729174877927,
+      "learning_rate": 4.612291663889567e-06,
+      "loss": 0.0333,
+      "step": 1091
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.413518111201729,
+      "learning_rate": 4.604314238008426e-06,
+      "loss": 0.0293,
+      "step": 1092
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.013397355900455,
+      "learning_rate": 4.596337825593969e-06,
+      "loss": 0.0198,
+      "step": 1093
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.8488149353686695,
+      "learning_rate": 4.588362447076115e-06,
+      "loss": 0.0308,
+      "step": 1094
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.808337943132434,
+      "learning_rate": 4.5803881228821375e-06,
+      "loss": 0.0277,
+      "step": 1095
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.9732316718127773,
+      "learning_rate": 4.572414873436606e-06,
+      "loss": 0.0268,
+      "step": 1096
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.6459252848175363,
+      "learning_rate": 4.564442719161338e-06,
+      "loss": 0.0279,
+      "step": 1097
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.6935117069628923,
+      "learning_rate": 4.556471680475348e-06,
+      "loss": 0.0239,
+      "step": 1098
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.607783773980419,
+      "learning_rate": 4.548501777794792e-06,
+      "loss": 0.0186,
+      "step": 1099
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.3877976209094554,
+      "learning_rate": 4.540533031532913e-06,
+      "loss": 0.0331,
+      "step": 1100
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.6712739416461726,
+      "learning_rate": 4.532565462099999e-06,
+      "loss": 0.0432,
+      "step": 1101
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.7879664908382655,
+      "learning_rate": 4.524599089903319e-06,
+      "loss": 0.028,
+      "step": 1102
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.7875123164738963,
+      "learning_rate": 4.516633935347075e-06,
+      "loss": 0.0301,
+      "step": 1103
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.9561176864842041,
+      "learning_rate": 4.508670018832353e-06,
+      "loss": 0.0316,
+      "step": 1104
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5519096705716723,
+      "learning_rate": 4.5007073607570674e-06,
+      "loss": 0.024,
+      "step": 1105
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.958646382749322,
+      "learning_rate": 4.492745981515907e-06,
+      "loss": 0.036,
+      "step": 1106
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.924537960903203,
+      "learning_rate": 4.48478590150029e-06,
+      "loss": 0.0306,
+      "step": 1107
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.852576984550701,
+      "learning_rate": 4.4768271410983e-06,
+      "loss": 0.0354,
+      "step": 1108
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.8059026291970833,
+      "learning_rate": 4.468869720694647e-06,
+      "loss": 0.022,
+      "step": 1109
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.8622128063886197,
+      "learning_rate": 4.460913660670604e-06,
+      "loss": 0.0302,
+      "step": 1110
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.0005211058131525,
+      "learning_rate": 4.452958981403963e-06,
+      "loss": 0.0244,
+      "step": 1111
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.49953654433337896,
+      "learning_rate": 4.445005703268981e-06,
+      "loss": 0.0246,
+      "step": 1112
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.5211757187596917,
+      "learning_rate": 4.4370538466363216e-06,
+      "loss": 0.0282,
+      "step": 1113
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.7049283173553994,
+      "learning_rate": 4.429103431873009e-06,
+      "loss": 0.0275,
+      "step": 1114
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.8760836408069194,
+      "learning_rate": 4.421154479342377e-06,
+      "loss": 0.0258,
+      "step": 1115
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5497290975156301,
+      "learning_rate": 4.413207009404012e-06,
+      "loss": 0.0234,
+      "step": 1116
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.43160638727821743,
+      "learning_rate": 4.4052610424137e-06,
+      "loss": 0.0204,
+      "step": 1117
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5232450640935504,
+      "learning_rate": 4.397316598723385e-06,
+      "loss": 0.0224,
+      "step": 1118
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.47825044777409836,
+      "learning_rate": 4.389373698681105e-06,
+      "loss": 0.0206,
+      "step": 1119
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.991085904207292,
+      "learning_rate": 4.381432362630942e-06,
+      "loss": 0.0321,
+      "step": 1120
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.987193721368169,
+      "learning_rate": 4.373492610912976e-06,
+      "loss": 0.0312,
+      "step": 1121
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.6093671176495833,
+      "learning_rate": 4.365554463863228e-06,
+      "loss": 0.0219,
+      "step": 1122
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.496517302197376,
+      "learning_rate": 4.3576179418136075e-06,
+      "loss": 0.0203,
+      "step": 1123
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.6130892595308474,
+      "learning_rate": 4.349683065091864e-06,
+      "loss": 0.0306,
+      "step": 1124
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.7776082447658934,
+      "learning_rate": 4.3417498540215325e-06,
+      "loss": 0.0398,
+      "step": 1125
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.3171830830330598,
+      "learning_rate": 4.33381832892188e-06,
+      "loss": 0.038,
+      "step": 1126
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.6685509894900467,
+      "learning_rate": 4.3258885101078565e-06,
+      "loss": 0.0302,
+      "step": 1127
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.1144124292702757,
+      "learning_rate": 4.317960417890043e-06,
+      "loss": 0.0275,
+      "step": 1128
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.501368764729424,
+      "learning_rate": 4.3100340725745934e-06,
+      "loss": 0.0374,
+      "step": 1129
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.2663880435731152,
+      "learning_rate": 4.3021094944631955e-06,
+      "loss": 0.0341,
+      "step": 1130
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.9252347942434317,
+      "learning_rate": 4.294186703853004e-06,
+      "loss": 0.0277,
+      "step": 1131
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.6717103063934182,
+      "learning_rate": 4.286265721036595e-06,
+      "loss": 0.027,
+      "step": 1132
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.1171647120173334,
+      "learning_rate": 4.27834656630192e-06,
+      "loss": 0.0428,
+      "step": 1133
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.7077621527686202,
+      "learning_rate": 4.270429259932243e-06,
+      "loss": 0.0268,
+      "step": 1134
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.1261178104190204,
+      "learning_rate": 4.262513822206095e-06,
+      "loss": 0.0346,
+      "step": 1135
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5321911431043505,
+      "learning_rate": 4.254600273397223e-06,
+      "loss": 0.0205,
+      "step": 1136
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.6395519526652998,
+      "learning_rate": 4.246688633774534e-06,
+      "loss": 0.0236,
+      "step": 1137
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.5594885034804874,
+      "learning_rate": 4.238778923602045e-06,
+      "loss": 0.014,
+      "step": 1138
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.8895582256492076,
+      "learning_rate": 4.230871163138831e-06,
+      "loss": 0.0257,
+      "step": 1139
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.8120185835677661,
+      "learning_rate": 4.2229653726389765e-06,
+      "loss": 0.0261,
+      "step": 1140
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.7699512160395745,
+      "learning_rate": 4.215061572351513e-06,
+      "loss": 0.0371,
+      "step": 1141
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.7904391026632073,
+      "learning_rate": 4.207159782520383e-06,
+      "loss": 0.0143,
+      "step": 1142
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.04855099373956,
+      "learning_rate": 4.199260023384376e-06,
+      "loss": 0.0431,
+      "step": 1143
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.2533318783486824,
+      "learning_rate": 4.1913623151770765e-06,
+      "loss": 0.0318,
+      "step": 1144
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.9220419806025165,
+      "learning_rate": 4.183466678126822e-06,
+      "loss": 0.0317,
+      "step": 1145
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.647797635933002,
+      "learning_rate": 4.175573132456644e-06,
+      "loss": 0.0221,
+      "step": 1146
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.5143353606003064,
+      "learning_rate": 4.167681698384211e-06,
+      "loss": 0.026,
+      "step": 1147
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.9659467748981236,
+      "learning_rate": 4.1597923961217935e-06,
+      "loss": 0.028,
+      "step": 1148
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.8694130191332563,
+      "learning_rate": 4.151905245876194e-06,
+      "loss": 0.0302,
+      "step": 1149
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.587958711576844,
+      "learning_rate": 4.144020267848707e-06,
+      "loss": 0.02,
+      "step": 1150
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.7156546748797903,
+      "learning_rate": 4.13613748223506e-06,
+      "loss": 0.0158,
+      "step": 1151
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.0851432279340858,
+      "learning_rate": 4.128256909225366e-06,
+      "loss": 0.0286,
+      "step": 1152
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.7096178232213832,
+      "learning_rate": 4.120378569004074e-06,
+      "loss": 0.0316,
+      "step": 1153
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5717606844829977,
+      "learning_rate": 4.112502481749911e-06,
+      "loss": 0.0194,
+      "step": 1154
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.0662768483868281,
+      "learning_rate": 4.104628667635835e-06,
+      "loss": 0.0272,
+      "step": 1155
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.0802239084531717,
+      "learning_rate": 4.0967571468289815e-06,
+      "loss": 0.0419,
+      "step": 1156
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 4.075912303145742,
+      "learning_rate": 4.0888879394906094e-06,
+      "loss": 0.0311,
+      "step": 1157
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.2670503288679043,
+      "learning_rate": 4.081021065776058e-06,
+      "loss": 0.0345,
+      "step": 1158
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5912311253811383,
+      "learning_rate": 4.073156545834685e-06,
+      "loss": 0.0306,
+      "step": 1159
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.7740612434514832,
+      "learning_rate": 4.065294399809819e-06,
+      "loss": 0.0248,
+      "step": 1160
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.6660766846869574,
+      "learning_rate": 4.057434647838713e-06,
+      "loss": 0.0272,
+      "step": 1161
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.4464175518732789,
+      "learning_rate": 4.049577310052482e-06,
+      "loss": 0.0187,
+      "step": 1162
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.6420961971799071,
+      "learning_rate": 4.041722406576062e-06,
+      "loss": 0.0178,
+      "step": 1163
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5933792075717889,
+      "learning_rate": 4.033869957528153e-06,
+      "loss": 0.023,
+      "step": 1164
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.7545983123520259,
+      "learning_rate": 4.026019983021168e-06,
+      "loss": 0.0279,
+      "step": 1165
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.0413887638237633,
+      "learning_rate": 4.018172503161179e-06,
+      "loss": 0.0371,
+      "step": 1166
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.0934504551905797,
+      "learning_rate": 4.010327538047877e-06,
+      "loss": 0.0328,
+      "step": 1167
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.36078229020799923,
+      "learning_rate": 4.002485107774503e-06,
+      "loss": 0.016,
+      "step": 1168
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.6444553615010122,
+      "learning_rate": 3.994645232427809e-06,
+      "loss": 0.0291,
+      "step": 1169
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.6594877072308999,
+      "learning_rate": 3.986807932088004e-06,
+      "loss": 0.0232,
+      "step": 1170
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.9328266110281278,
+      "learning_rate": 3.978973226828702e-06,
+      "loss": 0.0312,
+      "step": 1171
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5447087731555137,
+      "learning_rate": 3.971141136716866e-06,
+      "loss": 0.0189,
+      "step": 1172
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.9056993750867337,
+      "learning_rate": 3.963311681812768e-06,
+      "loss": 0.0228,
+      "step": 1173
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.7617244908266496,
+      "learning_rate": 3.955484882169923e-06,
+      "loss": 0.0207,
+      "step": 1174
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.053093794917535,
+      "learning_rate": 3.947660757835049e-06,
+      "loss": 0.0368,
+      "step": 1175
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.7773155500120877,
+      "learning_rate": 3.939839328848014e-06,
+      "loss": 0.0275,
+      "step": 1176
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.1309111614765357,
+      "learning_rate": 3.932020615241777e-06,
+      "loss": 0.0391,
+      "step": 1177
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.7772381331788334,
+      "learning_rate": 3.9242046370423434e-06,
+      "loss": 0.0228,
+      "step": 1178
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.0497661692021192,
+      "learning_rate": 3.9163914142687185e-06,
+      "loss": 0.0197,
+      "step": 1179
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.802443273282112,
+      "learning_rate": 3.9085809669328415e-06,
+      "loss": 0.0287,
+      "step": 1180
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.6422605212693915,
+      "learning_rate": 3.900773315039548e-06,
+      "loss": 0.0263,
+      "step": 1181
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.4827986025988822,
+      "learning_rate": 3.892968478586513e-06,
+      "loss": 0.0385,
+      "step": 1182
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.0726420346174323,
+      "learning_rate": 3.885166477564199e-06,
+      "loss": 0.0315,
+      "step": 1183
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.4868137126849361,
+      "learning_rate": 3.877367331955808e-06,
+      "loss": 0.0301,
+      "step": 1184
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.7793921575940534,
+      "learning_rate": 3.869571061737226e-06,
+      "loss": 0.0278,
+      "step": 1185
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.7423673491553581,
+      "learning_rate": 3.861777686876978e-06,
+      "loss": 0.0172,
+      "step": 1186
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.8445082531782259,
+      "learning_rate": 3.853987227336168e-06,
+      "loss": 0.0234,
+      "step": 1187
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.4866791378338005,
+      "learning_rate": 3.8461997030684386e-06,
+      "loss": 0.0413,
+      "step": 1188
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.696171863459194,
+      "learning_rate": 3.838415134019911e-06,
+      "loss": 0.0278,
+      "step": 1189
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.9645765538518006,
+      "learning_rate": 3.830633540129135e-06,
+      "loss": 0.0333,
+      "step": 1190
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.1588316179391511,
+      "learning_rate": 3.822854941327046e-06,
+      "loss": 0.0247,
+      "step": 1191
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.8964545735956162,
+      "learning_rate": 3.815079357536907e-06,
+      "loss": 0.0244,
+      "step": 1192
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.9118777787298978,
+      "learning_rate": 3.8073068086742514e-06,
+      "loss": 0.0275,
+      "step": 1193
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.777626923225908,
+      "learning_rate": 3.799537314646848e-06,
+      "loss": 0.0205,
+      "step": 1194
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.6685174415093665,
+      "learning_rate": 3.791770895354635e-06,
+      "loss": 0.0206,
+      "step": 1195
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.48232058435281616,
+      "learning_rate": 3.7840075706896824e-06,
+      "loss": 0.0207,
+      "step": 1196
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.0217841454148604,
+      "learning_rate": 3.776247360536127e-06,
+      "loss": 0.0393,
+      "step": 1197
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.6119942879107466,
+      "learning_rate": 3.768490284770131e-06,
+      "loss": 0.0203,
+      "step": 1198
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5206581259536639,
+      "learning_rate": 3.7607363632598305e-06,
+      "loss": 0.0196,
+      "step": 1199
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.9881570998597756,
+      "learning_rate": 3.7529856158652792e-06,
+      "loss": 0.0163,
+      "step": 1200
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.6382121125866328,
+      "learning_rate": 3.7452380624384026e-06,
+      "loss": 0.0231,
+      "step": 1201
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.4224758933081467,
+      "learning_rate": 3.7374937228229472e-06,
+      "loss": 0.0139,
+      "step": 1202
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.38564278891273984,
+      "learning_rate": 3.7297526168544253e-06,
+      "loss": 0.0173,
+      "step": 1203
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.0962566818321713,
+      "learning_rate": 3.722014764360067e-06,
+      "loss": 0.025,
+      "step": 1204
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.2246173963024596,
+      "learning_rate": 3.714280185158771e-06,
+      "loss": 0.0406,
+      "step": 1205
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.8603176514449102,
+      "learning_rate": 3.706548899061052e-06,
+      "loss": 0.026,
+      "step": 1206
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.2822870337099297,
+      "learning_rate": 3.6988209258689877e-06,
+      "loss": 0.0313,
+      "step": 1207
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.2715509865915087,
+      "learning_rate": 3.6910962853761738e-06,
+      "loss": 0.0355,
+      "step": 1208
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.1853601452520386,
+      "learning_rate": 3.683374997367668e-06,
+      "loss": 0.0321,
+      "step": 1209
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.902921247787625,
+      "learning_rate": 3.675657081619941e-06,
+      "loss": 0.0191,
+      "step": 1210
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.7821409311223014,
+      "learning_rate": 3.6679425579008278e-06,
+      "loss": 0.0326,
+      "step": 1211
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.7903619927425193,
+      "learning_rate": 3.6602314459694743e-06,
+      "loss": 0.0301,
+      "step": 1212
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.46174044405418857,
+      "learning_rate": 3.652523765576287e-06,
+      "loss": 0.022,
+      "step": 1213
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.6323491372648098,
+      "learning_rate": 3.6448195364628857e-06,
+      "loss": 0.0191,
+      "step": 1214
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.9099434352944868,
+      "learning_rate": 3.6371187783620486e-06,
+      "loss": 0.0241,
+      "step": 1215
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.9207530906993158,
+      "learning_rate": 3.6294215109976628e-06,
+      "loss": 0.0222,
+      "step": 1216
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5852047866123646,
+      "learning_rate": 3.6217277540846775e-06,
+      "loss": 0.0268,
+      "step": 1217
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.7561125846165206,
+      "learning_rate": 3.614037527329048e-06,
+      "loss": 0.0286,
+      "step": 1218
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.7062712342887891,
+      "learning_rate": 3.606350850427688e-06,
+      "loss": 0.0356,
+      "step": 1219
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.681472660161719,
+      "learning_rate": 3.5986677430684224e-06,
+      "loss": 0.0225,
+      "step": 1220
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.8267818175730947,
+      "learning_rate": 3.5909882249299287e-06,
+      "loss": 0.0318,
+      "step": 1221
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.7355695025915546,
+      "learning_rate": 3.583312315681693e-06,
+      "loss": 0.0223,
+      "step": 1222
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 2.833987497480372,
+      "learning_rate": 3.5756400349839603e-06,
+      "loss": 0.0337,
+      "step": 1223
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.9794172004233616,
+      "learning_rate": 3.567971402487679e-06,
+      "loss": 0.0218,
+      "step": 1224
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 3.6501686738425505,
+      "learning_rate": 3.5603064378344536e-06,
+      "loss": 0.0262,
+      "step": 1225
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5674790693617395,
+      "learning_rate": 3.552645160656497e-06,
+      "loss": 0.0265,
+      "step": 1226
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.0102812039517735,
+      "learning_rate": 3.544987590576574e-06,
+      "loss": 0.0226,
+      "step": 1227
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.5226619099252662,
+      "learning_rate": 3.537333747207955e-06,
+      "loss": 0.0183,
+      "step": 1228
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.0690310624262982,
+      "learning_rate": 3.529683650154368e-06,
+      "loss": 0.0367,
+      "step": 1229
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 2.019580208227227,
+      "learning_rate": 3.5220373190099428e-06,
+      "loss": 0.0379,
+      "step": 1230
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.6233956991684912,
+      "learning_rate": 3.5143947733591633e-06,
+      "loss": 0.0231,
+      "step": 1231
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.6281650494544383,
+      "learning_rate": 3.50675603277682e-06,
+      "loss": 0.024,
+      "step": 1232
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5383027091207303,
+      "learning_rate": 3.499121116827956e-06,
+      "loss": 0.0167,
+      "step": 1233
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.7215097341859587,
+      "learning_rate": 3.491490045067818e-06,
+      "loss": 0.0258,
+      "step": 1234
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.5910503130056679,
+      "learning_rate": 3.48386283704181e-06,
+      "loss": 0.0231,
+      "step": 1235
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 2.4693134066224136,
+      "learning_rate": 3.4762395122854336e-06,
+      "loss": 0.0363,
+      "step": 1236
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.166411864946178,
+      "learning_rate": 3.4686200903242497e-06,
+      "loss": 0.0259,
+      "step": 1237
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.4817375402839268,
+      "learning_rate": 3.461004590673819e-06,
+      "loss": 0.0344,
+      "step": 1238
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.046182158547702,
+      "learning_rate": 3.45339303283966e-06,
+      "loss": 0.0242,
+      "step": 1239
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.292224774416532,
+      "learning_rate": 3.445785436317193e-06,
+      "loss": 0.0375,
+      "step": 1240
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.861521851112122,
+      "learning_rate": 3.4381818205916907e-06,
+      "loss": 0.0218,
+      "step": 1241
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.5716383717347113,
+      "learning_rate": 3.4305822051382345e-06,
+      "loss": 0.0254,
+      "step": 1242
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.8052011902200021,
+      "learning_rate": 3.422986609421655e-06,
+      "loss": 0.0266,
+      "step": 1243
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.9384565798258717,
+      "learning_rate": 3.4153950528964867e-06,
+      "loss": 0.0399,
+      "step": 1244
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.9711098718625382,
+      "learning_rate": 3.4078075550069255e-06,
+      "loss": 0.0241,
+      "step": 1245
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.6655061872931324,
+      "learning_rate": 3.400224135186765e-06,
+      "loss": 0.0364,
+      "step": 1246
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.5502248667043002,
+      "learning_rate": 3.392644812859354e-06,
+      "loss": 0.0416,
+      "step": 1247
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.844181133418411,
+      "learning_rate": 3.385069607437552e-06,
+      "loss": 0.0354,
+      "step": 1248
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.1257355441450307,
+      "learning_rate": 3.3774985383236685e-06,
+      "loss": 0.0232,
+      "step": 1249
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.9739750889298195,
+      "learning_rate": 3.3699316249094195e-06,
+      "loss": 0.0395,
+      "step": 1250
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.7690071951479069,
+      "learning_rate": 3.36236888657588e-06,
+      "loss": 0.0344,
+      "step": 1251
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 2.6660183823158623,
+      "learning_rate": 3.3548103426934287e-06,
+      "loss": 0.0213,
+      "step": 1252
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.6154954723041898,
+      "learning_rate": 3.3472560126217004e-06,
+      "loss": 0.0192,
+      "step": 1253
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.6999616838858199,
+      "learning_rate": 3.3397059157095412e-06,
+      "loss": 0.0222,
+      "step": 1254
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.625168930364848,
+      "learning_rate": 3.33216007129495e-06,
+      "loss": 0.0281,
+      "step": 1255
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.7423624969087769,
+      "learning_rate": 3.3246184987050366e-06,
+      "loss": 0.0285,
+      "step": 1256
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5785260967877592,
+      "learning_rate": 3.3170812172559695e-06,
+      "loss": 0.0258,
+      "step": 1257
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.053003162298107,
+      "learning_rate": 3.309548246252925e-06,
+      "loss": 0.0259,
+      "step": 1258
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.9626579997361799,
+      "learning_rate": 3.3020196049900386e-06,
+      "loss": 0.0341,
+      "step": 1259
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.5378411567759094,
+      "learning_rate": 3.2944953127503593e-06,
+      "loss": 0.0229,
+      "step": 1260
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.40918902129839757,
+      "learning_rate": 3.2869753888057936e-06,
+      "loss": 0.0174,
+      "step": 1261
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.948390041474773,
+      "learning_rate": 3.2794598524170606e-06,
+      "loss": 0.021,
+      "step": 1262
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.9761585653303926,
+      "learning_rate": 3.271948722833643e-06,
+      "loss": 0.0245,
+      "step": 1263
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.1402260326532885,
+      "learning_rate": 3.264442019293734e-06,
+      "loss": 0.0289,
+      "step": 1264
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.8272407776944543,
+      "learning_rate": 3.2569397610241915e-06,
+      "loss": 0.022,
+      "step": 1265
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.1065240923106912,
+      "learning_rate": 3.249441967240489e-06,
+      "loss": 0.0387,
+      "step": 1266
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.7964507556545418,
+      "learning_rate": 3.241948657146663e-06,
+      "loss": 0.0258,
+      "step": 1267
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.9722127655862852,
+      "learning_rate": 3.2344598499352663e-06,
+      "loss": 0.0233,
+      "step": 1268
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.6702923565192391,
+      "learning_rate": 3.226975564787322e-06,
+      "loss": 0.02,
+      "step": 1269
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.6781473563830941,
+      "learning_rate": 3.2194958208722656e-06,
+      "loss": 0.027,
+      "step": 1270
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.8764587765192495,
+      "learning_rate": 3.2120206373479024e-06,
+      "loss": 0.0259,
+      "step": 1271
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.204359727983454,
+      "learning_rate": 3.204550033360362e-06,
+      "loss": 0.0261,
+      "step": 1272
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 4.3643673401914205,
+      "learning_rate": 3.1970840280440384e-06,
+      "loss": 0.0189,
+      "step": 1273
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.7324647980496716,
+      "learning_rate": 3.1896226405215503e-06,
+      "loss": 0.0221,
+      "step": 1274
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.8575165229725605,
+      "learning_rate": 3.1821658899036876e-06,
+      "loss": 0.0276,
+      "step": 1275
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.7541136711033276,
+      "learning_rate": 3.174713795289366e-06,
+      "loss": 0.0322,
+      "step": 1276
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.9834424977627805,
+      "learning_rate": 3.1672663757655707e-06,
+      "loss": 0.0441,
+      "step": 1277
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.3702494488264347,
+      "learning_rate": 3.1598236504073215e-06,
+      "loss": 0.0329,
+      "step": 1278
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.02349427189603,
+      "learning_rate": 3.152385638277603e-06,
+      "loss": 0.0442,
+      "step": 1279
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.0267724041865374,
+      "learning_rate": 3.144952358427339e-06,
+      "loss": 0.0265,
+      "step": 1280
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.4656151906951841,
+      "learning_rate": 3.137523829895326e-06,
+      "loss": 0.0178,
+      "step": 1281
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.9467462721261859,
+      "learning_rate": 3.1301000717081926e-06,
+      "loss": 0.0325,
+      "step": 1282
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.1709202769320373,
+      "learning_rate": 3.1226811028803514e-06,
+      "loss": 0.0333,
+      "step": 1283
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.3576000887150794,
+      "learning_rate": 3.115266942413946e-06,
+      "loss": 0.0301,
+      "step": 1284
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.7941095758553105,
+      "learning_rate": 3.107857609298802e-06,
+      "loss": 0.021,
+      "step": 1285
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.7929011356573532,
+      "learning_rate": 3.1004531225123857e-06,
+      "loss": 0.0256,
+      "step": 1286
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.8786043128992731,
+      "learning_rate": 3.093053501019748e-06,
+      "loss": 0.0258,
+      "step": 1287
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.260719000880376,
+      "learning_rate": 3.0856587637734783e-06,
+      "loss": 0.0282,
+      "step": 1288
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.0540374948932114,
+      "learning_rate": 3.0782689297136554e-06,
+      "loss": 0.0309,
+      "step": 1289
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.5725262960615425,
+      "learning_rate": 3.070884017767801e-06,
+      "loss": 0.022,
+      "step": 1290
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.959101589090888,
+      "learning_rate": 3.0635040468508294e-06,
+      "loss": 0.028,
+      "step": 1291
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5386872857085101,
+      "learning_rate": 3.0561290358649996e-06,
+      "loss": 0.0151,
+      "step": 1292
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.569838897864246,
+      "learning_rate": 3.048759003699866e-06,
+      "loss": 0.0207,
+      "step": 1293
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.581532645283473,
+      "learning_rate": 3.0413939692322304e-06,
+      "loss": 0.0225,
+      "step": 1294
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5028362927727279,
+      "learning_rate": 3.0340339513260976e-06,
+      "loss": 0.0235,
+      "step": 1295
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.8007831617356519,
+      "learning_rate": 3.0266789688326187e-06,
+      "loss": 0.0227,
+      "step": 1296
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.7695969527513412,
+      "learning_rate": 3.0193290405900494e-06,
+      "loss": 0.0196,
+      "step": 1297
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.0289670306604963,
+      "learning_rate": 3.0119841854237027e-06,
+      "loss": 0.0307,
+      "step": 1298
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.6835324484005986,
+      "learning_rate": 3.004644422145894e-06,
+      "loss": 0.0292,
+      "step": 1299
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.581528963254537,
+      "learning_rate": 2.9973097695558982e-06,
+      "loss": 0.0259,
+      "step": 1300
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.8906463659166122,
+      "learning_rate": 2.9899802464399018e-06,
+      "loss": 0.0441,
+      "step": 1301
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.8464019379319843,
+      "learning_rate": 2.982655871570952e-06,
+      "loss": 0.0156,
+      "step": 1302
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.3655786669874588,
+      "learning_rate": 2.9753366637089075e-06,
+      "loss": 0.0263,
+      "step": 1303
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.4302998086198895,
+      "learning_rate": 2.9680226416003986e-06,
+      "loss": 0.0331,
+      "step": 1304
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.6007442648446595,
+      "learning_rate": 2.960713823978769e-06,
+      "loss": 0.0188,
+      "step": 1305
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5396731652610417,
+      "learning_rate": 2.9534102295640305e-06,
+      "loss": 0.0216,
+      "step": 1306
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.2810585899025055,
+      "learning_rate": 2.9461118770628226e-06,
+      "loss": 0.0248,
+      "step": 1307
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.8106101777855437,
+      "learning_rate": 2.9388187851683537e-06,
+      "loss": 0.0223,
+      "step": 1308
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.6780089328890525,
+      "learning_rate": 2.9315309725603596e-06,
+      "loss": 0.0251,
+      "step": 1309
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5254825720095616,
+      "learning_rate": 2.9242484579050566e-06,
+      "loss": 0.0233,
+      "step": 1310
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5856391286139735,
+      "learning_rate": 2.9169712598550885e-06,
+      "loss": 0.0183,
+      "step": 1311
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.6144133333430777,
+      "learning_rate": 2.9096993970494825e-06,
+      "loss": 0.0207,
+      "step": 1312
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.7793656594295922,
+      "learning_rate": 2.9024328881136e-06,
+      "loss": 0.0275,
+      "step": 1313
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.8547383245915084,
+      "learning_rate": 2.895171751659093e-06,
+      "loss": 0.024,
+      "step": 1314
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.136725575523068,
+      "learning_rate": 2.887916006283849e-06,
+      "loss": 0.0296,
+      "step": 1315
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.000339457408409,
+      "learning_rate": 2.8806656705719492e-06,
+      "loss": 0.0216,
+      "step": 1316
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5570093686866413,
+      "learning_rate": 2.8734207630936195e-06,
+      "loss": 0.0209,
+      "step": 1317
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.4454859134820076,
+      "learning_rate": 2.86618130240518e-06,
+      "loss": 0.0147,
+      "step": 1318
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.563113334058305,
+      "learning_rate": 2.858947307049001e-06,
+      "loss": 0.0187,
+      "step": 1319
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.8038186022459026,
+      "learning_rate": 2.851718795553461e-06,
+      "loss": 0.0214,
+      "step": 1320
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.8962721319951323,
+      "learning_rate": 2.84449578643288e-06,
+      "loss": 0.0332,
+      "step": 1321
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.0943982653356987,
+      "learning_rate": 2.8372782981874964e-06,
+      "loss": 0.0203,
+      "step": 1322
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.6210789822485363,
+      "learning_rate": 2.830066349303401e-06,
+      "loss": 0.0207,
+      "step": 1323
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.7558588675119113,
+      "learning_rate": 2.8228599582524975e-06,
+      "loss": 0.0279,
+      "step": 1324
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.2117255873190556,
+      "learning_rate": 2.815659143492461e-06,
+      "loss": 0.0197,
+      "step": 1325
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.8905208919281724,
+      "learning_rate": 2.8084639234666753e-06,
+      "loss": 0.0254,
+      "step": 1326
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.5673813607386939,
+      "learning_rate": 2.8012743166042002e-06,
+      "loss": 0.0195,
+      "step": 1327
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.7629642074483579,
+      "learning_rate": 2.794090341319715e-06,
+      "loss": 0.0234,
+      "step": 1328
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.0756254635065226,
+      "learning_rate": 2.786912016013478e-06,
+      "loss": 0.0197,
+      "step": 1329
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.7397629259597135,
+      "learning_rate": 2.7797393590712706e-06,
+      "loss": 0.0286,
+      "step": 1330
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.3606802714238877,
+      "learning_rate": 2.7725723888643663e-06,
+      "loss": 0.0195,
+      "step": 1331
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.1951944803879382,
+      "learning_rate": 2.765411123749463e-06,
+      "loss": 0.0227,
+      "step": 1332
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.572616845327841,
+      "learning_rate": 2.758255582068651e-06,
+      "loss": 0.0214,
+      "step": 1333
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.6254891645360283,
+      "learning_rate": 2.75110578214936e-06,
+      "loss": 0.0226,
+      "step": 1334
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.8258197500603863,
+      "learning_rate": 2.7439617423043146e-06,
+      "loss": 0.0253,
+      "step": 1335
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5567925238200621,
+      "learning_rate": 2.736823480831482e-06,
+      "loss": 0.0103,
+      "step": 1336
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.9486102645398342,
+      "learning_rate": 2.7296910160140365e-06,
+      "loss": 0.036,
+      "step": 1337
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 5.9259932643996045,
+      "learning_rate": 2.7225643661203004e-06,
+      "loss": 0.0322,
+      "step": 1338
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.691222736049963,
+      "learning_rate": 2.7154435494037033e-06,
+      "loss": 0.0148,
+      "step": 1339
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.8575355131535669,
+      "learning_rate": 2.708328584102734e-06,
+      "loss": 0.0181,
+      "step": 1340
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.1963430964304536,
+      "learning_rate": 2.7012194884408942e-06,
+      "loss": 0.0282,
+      "step": 1341
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.9228574782155844,
+      "learning_rate": 2.6941162806266506e-06,
+      "loss": 0.0259,
+      "step": 1342
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.6683360136989352,
+      "learning_rate": 2.6870189788533953e-06,
+      "loss": 0.0233,
+      "step": 1343
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.6158335243969865,
+      "learning_rate": 2.679927601299386e-06,
+      "loss": 0.0224,
+      "step": 1344
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.4342317868225625,
+      "learning_rate": 2.6728421661277105e-06,
+      "loss": 0.0187,
+      "step": 1345
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.8702462153937798,
+      "learning_rate": 2.665762691486235e-06,
+      "loss": 0.0279,
+      "step": 1346
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.9076288865381044,
+      "learning_rate": 2.65868919550756e-06,
+      "loss": 0.0256,
+      "step": 1347
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.7546622654755015,
+      "learning_rate": 2.6516216963089698e-06,
+      "loss": 0.0239,
+      "step": 1348
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.5596958413844281,
+      "learning_rate": 2.6445602119923963e-06,
+      "loss": 0.0159,
+      "step": 1349
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.817812674260711,
+      "learning_rate": 2.637504760644359e-06,
+      "loss": 0.0173,
+      "step": 1350
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.9082917695452071,
+      "learning_rate": 2.630455360335929e-06,
+      "loss": 0.0284,
+      "step": 1351
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.0181782420057977,
+      "learning_rate": 2.623412029122675e-06,
+      "loss": 0.0235,
+      "step": 1352
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.4308595058667782,
+      "learning_rate": 2.616374785044624e-06,
+      "loss": 0.0195,
+      "step": 1353
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.0010421051313705,
+      "learning_rate": 2.60934364612621e-06,
+      "loss": 0.0405,
+      "step": 1354
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.1072856388623673,
+      "learning_rate": 2.602318630376234e-06,
+      "loss": 0.0346,
+      "step": 1355
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.570554903732537,
+      "learning_rate": 2.59529975578781e-06,
+      "loss": 0.02,
+      "step": 1356
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.763034168480365,
+      "learning_rate": 2.588287040338323e-06,
+      "loss": 0.028,
+      "step": 1357
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.9818915105741374,
+      "learning_rate": 2.5812805019893837e-06,
+      "loss": 0.0255,
+      "step": 1358
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.37647009133737225,
+      "learning_rate": 2.574280158686782e-06,
+      "loss": 0.0128,
+      "step": 1359
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.8141530396717688,
+      "learning_rate": 2.567286028360435e-06,
+      "loss": 0.0279,
+      "step": 1360
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.7619954885706852,
+      "learning_rate": 2.560298128924358e-06,
+      "loss": 0.0212,
+      "step": 1361
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.7786618256826614,
+      "learning_rate": 2.5533164782765974e-06,
+      "loss": 0.0244,
+      "step": 1362
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.8083725721324221,
+      "learning_rate": 2.5463410942991986e-06,
+      "loss": 0.0298,
+      "step": 1363
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.8305456421254139,
+      "learning_rate": 2.5393719948581507e-06,
+      "loss": 0.029,
+      "step": 1364
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.4458021272805112,
+      "learning_rate": 2.532409197803353e-06,
+      "loss": 0.0178,
+      "step": 1365
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.6645665194259693,
+      "learning_rate": 2.525452720968563e-06,
+      "loss": 0.0207,
+      "step": 1366
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.6310746208910438,
+      "learning_rate": 2.5185025821713454e-06,
+      "loss": 0.0175,
+      "step": 1367
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.0364074720417116,
+      "learning_rate": 2.5115587992130342e-06,
+      "loss": 0.0275,
+      "step": 1368
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.6205551171830646,
+      "learning_rate": 2.504621389878682e-06,
+      "loss": 0.0301,
+      "step": 1369
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.49694932253221913,
+      "learning_rate": 2.4976903719370193e-06,
+      "loss": 0.0142,
+      "step": 1370
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.5103829516719648,
+      "learning_rate": 2.4907657631404037e-06,
+      "loss": 0.023,
+      "step": 1371
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.6578952733958501,
+      "learning_rate": 2.483847581224782e-06,
+      "loss": 0.0266,
+      "step": 1372
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.124638769126836,
+      "learning_rate": 2.4769358439096347e-06,
+      "loss": 0.0318,
+      "step": 1373
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.8613357003477377,
+      "learning_rate": 2.470030568897938e-06,
+      "loss": 0.0329,
+      "step": 1374
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.7481385202102532,
+      "learning_rate": 2.4631317738761155e-06,
+      "loss": 0.0261,
+      "step": 1375
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5082467310959725,
+      "learning_rate": 2.456239476513994e-06,
+      "loss": 0.0202,
+      "step": 1376
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.34702336652624877,
+      "learning_rate": 2.4493536944647566e-06,
+      "loss": 0.0202,
+      "step": 1377
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.6745601361393228,
+      "learning_rate": 2.442474445364904e-06,
+      "loss": 0.0223,
+      "step": 1378
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.5011297804780708,
+      "learning_rate": 2.4356017468341977e-06,
+      "loss": 0.0338,
+      "step": 1379
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.934845789299032,
+      "learning_rate": 2.4287356164756244e-06,
+      "loss": 0.0204,
+      "step": 1380
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5768390021313147,
+      "learning_rate": 2.421876071875347e-06,
+      "loss": 0.0189,
+      "step": 1381
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.4666835633560529,
+      "learning_rate": 2.415023130602661e-06,
+      "loss": 0.0217,
+      "step": 1382
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.7251505144838705,
+      "learning_rate": 2.408176810209946e-06,
+      "loss": 0.0245,
+      "step": 1383
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.7030905046001982,
+      "learning_rate": 2.401337128232631e-06,
+      "loss": 0.0238,
+      "step": 1384
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.7104189569845816,
+      "learning_rate": 2.3945041021891335e-06,
+      "loss": 0.0212,
+      "step": 1385
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.2295751661808245,
+      "learning_rate": 2.387677749580828e-06,
+      "loss": 0.0355,
+      "step": 1386
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.6330065194593735,
+      "learning_rate": 2.3808580878919948e-06,
+      "loss": 0.0348,
+      "step": 1387
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.918874073383736,
+      "learning_rate": 2.3740451345897773e-06,
+      "loss": 0.0164,
+      "step": 1388
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.582892515629951,
+      "learning_rate": 2.3672389071241354e-06,
+      "loss": 0.0195,
+      "step": 1389
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.0182000574709098,
+      "learning_rate": 2.3604394229278064e-06,
+      "loss": 0.0366,
+      "step": 1390
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.9912460164229965,
+      "learning_rate": 2.3536466994162522e-06,
+      "loss": 0.0272,
+      "step": 1391
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.9809821583130328,
+      "learning_rate": 2.3468607539876186e-06,
+      "loss": 0.0209,
+      "step": 1392
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.6201084241577994,
+      "learning_rate": 2.3400816040226925e-06,
+      "loss": 0.0199,
+      "step": 1393
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.8996803265244651,
+      "learning_rate": 2.3333092668848544e-06,
+      "loss": 0.0333,
+      "step": 1394
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.4792146683688987,
+      "learning_rate": 2.326543759920034e-06,
+      "loss": 0.0203,
+      "step": 1395
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.45049080136019715,
+      "learning_rate": 2.3197851004566723e-06,
+      "loss": 0.0215,
+      "step": 1396
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.2387095657348046,
+      "learning_rate": 2.313033305805667e-06,
+      "loss": 0.0269,
+      "step": 1397
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.5598441840881433,
+      "learning_rate": 2.3062883932603326e-06,
+      "loss": 0.0252,
+      "step": 1398
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.7873601630358676,
+      "learning_rate": 2.2995503800963593e-06,
+      "loss": 0.0215,
+      "step": 1399
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.907763784643883,
+      "learning_rate": 2.2928192835717642e-06,
+      "loss": 0.0264,
+      "step": 1400
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.8360874450536064,
+      "learning_rate": 2.2860951209268485e-06,
+      "loss": 0.0254,
+      "step": 1401
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.1729734710689084,
+      "learning_rate": 2.2793779093841565e-06,
+      "loss": 0.0172,
+      "step": 1402
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.894340993070462,
+      "learning_rate": 2.2726676661484265e-06,
+      "loss": 0.0138,
+      "step": 1403
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.609965868879223,
+      "learning_rate": 2.2659644084065483e-06,
+      "loss": 0.0214,
+      "step": 1404
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.4737416535129027,
+      "learning_rate": 2.2592681533275223e-06,
+      "loss": 0.0187,
+      "step": 1405
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.0187050959181063,
+      "learning_rate": 2.2525789180624096e-06,
+      "loss": 0.0157,
+      "step": 1406
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.9494581574836511,
+      "learning_rate": 2.2458967197442942e-06,
+      "loss": 0.0264,
+      "step": 1407
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.7222904042380601,
+      "learning_rate": 2.2392215754882363e-06,
+      "loss": 0.0316,
+      "step": 1408
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.5612517935965755,
+      "learning_rate": 2.232553502391227e-06,
+      "loss": 0.0208,
+      "step": 1409
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 2.564448457114273,
+      "learning_rate": 2.2258925175321476e-06,
+      "loss": 0.0358,
+      "step": 1410
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.9895281557974522,
+      "learning_rate": 2.2192386379717214e-06,
+      "loss": 0.0322,
+      "step": 1411
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.5522292721758334,
+      "learning_rate": 2.2125918807524795e-06,
+      "loss": 0.0238,
+      "step": 1412
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.0731253951864272,
+      "learning_rate": 2.205952262898704e-06,
+      "loss": 0.0333,
+      "step": 1413
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.48693201879768505,
+      "learning_rate": 2.199319801416394e-06,
+      "loss": 0.0185,
+      "step": 1414
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.8312161420257296,
+      "learning_rate": 2.1926945132932188e-06,
+      "loss": 0.0296,
+      "step": 1415
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.6626417816503829,
+      "learning_rate": 2.1860764154984736e-06,
+      "loss": 0.0207,
+      "step": 1416
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.6049530155376529,
+      "learning_rate": 2.179465524983036e-06,
+      "loss": 0.0172,
+      "step": 1417
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.8572149751190988,
+      "learning_rate": 2.1728618586793292e-06,
+      "loss": 0.0263,
+      "step": 1418
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.553519448666031,
+      "learning_rate": 2.1662654335012684e-06,
+      "loss": 0.0133,
+      "step": 1419
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.7689498757925647,
+      "learning_rate": 2.159676266344222e-06,
+      "loss": 0.0232,
+      "step": 1420
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.5344206051884095,
+      "learning_rate": 2.1530943740849696e-06,
+      "loss": 0.0204,
+      "step": 1421
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.7641828859197944,
+      "learning_rate": 2.1465197735816585e-06,
+      "loss": 0.0246,
+      "step": 1422
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.5707472485620008,
+      "learning_rate": 2.139952481673756e-06,
+      "loss": 0.0209,
+      "step": 1423
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.5335231672331546,
+      "learning_rate": 2.1333925151820175e-06,
+      "loss": 0.0184,
+      "step": 1424
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.37509628954758883,
+      "learning_rate": 2.126839890908428e-06,
+      "loss": 0.0132,
+      "step": 1425
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.4228335969477914,
+      "learning_rate": 2.120294625636171e-06,
+      "loss": 0.0198,
+      "step": 1426
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.5467505342742434,
+      "learning_rate": 2.113756736129581e-06,
+      "loss": 0.0232,
+      "step": 1427
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.6241443244997975,
+      "learning_rate": 2.1072262391340996e-06,
+      "loss": 0.0226,
+      "step": 1428
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.914095950785102,
+      "learning_rate": 2.100703151376234e-06,
+      "loss": 0.037,
+      "step": 1429
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.5764144696089817,
+      "learning_rate": 2.0941874895635184e-06,
+      "loss": 0.0216,
+      "step": 1430
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.5566137561193788,
+      "learning_rate": 2.087679270384461e-06,
+      "loss": 0.0139,
+      "step": 1431
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.9897236771294801,
+      "learning_rate": 2.081178510508512e-06,
+      "loss": 0.0279,
+      "step": 1432
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.9771936407490407,
+      "learning_rate": 2.0746852265860123e-06,
+      "loss": 0.0223,
+      "step": 1433
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 2.291363849454442,
+      "learning_rate": 2.0681994352481575e-06,
+      "loss": 0.015,
+      "step": 1434
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.3268806800493107,
+      "learning_rate": 2.0617211531069503e-06,
+      "loss": 0.0294,
+      "step": 1435
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.7891869090943812,
+      "learning_rate": 2.0552503967551634e-06,
+      "loss": 0.0247,
+      "step": 1436
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.0008723260609602,
+      "learning_rate": 2.0487871827662913e-06,
+      "loss": 0.0326,
+      "step": 1437
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.43601598237058325,
+      "learning_rate": 2.0423315276945103e-06,
+      "loss": 0.02,
+      "step": 1438
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.6944295941030163,
+      "learning_rate": 2.0358834480746363e-06,
+      "loss": 0.0269,
+      "step": 1439
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.6707427363028934,
+      "learning_rate": 2.0294429604220833e-06,
+      "loss": 0.0203,
+      "step": 1440
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.685925821741964,
+      "learning_rate": 2.0230100812328167e-06,
+      "loss": 0.023,
+      "step": 1441
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.5147866054111775,
+      "learning_rate": 2.0165848269833215e-06,
+      "loss": 0.0203,
+      "step": 1442
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.776632642956283,
+      "learning_rate": 2.010167214130546e-06,
+      "loss": 0.0246,
+      "step": 1443
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.0411426767089536,
+      "learning_rate": 2.003757259111869e-06,
+      "loss": 0.0282,
+      "step": 1444
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.957824081037812,
+      "learning_rate": 1.9973549783450563e-06,
+      "loss": 0.0188,
+      "step": 1445
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.134533925904978,
+      "learning_rate": 1.9909603882282156e-06,
+      "loss": 0.023,
+      "step": 1446
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.7392234352149661,
+      "learning_rate": 1.984573505139758e-06,
+      "loss": 0.0185,
+      "step": 1447
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.8419016776814038,
+      "learning_rate": 1.9781943454383583e-06,
+      "loss": 0.0292,
+      "step": 1448
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.5482575841766507,
+      "learning_rate": 1.9718229254629018e-06,
+      "loss": 0.0201,
+      "step": 1449
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.7895825808166198,
+      "learning_rate": 1.9654592615324542e-06,
+      "loss": 0.0261,
+      "step": 1450
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.5391985624677404,
+      "learning_rate": 1.9591033699462165e-06,
+      "loss": 0.0166,
+      "step": 1451
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.0509339173777723,
+      "learning_rate": 1.9527552669834797e-06,
+      "loss": 0.0317,
+      "step": 1452
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.6975292110442908,
+      "learning_rate": 1.9464149689035912e-06,
+      "loss": 0.0259,
+      "step": 1453
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.6687890100923896,
+      "learning_rate": 1.940082491945902e-06,
+      "loss": 0.0177,
+      "step": 1454
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.086040887866882,
+      "learning_rate": 1.933757852329734e-06,
+      "loss": 0.0285,
+      "step": 1455
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.6219525438540332,
+      "learning_rate": 1.927441066254334e-06,
+      "loss": 0.0237,
+      "step": 1456
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.7987299511140524,
+      "learning_rate": 1.921132149898834e-06,
+      "loss": 0.0152,
+      "step": 1457
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.8747595635173462,
+      "learning_rate": 1.9148311194222084e-06,
+      "loss": 0.0267,
+      "step": 1458
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.1033645254183688,
+      "learning_rate": 1.908537990963238e-06,
+      "loss": 0.0252,
+      "step": 1459
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.1394276850699414,
+      "learning_rate": 1.9022527806404583e-06,
+      "loss": 0.0399,
+      "step": 1460
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.443076629359654,
+      "learning_rate": 1.8959755045521283e-06,
+      "loss": 0.0169,
+      "step": 1461
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.6779496548512952,
+      "learning_rate": 1.8897061787761823e-06,
+      "loss": 0.0272,
+      "step": 1462
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.9373551720667195,
+      "learning_rate": 1.883444819370193e-06,
+      "loss": 0.0384,
+      "step": 1463
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.645241163161573,
+      "learning_rate": 1.8771914423713277e-06,
+      "loss": 0.0186,
+      "step": 1464
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.4385818493699251,
+      "learning_rate": 1.8709460637963123e-06,
+      "loss": 0.0137,
+      "step": 1465
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.0003881903051037,
+      "learning_rate": 1.8647086996413816e-06,
+      "loss": 0.0257,
+      "step": 1466
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.5380060560641039,
+      "learning_rate": 1.8584793658822452e-06,
+      "loss": 0.0245,
+      "step": 1467
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.5642531276125784,
+      "learning_rate": 1.8522580784740452e-06,
+      "loss": 0.0202,
+      "step": 1468
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.5922481529196789,
+      "learning_rate": 1.8460448533513126e-06,
+      "loss": 0.0236,
+      "step": 1469
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.3489071504947676,
+      "learning_rate": 1.8398397064279282e-06,
+      "loss": 0.0146,
+      "step": 1470
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.46384690248928895,
+      "learning_rate": 1.833642653597088e-06,
+      "loss": 0.0188,
+      "step": 1471
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.9116631997505819,
+      "learning_rate": 1.8274537107312502e-06,
+      "loss": 0.0182,
+      "step": 1472
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.40347487957056644,
+      "learning_rate": 1.8212728936821034e-06,
+      "loss": 0.0099,
+      "step": 1473
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.9256491401658001,
+      "learning_rate": 1.8151002182805226e-06,
+      "loss": 0.0196,
+      "step": 1474
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 2.378903999596482,
+      "learning_rate": 1.8089357003365316e-06,
+      "loss": 0.0243,
+      "step": 1475
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 5.026642187164877,
+      "learning_rate": 1.802779355639257e-06,
+      "loss": 0.0085,
+      "step": 1476
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.8339934893740829,
+      "learning_rate": 1.7966311999568974e-06,
+      "loss": 0.0338,
+      "step": 1477
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.1253234493067137,
+      "learning_rate": 1.7904912490366723e-06,
+      "loss": 0.0295,
+      "step": 1478
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.4327414591473575,
+      "learning_rate": 1.7843595186047863e-06,
+      "loss": 0.0271,
+      "step": 1479
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.671546053350665,
+      "learning_rate": 1.7782360243663905e-06,
+      "loss": 0.0206,
+      "step": 1480
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.7710182655596616,
+      "learning_rate": 1.7721207820055402e-06,
+      "loss": 0.0232,
+      "step": 1481
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 25.15180406399679,
+      "learning_rate": 1.7660138071851545e-06,
+      "loss": 0.0232,
+      "step": 1482
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.8164567044306239,
+      "learning_rate": 1.7599151155469802e-06,
+      "loss": 0.0243,
+      "step": 1483
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.6379600019853553,
+      "learning_rate": 1.753824722711544e-06,
+      "loss": 0.02,
+      "step": 1484
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.8499432372174188,
+      "learning_rate": 1.7477426442781198e-06,
+      "loss": 0.0261,
+      "step": 1485
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.6261917549682076,
+      "learning_rate": 1.7416688958246847e-06,
+      "loss": 0.0193,
+      "step": 1486
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.5082059473411364,
+      "learning_rate": 1.7356034929078803e-06,
+      "loss": 0.0111,
+      "step": 1487
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.4353187166263084,
+      "learning_rate": 1.7295464510629718e-06,
+      "loss": 0.0193,
+      "step": 1488
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.5191009047621435,
+      "learning_rate": 1.7234977858038137e-06,
+      "loss": 0.0315,
+      "step": 1489
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.6488737261899189,
+      "learning_rate": 1.7174575126228016e-06,
+      "loss": 0.0227,
+      "step": 1490
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.9264160824660079,
+      "learning_rate": 1.711425646990838e-06,
+      "loss": 0.0254,
+      "step": 1491
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.4176930071798314,
+      "learning_rate": 1.7054022043572882e-06,
+      "loss": 0.0153,
+      "step": 1492
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.3590710184553117,
+      "learning_rate": 1.6993872001499461e-06,
+      "loss": 0.0193,
+      "step": 1493
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.9707860340678295,
+      "learning_rate": 1.6933806497749955e-06,
+      "loss": 0.023,
+      "step": 1494
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.4370541782227388,
+      "learning_rate": 1.6873825686169625e-06,
+      "loss": 0.0171,
+      "step": 1495
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.6244044034847467,
+      "learning_rate": 1.681392972038684e-06,
+      "loss": 0.0192,
+      "step": 1496
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.8926815180721247,
+      "learning_rate": 1.6754118753812632e-06,
+      "loss": 0.0242,
+      "step": 1497
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.7593507350335365,
+      "learning_rate": 1.669439293964033e-06,
+      "loss": 0.0262,
+      "step": 1498
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.5578039243484335,
+      "learning_rate": 1.6634752430845196e-06,
+      "loss": 0.0297,
+      "step": 1499
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.584385071528929,
+      "learning_rate": 1.6575197380183965e-06,
+      "loss": 0.0143,
+      "step": 1500
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.56402011021067,
+      "learning_rate": 1.6515727940194493e-06,
+      "loss": 0.0195,
+      "step": 1501
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.44268730746617535,
+      "learning_rate": 1.6456344263195374e-06,
+      "loss": 0.013,
+      "step": 1502
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.4961967681210577,
+      "learning_rate": 1.639704650128552e-06,
+      "loss": 0.0193,
+      "step": 1503
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.970299813916526,
+      "learning_rate": 1.6337834806343783e-06,
+      "loss": 0.019,
+      "step": 1504
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.6257331643439866,
+      "learning_rate": 1.6278709330028636e-06,
+      "loss": 0.0206,
+      "step": 1505
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.9686357872123381,
+      "learning_rate": 1.621967022377765e-06,
+      "loss": 0.0284,
+      "step": 1506
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 2.274058130322033,
+      "learning_rate": 1.6160717638807205e-06,
+      "loss": 0.0226,
+      "step": 1507
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.7747602086226845,
+      "learning_rate": 1.6101851726112067e-06,
+      "loss": 0.0116,
+      "step": 1508
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.8105945738690109,
+      "learning_rate": 1.6043072636465017e-06,
+      "loss": 0.0217,
+      "step": 1509
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 2.6793023565390253,
+      "learning_rate": 1.598438052041643e-06,
+      "loss": 0.0285,
+      "step": 1510
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.5613810363225675,
+      "learning_rate": 1.5925775528293985e-06,
+      "loss": 0.025,
+      "step": 1511
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.688507515991181,
+      "learning_rate": 1.586725781020214e-06,
+      "loss": 0.0186,
+      "step": 1512
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.3351929076892877,
+      "learning_rate": 1.5808827516021851e-06,
+      "loss": 0.0199,
+      "step": 1513
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 2.1889487189572696,
+      "learning_rate": 1.5750484795410143e-06,
+      "loss": 0.0307,
+      "step": 1514
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.48865042364173655,
+      "learning_rate": 1.5692229797799752e-06,
+      "loss": 0.0215,
+      "step": 1515
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.8347701023901559,
+      "learning_rate": 1.5634062672398697e-06,
+      "loss": 0.0225,
+      "step": 1516
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.800077144695363,
+      "learning_rate": 1.557598356819e-06,
+      "loss": 0.0303,
+      "step": 1517
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.3919408000387406,
+      "learning_rate": 1.5517992633931178e-06,
+      "loss": 0.014,
+      "step": 1518
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.6177121499869433,
+      "learning_rate": 1.5460090018153923e-06,
+      "loss": 0.0227,
+      "step": 1519
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.6349338552380158,
+      "learning_rate": 1.5402275869163736e-06,
+      "loss": 0.0283,
+      "step": 1520
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.5126033869999117,
+      "learning_rate": 1.5344550335039521e-06,
+      "loss": 0.0155,
+      "step": 1521
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.856223292939775,
+      "learning_rate": 1.5286913563633194e-06,
+      "loss": 0.017,
+      "step": 1522
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.0083250205891625,
+      "learning_rate": 1.522936570256937e-06,
+      "loss": 0.0217,
+      "step": 1523
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.4994108438051356,
+      "learning_rate": 1.517190689924491e-06,
+      "loss": 0.0184,
+      "step": 1524
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.37321497836596046,
+      "learning_rate": 1.5114537300828558e-06,
+      "loss": 0.0173,
+      "step": 1525
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.7701752941065383,
+      "learning_rate": 1.505725705426061e-06,
+      "loss": 0.0244,
+      "step": 1526
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.1281368029828174,
+      "learning_rate": 1.5000066306252475e-06,
+      "loss": 0.0166,
+      "step": 1527
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.7338424153324811,
+      "learning_rate": 1.4942965203286337e-06,
+      "loss": 0.0177,
+      "step": 1528
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.41491260429600013,
+      "learning_rate": 1.4885953891614813e-06,
+      "loss": 0.0153,
+      "step": 1529
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.7844248268278589,
+      "learning_rate": 1.482903251726049e-06,
+      "loss": 0.0305,
+      "step": 1530
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.8885898820999001,
+      "learning_rate": 1.4772201226015615e-06,
+      "loss": 0.0309,
+      "step": 1531
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 2.0705376041540378,
+      "learning_rate": 1.4715460163441703e-06,
+      "loss": 0.0492,
+      "step": 1532
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.60630026470556,
+      "learning_rate": 1.4658809474869174e-06,
+      "loss": 0.0163,
+      "step": 1533
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.7257005504788903,
+      "learning_rate": 1.4602249305396966e-06,
+      "loss": 0.0146,
+      "step": 1534
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.2429270738071347,
+      "learning_rate": 1.4545779799892179e-06,
+      "loss": 0.0298,
+      "step": 1535
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.5803068356342167,
+      "learning_rate": 1.4489401102989698e-06,
+      "loss": 0.0143,
+      "step": 1536
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.9334893520475549,
+      "learning_rate": 1.4433113359091805e-06,
+      "loss": 0.029,
+      "step": 1537
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.8061341751112961,
+      "learning_rate": 1.4376916712367838e-06,
+      "loss": 0.0186,
+      "step": 1538
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.9039562950084497,
+      "learning_rate": 1.4320811306753796e-06,
+      "loss": 0.0212,
+      "step": 1539
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.4714931876115984,
+      "learning_rate": 1.426479728595202e-06,
+      "loss": 0.02,
+      "step": 1540
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.7906593826204635,
+      "learning_rate": 1.4208874793430738e-06,
+      "loss": 0.0221,
+      "step": 1541
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.7280596395913382,
+      "learning_rate": 1.415304397242378e-06,
+      "loss": 0.0207,
+      "step": 1542
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.8117288716060429,
+      "learning_rate": 1.409730496593016e-06,
+      "loss": 0.0312,
+      "step": 1543
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.7925630934679432,
+      "learning_rate": 1.4041657916713741e-06,
+      "loss": 0.0181,
+      "step": 1544
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.3841331652094588,
+      "learning_rate": 1.398610296730284e-06,
+      "loss": 0.0262,
+      "step": 1545
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.620345714196815,
+      "learning_rate": 1.3930640259989914e-06,
+      "loss": 0.0215,
+      "step": 1546
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.72655687412436,
+      "learning_rate": 1.3875269936831133e-06,
+      "loss": 0.0246,
+      "step": 1547
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.6574460872584007,
+      "learning_rate": 1.3819992139646037e-06,
+      "loss": 0.0222,
+      "step": 1548
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.7069930906531742,
+      "learning_rate": 1.37648070100172e-06,
+      "loss": 0.0266,
+      "step": 1549
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.6845802946247612,
+      "learning_rate": 1.3709714689289844e-06,
+      "loss": 0.0196,
+      "step": 1550
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.6832692791063089,
+      "learning_rate": 1.3654715318571455e-06,
+      "loss": 0.0161,
+      "step": 1551
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.8309536486895222,
+      "learning_rate": 1.3599809038731503e-06,
+      "loss": 0.026,
+      "step": 1552
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.861610800620917,
+      "learning_rate": 1.354499599040097e-06,
+      "loss": 0.0218,
+      "step": 1553
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.793696990792352,
+      "learning_rate": 1.3490276313972073e-06,
+      "loss": 0.0232,
+      "step": 1554
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.6407327606080412,
+      "learning_rate": 1.3435650149597862e-06,
+      "loss": 0.0265,
+      "step": 1555
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.8759165356024202,
+      "learning_rate": 1.3381117637191887e-06,
+      "loss": 0.0225,
+      "step": 1556
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.0577346876127711,
+      "learning_rate": 1.3326678916427803e-06,
+      "loss": 0.0298,
+      "step": 1557
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.4542805258478709,
+      "learning_rate": 1.3272334126739094e-06,
+      "loss": 0.0174,
+      "step": 1558
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.5182354308636529,
+      "learning_rate": 1.3218083407318606e-06,
+      "loss": 0.0179,
+      "step": 1559
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.3954805565498662,
+      "learning_rate": 1.3163926897118252e-06,
+      "loss": 0.0192,
+      "step": 1560
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.811580137556446,
+      "learning_rate": 1.310986473484866e-06,
+      "loss": 0.0182,
+      "step": 1561
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.1921238444026916,
+      "learning_rate": 1.3055897058978806e-06,
+      "loss": 0.0358,
+      "step": 1562
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 2.0037836419531394,
+      "learning_rate": 1.300202400773563e-06,
+      "loss": 0.0114,
+      "step": 1563
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.9513953738579717,
+      "learning_rate": 1.2948245719103774e-06,
+      "loss": 0.0268,
+      "step": 1564
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.2881756130443442,
+      "learning_rate": 1.2894562330825106e-06,
+      "loss": 0.022,
+      "step": 1565
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.9456261923512207,
+      "learning_rate": 1.2840973980398446e-06,
+      "loss": 0.0251,
+      "step": 1566
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.7087238217511729,
+      "learning_rate": 1.27874808050792e-06,
+      "loss": 0.0205,
+      "step": 1567
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.8087893817399272,
+      "learning_rate": 1.273408294187899e-06,
+      "loss": 0.0249,
+      "step": 1568
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.2547387247611936,
+      "learning_rate": 1.2680780527565313e-06,
+      "loss": 0.024,
+      "step": 1569
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.0368957302101323,
+      "learning_rate": 1.2627573698661228e-06,
+      "loss": 0.0374,
+      "step": 1570
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.5301292447138974,
+      "learning_rate": 1.257446259144494e-06,
+      "loss": 0.0243,
+      "step": 1571
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.46242128257381343,
+      "learning_rate": 1.2521447341949494e-06,
+      "loss": 0.0174,
+      "step": 1572
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.0290998698478622,
+      "learning_rate": 1.24685280859624e-06,
+      "loss": 0.0329,
+      "step": 1573
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.7094916079841542,
+      "learning_rate": 1.2415704959025321e-06,
+      "loss": 0.0197,
+      "step": 1574
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.0464668946048912,
+      "learning_rate": 1.236297809643368e-06,
+      "loss": 0.0259,
+      "step": 1575
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 2.2412781545389158,
+      "learning_rate": 1.2310347633236402e-06,
+      "loss": 0.0296,
+      "step": 1576
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.3720843586226656,
+      "learning_rate": 1.225781370423541e-06,
+      "loss": 0.0398,
+      "step": 1577
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.011296703595368,
+      "learning_rate": 1.2205376443985444e-06,
+      "loss": 0.0193,
+      "step": 1578
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.5876467993872176,
+      "learning_rate": 1.2153035986793627e-06,
+      "loss": 0.0331,
+      "step": 1579
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.5603223195065772,
+      "learning_rate": 1.2100792466719118e-06,
+      "loss": 0.0208,
+      "step": 1580
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.6448719403345857,
+      "learning_rate": 1.2048646017572857e-06,
+      "loss": 0.0222,
+      "step": 1581
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.28342050570494437,
+      "learning_rate": 1.1996596772917091e-06,
+      "loss": 0.0092,
+      "step": 1582
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.39879018942234634,
+      "learning_rate": 1.1944644866065125e-06,
+      "loss": 0.0176,
+      "step": 1583
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.42845470718418127,
+      "learning_rate": 1.1892790430080952e-06,
+      "loss": 0.0163,
+      "step": 1584
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.504022504721399,
+      "learning_rate": 1.1841033597778905e-06,
+      "loss": 0.0155,
+      "step": 1585
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.8398692539657486,
+      "learning_rate": 1.1789374501723328e-06,
+      "loss": 0.0227,
+      "step": 1586
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.0945442042275597,
+      "learning_rate": 1.173781327422826e-06,
+      "loss": 0.0203,
+      "step": 1587
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.7836709704029713,
+      "learning_rate": 1.1686350047357036e-06,
+      "loss": 0.0323,
+      "step": 1588
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.6375367175543545,
+      "learning_rate": 1.1634984952921996e-06,
+      "loss": 0.0196,
+      "step": 1589
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.7287547471933316,
+      "learning_rate": 1.1583718122484128e-06,
+      "loss": 0.0208,
+      "step": 1590
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.7912568103894504,
+      "learning_rate": 1.1532549687352752e-06,
+      "loss": 0.021,
+      "step": 1591
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.5528882367856399,
+      "learning_rate": 1.1481479778585142e-06,
+      "loss": 0.0193,
+      "step": 1592
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.8246650683849523,
+      "learning_rate": 1.1430508526986261e-06,
+      "loss": 0.0216,
+      "step": 1593
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.3077247280046171,
+      "learning_rate": 1.137963606310834e-06,
+      "loss": 0.0232,
+      "step": 1594
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.4102947057515644,
+      "learning_rate": 1.132886251725061e-06,
+      "loss": 0.0227,
+      "step": 1595
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.6409286483577383,
+      "learning_rate": 1.127818801945893e-06,
+      "loss": 0.0209,
+      "step": 1596
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.6933263716604913,
+      "learning_rate": 1.1227612699525475e-06,
+      "loss": 0.0255,
+      "step": 1597
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.6940363977524191,
+      "learning_rate": 1.1177136686988383e-06,
+      "loss": 0.0205,
+      "step": 1598
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.7791018362615008,
+      "learning_rate": 1.1126760111131474e-06,
+      "loss": 0.0186,
+      "step": 1599
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.473217425311028,
+      "learning_rate": 1.1076483100983843e-06,
+      "loss": 0.0193,
+      "step": 1600
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.859716563176724,
+      "learning_rate": 1.1026305785319585e-06,
+      "loss": 0.0276,
+      "step": 1601
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.547744977616838,
+      "learning_rate": 1.0976228292657447e-06,
+      "loss": 0.0231,
+      "step": 1602
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.7192227614672417,
+      "learning_rate": 1.0926250751260492e-06,
+      "loss": 0.0217,
+      "step": 1603
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.4437076760782891,
+      "learning_rate": 1.0876373289135778e-06,
+      "loss": 0.0214,
+      "step": 1604
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.6198481679854476,
+      "learning_rate": 1.0826596034034066e-06,
+      "loss": 0.0274,
+      "step": 1605
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.6905795095781041,
+      "learning_rate": 1.0776919113449407e-06,
+      "loss": 0.0253,
+      "step": 1606
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.9346647468986687,
+      "learning_rate": 1.0727342654618905e-06,
+      "loss": 0.0297,
+      "step": 1607
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.6448428595540737,
+      "learning_rate": 1.0677866784522317e-06,
+      "loss": 0.0211,
+      "step": 1608
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.7632216839916288,
+      "learning_rate": 1.0628491629881794e-06,
+      "loss": 0.0285,
+      "step": 1609
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.043944368613153,
+      "learning_rate": 1.0579217317161494e-06,
+      "loss": 0.0236,
+      "step": 1610
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.6420776815388155,
+      "learning_rate": 1.0530043972567339e-06,
+      "loss": 0.0286,
+      "step": 1611
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.36874244760640196,
+      "learning_rate": 1.0480971722046602e-06,
+      "loss": 0.0119,
+      "step": 1612
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.698553180562803,
+      "learning_rate": 1.0432000691287619e-06,
+      "loss": 0.0232,
+      "step": 1613
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.4035736417059915,
+      "learning_rate": 1.0383131005719505e-06,
+      "loss": 0.0126,
+      "step": 1614
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.516220342713733,
+      "learning_rate": 1.0334362790511776e-06,
+      "loss": 0.0182,
+      "step": 1615
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.39365931838262347,
+      "learning_rate": 1.0285696170574045e-06,
+      "loss": 0.0126,
+      "step": 1616
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.5480992689342562,
+      "learning_rate": 1.023713127055575e-06,
+      "loss": 0.0179,
+      "step": 1617
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.5196346398987512,
+      "learning_rate": 1.0188668214845765e-06,
+      "loss": 0.0155,
+      "step": 1618
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.4489352579236505,
+      "learning_rate": 1.0140307127572125e-06,
+      "loss": 0.0184,
+      "step": 1619
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.690103014170062,
+      "learning_rate": 1.009204813260164e-06,
+      "loss": 0.0202,
+      "step": 1620
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.3570709210532113,
+      "learning_rate": 1.004389135353972e-06,
+      "loss": 0.0365,
+      "step": 1621
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.0257681719369882,
+      "learning_rate": 9.995836913729918e-07,
+      "loss": 0.0237,
+      "step": 1622
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.6168406563119275,
+      "learning_rate": 9.947884936253666e-07,
+      "loss": 0.0196,
+      "step": 1623
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.7692543213723014,
+      "learning_rate": 9.90003554392997e-07,
+      "loss": 0.0232,
+      "step": 1624
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.2328193733615247,
+      "learning_rate": 9.852288859315096e-07,
+      "loss": 0.0431,
+      "step": 1625
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.4613027425912049,
+      "learning_rate": 9.804645004702208e-07,
+      "loss": 0.0248,
+      "step": 1626
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.7094673929023848,
+      "learning_rate": 9.757104102121152e-07,
+      "loss": 0.0189,
+      "step": 1627
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.7652661275955153,
+      "learning_rate": 9.709666273338037e-07,
+      "loss": 0.0227,
+      "step": 1628
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.6515555057917235,
+      "learning_rate": 9.662331639854977e-07,
+      "loss": 0.0214,
+      "step": 1629
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.4895831342399959,
+      "learning_rate": 9.615100322909787e-07,
+      "loss": 0.018,
+      "step": 1630
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.609954132677896,
+      "learning_rate": 9.567972443475648e-07,
+      "loss": 0.0203,
+      "step": 1631
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.5067204919211488,
+      "learning_rate": 9.520948122260792e-07,
+      "loss": 0.0214,
+      "step": 1632
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.2597206502873675,
+      "learning_rate": 9.474027479708254e-07,
+      "loss": 0.0206,
+      "step": 1633
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.3841045305323135,
+      "learning_rate": 9.427210635995482e-07,
+      "loss": 0.0152,
+      "step": 1634
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.5821817358914164,
+      "learning_rate": 9.380497711034065e-07,
+      "loss": 0.0203,
+      "step": 1635
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.6100463174092755,
+      "learning_rate": 9.33388882446944e-07,
+      "loss": 0.0184,
+      "step": 1636
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.9490671043170982,
+      "learning_rate": 9.287384095680558e-07,
+      "loss": 0.0257,
+      "step": 1637
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.41866063189682606,
+      "learning_rate": 9.240983643779589e-07,
+      "loss": 0.0099,
+      "step": 1638
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.916192123980324,
+      "learning_rate": 9.19468758761165e-07,
+      "loss": 0.0196,
+      "step": 1639
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.6291804775621946,
+      "learning_rate": 9.148496045754441e-07,
+      "loss": 0.0219,
+      "step": 1640
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.46506938594026603,
+      "learning_rate": 9.10240913651797e-07,
+      "loss": 0.0209,
+      "step": 1641
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3763574887343969,
+      "learning_rate": 9.056426977944272e-07,
+      "loss": 0.0175,
+      "step": 1642
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.6263963200007582,
+      "learning_rate": 9.010549687807058e-07,
+      "loss": 0.026,
+      "step": 1643
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.8782893052468342,
+      "learning_rate": 8.964777383611445e-07,
+      "loss": 0.0147,
+      "step": 1644
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.8499726766140078,
+      "learning_rate": 8.919110182593682e-07,
+      "loss": 0.0307,
+      "step": 1645
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.870144300950776,
+      "learning_rate": 8.873548201720788e-07,
+      "loss": 0.0259,
+      "step": 1646
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.7272528604695497,
+      "learning_rate": 8.828091557690288e-07,
+      "loss": 0.0296,
+      "step": 1647
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.676027817512814,
+      "learning_rate": 8.78274036692991e-07,
+      "loss": 0.0284,
+      "step": 1648
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.856038483122214,
+      "learning_rate": 8.73749474559728e-07,
+      "loss": 0.0297,
+      "step": 1649
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 2.399332352098762,
+      "learning_rate": 8.692354809579634e-07,
+      "loss": 0.0443,
+      "step": 1650
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.8513838144339223,
+      "learning_rate": 8.647320674493531e-07,
+      "loss": 0.0233,
+      "step": 1651
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 2.2022612364336753,
+      "learning_rate": 8.602392455684522e-07,
+      "loss": 0.0288,
+      "step": 1652
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.6318734527708079,
+      "learning_rate": 8.557570268226889e-07,
+      "loss": 0.0227,
+      "step": 1653
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.4986624707680987,
+      "learning_rate": 8.512854226923328e-07,
+      "loss": 0.012,
+      "step": 1654
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.9490939870975297,
+      "learning_rate": 8.468244446304664e-07,
+      "loss": 0.0232,
+      "step": 1655
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.5687247694401628,
+      "learning_rate": 8.423741040629557e-07,
+      "loss": 0.0171,
+      "step": 1656
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.9342752487631834,
+      "learning_rate": 8.379344123884231e-07,
+      "loss": 0.0256,
+      "step": 1657
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.4259907942214256,
+      "learning_rate": 8.335053809782129e-07,
+      "loss": 0.0208,
+      "step": 1658
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.9361781880015428,
+      "learning_rate": 8.29087021176368e-07,
+      "loss": 0.0143,
+      "step": 1659
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.8441919046857431,
+      "learning_rate": 8.246793442995954e-07,
+      "loss": 0.0167,
+      "step": 1660
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.6969608754714764,
+      "learning_rate": 8.20282361637243e-07,
+      "loss": 0.0171,
+      "step": 1661
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.8202955595934445,
+      "learning_rate": 8.158960844512654e-07,
+      "loss": 0.0195,
+      "step": 1662
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.3779435824059567,
+      "learning_rate": 8.115205239761986e-07,
+      "loss": 0.029,
+      "step": 1663
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 2.1595061577911956,
+      "learning_rate": 8.071556914191298e-07,
+      "loss": 0.0151,
+      "step": 1664
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.3886150974844245,
+      "learning_rate": 8.028015979596681e-07,
+      "loss": 0.0128,
+      "step": 1665
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.715937707945813,
+      "learning_rate": 7.984582547499176e-07,
+      "loss": 0.0219,
+      "step": 1666
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.8299951514667157,
+      "learning_rate": 7.941256729144464e-07,
+      "loss": 0.0256,
+      "step": 1667
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.101475048961094,
+      "learning_rate": 7.898038635502631e-07,
+      "loss": 0.0303,
+      "step": 1668
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.6129852664782461,
+      "learning_rate": 7.854928377267812e-07,
+      "loss": 0.023,
+      "step": 1669
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.654917809207751,
+      "learning_rate": 7.81192606485795e-07,
+      "loss": 0.0233,
+      "step": 1670
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.7485343464875281,
+      "learning_rate": 7.769031808414523e-07,
+      "loss": 0.0197,
+      "step": 1671
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.4487323641792147,
+      "learning_rate": 7.726245717802233e-07,
+      "loss": 0.0375,
+      "step": 1672
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.3217437634939568,
+      "learning_rate": 7.68356790260873e-07,
+      "loss": 0.0122,
+      "step": 1673
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.4868540639820885,
+      "learning_rate": 7.640998472144373e-07,
+      "loss": 0.0196,
+      "step": 1674
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.8908143071660273,
+      "learning_rate": 7.59853753544188e-07,
+      "loss": 0.0222,
+      "step": 1675
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.9843729139133108,
+      "learning_rate": 7.556185201256105e-07,
+      "loss": 0.0158,
+      "step": 1676
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.8812079648965719,
+      "learning_rate": 7.513941578063732e-07,
+      "loss": 0.023,
+      "step": 1677
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 2.041052471945282,
+      "learning_rate": 7.471806774063e-07,
+      "loss": 0.0214,
+      "step": 1678
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.36991466567386605,
+      "learning_rate": 7.429780897173427e-07,
+      "loss": 0.015,
+      "step": 1679
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3146801134651237,
+      "learning_rate": 7.387864055035571e-07,
+      "loss": 0.013,
+      "step": 1680
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.5791780869947971,
+      "learning_rate": 7.346056355010683e-07,
+      "loss": 0.0205,
+      "step": 1681
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.4843910836561715,
+      "learning_rate": 7.304357904180475e-07,
+      "loss": 0.0183,
+      "step": 1682
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.9352327802951779,
+      "learning_rate": 7.262768809346849e-07,
+      "loss": 0.0207,
+      "step": 1683
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.6355901282179665,
+      "learning_rate": 7.221289177031609e-07,
+      "loss": 0.0215,
+      "step": 1684
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 2.827734270635811,
+      "learning_rate": 7.179919113476192e-07,
+      "loss": 0.0574,
+      "step": 1685
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.26441919617238413,
+      "learning_rate": 7.138658724641417e-07,
+      "loss": 0.0091,
+      "step": 1686
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.9709198286425481,
+      "learning_rate": 7.097508116207169e-07,
+      "loss": 0.0261,
+      "step": 1687
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.3166787822548412,
+      "learning_rate": 7.056467393572158e-07,
+      "loss": 0.0178,
+      "step": 1688
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.721525302234989,
+      "learning_rate": 7.015536661853656e-07,
+      "loss": 0.0241,
+      "step": 1689
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.5768351297093026,
+      "learning_rate": 6.974716025887207e-07,
+      "loss": 0.0194,
+      "step": 1690
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.3058055162169507,
+      "learning_rate": 6.934005590226361e-07,
+      "loss": 0.0137,
+      "step": 1691
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.6765565581741332,
+      "learning_rate": 6.893405459142439e-07,
+      "loss": 0.02,
+      "step": 1692
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.46982186330771497,
+      "learning_rate": 6.85291573662421e-07,
+      "loss": 0.0157,
+      "step": 1693
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.8793025805347607,
+      "learning_rate": 6.812536526377673e-07,
+      "loss": 0.0214,
+      "step": 1694
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.6662828745104178,
+      "learning_rate": 6.772267931825765e-07,
+      "loss": 0.0226,
+      "step": 1695
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.491202037016575,
+      "learning_rate": 6.732110056108105e-07,
+      "loss": 0.02,
+      "step": 1696
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.676251852333797,
+      "learning_rate": 6.692063002080717e-07,
+      "loss": 0.0303,
+      "step": 1697
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.6695814761840161,
+      "learning_rate": 6.652126872315812e-07,
+      "loss": 0.0236,
+      "step": 1698
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.33574986126203843,
+      "learning_rate": 6.612301769101464e-07,
+      "loss": 0.012,
+      "step": 1699
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.4367587595934483,
+      "learning_rate": 6.572587794441381e-07,
+      "loss": 0.0172,
+      "step": 1700
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.62403592781587,
+      "learning_rate": 6.532985050054635e-07,
+      "loss": 0.0208,
+      "step": 1701
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.0886268911836297,
+      "learning_rate": 6.493493637375414e-07,
+      "loss": 0.0266,
+      "step": 1702
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.7851741637225625,
+      "learning_rate": 6.45411365755273e-07,
+      "loss": 0.0186,
+      "step": 1703
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.39687201481838075,
+      "learning_rate": 6.414845211450243e-07,
+      "loss": 0.0168,
+      "step": 1704
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.858159783352387,
+      "learning_rate": 6.375688399645863e-07,
+      "loss": 0.0226,
+      "step": 1705
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.569237008902663,
+      "learning_rate": 6.336643322431624e-07,
+      "loss": 0.0224,
+      "step": 1706
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.092683438598888,
+      "learning_rate": 6.297710079813346e-07,
+      "loss": 0.0296,
+      "step": 1707
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.7351588601818313,
+      "learning_rate": 6.258888771510441e-07,
+      "loss": 0.0195,
+      "step": 1708
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.6404335577191063,
+      "learning_rate": 6.220179496955608e-07,
+      "loss": 0.0295,
+      "step": 1709
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.170165515233352,
+      "learning_rate": 6.181582355294579e-07,
+      "loss": 0.0175,
+      "step": 1710
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.9322068050386567,
+      "learning_rate": 6.143097445385904e-07,
+      "loss": 0.0202,
+      "step": 1711
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.7698760246873417,
+      "learning_rate": 6.104724865800665e-07,
+      "loss": 0.0218,
+      "step": 1712
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.7113173705764911,
+      "learning_rate": 6.066464714822224e-07,
+      "loss": 0.017,
+      "step": 1713
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.42364594714097703,
+      "learning_rate": 6.028317090446006e-07,
+      "loss": 0.0146,
+      "step": 1714
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.3595453398378823,
+      "learning_rate": 5.990282090379201e-07,
+      "loss": 0.0182,
+      "step": 1715
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.6382144279458921,
+      "learning_rate": 5.952359812040548e-07,
+      "loss": 0.0266,
+      "step": 1716
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.2344671026355398,
+      "learning_rate": 5.91455035256005e-07,
+      "loss": 0.0266,
+      "step": 1717
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.44343690963147425,
+      "learning_rate": 5.876853808778782e-07,
+      "loss": 0.0164,
+      "step": 1718
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.3139772726167909,
+      "learning_rate": 5.839270277248565e-07,
+      "loss": 0.0297,
+      "step": 1719
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.5404404381299035,
+      "learning_rate": 5.801799854231826e-07,
+      "loss": 0.0218,
+      "step": 1720
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.9120055408801463,
+      "learning_rate": 5.764442635701229e-07,
+      "loss": 0.0263,
+      "step": 1721
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.7776723247223076,
+      "learning_rate": 5.727198717339511e-07,
+      "loss": 0.0166,
+      "step": 1722
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.9105654312944207,
+      "learning_rate": 5.690068194539217e-07,
+      "loss": 0.0259,
+      "step": 1723
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.5182676413757116,
+      "learning_rate": 5.653051162402445e-07,
+      "loss": 0.021,
+      "step": 1724
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.9792005660232671,
+      "learning_rate": 5.616147715740611e-07,
+      "loss": 0.0161,
+      "step": 1725
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.37941653477897735,
+      "learning_rate": 5.579357949074221e-07,
+      "loss": 0.0119,
+      "step": 1726
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.1030201633413343,
+      "learning_rate": 5.542681956632601e-07,
+      "loss": 0.0253,
+      "step": 1727
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.5322457268616182,
+      "learning_rate": 5.506119832353662e-07,
+      "loss": 0.0201,
+      "step": 1728
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.787967539652566,
+      "learning_rate": 5.469671669883675e-07,
+      "loss": 0.0146,
+      "step": 1729
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.3041562263159943,
+      "learning_rate": 5.433337562577018e-07,
+      "loss": 0.0244,
+      "step": 1730
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.33548978851264033,
+      "learning_rate": 5.397117603495927e-07,
+      "loss": 0.0163,
+      "step": 1731
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.2796190944664051,
+      "learning_rate": 5.361011885410311e-07,
+      "loss": 0.0142,
+      "step": 1732
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.4321550120605657,
+      "learning_rate": 5.325020500797434e-07,
+      "loss": 0.0251,
+      "step": 1733
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.2378411684030717,
+      "learning_rate": 5.289143541841735e-07,
+      "loss": 0.0239,
+      "step": 1734
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.6006956440056487,
+      "learning_rate": 5.253381100434574e-07,
+      "loss": 0.0171,
+      "step": 1735
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.8446344159688172,
+      "learning_rate": 5.217733268173996e-07,
+      "loss": 0.0167,
+      "step": 1736
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.3860698594231629,
+      "learning_rate": 5.182200136364491e-07,
+      "loss": 0.0116,
+      "step": 1737
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.6743659201298501,
+      "learning_rate": 5.146781796016798e-07,
+      "loss": 0.0228,
+      "step": 1738
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.6578276056168991,
+      "learning_rate": 5.111478337847603e-07,
+      "loss": 0.0145,
+      "step": 1739
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.4444967706503058,
+      "learning_rate": 5.076289852279375e-07,
+      "loss": 0.0199,
+      "step": 1740
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.8905759201198322,
+      "learning_rate": 5.041216429440088e-07,
+      "loss": 0.0202,
+      "step": 1741
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.8652719179398571,
+      "learning_rate": 5.006258159163007e-07,
+      "loss": 0.026,
+      "step": 1742
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.920438619981302,
+      "learning_rate": 4.971415130986457e-07,
+      "loss": 0.0256,
+      "step": 1743
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.4690467940337577,
+      "learning_rate": 4.936687434153619e-07,
+      "loss": 0.0139,
+      "step": 1744
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.0857169067280574,
+      "learning_rate": 4.902075157612241e-07,
+      "loss": 0.0268,
+      "step": 1745
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.9071954914058828,
+      "learning_rate": 4.867578390014466e-07,
+      "loss": 0.021,
+      "step": 1746
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 2.0027348201423516,
+      "learning_rate": 4.833197219716595e-07,
+      "loss": 0.0242,
+      "step": 1747
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.7208333147985326,
+      "learning_rate": 4.798931734778794e-07,
+      "loss": 0.0225,
+      "step": 1748
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.3340296826164514,
+      "learning_rate": 4.764782022965014e-07,
+      "loss": 0.014,
+      "step": 1749
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.41348222258487555,
+      "learning_rate": 4.730748171742611e-07,
+      "loss": 0.0234,
+      "step": 1750
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.783608172436412,
+      "learning_rate": 4.696830268282204e-07,
+      "loss": 0.0207,
+      "step": 1751
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.5594126597283506,
+      "learning_rate": 4.6630283994574475e-07,
+      "loss": 0.0188,
+      "step": 1752
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.4130436506101321,
+      "learning_rate": 4.629342651844787e-07,
+      "loss": 0.0149,
+      "step": 1753
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.47117699815033703,
+      "learning_rate": 4.595773111723245e-07,
+      "loss": 0.0157,
+      "step": 1754
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.06468324325383,
+      "learning_rate": 4.562319865074222e-07,
+      "loss": 0.0218,
+      "step": 1755
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.4979626901475594,
+      "learning_rate": 4.528982997581233e-07,
+      "loss": 0.0201,
+      "step": 1756
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.988935035787326,
+      "learning_rate": 4.4957625946297266e-07,
+      "loss": 0.0178,
+      "step": 1757
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.637273291889344,
+      "learning_rate": 4.462658741306847e-07,
+      "loss": 0.016,
+      "step": 1758
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.7542039279222552,
+      "learning_rate": 4.4296715224012187e-07,
+      "loss": 0.0169,
+      "step": 1759
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.7968353487232488,
+      "learning_rate": 4.3968010224027247e-07,
+      "loss": 0.0147,
+      "step": 1760
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.9266624597530224,
+      "learning_rate": 4.364047325502324e-07,
+      "loss": 0.0193,
+      "step": 1761
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.4242221780268279,
+      "learning_rate": 4.331410515591783e-07,
+      "loss": 0.0136,
+      "step": 1762
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 2.2405700148688243,
+      "learning_rate": 4.298890676263495e-07,
+      "loss": 0.0238,
+      "step": 1763
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 2.236346074001916,
+      "learning_rate": 4.2664878908102556e-07,
+      "loss": 0.0355,
+      "step": 1764
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.8651345063698643,
+      "learning_rate": 4.2342022422250553e-07,
+      "loss": 0.0258,
+      "step": 1765
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.3813222697994349,
+      "learning_rate": 4.2020338132008454e-07,
+      "loss": 0.0121,
+      "step": 1766
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.7925799529634308,
+      "learning_rate": 4.1699826861303804e-07,
+      "loss": 0.0249,
+      "step": 1767
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.9035223172963347,
+      "learning_rate": 4.138048943105938e-07,
+      "loss": 0.0237,
+      "step": 1768
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.7287711613306799,
+      "learning_rate": 4.106232665919152e-07,
+      "loss": 0.0109,
+      "step": 1769
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.6004627965162954,
+      "learning_rate": 4.0745339360607927e-07,
+      "loss": 0.0136,
+      "step": 1770
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.5499328026562983,
+      "learning_rate": 4.042952834720548e-07,
+      "loss": 0.0203,
+      "step": 1771
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.44565309449455925,
+      "learning_rate": 4.011489442786831e-07,
+      "loss": 0.0147,
+      "step": 1772
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.3121101306415952,
+      "learning_rate": 3.9801438408465895e-07,
+      "loss": 0.0136,
+      "step": 1773
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.9336561135203006,
+      "learning_rate": 3.9489161091850413e-07,
+      "loss": 0.0228,
+      "step": 1774
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.1439685088962397,
+      "learning_rate": 3.917806327785517e-07,
+      "loss": 0.0194,
+      "step": 1775
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.7667463909847458,
+      "learning_rate": 3.886814576329245e-07,
+      "loss": 0.0343,
+      "step": 1776
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.39481524338438234,
+      "learning_rate": 3.855940934195146e-07,
+      "loss": 0.0182,
+      "step": 1777
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.4682956312455625,
+      "learning_rate": 3.825185480459614e-07,
+      "loss": 0.0142,
+      "step": 1778
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.49595668042956637,
+      "learning_rate": 3.794548293896355e-07,
+      "loss": 0.0132,
+      "step": 1779
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.21011637441168096,
+      "learning_rate": 3.7640294529761424e-07,
+      "loss": 0.0081,
+      "step": 1780
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 5.79580408555057,
+      "learning_rate": 3.7336290358666206e-07,
+      "loss": 0.0291,
+      "step": 1781
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.4241950547346609,
+      "learning_rate": 3.703347120432138e-07,
+      "loss": 0.0164,
+      "step": 1782
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.5231212609907914,
+      "learning_rate": 3.6731837842335085e-07,
+      "loss": 0.0233,
+      "step": 1783
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.9115936016118895,
+      "learning_rate": 3.6431391045278263e-07,
+      "loss": 0.028,
+      "step": 1784
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.7705099493152656,
+      "learning_rate": 3.6132131582683086e-07,
+      "loss": 0.0178,
+      "step": 1785
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.7642604555944872,
+      "learning_rate": 3.583406022104019e-07,
+      "loss": 0.0203,
+      "step": 1786
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.3350098895459638,
+      "learning_rate": 3.5537177723797335e-07,
+      "loss": 0.0106,
+      "step": 1787
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.6748864749922056,
+      "learning_rate": 3.52414848513572e-07,
+      "loss": 0.0232,
+      "step": 1788
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.7794439927916365,
+      "learning_rate": 3.4946982361075524e-07,
+      "loss": 0.0146,
+      "step": 1789
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.0614980884129646,
+      "learning_rate": 3.4653671007259084e-07,
+      "loss": 0.0259,
+      "step": 1790
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 3.4421989854447395,
+      "learning_rate": 3.436155154116383e-07,
+      "loss": 0.0122,
+      "step": 1791
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.9290405002699073,
+      "learning_rate": 3.407062471099298e-07,
+      "loss": 0.021,
+      "step": 1792
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.1172006958050862,
+      "learning_rate": 3.3780891261895043e-07,
+      "loss": 0.0281,
+      "step": 1793
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 2.227579498950002,
+      "learning_rate": 3.349235193596184e-07,
+      "loss": 0.0226,
+      "step": 1794
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.7512967509133619,
+      "learning_rate": 3.320500747222677e-07,
+      "loss": 0.0172,
+      "step": 1795
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.4290770474862749,
+      "learning_rate": 3.291885860666294e-07,
+      "loss": 0.0173,
+      "step": 1796
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.250875314305571,
+      "learning_rate": 3.263390607218103e-07,
+      "loss": 0.025,
+      "step": 1797
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.944211346524395,
+      "learning_rate": 3.235015059862767e-07,
+      "loss": 0.0241,
+      "step": 1798
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.4028666512919997,
+      "learning_rate": 3.206759291278333e-07,
+      "loss": 0.0133,
+      "step": 1799
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.8746495137789382,
+      "learning_rate": 3.178623373836076e-07,
+      "loss": 0.0209,
+      "step": 1800
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.915974115336977,
+      "learning_rate": 3.1506073796002734e-07,
+      "loss": 0.0196,
+      "step": 1801
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.713113489833011,
+      "learning_rate": 3.1227113803280863e-07,
+      "loss": 0.0423,
+      "step": 1802
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.5203902674476263,
+      "learning_rate": 3.0949354474692937e-07,
+      "loss": 0.0155,
+      "step": 1803
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.37043619740618705,
+      "learning_rate": 3.0672796521661663e-07,
+      "loss": 0.0131,
+      "step": 1804
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.600749381569276,
+      "learning_rate": 3.0397440652532585e-07,
+      "loss": 0.0246,
+      "step": 1805
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.5272728545861523,
+      "learning_rate": 3.0123287572572545e-07,
+      "loss": 0.0168,
+      "step": 1806
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.255654485553308,
+      "learning_rate": 2.985033798396736e-07,
+      "loss": 0.0196,
+      "step": 1807
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.9078529994683149,
+      "learning_rate": 2.9578592585820856e-07,
+      "loss": 0.0237,
+      "step": 1808
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.40880978967914217,
+      "learning_rate": 2.930805207415205e-07,
+      "loss": 0.0138,
+      "step": 1809
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.8889728807411933,
+      "learning_rate": 2.9038717141894266e-07,
+      "loss": 0.023,
+      "step": 1810
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.887903401671168,
+      "learning_rate": 2.8770588478892805e-07,
+      "loss": 0.0218,
+      "step": 1811
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.5935263702090745,
+      "learning_rate": 2.850366677190336e-07,
+      "loss": 0.0127,
+      "step": 1812
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.8674794845738173,
+      "learning_rate": 2.823795270459029e-07,
+      "loss": 0.0148,
+      "step": 1813
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.4074621926720314,
+      "learning_rate": 2.797344695752491e-07,
+      "loss": 0.0364,
+      "step": 1814
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.0369490988278562,
+      "learning_rate": 2.771015020818363e-07,
+      "loss": 0.0206,
+      "step": 1815
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.42829931436334806,
+      "learning_rate": 2.7448063130946224e-07,
+      "loss": 0.0153,
+      "step": 1816
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.23602416053185918,
+      "learning_rate": 2.718718639709411e-07,
+      "loss": 0.0074,
+      "step": 1817
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 5.629591449198259,
+      "learning_rate": 2.69275206748088e-07,
+      "loss": 0.0207,
+      "step": 1818
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.7405988875650557,
+      "learning_rate": 2.666906662916985e-07,
+      "loss": 0.0154,
+      "step": 1819
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.461232966739589,
+      "learning_rate": 2.641182492215361e-07,
+      "loss": 0.0116,
+      "step": 1820
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.7075154031682652,
+      "learning_rate": 2.615579621263109e-07,
+      "loss": 0.0211,
+      "step": 1821
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.1371845028463687,
+      "learning_rate": 2.590098115636658e-07,
+      "loss": 0.0316,
+      "step": 1822
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.6774639388623163,
+      "learning_rate": 2.5647380406015665e-07,
+      "loss": 0.0187,
+      "step": 1823
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.9274279811411574,
+      "learning_rate": 2.5394994611123934e-07,
+      "loss": 0.0179,
+      "step": 1824
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.306866211141696,
+      "learning_rate": 2.514382441812502e-07,
+      "loss": 0.0184,
+      "step": 1825
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.603304233210114,
+      "learning_rate": 2.489387047033909e-07,
+      "loss": 0.015,
+      "step": 1826
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.5328595053040174,
+      "learning_rate": 2.464513340797114e-07,
+      "loss": 0.0226,
+      "step": 1827
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.866341356192597,
+      "learning_rate": 2.439761386810935e-07,
+      "loss": 0.0172,
+      "step": 1828
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.5610663210635402,
+      "learning_rate": 2.4151312484723465e-07,
+      "loss": 0.0133,
+      "step": 1829
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.2493186264229105,
+      "learning_rate": 2.39062298886632e-07,
+      "loss": 0.0128,
+      "step": 1830
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.6844264402500051,
+      "learning_rate": 2.3662366707656537e-07,
+      "loss": 0.0176,
+      "step": 1831
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.4233840001854265,
+      "learning_rate": 2.341972356630845e-07,
+      "loss": 0.0144,
+      "step": 1832
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.7240753632477378,
+      "learning_rate": 2.3178301086098532e-07,
+      "loss": 0.0143,
+      "step": 1833
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.866915589742617,
+      "learning_rate": 2.293809988538037e-07,
+      "loss": 0.0246,
+      "step": 1834
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.1967068684214617,
+      "learning_rate": 2.2699120579379174e-07,
+      "loss": 0.0261,
+      "step": 1835
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.1879307349406338,
+      "learning_rate": 2.246136378019087e-07,
+      "loss": 0.0268,
+      "step": 1836
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.5166219459046876,
+      "learning_rate": 2.2224830096779837e-07,
+      "loss": 0.0195,
+      "step": 1837
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.1229789289094003,
+      "learning_rate": 2.198952013497796e-07,
+      "loss": 0.0213,
+      "step": 1838
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.8698757679037399,
+      "learning_rate": 2.175543449748263e-07,
+      "loss": 0.0171,
+      "step": 1839
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.0318428935613762,
+      "learning_rate": 2.1522573783855473e-07,
+      "loss": 0.0094,
+      "step": 1840
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.4064528369110108,
+      "learning_rate": 2.129093859052067e-07,
+      "loss": 0.03,
+      "step": 1841
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.961657681832751,
+      "learning_rate": 2.106052951076365e-07,
+      "loss": 0.0211,
+      "step": 1842
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.9449980636138309,
+      "learning_rate": 2.083134713472923e-07,
+      "loss": 0.0192,
+      "step": 1843
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.7800742218320303,
+      "learning_rate": 2.0603392049420357e-07,
+      "loss": 0.0211,
+      "step": 1844
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.0808053808935367,
+      "learning_rate": 2.0376664838696546e-07,
+      "loss": 0.0109,
+      "step": 1845
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.6351185948935967,
+      "learning_rate": 2.0151166083272222e-07,
+      "loss": 0.0137,
+      "step": 1846
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.5968180495011286,
+      "learning_rate": 1.9926896360715542e-07,
+      "loss": 0.0161,
+      "step": 1847
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.5208311636465429,
+      "learning_rate": 1.9703856245446795e-07,
+      "loss": 0.0146,
+      "step": 1848
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.5894146185232907,
+      "learning_rate": 1.948204630873668e-07,
+      "loss": 0.0181,
+      "step": 1849
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.2825054473544255,
+      "learning_rate": 1.9261467118705246e-07,
+      "loss": 0.0106,
+      "step": 1850
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.2878943348884669,
+      "learning_rate": 1.9042119240320067e-07,
+      "loss": 0.0103,
+      "step": 1851
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.5635184171884943,
+      "learning_rate": 1.8824003235395128e-07,
+      "loss": 0.0223,
+      "step": 1852
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.5714917480875857,
+      "learning_rate": 1.8607119662589045e-07,
+      "loss": 0.0214,
+      "step": 1853
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.5819822583931356,
+      "learning_rate": 1.8391469077404134e-07,
+      "loss": 0.0222,
+      "step": 1854
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.7490706862618906,
+      "learning_rate": 1.8177052032184285e-07,
+      "loss": 0.014,
+      "step": 1855
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.9170654518824781,
+      "learning_rate": 1.7963869076114193e-07,
+      "loss": 0.018,
+      "step": 1856
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.9597599706808101,
+      "learning_rate": 1.7751920755217532e-07,
+      "loss": 0.0235,
+      "step": 1857
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.8393265995727257,
+      "learning_rate": 1.7541207612355894e-07,
+      "loss": 0.0237,
+      "step": 1858
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.5720599297711036,
+      "learning_rate": 1.7331730187227003e-07,
+      "loss": 0.0159,
+      "step": 1859
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.38535609643181973,
+      "learning_rate": 1.7123489016363793e-07,
+      "loss": 0.0123,
+      "step": 1860
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.6143963637658444,
+      "learning_rate": 1.6916484633132558e-07,
+      "loss": 0.0224,
+      "step": 1861
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.5189176558087583,
+      "learning_rate": 1.671071756773196e-07,
+      "loss": 0.0171,
+      "step": 1862
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.6668235969578423,
+      "learning_rate": 1.6506188347191477e-07,
+      "loss": 0.0215,
+      "step": 1863
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.1154353763845224,
+      "learning_rate": 1.6302897495370175e-07,
+      "loss": 0.0282,
+      "step": 1864
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.6912898158788077,
+      "learning_rate": 1.61008455329551e-07,
+      "loss": 0.0214,
+      "step": 1865
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.755595253610713,
+      "learning_rate": 1.59000329774604e-07,
+      "loss": 0.0243,
+      "step": 1866
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.43189963611385784,
+      "learning_rate": 1.5700460343225644e-07,
+      "loss": 0.0144,
+      "step": 1867
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.7980037217430461,
+      "learning_rate": 1.5502128141414496e-07,
+      "loss": 0.0232,
+      "step": 1868
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.42317850471550683,
+      "learning_rate": 1.5305036880013612e-07,
+      "loss": 0.0144,
+      "step": 1869
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3440586152342501,
+      "learning_rate": 1.5109187063831243e-07,
+      "loss": 0.0151,
+      "step": 1870
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.9689038545074031,
+      "learning_rate": 1.4914579194495794e-07,
+      "loss": 0.025,
+      "step": 1871
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.0026706796663856,
+      "learning_rate": 1.4721213770454988e-07,
+      "loss": 0.0193,
+      "step": 1872
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.3776628922159566,
+      "learning_rate": 1.4529091286973994e-07,
+      "loss": 0.0323,
+      "step": 1873
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.1160811515237832,
+      "learning_rate": 1.4338212236134518e-07,
+      "loss": 0.0182,
+      "step": 1874
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.27113768389532256,
+      "learning_rate": 1.414857710683365e-07,
+      "loss": 0.0141,
+      "step": 1875
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.4976224971778871,
+      "learning_rate": 1.3960186384782025e-07,
+      "loss": 0.0174,
+      "step": 1876
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.5627076451492086,
+      "learning_rate": 1.3773040552503447e-07,
+      "loss": 0.0139,
+      "step": 1877
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.6144894015224014,
+      "learning_rate": 1.3587140089332984e-07,
+      "loss": 0.0235,
+      "step": 1878
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.37920150421252663,
+      "learning_rate": 1.3402485471415871e-07,
+      "loss": 0.0115,
+      "step": 1879
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.3543116999305515,
+      "learning_rate": 1.3219077171706563e-07,
+      "loss": 0.0114,
+      "step": 1880
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.6932990480913824,
+      "learning_rate": 1.303691565996712e-07,
+      "loss": 0.0214,
+      "step": 1881
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.48280266582056625,
+      "learning_rate": 1.2856001402766384e-07,
+      "loss": 0.0123,
+      "step": 1882
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.3600213634007312,
+      "learning_rate": 1.267633486347858e-07,
+      "loss": 0.018,
+      "step": 1883
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.6233047924767592,
+      "learning_rate": 1.2497916502282104e-07,
+      "loss": 0.0118,
+      "step": 1884
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.6557622650641682,
+      "learning_rate": 1.232074677615841e-07,
+      "loss": 0.0144,
+      "step": 1885
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.68680008251895,
+      "learning_rate": 1.2144826138890897e-07,
+      "loss": 0.014,
+      "step": 1886
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.6753520125550132,
+      "learning_rate": 1.1970155041063636e-07,
+      "loss": 0.0185,
+      "step": 1887
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.8061089749043806,
+      "learning_rate": 1.1796733930060256e-07,
+      "loss": 0.027,
+      "step": 1888
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.9511108204414935,
+      "learning_rate": 1.1624563250062836e-07,
+      "loss": 0.0249,
+      "step": 1889
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.5887898728183794,
+      "learning_rate": 1.1453643442050744e-07,
+      "loss": 0.0211,
+      "step": 1890
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.4905414507949429,
+      "learning_rate": 1.1283974943799403e-07,
+      "loss": 0.0138,
+      "step": 1891
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.9247168896017007,
+      "learning_rate": 1.1115558189879416e-07,
+      "loss": 0.0122,
+      "step": 1892
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.6215487472298206,
+      "learning_rate": 1.0948393611655229e-07,
+      "loss": 0.0163,
+      "step": 1893
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.7473599642814582,
+      "learning_rate": 1.0782481637284014e-07,
+      "loss": 0.0157,
+      "step": 1894
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.0402363260233884,
+      "learning_rate": 1.0617822691714796e-07,
+      "loss": 0.0153,
+      "step": 1895
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.5761737409888579,
+      "learning_rate": 1.0454417196687216e-07,
+      "loss": 0.0186,
+      "step": 1896
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.5818796082089626,
+      "learning_rate": 1.0292265570730431e-07,
+      "loss": 0.0173,
+      "step": 1897
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.4937356890575318,
+      "learning_rate": 1.0131368229161997e-07,
+      "loss": 0.0116,
+      "step": 1898
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.595595287827361,
+      "learning_rate": 9.971725584086933e-08,
+      "loss": 0.0182,
+      "step": 1899
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.704730771852314,
+      "learning_rate": 9.813338044396714e-08,
+      "loss": 0.0166,
+      "step": 1900
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.9696721850434132,
+      "learning_rate": 9.656206015768e-08,
+      "loss": 0.0162,
+      "step": 1901
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.6552313791421823,
+      "learning_rate": 9.500329900661742e-08,
+      "loss": 0.0185,
+      "step": 1902
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.6329810385027012,
+      "learning_rate": 9.345710098322247e-08,
+      "loss": 0.0141,
+      "step": 1903
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.47133660028286867,
+      "learning_rate": 9.192347004775781e-08,
+      "loss": 0.0258,
+      "step": 1904
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.37229356852218276,
+      "learning_rate": 9.040241012830131e-08,
+      "loss": 0.0172,
+      "step": 1905
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.55541532428076,
+      "learning_rate": 8.889392512072992e-08,
+      "loss": 0.0225,
+      "step": 1906
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.5276375856301501,
+      "learning_rate": 8.739801888871468e-08,
+      "loss": 0.0211,
+      "step": 1907
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.4956939691451788,
+      "learning_rate": 8.591469526370799e-08,
+      "loss": 0.0156,
+      "step": 1908
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.7859967160273817,
+      "learning_rate": 8.444395804493411e-08,
+      "loss": 0.0197,
+      "step": 1909
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.5770278583139444,
+      "learning_rate": 8.298581099937975e-08,
+      "loss": 0.0161,
+      "step": 1910
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.1811644683909457,
+      "learning_rate": 8.154025786178577e-08,
+      "loss": 0.0281,
+      "step": 1911
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.48198193891981583,
+      "learning_rate": 8.010730233463493e-08,
+      "loss": 0.0212,
+      "step": 1912
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.8232670905597198,
+      "learning_rate": 7.86869480881447e-08,
+      "loss": 0.027,
+      "step": 1913
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.7847367471690805,
+      "learning_rate": 7.727919876025669e-08,
+      "loss": 0.0224,
+      "step": 1914
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.3673488997507827,
+      "learning_rate": 7.588405795662779e-08,
+      "loss": 0.0147,
+      "step": 1915
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.5712286156889304,
+      "learning_rate": 7.450152925062015e-08,
+      "loss": 0.016,
+      "step": 1916
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.584801234518443,
+      "learning_rate": 7.3131616183294e-08,
+      "loss": 0.0221,
+      "step": 1917
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.1547907543640876,
+      "learning_rate": 7.177432226339542e-08,
+      "loss": 0.0326,
+      "step": 1918
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.5996020279426367,
+      "learning_rate": 7.042965096735076e-08,
+      "loss": 0.0173,
+      "step": 1919
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.9630804667184417,
+      "learning_rate": 6.909760573925561e-08,
+      "loss": 0.0194,
+      "step": 1920
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.7923608884650656,
+      "learning_rate": 6.777818999086582e-08,
+      "loss": 0.0242,
+      "step": 1921
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.6136669614491637,
+      "learning_rate": 6.647140710159039e-08,
+      "loss": 0.0231,
+      "step": 1922
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 2.444098778295829,
+      "learning_rate": 6.51772604184825e-08,
+      "loss": 0.0227,
+      "step": 1923
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.7236056147422347,
+      "learning_rate": 6.389575325622787e-08,
+      "loss": 0.0245,
+      "step": 1924
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.5610244487794093,
+      "learning_rate": 6.262688889714152e-08,
+      "loss": 0.0189,
+      "step": 1925
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.6533403737022699,
+      "learning_rate": 6.137067059115431e-08,
+      "loss": 0.0223,
+      "step": 1926
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.2250995109120701,
+      "learning_rate": 6.012710155580858e-08,
+      "loss": 0.0266,
+      "step": 1927
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.5322808366260482,
+      "learning_rate": 5.889618497624649e-08,
+      "loss": 0.0142,
+      "step": 1928
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.6284323476920674,
+      "learning_rate": 5.767792400520556e-08,
+      "loss": 0.0219,
+      "step": 1929
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.5611338062761357,
+      "learning_rate": 5.647232176300754e-08,
+      "loss": 0.0192,
+      "step": 1930
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.5508596623052354,
+      "learning_rate": 5.5279381337551286e-08,
+      "loss": 0.0186,
+      "step": 1931
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.6227099224632395,
+      "learning_rate": 5.409910578430488e-08,
+      "loss": 0.0132,
+      "step": 1932
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.5567745537204065,
+      "learning_rate": 5.2931498126298495e-08,
+      "loss": 0.0097,
+      "step": 1933
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.5576063453236663,
+      "learning_rate": 5.177656135411657e-08,
+      "loss": 0.0161,
+      "step": 1934
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.3370801447420126,
+      "learning_rate": 5.063429842588841e-08,
+      "loss": 0.0265,
+      "step": 1935
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.7544241187243449,
+      "learning_rate": 4.950471226728371e-08,
+      "loss": 0.0182,
+      "step": 1936
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.5253352786844406,
+      "learning_rate": 4.838780577150093e-08,
+      "loss": 0.0174,
+      "step": 1937
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.7944907580697882,
+      "learning_rate": 4.728358179926451e-08,
+      "loss": 0.0189,
+      "step": 1938
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.9651599809552647,
+      "learning_rate": 4.619204317881376e-08,
+      "loss": 0.0278,
+      "step": 1939
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.6170478992323329,
+      "learning_rate": 4.511319270589731e-08,
+      "loss": 0.0187,
+      "step": 1940
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.5413410426193749,
+      "learning_rate": 4.404703314376646e-08,
+      "loss": 0.0212,
+      "step": 1941
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.3773729241544756,
+      "learning_rate": 4.299356722316683e-08,
+      "loss": 0.0157,
+      "step": 1942
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.5028399673629241,
+      "learning_rate": 4.1952797642331736e-08,
+      "loss": 0.0161,
+      "step": 1943
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.7421325043346971,
+      "learning_rate": 4.092472706697603e-08,
+      "loss": 0.02,
+      "step": 1944
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.5828564139036714,
+      "learning_rate": 3.99093581302884e-08,
+      "loss": 0.0159,
+      "step": 1945
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.3703638920114571,
+      "learning_rate": 3.890669343292464e-08,
+      "loss": 0.0153,
+      "step": 1946
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.7208034989055797,
+      "learning_rate": 3.791673554300157e-08,
+      "loss": 0.0166,
+      "step": 1947
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.7710647513091973,
+      "learning_rate": 3.6939486996090953e-08,
+      "loss": 0.0229,
+      "step": 1948
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.6997720856631476,
+      "learning_rate": 3.597495029521059e-08,
+      "loss": 0.0223,
+      "step": 1949
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.44352812503571115,
+      "learning_rate": 3.5023127910820966e-08,
+      "loss": 0.0106,
+      "step": 1950
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.7756413157400694,
+      "learning_rate": 3.408402228081642e-08,
+      "loss": 0.0233,
+      "step": 1951
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.5743501883431086,
+      "learning_rate": 3.315763581052067e-08,
+      "loss": 0.0136,
+      "step": 1952
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.662378100288348,
+      "learning_rate": 3.224397087267961e-08,
+      "loss": 0.0172,
+      "step": 1953
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.45464868521172774,
+      "learning_rate": 3.1343029807456296e-08,
+      "loss": 0.0162,
+      "step": 1954
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.3151847873751775,
+      "learning_rate": 3.045481492242319e-08,
+      "loss": 0.0088,
+      "step": 1955
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.9179713108290374,
+      "learning_rate": 2.9579328492557734e-08,
+      "loss": 0.0188,
+      "step": 1956
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.7442297593300125,
+      "learning_rate": 2.8716572760236205e-08,
+      "loss": 0.0176,
+      "step": 1957
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.6467132296202227,
+      "learning_rate": 2.7866549935227638e-08,
+      "loss": 0.0262,
+      "step": 1958
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.4271184977588041,
+      "learning_rate": 2.702926219468882e-08,
+      "loss": 0.0181,
+      "step": 1959
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.34858153463525593,
+      "learning_rate": 2.620471168315819e-08,
+      "loss": 0.0114,
+      "step": 1960
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.5191191049249508,
+      "learning_rate": 2.5392900512549168e-08,
+      "loss": 0.0148,
+      "step": 1961
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.7904320889688824,
+      "learning_rate": 2.459383076214794e-08,
+      "loss": 0.0227,
+      "step": 1962
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.40664887186710424,
+      "learning_rate": 2.3807504478604583e-08,
+      "loss": 0.014,
+      "step": 1963
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 2.404072714467341,
+      "learning_rate": 2.303392367593027e-08,
+      "loss": 0.0302,
+      "step": 1964
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.8069295138522137,
+      "learning_rate": 2.2273090335491744e-08,
+      "loss": 0.0241,
+      "step": 1965
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3797343238694821,
+      "learning_rate": 2.152500640600519e-08,
+      "loss": 0.015,
+      "step": 1966
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.8463632953185086,
+      "learning_rate": 2.0789673803530696e-08,
+      "loss": 0.0201,
+      "step": 1967
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.0329783325265296,
+      "learning_rate": 2.006709441147059e-08,
+      "loss": 0.0236,
+      "step": 1968
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.4942200195158771,
+      "learning_rate": 1.9357270080561654e-08,
+      "loss": 0.0204,
+      "step": 1969
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.4577934210496395,
+      "learning_rate": 1.866020262887014e-08,
+      "loss": 0.0214,
+      "step": 1970
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.9348702790058298,
+      "learning_rate": 1.7975893841790105e-08,
+      "loss": 0.0175,
+      "step": 1971
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.563106283801156,
+      "learning_rate": 1.7304345472035634e-08,
+      "loss": 0.0142,
+      "step": 1972
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.8118174545670076,
+      "learning_rate": 1.6645559239638066e-08,
+      "loss": 0.0139,
+      "step": 1973
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.7418989057261591,
+      "learning_rate": 1.5999536831941e-08,
+      "loss": 0.0152,
+      "step": 1974
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.0174044425443258,
+      "learning_rate": 1.536627990359585e-08,
+      "loss": 0.0196,
+      "step": 1975
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.5660161998151887,
+      "learning_rate": 1.474579007655963e-08,
+      "loss": 0.0149,
+      "step": 1976
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.6964625660138856,
+      "learning_rate": 1.413806894008718e-08,
+      "loss": 0.0278,
+      "step": 1977
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.7449047694433539,
+      "learning_rate": 1.3543118050730053e-08,
+      "loss": 0.0095,
+      "step": 1978
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.5775894499740872,
+      "learning_rate": 1.2960938932329858e-08,
+      "loss": 0.0194,
+      "step": 1979
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.5028520606659571,
+      "learning_rate": 1.2391533076018258e-08,
+      "loss": 0.0185,
+      "step": 1980
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.8258924922257007,
+      "learning_rate": 1.1834901940209752e-08,
+      "loss": 0.0302,
+      "step": 1981
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.404199557513869,
+      "learning_rate": 1.129104695059835e-08,
+      "loss": 0.0208,
+      "step": 1982
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.7722768594707403,
+      "learning_rate": 1.0759969500155897e-08,
+      "loss": 0.018,
+      "step": 1983
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.2411521975184556,
+      "learning_rate": 1.0241670949127091e-08,
+      "loss": 0.0228,
+      "step": 1984
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.40738726676339176,
+      "learning_rate": 9.73615262502503e-09,
+      "loss": 0.0166,
+      "step": 1985
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.5418543637769685,
+      "learning_rate": 9.243415822629553e-09,
+      "loss": 0.0137,
+      "step": 1986
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.5457350500278516,
+      "learning_rate": 8.763461803983907e-09,
+      "loss": 0.0168,
+      "step": 1987
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 2.046562261481058,
+      "learning_rate": 8.29629179839031e-09,
+      "loss": 0.0236,
+      "step": 1988
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.5594564303475206,
+      "learning_rate": 7.841907002407723e-09,
+      "loss": 0.0188,
+      "step": 1989
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.4520631765372096,
+      "learning_rate": 7.40030857984686e-09,
+      "loss": 0.0175,
+      "step": 1990
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.7392882226317737,
+      "learning_rate": 6.971497661771854e-09,
+      "loss": 0.0201,
+      "step": 1991
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.7067878390043856,
+      "learning_rate": 6.555475346491369e-09,
+      "loss": 0.019,
+      "step": 1992
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.4829992263795246,
+      "learning_rate": 6.152242699560273e-09,
+      "loss": 0.023,
+      "step": 1993
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.557053916058929,
+      "learning_rate": 5.761800753775193e-09,
+      "loss": 0.0181,
+      "step": 1994
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.4902359995960106,
+      "learning_rate": 5.384150509171737e-09,
+      "loss": 0.0106,
+      "step": 1995
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.6514146661773347,
+      "learning_rate": 5.019292933022279e-09,
+      "loss": 0.022,
+      "step": 1996
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.6339899864140751,
+      "learning_rate": 4.6672289598337364e-09,
+      "loss": 0.0182,
+      "step": 1997
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.563678714006939,
+      "learning_rate": 4.327959491344791e-09,
+      "loss": 0.019,
+      "step": 1998
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.46035552482503506,
+      "learning_rate": 4.001485396523119e-09,
+      "loss": 0.0158,
+      "step": 1999
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.8784396812135417,
+      "learning_rate": 3.6878075115642785e-09,
+      "loss": 0.0144,
+      "step": 2000
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.5350880147507716,
+      "learning_rate": 3.3869266398889323e-09,
+      "loss": 0.0127,
+      "step": 2001
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.1067951490975092,
+      "learning_rate": 3.098843552140629e-09,
+      "loss": 0.0216,
+      "step": 2002
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.7438453402090256,
+      "learning_rate": 2.8235589861846935e-09,
+      "loss": 0.0223,
+      "step": 2003
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.562580216953502,
+      "learning_rate": 2.5610736471043398e-09,
+      "loss": 0.0135,
+      "step": 2004
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.7769503250226634,
+      "learning_rate": 2.311388207201781e-09,
+      "loss": 0.0201,
+      "step": 2005
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.48698680842107656,
+      "learning_rate": 2.0745033059943463e-09,
+      "loss": 0.0159,
+      "step": 2006
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.5185885006351848,
+      "learning_rate": 1.850419550212812e-09,
+      "loss": 0.0197,
+      "step": 2007
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.5632801114175002,
+      "learning_rate": 1.6391375138019583e-09,
+      "loss": 0.0153,
+      "step": 2008
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.5604911746221906,
+      "learning_rate": 1.4406577379155739e-09,
+      "loss": 0.0155,
+      "step": 2009
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 1.0519838330608335,
+      "learning_rate": 1.2549807309192308e-09,
+      "loss": 0.0213,
+      "step": 2010
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.44923896577940386,
+      "learning_rate": 1.082106968385288e-09,
+      "loss": 0.0178,
+      "step": 2011
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.5734155390579543,
+      "learning_rate": 9.220368930945578e-10,
+      "loss": 0.0285,
+      "step": 2012
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.8907738585219976,
+      "learning_rate": 7.74770915032419e-10,
+      "loss": 0.024,
+      "step": 2013
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.2489096332464878,
+      "learning_rate": 6.403094113904828e-10,
+      "loss": 0.012,
+      "step": 2014
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.664084472215052,
+      "learning_rate": 5.186527265638175e-10,
+      "loss": 0.0275,
+      "step": 2015
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.419111160239852,
+      "learning_rate": 4.098011721503925e-10,
+      "loss": 0.0192,
+      "step": 2016
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.745323057773564,
+      "learning_rate": 3.137550269516343e-10,
+      "loss": 0.0267,
+      "step": 2017
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.8704302892058308,
+      "learning_rate": 2.305145369685402e-10,
+      "loss": 0.0193,
+      "step": 2018
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.6153612302634468,
+      "learning_rate": 1.6007991540556434e-10,
+      "loss": 0.0191,
+      "step": 2019
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.8365975506490403,
+      "learning_rate": 1.0245134266562152e-10,
+      "loss": 0.0139,
+      "step": 2020
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.6040430941452167,
+      "learning_rate": 5.762896635175263e-11,
+      "loss": 0.0178,
+      "step": 2021
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.5858775848368828,
+      "learning_rate": 2.56129012682349e-11,
+      "loss": 0.0197,
+      "step": 2022
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.0557910188186006,
+      "learning_rate": 6.403229417251134e-12,
+      "loss": 0.0252,
+      "step": 2023
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.3665471722262132,
+      "learning_rate": 0.0,
+      "loss": 0.0111,
+      "step": 2024
+    },
+    {
+      "epoch": 2.0,
+      "step": 2024,
+      "total_flos": 86476241235968.0,
+      "train_loss": 0.05223587417046506,
+      "train_runtime": 8272.5955,
+      "train_samples_per_second": 1.958,
+      "train_steps_per_second": 0.245
     }
   ],
   "logging_steps": 1.0,
@@ -7014,7 +14191,7 @@
   "num_input_tokens_seen": 0,
   "num_train_epochs": 2,
   "save_steps": 1000,
-  "total_flos": 42732724217856.0,
+  "total_flos": 86476241235968.0,
   "train_batch_size": 1,
   "trial_name": null,
   "trial_params": null